Re-evaluation of the traditional diet-heart hypothesis: analysis of recovered data from Minnesota Coronary Experiment (1968-73)
BMJ 2016; 353 doi: https://doi.org/10.1136/bmj.i1246 (Published 12 April 2016) Cite this as: BMJ 2016;353:i1246
All rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
I agree with Professor Willett's response to the re-evaluation of the traditional diet-heart hypothesis.1 An adequate intake of N-3 and N-6 polyunsaturated fatty acids is very important as deficiencies in essential fatty acid pathways are very common.
For some years my patients were tested using John McLaren-Howard's red cell essential fatty acid (EFA) profile. N-3 and N-6 pathway deficiencies were particularly likely in patients who also had zinc and B vitamin deficiencies which resulted in blocks in these pathways, The commonest abnormalities were too high levels of 18:0 stearic acid from animal fat and 16:0 palmitolic acids from palm oil but too low levels of 20:5, 22:5 and 22:6 N-3 EFAs especially in men, and, too low levels of 20:3, 20:4, 22:5 and 22:6N-6 EFAs especially women.
These common abnormalities can be corrected with N-3 and N-6 supplements along with mineral and vitamin supplements and a high protein low allergy rotation diet. 2
1 Re-evaluation of the traditional diet-heart hypothesis: analysis of recovered data from Minnesota Coronary Experiment (1968-73) BMJ 2016;353:i1246
2 Grant ECG. Food allergies and migraine. Lancet 1979;1:966-69.
Competing interests: No competing interests
To the Editors
The recent paper by Ramsden et al. (1) on the Minnesota coronary heart disease prevention trial is an interesting historical footnote that has no relevance to current dietary recommendations. We have known for many years that the classical diet-heart is incomplete, including recognition of the importance of both N-3 fatty acids and N-6 dietary fatty acids, the roles of different cholesterol fractions, and the many other pathways that connect diet with heart disease (2).
The diet used in the Minnesota trial was never consumed by any appreciable number of Americans and is not recommended by the American Heart Association (AHA) or any group. The level of linoleic acid was well above the range recommended by the AHA; to get to this level the investigators created fake meat, cheese, and milk by removing the natural fats as much as possible and replacing these with corn oil. Whatever small amounts of N-3 fatty were present would have been removed. They also created a special corn oil margarine that was lower in trans fat than the standard margarine, but we now know that the most dangerous types of trans fat (18:2 trans isomers) are likely to be higher in these lightly hydrogenated products than in the more heavily hydrogenated forms (3).
Even in its time, the Minnesota trial, conducted among patients in mental hospitals, was a major failure due to the massive dropouts and very short duration on the assigned diets. This trial was the victim of the deinstitutionalization of patients in these hospitals that occurred in the 1960’s and 70’s. The original authors enrolled 9423 assuming that at least three years would be needed to detect a benefit of the diet. At that time patients hospitalized with chronic mental illness seemed to be a good population to study because they were a “captive audience” that could be kept on assigned diets for many years. However, largely because of patients being discharged, 75% of the subjects were lost within the first year, and only 17% of those enrolled remained in the study at two years. Many fewer would have stayed a full three years, which is still a short time to study the effects of diet on atherosclerosis. The study was clearly a failure for reasons beyond the control of the investigators, and it adds very minimal information, if any, about the long term effects of diet on risk of heart disease.
The main results of this study were published by Frantz et al. in 1989 (4), with all of the above limitations, and reported no benefit on myocardial infarctions, sudden deaths, or all-cause mortality. Any implications that the authors failed to report their findings appropriately is simply incorrect, and the present BMJ paper adds no new data to this conclusion. Ramsden et al. show a figure describing no effect on total mortality, which was reported by Frantz et al., and would be expected if there is no effect on risk of heart disease. The causes of deaths in the BMJ paper are not known, which makes this endpoint particularly uninformative, especially in these hospitalized patients who likely had complex conditions. A small amount of new but incomplete data on autopsies was included in the BMJ report. However, the average follow up before death was less than one year from baseline, making it unlikely that any effect of diet on atherosclerosis would be seen in gross autopsy findings.
The authors include a meta-analysis of published data examining mortality from coronary heart disease. In addition to the major limitations of the Minnesota and other studies included, the statistical power was low for cardiac mortality and the confidence interval includes a potential important benefit. Importantly, the authors fail to mention that they earlier reported a benefit for incidence of coronary heart disease in a meta-analysis of randomized trials of trials in which saturated fat was replaced by a vegetable oils high in linoleic acid with a small amount of N-3 fatty acids, usually as soybean oil (5). Notably, N-3 fatty acids can be obtained from seafood as well as plant oils, but seafood intake was likely to be low in Minnesota in that period of time.
Overall, this report adds no useful new information and is irrelevant to current dietary recommendations that emphasize overall dietary patterns and replacing saturated fat with polyunsaturated fat, including sources of both N-3 and N-6 fatty acids. Many lines of evidence support this conclusion, including beneficial effects on blood lipids (6), and summaries of prospective cohort studies (7) and randomized trials (8, 5), including the meta-analysis previously published by Ramsden and others. Notably, since the 1960’s the US diet has changed in this direction; intake of linoleic acid has approximately doubled, and this has corresponded to a greater than 60% decline in coronary heart disease mortality and a major increase in life expectency. Although multiple factors have contributed to this decline, none of the other factors can explain this huge improvement in health, and the replacement of saturated fat with polyunsaturated fat (both N-6 and N-3) is almost certainly a major, probably most important, factor. Reversing these changes would almost certainly result in great harm.
Although we have a high level of evidence that the above direction of dietary change has been beneficial, more research is desirable to determine optimal intakes of different types and sources of dietary fat and their effects on all major health outcomes.
Sincerely,
Walter Willett, MD DrPH
Harvard University
1. Ramsden CE, Zamaora 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;352:i1246
2. Willett W. Nutritional Epidemiology, Oxford University Press, 2012
3. Sun Q, Ma J, Campos H, Hankinson SE, Manson JE, Stampfer MJ, Rexrode KM, Willett WC, Hu FB. A prospective study of trans fatty acids in erythrocytes and risk of coronary heart disease. Circulation 2007;115:1858-65.
4. Frantz ID Jr., Dawson EA, Ashman PL, et al. Test of effect of lipid lowering by diet on cardiovascular risk. The Minnesota Coronary Survey. Arteriosclerosis 1989;9:129-35.
5. Ramsden CE, Zamora D, Leelarthaepin B, Majchrzak-Hong SF, Faurot KR, Suchindran CM, Ringel A, Davis JM, Hibbeln JR. Use of dietary linoleic acid for secondary prevention of coronary heart disease and death: evaluation of recovered data from the Sydney Diet Heart Study and updated meta-analysis. BMJ. 2013 Feb 4;346:e8707
6. Mensink, R.P., et al., Effects of dietary fatty acids and carbohydrates on the ratio of serum total to HDL cholesterol and on serum lipids and apolipoproteins: a meta-analysis of 60 controlled trials. Am J Clin Nutr, 2003. 77(5): p. 1146-55
7. Farvid MS, Ding M, Pan A, et al. Dietary linoleic acid and risk of coronary heart disease:a systematic review and meta-analysis of prospective cohort studies. Circulation 2014;130:1568-78.
8. http://health.gov/dietaryguidelines/2015-scientific-report/
Competing interests: No competing interests
Sir, the diet-heart hypothesis is firmly entrenched in the the minds of the lay people as the most intuitive explanation for CHD. It is the bedrock of standard cardiology practice and the basis of the statin industry. It is very difficult to imagine a cardiologist advising a patient to eat more butter and red meat even though you have showed that, that is what he should do! popular acceptance of the study findings face overwhelming odds.Thank God Ancel Keys is not around anymore...!
Competing interests: No competing interests
There is no mention of cigarette smoking in the report. There is a high level of cigarette smoking among patients in psychiatric hospitals and it could be that this effect swamps anythig that the lesser effect of dietary change may have shown.
Competing interests: No competing interests
The analysis of recovered data from the Minnesota Coronary Experiment (MCE) by Ramsden et al. ignores a basic flaw in the MCE (1). The 2015 Cochrane review of randomized controlled trials on saturated fat and cardiovascular disease excluded data from the MCE because “although the study proceeded for over 4 years participants (patients) came and went and mean follow-up was only 1 year” (2). The new (and incomplete) “recovered” data doesn’t change that fundamental limitation of the study.
The Cochrane review of 15 well-designed randomized controlled trials concluded that reducing saturated fat reduces the risk of cardiovascular events by 17 percent (2). In trials that replaced saturated with polyunsaturated fats, the risk was reduced by 27 percent.(2,3) As an editorial by the Cochrane authors noted, although the review found no effect of reducing saturated fat on mortality, “this perhaps was not surprising with mean trial durations of 4–5 years.” (3) The authors’ conclusions still hold: “Lifestyle advice to all those with, or at risk of CVD and to lower-risk population groups should continue to include permanent reduction of dietary saturated fat and partial replacement by polyunsaturated fats.”
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: http://dx.doi.org/10.1136/bmj.i1246.
2. Hooper L, Martin N, Abdelhamid A, Davey Smith G. Reduction in saturated fat intake for cardiovascular disease. Cochrane Database Syst Rev. 2015;6:CD011737. doi: 10.1002/14651858.CD011737.
3. Hooper L, Martin N, Abdelhamid A. Cochrane corner: what are the effects of reducing saturated fat intake on cardiovascular disease and mortality? Heart 2015;101(24):1938-40. doi: 10.1136/heartjnl-2015-308521.
Competing interests: No competing interests
It is high time that we quickly move away from the fat debate. Further, this debate often provides mixed messages that are based not merely on strong evidence or science. It is unfortunately not even based on common sense. However, it is evident that these confusing massages do more harm to our community health and well-being. The importance of maintaining a healthy balanced diet based on common sense evidence needs to be promoted in order to enhance community overall health and well-being. It is equally important to promote a healthy balanced diet if we sincerely want to prevent some chronic diseases in our communities. It makes sense that clinicians' advice to their patients on the right diet needs to be based on a common sense approach rather than ever changing science and research. It is very important that clinicians advise their patients to embrace a healthy diet as part of their daily routine (1, 2).This makes sense as new research shows that well balanced diets could enhance our children’s school performance to varying degrees (3). Further, research indicates that eating a well balanced diet throughout the day enhances our daily work and the productivity (4).
A daily well balanced diet could include but is not limited to the following: fruits, vegetables, nuts, proteins (plant and animal based), moderate amount of carbohydrates, wholegrain food, dairy products, dietary fiber, healthy level of most vitamins and minerals (often available from a diet rich in fruits and vegetables, protein and dairy products). We also need to promote, adequate daily water intake, avoid excessive salt, refined sugar, sugary drinks.
REFERENCES:
(1). Al-Isa AN, Campbell J, Desapriya E, Wijesinghe N. J Obes. 2011;2011:512363. doi: 10.1155/2011/512363. Epub 2011 Apr 27.
(2). Al-Isa AN, Campbell J, Desapriya E. Factors Associated with Overweight and Obesity among Kuwaiti Elementary Male School Children Aged 6-10 Years. Int J Pediatr. 2010;2010. pii: 459261. doi: 10.1155/2010/459261. Epub 2010 Sep 22.
(3). Kim SY, Sim S, Park B, Kong IG, Kim JH, Choi HG. Dietary Habits Are Associated With School Performance in Adolescents. Medicine (Baltimore). 2016 Mar;95(12):e3096. doi: 10.1097/MD.0000000000003096.
(4). Conner TS, Brookie KL, Richardson AC, Polak MA. On carrots and curiosity: eating fruit and vegetables is associated with greater flourishing in daily life. Br J Health Psychol. 2015 May;20(2):413-27. doi: 10.1111/bjhp.12113. Epub 2014 Jul 30.
Competing interests: No competing interests
Dear Sir,
The analysis of recovered data from the Minnesota Coronary Experiment by Ramsden et al has revealed results consistent with other analyses of diet-heart RCTs that excluded trials involving confounding by mixed interventions, that the experimental lowering of saturated fat results in no reduction in cardiovascular or all-cause mortality.[1,2,]
The possible adverse effect of the intervention in elderly subjects described by Ramsden et al is also consistent with observational studies of cholesterol and mortality risk in the elderly, including a study in which changes in cholesterol were tracked. Elderly subjects with low cholesterol whose cholesterol remained low in subsequent tests had the highest mortality; those in whom cholesterol increased again were not exposed to increased risk.[3] With all that has long been known about the correlations between cholesterol and risk in the elderly, it is not frivolous to ask why an intervention intended to raise cholesterol by replacing an unhealthful food (such as corn oil) with a healthful food rich in saturated fat (such as sour cream) has never been trialed.
The recent results from the EPIC-Netherlands prospective observational study of 35,597 subjects followed for 12 years describe a population in which the consumption of higher intakes of dairy fat is associated with a lower risk of ischemic heart disease (IHD). An energy substitution analysis based on the EPIC-N data even predicts that the replacement of energy from dairy saturated fat with energy from monounsaturated or polyunsaturated fat will increase IHD risk.[4]
One suspects that had the diet-heart hypothesis been delayed until the 21st century, its postulates may just as easily been the inverse of the traditional ones, and that dairy fats and fish oils, which do not generally lower cholesterol, might have been emphasized.
Although studies that seem to support the diet-heart hypothesis have always been published early and often, the null result from the Minnesota Coronary Experiment was not published for 16 years, until well after dietary guidelines to lower saturated fat became fixed. However, Ancel Keys was aware of the result by 1973. This passage in the 1980 Seven Countries Study, for which Keys was the lead author, seems to have largely escaped the notice of those promoting refined oils and spreads.
“The coronary incidence rate of the cohorts was not significantly correlated with the percentage of calories in the diet provided by proteins or by polyunsaturated fatty acids. In regard to the latter, the averages for the cohorts ranged from 3 percent to seven percent of calories from linoleic acid, with only trivial contributions from other polyenes. These findings conform to the general picture that in no natural diet of man so far studied do polyunsaturated fatty acids contribute more than a very small fraction of the total calories. Accordingly, it must be expected that in such natural diets variations in the amount of polyunsaturated fatty acids will have at most only a trivial effect on the concentration of cholesterol in the blood serum or risk associated with it.”[5]
The questions raised by Marsden et al are not merely ones of historical interest. Whatever the classical causes of heart disease, the risk today is increasingly one associated with diabetes, overweight, and non-alcoholic fatty liver disease. Low carbohydrate, high fat (LCHF) diets are one of the most effective interventions for these conditions and tend to produce beneficial changes in multiple metabolic risk factors, predicting long-term lowering of cardiovascular risk, in these patients.[6] However, many clinical LCHF interventions are also reduced saturated-fat interventions, and some of these may use levels of linoleic acid comparable to that in the MSE study. There is a scarcity of evidence that saturated fat restriction or an intake of linoleic acid beyond the natural range described by Keys et al produces superior metabolic benefits in LCHF diet trials. A comparative meta-analysis of the trial data that bears on this question would be welcomed.
Essential fatty acids are essential, but they are available in all naturally fatty foods, a grouping which does not usually include corn or soy. It is questionable in nutritional terms whether anyone who eats a high-fat diet which includes some of the natural sources of linoleic acid - nuts, seeds, meat, lard, fish and dairy - can ever benefit from an even higher intake via the replacement of any of these foods with refined oils and spreads. No-one has yet proved that there is anything wrong with a nutritious diet of real foods which have been minimally processed and refined, and we suggest that the default prescription for any test of a diet and disease hypothesis in future should be a diet of foods with a low Human Interference Factor (HIF), rather than foods that can only be produced by industrial processing, such as corn oil.
References
[1] Hooper L, Summerbell CD, Thompson R, Sills D, Roberts FG, Moore HJ, Davey Smith G. Reduced or modified dietary fat for preventing cardiovascular disease. Cochrane Database Syst Rev. 2012 May 16;5:CD002137.
[2] Hooper L, Martin N, Abdelhamid A, Davey Smith G. Reduction in saturated fat intake for cardiovascular disease. Cochrane Database Syst Rev. 2015 Jun 10;6:CD011737.
[3] Schatz IJ, Masaki K, Yano K, Chen R, Rodriguez BL, Curb JD. Cholesterol and all-cause mortality in elderly people from the Honolulu Heart Program: a cohort study. Lancet. 2001 Aug 4;358(9279):351-5.
[4] Praagman J, Beulens JW, Alssema M, Zock PL, Wanders AJ4, Sluijs I, van der Schouw YT. The association between dietary saturated fatty acids and ischemic heart disease depends on the type and source of fatty acid in the European Prospective Investigation into Cancer and Nutrition-Netherlands cohort. Am J Clin Nutr. 2016 Feb;103(2):356-65. doi: 10.3945/ajcn.115.122671. Epub 2016 Jan 20.
[5] Keys, A., Aravanis, C., Blackburn, H., Buzina, R., Djordjević, B.S., Dontas, A.S., Fidanza, F., Karvonen, M.J., Kimura, N., Menotti, A. et al. Seven Countries. A Multivariate Analysis of Death and Coronary Heart Disease. p253 ch14. 1980. Harvard University Press, Cambridge, Massachusetts.
[6] Schofield G, Henderson G, Thornley S, Crofts C. Very low-carbohydrate diets in the management of diabetes revisited. NZMJ. 2016;129(1432).
https://www.nzma.org.nz/journal/read-the-journal/all-issues/2010-2019/20...
Competing interests: No competing interests
In the systematic review and meta-analysis included in Ramsden et al, randomized controlled trials that specifically tested replacement of saturated fat with vegetable oil rich in linoleic acid provide no indication of benefit, even though the interventions effectively lowered serum cholesterol.
We invite readers seeking the broader context that is needed to evaluate this conclusion to read the detailed appendix (p. 15-37) at
http://www.bmj.com/content/bmj/suppl/2016/04/12/bmj.i1246.DC1/ramc027623...
Some key points are summarized below.
---
INTRODUCTION
The traditional diet-heart hypothesis predicts that replacing dietary saturated fat with vegetable oils rich in linoleic acid (LA) will reduce coronary heart disease and deaths by lowering serum cholesterol. Advice to replace saturated fat with LA-rich vegetable oils (e.g., corn oil, sunflower oil, safflower oil, cottonseed oil, or soybean oil) has been a cornerstone of dietary guidelines for the past half-century (see main paper, Figure 10).
However, the lack of supporting evidence from randomized controlled trials for such advice has been a source of controversy [13, 19]. Several diet-heart meta-analyses have been published, but they have not specifically examined the effects of replacing saturated fat with LA-rich vegetable oils. For example, a meta-analysis by Hooper et al [20] included randomized controlled trials that lowered saturated fat but did not distinguish between trials that replaced saturated fat with LA-rich oils, from those that replaced total and saturated fat with carbohydrates, and also included randomized controlled trials that markedly increased dietary n-3 EPA+DHA alongside LA. Similarly, Mozaffarian et al [21] included randomized controlled trials that markedly increased dietary EPA+DHA from seafood and cod liver oil, among other diet changes (e.g. sugar restriction, increase in fiber). EPA and DHA are not present in vegetable oils and are reported to influence coronary heart disease by mechanisms independent of cholesterol lowering. Therefore, it is not clear whether the results of previous meta-analyses are driven by 1) reductions in saturated fat, 2) replacement of saturated fat with LA-rich vegetable oil, 3) or to increases in EPA+DHA. The objective of this systematic review and meta-analysis is to determine whether randomized controlled trials that specifically replaced saturated fat with LA-rich vegetable oils provide evidence to support the traditional diet-heart hypothesis.
-----
RANDOMIZED CONTROLLED TRIALS INCLUDED IN MAIN ANALYSIS
Only five randomized controlled trials met the inclusion criteria: the Minnesota Coronary Experiment (MCE), the Sydney Diet Heart Study (SDHS), the Rose Corn Oil Trial (RCOT), the Los Angeles Veterans study (LA Vet), and the Medical Research Council Soy study (MRC Soy) (table K). These are the five known trials that randomly assigned individual participants to a diet intervention that provided vegetable oil rich in linoleic acid in place of saturated fat compared to a usual care control diet, reported deaths from coronary heart disease or all-causes, and had no between-group differences in major concomitant interventions. Compared to control groups, all five intervention groups lowered serum cholesterol (mean range from 8 to 14% lower). They represent a total of 10,808 individuals.
---
EXPLANATION OF STUDIES NOT INCLUDED IN MAIN ANALYSIS
Exclusion of the Finnish Mental Hospital Study (FMHS)
The FMHS has been included in some previous meta-analyses of randomized controlled trials [21]. However, the FMHS is not a randomized controlled trial and has other critical limitations [Reviewed in 109, 137]. For example, there was disproportionate exposure to the cardiotoxic drug thioridazine in the control arm. The FMHS was a 12-year crossover study that assigned two hospitals (Hospital N and Hospital K) of mostly schizophrenic patients (77% in K and 69% in N) in 1959 to either a soybean oil based serum cholesterol-lowering diet (N) or the usual hospital diet (K) for 6 years. After this initial 6-year phase, diets were switched so that Hospital N patients received the Hospital N usual diet and Hospital K patients received a soybean oil diet. Issues related to within-hospital diet exposures at crossover in 1965 also confounded this unusual design. For example, study populations were ‘rejuvenated by discarding the six oldest annual cohorts and admitting six new annual cohorts on the younger end of the age range’ in 1965 [26].
Randomized controlled trials included in sensitivity analyses
Three additional randomized controlled trials were included in a sensitivity analysis: the Oslo Diet Heart Study, the St. Thomas Atherosclerosis Regression Study, and the Diet and Re-infarction Trial (tableK). These diet-heart trials, which randomly assigned individual participants and reported deaths from coronary heart disease and all-causes, have been included in previous meta-analyses on this topic. However, two of these trials were confounded by unequal application of other dietary factors, and another achieved only a very modest change in dietary LA without provision of study oils.
Oslo Diet Heart Study (ODHS)
The ODHS has often been represented as a test of the replacement of saturated fat with an LA-rich vegetable oil (soybean oil). However, in addition to soybean oil, the intervention group received a very large dose (~5 grams per day) of n-3 EPA+DHA from provision of sardines canned in cod liver oil, and was advised to restrict sugar intake and to replace refined carbohydrates with less processed selections [125-127]. Since EPA+DHA (and sugar) are reported to influence coronary heart disease risk by mechanisms independent of serum cholesterol lowering, it is not possible to determine which intervention components were responsible for study results.
St. Thomas Atherosclerosis Regression Study (STARS)
STARS [128-130] has often been represented as a test of the replacement of saturated fat with an LA-rich vegetable oil. However, unequal administration of multiple dietary factors do not allow for determination of the effects of LA. For example, the STARS intervention group received advice (and some study foods) to: (1) reduce total fat and saturated fat; (2) increase n-6 and n-3 polyunsaturated fatty acids, (3) avoid processed foods, and (4) increase dietary fiber (especially the soluble fiber polygalacturonate). While this intervention had only a very modest (non-statistically significant) increase in n-6 LA (+1.6%E), it doubled EPA+DHA from (100 to 210 mg per day), lowered total fat by 27%, and increased fiber by 53%. Given these multifaceted changes, the very modest increase in LA likely accounted for only a small fraction of the observed 12.2% serum cholesterol lowering achieved. Taken together, this study design provides little insight into whether replacement of saturated fat with LA rich vegetable oils can reduce coronary heart disease and death. Nevertheless, since other meta-analyses on this topic included STARS, we include it here in sensitivity analysis.
Diet and Re-infarction Trial (DART)
The DART ‘fat advice’ intervention group achieved only a very modest reduction in serum cholesterol (-4%), and only modestly increased total polyunsaturated fatty acids, without providing study oils or reporting the specific n-6 and n-3 content of such increases. Given these limitations, it is not clear how much the DART results can be expected to help determine whether replacement of saturated fat with LA-rich oil [26] reduces risk of coronary heart disease and death (the traditional diet heart hypothesis) 131-136. Nevertheless, since other meta-analyses on this topic included DART, we include it here a sensitivity analysis.
---
SUMMARY OF EVIDENCE
In our main meta-analyses, based on the five randomized controlled trials that provided LA-rich vegetable oil in place of saturated fat, we found no evidence for reductions in either coronary heart disease mortality or all-cause mortality. This conclusion was unchanged after sensitivity analyses that either 1) included randomized controlled trials that offered advice only or that, in addition to LA sources, also provided n-3 EPA and DHA, or 2) included composite or nonfatal endpoints. However, evidence of moderate heterogeneity weakens the conclusions we can make about the coronary heart disease mortality findings and their ability to translate into recommendations for the population. Exploratory analyses suggest that neither the between-group differences in serum cholesterol lowering nor the doses of LA provided help to explain this heterogeneity.
LIMITATIONS
The small number of randomized controlled trials that have tested the traditional diet-heart hypothesis replacing saturated fat with LA-rich vegetable oil is an important limitation of our meta-analysis. Remarkably, only five diet-heart randomized controlled trials have specifically tested whether provision of an LA-rich vegetable oil in place of saturated fat reduced risk of coronary heart disease mortality or all-cause mortality. The fact that the Minnesota Coronary Experiment (MCE) accounted for about 80% of all randomized participants in these trials highlights the paucity of causal evidence supporting the traditional diet heart hypothesis and the importance of the MCE in assessing the evidence base for LA-rich interventions. Even with inclusion of advice-only trials (with only modest diet changes and other limitations) and trials confounded by provision of large quantities of n-3 EPA+DHA in sensitivity analyses, MCE still accounted for 68% of all randomized participants. The small number of randomized controlled trials, coupled with differences in methodological quality and design, and population characteristics of the individual trials (tables K and L) [reviewed in 13, 109] indicate that more research is needed in this area before evidence-based recommendations can be supported.
Competing interests: No competing interests
Summary of the article
This is a secondary analysis of data collected as part of a study of institutionalized individuals (mental patients and nursing home patients) that took place in the late 1960s and early 1970s. The authors of the current article re-analyzed data on 2,355 of the original 9,570 participants. Although the original sample was randomized, authors of the current article had incomplete records of randomization information and so analyzed these data longitudinally. The also investigated autopsy records for 149 individuals.
Individuals either received the intervention exposure in which hospitals served more polyunsaturated fatty acids and less saturated fatty acids, or the control exposure in which hospitals also served more polyunsaturated fatty acids (though not as much as the intervention exposure) and no change in saturated fatty acids, compared to baseline diets. Though the authors briefly mention trans-unsaturated fatty acids, these were not accounted for in the original study. Authors of the current article reported results for 2,355 participants who were followed for an average of 2.9 years.
The authors found that the intervention did reduce serum cholesterol compared to the control group (p<0.001), as expected. The mean change was -31.2 mg/dl for the intervention group.
They then discuss findings on relationship between serum cholesterol and mortality (death by any cause). They found that, among those 65 year old or older, lower serum cholesterol was associated with higher risk of mortality. There was no association among those less than age 65 years old. They did not look at death from cardiovascular disease or even incidence of cardiovascular disease due to limitations in the dataset.
They also looked at autopsies of 149 individuals for signs of heart disease. After adjusting for confounders, there was no association between serum cholesterol and signs of heart disease (e.g., myocardial infarction, coronary atherosclerosis, or aortic atherosclerosis).
Additionally, they conducted a systematic review which included five randomized controlled trials. In this review, they found no relationship between saturated fat intake and heart disease.
Critique of the article
The authors set out to reinvestigate data to look at the relationship between serum cholesterol and heart disease in a randomized controlled trial. Unfortunately, due to limitations and data availability they were not able to fully accomplish this. Instead, what they actually did was conduct a longitudinal analysis of the relationship between serum cholesterol and all-cause mortality over an average follow-up period of 2.9 years.
The only finding in this article that goes against conventional wisdom is the finding that lower serum cholesterol is associated with increased mortality rate in those above age 64 years old. However, there is no accounting for many co-morbidities (e.g., alcoholism, drug abuse, depression, etc.) that may be present in this high risk population (e.g., mental patients and institutionalized older adults). Additionally, authors investigated all-cause mortality, not heart disease, and so this finding has limited value in a heart disease context.
The findings from the autopsies were from a small sample and unsurprisingly showed no relationship between serum cholesterol and heart disease after controlling for confounders.
Also, very importantly, actual dietary intake was not assessed, so findings are only relevant to serum cholesterol levels, not diet. Both exposure condition (I.e., intervention vs. control) provided trans unsaturated fatty acids (shown to increase heart disease risk), and due to not accounting for this in the meals served or food consumed, trans unsaturated fatty acid intake can be a big confounder to these findings.
The findings from the authors’ meta-analysis found no relationship between saturated fat intake and heart disease, although others have, such as this one (http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2843598/) by Mozaffarian et al., 2010. A careful review of the meta-analysis, and comparison with contradictory examples would be needed to determine the relevance of this finding.
Relevance to dietary guidance
Dietary guidance is made based on the preponderance of the evidence. Not only is one article not sufficient to influence dietary guidance, this particular article provides no new information on the relationship between diet and heart disease, as neither diet nor heart disease was assessed in the sample of 2,355 institutionalized individuals. There are some interesting findings presented related to serum cholesterol and all-cause mortality in older adults that may be an avenue for further research, but due to the reported limitations in the data and analyses in this article, implications of this finding are uncertain.
Competing interests: No competing interests
Does quality of dietary fats still matter?
I applaud Ramsden and co-workers for their effort (1). Based on their meta-analysis of 8 RCTs, intervention effect estimate (hazard ratio [HR] and 95% confidence interval [CI]) was 1.00 (0.81 to 1.24) for coronary heart disease (CHD) mortality and 1.00 (0.87 to 1.15) for all-cause mortality (1).
Their results are in line with Cochrane review subgroup analyses in which saturated fats were replaced with polyunsaturated fats (2). In addition, risk ratio (RR) for cardiovascular disease (CVD) events was 0.73 (95% CI 0.58 to 0.92, I^2 69%) based on 7 RCTs (2, 3). Minnesota Coronary Experiment, however, was excluded from the Cochrane review based on its duration (2).
One of the RCTs also included in these meta-analyses (1-3) was Los Angeles Veterans study (4). According to Dayton et al (page II-6):
"The primary end point was coronary heart disease manifested either as sudden death or as acute myocardial infarction." (4).
This RCT showed statistically non-significant result for its primary outcome (4).
I have earlier re-analyzed the aforementioned Cochrane review subgroup analysis considering CVD events in this very same journal (rapid response) (5). Briefly, I used t-distribution based method by Hartung-Knapp-Sidik-Jonkman (HKSJ) for random-effects meta-analysis. RR for CVD events was 0.73 (95% CI 0.50 to 1.08) with the HKSJ method (see ref. 5 for details).
Alternative meta-analysis methods than the standard DerSimonian-Laird random-effects meta-analysis should be preferred especially when number of studies is small and statistical variability is evident between studies (6). Different analytical choices might also make a difference when focus is on subjective or objective outcomes (7).
Panagiotou and Ioannidis showed that interpretation of meta-analysis depended on whom were asked (8). For example, researchers who had published more in the surveyed field, believed that stronger association (higher estimates of true odds ratios) existed based on a meta-analysis which included their own study (8).
Naturally, we will never know what the truth is with any given study (or combination of studies), but we should put our faith in research methodologists and statisticians for their capability of evaluating research evidence in a least biased way.
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.
2. Hooper L, Martin N, Abdelhamid A, Davey Smith G. Reduction in saturated fat intake for cardiovascular disease. Cochrane Database Syst Rev 2015; 6: CD011737.
3. Hooper L, Martin N, Abdelhamid A. Cochrane corner: what are the effects of reducing saturated fat intake on cardiovascular disease and mortality? Heart 2015; 101: 1938-40.
4. Dayton S, Pearce ML, Hashimoto S, Dixon WJ, Tomiyasu U. A Controlled Clinical Trial of a Diet High in Unsaturated Fat in Preventing Complications of Atherosclerosis. Circulation 1969; 40: II-1-II-63.
5. Kivelä JM. Does quality of dietary fats matter? http://www.bmj.com/content/351/bmj.h3978/rr-19 Cited April 16, 2016
6. Cornell JE, Mulrow CD, Localio R, et al. Random-effects meta-analysis of inconsistent effects: a time for change. Ann Intern Med 2014; 160: 267-70.
7. Dechartres A, Altman DG, Trinquart L, Boutron I, Ravaud P. Association between analytic strategy and estimates of treatment outcomes in meta-analyses. JAMA 2014; 312: 623-30.
8. Panagiotou OA, Ioannidis JP. Primary study authors of significant studies are more likely to believe that a strong association exists in a heterogeneous meta-analysis compared with methodologists. J Clin Epidemiol 2012; 65: 740-7.
Email: jesper.m.kivela@helsinki.fi
Competing interests: I like statistics.