Fat and the heart

Between 1968 and 1973, in six Minnesota state mental hospitals and a nursing home, 9,423 women and men were randomly assigned to one of two diets. One replaced saturated fat with corn oil and a corn-oil margarine; the other kept the animal fats, common margarines and shortenings the hospitals already served. The trial was double blind, and its co-principal investigator was Ancel Keys, whose Seven Countries Study this lesson meets later. The corn-oil diet cut blood cholesterol by 13.8 percent, against 1 percent in the controls. It didn't cut deaths. Much of it went unpublished until Christopher Ramsden's team recovered it in 2016.¹ Supporters of saturated fat limits have serious answers to that trial. This lesson is about how "eat less saturated fat" breaks into claims of very different strength, and how to tell which one somebody is making.

Three claims that travel together

"Fat is bad for your heart" usually runs three claims together.

1. Industrial trans fat harms the heart. Established. Every side of the dispute accepts it.
2. LDL cholesterol causes atherosclerosis, the build-up of fatty plaque in artery walls, and saturated fat raises LDL. Established, with a small dissent this lesson names.
3. Eating less saturated fat cuts heart attacks, strokes and deaths at the intakes people actually eat. Contested, by serious people on both sides.

The commonest mistake here is getting that sort wrong: treating claim 2 as if it were as shaky as claim 3, or claim 3 as if it were as solid as claim 1.

First, the names. Saturated fat dominates in fatty meat and full-fat dairy, butter included; the 2015 US guidelines note that "foods that are higher in dietary cholesterol, such as fatty meats and high-fat dairy products, are also higher in saturated fats."²⁰ Unsaturated fats are monounsaturated (MUFA) or polyunsaturated (PUFA). Linoleic acid, the omega-6 fat lesson 1 met, is in Ramsden's words "the most abundant polyunsaturated fatty acid", supplied by the old trials' corn and safflower oils.²

Trans fat: what agreement looks like

Trans fats, in the opening lines of a 2006 review by Mozaffarian and colleagues, "are formed during the partial hydrogenation of vegetable oils, a process that converts vegetable oils into semisolid fats for use in margarines, commercial cooking, and manufacturing processes."⁶ This course read only that opening paragraph.

Chowdhury's own authors drew a wider conclusion: the evidence "does not clearly support cardiovascular guidelines that encourage high consumption of polyunsaturated fatty acids and low consumption of total saturated fats."⁷ The paper was corrected after publication, and Walter Willett published a criticism of errors in it, chiefly on omega-6 fat. Its saturated fat figure also compares the top third with whatever the bottom third ate instead, a point this lesson comes back to.

In 2015 the US Food and Drug Administration made "a final determination that there is no longer a consensus among qualified experts that partially hydrogenated oils (PHOs), which are the primary dietary source of industrially-produced trans fatty acids (IP-TFA) are generally recognized as safe (GRAS) for any use in human food".⁵ This course read the Federal Register's summary. WHO's trans fat fact sheet of January 2024, read in full, says "high intake of trans fat increases the risk of death from any cause by 34%, coronary heart disease deaths by 28%, and coronary heart disease by 21%. Trans fat has no known health benefits", and attributes "More than 278 000 deaths each year globally" to the industrial kind.⁴

This course didn't read the feeding trials on trans fat, so what it shows you here is cohort evidence, a regulator's ruling and WHO's figures. What makes the claim established is that they agree and that nobody in the saturated fat fight disputes it. The dissenters use it themselves. Ramsden's team asked whether trans fat in Minnesota's corn-oil margarine could explain the missing benefit, and concluded that the control diet, full of "common margarines and shortenings (major sources of trans fat)", held more.¹ The dissenting document behind the 2026 US guidelines, met below, counts a trial's cut in hydrogenated margarines as a disqualifying confounder.¹⁵

LDL and the artery

Blood carries cholesterol in particles, one kind being low-density lipoprotein, LDL. In 2017 the European Atherosclerosis Society published a consensus statement whose title is its claim: "Low-density lipoproteins cause atherosclerotic cardiovascular disease."⁸ This course read the abstract and, from the full text, the sections on Mendelian randomisation, statin trials and funding.

The case is that evidence with different weaknesses agrees. Pooling cohorts, Mendelian randomisation studies and trials with "more than 2 million participants with over 20 million person-years of follow-up", the panel found "a remarkably consistent dose-dependent log-linear association between the absolute magnitude of exposure of the vasculature to LDL-C and the risk of ASCVD". In plainer words: the more LDL the arteries meet, the higher the risk, rising steadily. (LDL-C is the cholesterol carried in LDL; ASCVD is atherosclerotic cardiovascular disease; person-years, from lesson 4, count each person once per year followed.) The clearest case is genetic: "Rare genetic mutations that cause reduced LDL receptor function lead to markedly higher LDL-C and a dose-dependent increase in the risk of ASCVD, whereas rare variants leading to lower LDL-C are associated with a correspondingly lower risk of ASCVD." Its conclusion: the evidence "unequivocally establishes that LDL causes ASCVD."⁸

Two figures from the full text matter later. In statin trials (statins are drugs that lower LDL), each millimole per litre less LDL-C brought "a log-linear 22% proportional reduction in the risk of major cardiovascular events per millimole per litre reduction in LDL-C over a median of 5 years of treatment". (A millimole per litre is the other common unit for blood cholesterol; the 30 mg/dL met below is 0.78 of one.) In Mendelian randomisation, the gene-based natural experiment of lesson 1, the same drop carried "an expected 54% reduction in risk per millimole per litre reduction in LDL-C over 40 years of exposure".⁸ The panel says the effect "appears to increase with increasing duration of exposure to LDL-C". If so, a diet trial of a few years catches only part of a lifetime's effect, which is this course's reading of what that means for the diet trials below.

The panel adds a condition: lowering LDL by any route "should reduce the risk of ASCVD events proportional to the absolute reduction in LDL-C and the cumulative duration of exposure to lower LDL-C", provided "the achieved reduction in LDL-C is concordant with the reduction in LDL particle number and that there are no competing deleterious off-target effects" (no harm by some other route).⁸ Hold on to "particle number".

A small group led by Uffe Ravnskov disputes the whole claim. Their 2016 review of cohorts of people aged 60 and over concluded: "High LDL-C is inversely associated with mortality in most people over 60 years. This finding is inconsistent with the cholesterol hypothesis".⁹ In 2018 they argued the hypothesis "is unable to satisfy any of the Bradford Hill criteria for causality", a standard checklist for judging cause.¹⁰ This course read both abstracts.

Why isn't this a live dispute, when saturated fat is? Ravnskov's evidence is observational, in older people, and the EAS full text warns that cohort studies "are not randomized and are therefore unavoidably vulnerable to confounding, reverse causation, and other forms of bias".⁸ This course didn't find a published reply to Ravnskov, so the checkpoint below is its own reasoning.

If you take a medicine to lower your cholesterol, nothing in this lesson is a reason to change it without your doctor.

Saturated fat: what the trials found

The best-known synthesis is the 2020 Cochrane review by Lee Hooper and colleagues: 15 randomised trials of at least two years, 56,675 people.³ This course read the abstract and named parts of the full text, listed in the Sources.

Its headline: the long-term trials "suggested that reducing dietary saturated fat reduced the risk of combined cardiovascular events by 17% (risk ratio (RR) 0.83; 95% confidence interval (CI) 0.70 to 0.98, 12 trials, 53,758 participants of whom 8% had a cardiovascular event, I² = 67%, GRADE moderate-quality evidence)". GRADE is the standard system for rating how certain evidence is, from high through moderate and low to very low; the I² says the trials disagreed a good deal, as lesson 1 explained.

Three findings shape how far the 17 percent carries. It grows with the size of the cut: meta-regression "suggested that greater reductions in saturated fat (reflected in greater reductions in serum cholesterol) resulted in greater reductions in risk of CVD events, explaining most heterogeneity between trials". (Serum cholesterol is cholesterol measured in the blood; CVD is cardiovascular disease.) That's a dose-response pattern, and the review's account of that I². It shrinks in the best trials, as a footnote to the review's own summary table says: "Limiting trials to those at low summary risk of bias suggested a smaller and non‐statistically significant effect (RR 0.96, 95% CI 0.76 to 1.20) suggesting little or no effect on risk of CVD events. Downgraded once (along with publication bias)." So the review counts this against its own headline: it's part of why the 17 percent is graded moderate, not high. And one recovered trial moved it: adding Sydney's data in 2020 took the headline from 21 percent to 17, which the authors say "has not changed the overall conclusion."³

The authors conclude that the findings "suggest that reducing saturated fat intake for at least two years causes a potentially important reduction in combined cardiovascular events", and found "no evidence of harmful effects of reducing saturated fat intakes."³

The chart below lines the five estimates up on one scale.

Only the top interval sits wholly below the dashed line. Every estimate for death crosses 1, whoever pooled. The bottom row is Ramsden's 2016 meta-analysis of five linoleic acid trials, a narrower set chosen by different rules.¹

The newest review, by Steen and colleagues, published in December 2025, pooled 17 trials with 66,337 people and worked out absolute effects by baseline risk.¹³ This course read the abstract. For deaths from any cause it reports "low to moderate certainty that reducing saturated fat intake may result in a reduction in all-cause mortality (risk ratio [RR], 0.96 [95% CI, 0.88 to 1.06])", an interval that crosses 1. Its conclusion: "For persons at low cardiovascular risk, reducing or modifying saturated fat intake has little or no benefit over a period of 5 years. Among persons at high cardiovascular risk, low- to moderate-certainty evidence was found for important reductions in mortality and major cardiovascular events, particularly for MI, with respect to replacing saturated fat with polyunsaturated fat." (MI is a heart attack.) As of September 2026 neither WHO's 2023 guideline nor the Cochrane review reflects it.

The case for the limit, in its own words

The mainstream case is a chain, and the trials are its last link. The American Heart Association's 2017 advisory, read at abstract level: "Taking into consideration the totality of the scientific evidence, satisfying rigorous criteria for causality, we conclude strongly that lowering intake of saturated fat and replacing it with unsaturated fats, especially polyunsaturated fats, will lower the incidence of CVD." Trials that "replaced it with polyunsaturated vegetable oil reduced CVD by ≈30%, similar to the reduction achieved by statin treatment", while "replacement of saturated fat with mostly refined carbohydrates and sugars is not associated with lower rates of CVD and did not reduce CVD in clinical trials."¹¹ The abstract doesn't name those trials, and this course couldn't read the full text.

Christensen and colleagues, in a 2024 review read at abstract level, lay the chain out: "the link between low-density lipoprotein (LDL) particles and ASCVD", then "the link between dietary fat quality and LDL particles", then fat quality through LDL to disease.¹⁷ The first is the EAS consensus. The second rests on feeding trials: WHO, drawing on a review by Mensink, reports that replacing saturated fat with PUFA, MUFA or carbohydrate "all resulted in reductions in low-density lipoprotein (LDL) cholesterol in adults (high certainty evidence)", PUFA most.¹² If those hold, the mainstream argues, the third follows. Sun and Hu of Harvard, in a 2026 review read at abstract level, read those trials as confirming it: "longitudinal observational studies and long-term high-quality dietary intervention trials demonstrated that isocalorically replacing SFA with PUFA led to a lower risk of developing CAD."¹⁶ (SFA is saturated fat, CAD coronary artery disease, and isocalorically means at equal calories.)

WHO's 10 percent limit, in its 2023 guideline (this course read the summary in full), is a strong recommendation at moderate certainty, and WHO gives its reason: "evidence of moderate certainty overall from different study types assessing both risk factors and incidence of CVDs suggested reduced risk of CVDs with lower SFA intake. No undesirable effects or other mitigating factors were identified that would argue against a lower SFA intake."¹² The first sentence is the chain again, evidence of different kinds pointing one way. The second is a judgement about acting on evidence short of certainty: a benefit graded moderate, no harm found, so advise it.

The case against the limit, in its own words

The dissent opens with a concession. Astrup and eleven colleagues, in a 2020 cardiology journal review, accept that "SFAs increase low-density lipoprotein (LDL) cholesterol".¹⁴ This course read the abstract and, from a full-text copy, the conclusions and funding statement. Their case: "The recommendation to limit dietary saturated fatty acid (SFA) intake has persisted despite mounting evidence to the contrary. Most recent meta-analyses of randomized trials and observational studies found no beneficial effects of reducing SFA intake on cardiovascular disease (CVD) and total mortality, and instead found protective effects against stroke." Against the first half, Cochrane, published the same year, found 17 percent fewer combined events; on stroke it called the effect "unclear as the evidence was of very low quality".³

On LDL, the particle matters: in most individuals, they write, the rise "is not due to increasing levels of small, dense LDL particles, but rather larger LDL particles, which are much less strongly related to CVD risk." This is where the EAS proviso about particle number meets the dissent. The EAS panel notes that "plasma LDL burden is usually estimated by determination of plasma LDL cholesterol level (LDL-C)":⁸ the cholesterol the particles carry, not a count of them. Its promise of proportional benefit assumes the count falls with LDL-C. If saturated fat's rise comes in larger particles, the same cholesterol rides in fewer of them, so on Astrup's reading the LDL-C rise overstates the risk. Joining the two papers this way is this course's reading; neither answers the other. On food: "Whole-fat dairy, unprocessed meat, and dark chocolate are SFA-rich foods with a complex matrix that are not associated with increased risk of CVD. The totality of available evidence does not support further limiting the intake of such foods." "Further" is their own limit. And their conclusion: "The long-standing bias against foods rich in saturated fats should be replaced with a view toward recommending diets consisting of healthy foods."¹⁴

Ramsden's recovered trials are the dissent's trial evidence. Minnesota found "no mortality benefit for the intervention group in the full randomized cohort or for any prespecified subgroup" (a subgroup named before the analysis), and "There was a 22% higher risk of death for each 30 mg/dL (0.78 mmol/L) reduction in serum cholesterol".¹ The conclusion: replacing saturated fat with linoleic acid "effectively lowers serum cholesterol but does not support the hypothesis that this translates to a lower risk of death from coronary heart disease or all causes", and "incomplete publication has contributed to overestimation" of its benefits.¹ The Sydney trial gave 458 men who'd recently had a coronary event safflower oil and margarine in place of animal fats, or no instruction, and the oil group died more often: "all cause 17.6% v 11.8%, hazard ratio 1.62 (95% confidence interval 1.00 to 2.64), P=0.05".² (Lessons 2 and 3 glossed hazard ratios and P values.) This course read the abstract.

The dissent's newest statement is The Scientific Foundation for the Dietary Guidelines for Americans, 2025–2030, published in January 2026, compiled by Ramsden; this course read its fats chapter in full.¹⁵ On the trials: "none of the individual RCTs demonstrated the anticipated benefit, and a couple suggested potential for increased risk in some populations despite greater cholesterol lowering". It concludes that half a century of research "has not confirmed" that lowering saturated fat below 10 percent of energy, or replacing it with oils rich in linoleic acid, "reduces coronary heart disease or mortality risk", and: "Within typical intake ranges, saturated fat appears neither uniquely harmful nor protective. The evidence therefore supports a neutral stance: Foods containing saturated fat can be part of healthy dietary patterns when consumed in reasonable amounts and within minimally processed contexts".¹⁵ It doesn't call saturated fat good for you, and "in reasonable amounts" is its own qualifier.

A cohort meta-analysis often cited for the dissent is narrower in its own words. Siri-Tarino and colleagues, in 2010, found "no significant evidence for concluding that dietary saturated fat is associated with an increased risk of CHD or CVD", and added at once: "More data are needed to elucidate whether CVD risks are likely to be influenced by the specific nutrients used to replace saturated fat."¹⁸ This course read the abstract and parts of the full text. Two of its authors, Qi Sun and Frank Hu, wrote the 2026 mainstream review quoted above.

One trial, two readings

The dissent's reading of Minnesota. A big double-blind trial of the very oil the guidelines favour cut cholesterol and saved nobody.

The objections, several from the paper. "Participants were followed only while in hospital, and only about a quarter of randomized participants remained in the study for a year or longer."¹ The paper states that limit itself, and the objection it supports is plain: heart disease takes years, and most people ate the diet for months. On the 22 percent, the authors warn that "the analysis of the association between serum cholesterol and death is observational in nature. Therefore, it is not possible to examine causality".¹ It compares people within one arm, and those whose cholesterol fell most weren't chosen at random. Sydney was small, its intervals brush 1, and its men were all heart patients; Ramsden's team said Minnesota's margarine "likely contained some trans fat", and whether Sydney's did is a question its abstract doesn't answer.¹²

Now the other direction. The Scientific Foundation sets aside the Oslo Diet-Heart Study, which it says "markedly reduced partially hydrogenated fish oil (PHFO) and PHVO margarines ..., markedly increased fish-derived omega-3 fatty acids via provision of sardines and cod liver oil ..., and improved overall dietary quality ... changes that make it impossible to isolate the effects of high-linoleic-acid oils. When RCTs with dominant confounders such as ODHS are appropriately excluded, there is no indication of benefit."¹⁵ (PHVO is partially hydrogenated vegetable oil; ODHS is Oslo.)

Ramsden's own meta-analysis admitted only trials run "without confounding by concomitant interventions".¹ The AHA's ≈30 percent came from trials its abstract doesn't name. Reports say they include Oslo and leave out Minnesota, but this course couldn't read the full text to confirm it. If those reports are right, each side sets aside a trial the other counts, and each has a reason: Minnesota was short and leaky, Oslo changed several things at once. Reading its sources, this course thinks the gap between the AHA's 30 percent and Ramsden's null is largely a gap in which trials count. Yamada and colleagues, pooling nine trials in 2025, found no significant effect on cardiovascular deaths and concluded that "a reduction in saturated fats cannot be recommended at present", asking for trials "under the best possible medical care, including statin administration."¹⁹ That request implies a modern trial would run alongside statins, leaving a smaller extra effect for diet to show: this course's reading, not Yamada's words.

Who is at risk decides how much it matters

Steen's review moved the dispute from relative to absolute risk. Cochrane found that the reduction in combined events "did not alter by study duration, sex or baseline level of cardiovascular risk".³ So take 17 percent as fixed and ask what it buys. The arithmetic below is this course's, on round figures.

Step 1. In the trials, 8 percent had an event. Seventeen percent of 8 percent is about 1.4 percentage points: roughly 14 fewer events per 1,000 people over the few years the trials ran.

Step 2. Now take healthy younger adults, 2 in 100 of whom would have an event over those years. Steen's review set thresholds of importance, "5 and 10 per 1000 persons followed over 5 years for fatal and nonfatal outcomes, respectively".¹³ Work out the gain before reading on.

Step 3. Steen found that pattern in the trials: "For persons at low baseline cardiovascular risk, absolute reductions were below our thresholds of importance", while "for those at high risk, the benefits were above our thresholds, suggesting there may be important absolute reductions."¹³ The thresholds are the authors' judgement of how many fewer events are worth a change of diet; another panel could set them elsewhere. That's a value laid on the evidence, and lesson 8 shows two panels reaching opposite advice on meat through choices like it. Both camps can cite Steen.

What replaces it

Lesson 1 printed Li and colleagues' table: replacing 5 percent of energy from saturated fat with PUFA went with 25 percent less coronary heart disease, MUFA 15 percent less, whole grains 9 percent less, refined starches and sugars no significant difference.²² Jakobsen and colleagues, in 2009, pooled eleven cohorts, 344,696 people; this course read the abstract.²¹

WHO's recommendations follow the same grading. It "recommends replacing saturated fatty acids in the diet with polyunsaturated fatty acids (strong recommendation); monounsaturated fatty acids from plant sources (conditional recommendation); or carbohydrates from foods containing naturally occurring dietary fibre, such as whole grains, vegetables, fruits and pulses (conditional recommendation)".¹²

The trials are less tidy. Cochrane's subgroups "did not suggest significant differences between replacement of saturated fat calories with polyunsaturated fat or carbohydrate", and its authors say both "appear to be useful strategies".³ Steen, later and larger, found the effect "more pronounced when replacing saturated fat with polyunsaturated fat for nonfatal MI", a risk ratio of 0.75 (0.58 to 0.99), and rated its confidence that this subgroup difference is real as moderate.¹³ So the PUFA advantage is clearest in cohorts and thinner in trials.

And the dissent contests the very swap the mainstream favours. Ramsden's Sydney paper ends: "These findings could have important implications for worldwide dietary advice to substitute omega 6 linoleic acid, or polyunsaturated fats in general, for saturated fats."² The Scientific Foundation says "modern intake levels from refined oils now exceed physiological requirements severalfold", and asks for trials rather than asserting harm: "rigorous RCTs are needed to definitively determine whether reduced dietary exposures translate to improved clinical outcomes."¹⁵ Its objection to cohort substitution models, quoted in lesson 1, lands on the cohort half of the PUFA case; the trial half turns on which trials count.

Cheese, eggs and the food itself

Does a food behave like its fat? An expert panel at "a closed workshop", reported by Thorning and colleagues in 2017 with Astrup among the authors (its funding wasn't checked for this course), concluded that fermented dairy products "such as cheese and yogurt, generally show inverse associations" with cardiovascular disease and type 2 diabetes, and that "the nutritional values of dairy products should not be considered equivalent to their nutrient contents".²³ Blood markers of dairy fat point the same way in a 2021 meta-analysis, whose authors name its limit, "the inability of the biomarkers to distinguish different types of dairy foods".²⁴ This course read both abstracts. The mainstream's reply is an "instead of what?" answer. Sun and Hu write that evidence from feeding trials and observational studies "suggests that animal fat, including dairy fat, leads to adverse lipid profiles and elevated CAD risk when compared with nontropical plant-based oils rich in MUFA and PUFA."¹⁶ WHO's own answer to the food question: its recommendations "do not preclude consumption of particular foods. However, foods containing high levels of SFA should be consumed sparingly to meet the recommended level of intake."¹² Both sides can be right: dairy fat can look neutral against the average diet and worse against plant oils. That reconciliation is this course's hypothesis, which neither side has signed.

Eggs. Does cholesterol in food raise the LDL in your blood? The American Heart Association's 2020 advisory on dietary cholesterol, read at abstract level, says meta-analyses of intervention studies "differ in their findings", but "most associate intakes of cholesterol that exceed current average levels with elevated total or low-density lipoprotein cholesterol concentrations."³¹ In 2015 the US guidelines dropped their 300 mg limit and said "this change does not suggest that dietary cholesterol is no longer important to consider when building healthy eating patterns", adding that people "should eat as little dietary cholesterol as possible while consuming a healthy eating pattern."²⁰ Then two large mainstream analyses disagreed. Zhong and colleagues, in 2019, read at abstract level plus the full text's limitations section, pooled six US cohorts and found each extra half egg a day associated with more cardiovascular disease, "adjusted HR, 1.06 [95% CI, 1.03-1.10]". The egg associations "were no longer significant after adjusting for dietary cholesterol consumption", and the authors name their own limit: "this study relied on single measurement of egg and dietary cholesterol consumption."²⁵ Drouin-Chartier and colleagues, in 2020, read at abstract level, followed three cohorts for up to 32 years, adjusting for "updated lifestyle and dietary factors", and with a meta-analysis concluded that "moderate egg consumption (up to one egg per day) is not associated with cardiovascular disease risk overall".²⁶ This course's reading of the two is that measurement and adjustment choices can flip a single-food result: Zhong names its single measurement as a limit, and Drouin-Chartier adjusted for updated diet.

Keys, and the sugar papers

The Seven Countries Study began in 1958. Its website, run by people connected to it, describes "16 cohorts of middle-aged men from seven countries", in which "The average saturated fat intake at baseline was strongly correlated with 25-year CHD mortality rates."²⁷ That compares populations, not people, and the site answers that objection itself: "Ecological or cross-cultural associations are frequently viewed as the lowest level of evidence about causality. However, to demean ecological associations is to fail to recognize the powerful indication they provide about the population burden of disease." The honest criticism is on the same site: the cohorts "were not statistically representative of the correspondent countries and the choice was partly made on the basis of convenience." And in one Dutch cohort, Zutphen, "Dietary saturated fat was not associated with serum cholesterol and 10-year CHD mortality", which the site puts down to too little variation between the men and too much from day to day, lesson 2's measurement error.²⁷ The popular claim that Keys "cherry-picked" seven of twenty-two countries was traced, in this course's research, to a 1953 graph, not this study; that paper wasn't read, so the lesson leaves the detail there.

The sugar papers. In 2016 Kearns, Schmidt and Glantz published internal documents of the Sugar Research Foundation; this course read the abstract and named sections of the full text. The Foundation's first heart disease project, in 1965, was "a literature review published in the New England Journal of Medicine, which singled out fat and cholesterol as the dietary causes of CHD and downplayed evidence that sucrose consumption was also a risk factor. The SRF set the review's objective, contributed articles for inclusion, and received drafts. The SRF's funding and role was not disclosed."²⁸ Their reading of it: "our findings suggest the industry sponsored a research program in the 1960s and 1970s that successfully cast doubt about the hazards of sucrose while promoting fat as the dietary culprit in CHD."²⁸ And their limitations are frank: "There is no direct evidence that the sugar industry wrote or changed the NEJM review manuscript; the evidence that the industry shaped the review's conclusions is circumstantial."²⁸

The historians David Johns and Gerald Oppenheimer replied in Science in 2018. This course didn't read the paper, only Columbia University's press release, which quotes it: "There was no 'smoking gun.' There was no 'sugar conspiracy'-at least not one which we have identified", and "We think it is an error to demonize, almost as a reflex, scientists and their research when there is evidence of private funding".²⁹ The release adds that dairy money funded work finding saturated fat raised cholesterol. Both readings accept the documents; they part over how much the documents explain.

Who paid

Conflicts of interest run through every side. None is a reason to dismiss a finding; judge each by its design and data.

• Astrup's review came from a workshop "funded by the Nutrition Coalition ... in part with a generous grant from philanthropists Robert G. and Sue Douthit O'Donnell", and the statement adds: "The sponsors had no role in preparing or reviewing the manuscript before submission." Astrup "has received research funding from Danish Dairy Foundation, Arla Foods Amba, and the European Milk Foundation".¹⁴ Siri-Tarino's meta-analysis was "Supported by the National Dairy Council", and one author held "a Postdoctoral Fellowship from Unilever Corporate Research".¹⁸ The Ravnskov group's interests weren't checked for this course.
• The EAS consensus's "Travel and meeting logistics were supported by unrestricted educational grants from MSD and Amgen", makers of cholesterol drugs, and many of its authors disclose funding from drug makers, including Amgen, AstraZeneca, Sanofi-Regeneron and Pfizer.⁸
• "WHO funded the most recent update" of the Cochrane review, and Hooper was then a member of WHO's nutrition guidance advisory group; the review is one of four WHO's 2023 guideline rests on.³
• The 2026 Scientific Foundation was compiled by Ramsden, and its review authors include his co-author Daisy Zamora and two of Astrup's co-authors, J. Thomas Brenna and Jeff Volek. Brenna's published disclosures include the National Cattlemen's Beef Association.¹⁵
• The Seven Countries Study's sponsors included the American Heart Association and, for its Greek cohorts, an olive oil company.²⁷

Each side informed the body whose advice it defends.

What would settle it

Steen's review says "new trials are needed to clarify uncertainty".¹³ Ramsden says of Minnesota's lost records that "it is essential that these missing autopsy files are recovered and analyzed".¹ In this course's summary, it would take a long, large trial of PUFA for saturated fat counting heart attacks and deaths; trials of foods, such as full-fat against low-fat dairy; and agreed rules for which old trials count.

One dated line on policy. As of September 2026, the 2025–2030 Dietary Guidelines for Americans, released in January 2026, keep the ceiling, "In general, saturated fat consumption should not exceed 10% of total daily calories", and on cooking fats say "prioritize oils with essential fatty acids, such as olive oil. Other options can include butter or beef tallow."³⁰ Lesson 9 describes those guidelines and the dispute over them.

What people get wrong

"Science has proved saturated fat is harmless." No trial synthesis says so. Cochrane found 17 percent fewer combined events at moderate certainty, and Steen possibly important benefits at high risk.³¹³ Even the Scientific Foundation's "neutral stance" is qualified: "in reasonable amounts".

"Cutting saturated fat is proven to save lives." No pooled estimate here shows it: every interval for death from any cause crosses 1, and the events benefit fades in the trials at low risk of bias.³ Steen found possibly important reductions in deaths for people at high risk, at low to moderate certainty, and little or no benefit over five years at low risk.¹³ The mainstream's case rests on the LDL chain and fewer events, not on proven lives saved.

"The US government now says saturated fat is fine." As of September 2026 its guidelines keep the 10 percent ceiling; the "neutral stance" is the Scientific Foundation's, not the guidelines' own words.¹⁵³⁰

"Full-fat dairy is bad for your heart." Against the average diet, Thorning's panel and blood markers of dairy fat point the other way; the case against it is a comparison with plant oils.¹⁶²³²⁴

"Any cut in saturated fat helps." The swap decides it. Cohorts found no benefit when refined starches and sugars replaced it, and Jakobsen found slightly more coronary events for carbohydrate as a whole, though Li found whole grains went with 9 percent less; the AHA says the refined swap "did not reduce CVD in clinical trials".¹¹²¹²²

"LDL doesn't cause heart disease." That's the Ravnskov group's minority position.⁹ Genetic, Mendelian randomisation and drug-trial evidence point the other way,⁸ and Astrup's group, the leading saturated fat dissent, argues about which LDL particles carry the risk rather than against LDL causality.¹⁴

"The 2015 guidelines said cholesterol in food doesn't matter," and its mirror, "eggs are proven to harm the heart." The guidelines dropped a number, not the advice, and two large cohort analyses disagree about eggs.²⁰²⁵²⁶

"The sugar industry is why we feared fat." The documents are real, and their own authors call the evidence of influence on the review's conclusions "circumstantial".²⁸

"Only industrial trans fat is harmful." WHO "determined that the available evidence did not support making a distinction between industrially produced and ruminant TFA", ruminant meaning the natural kind in meat and milk.¹² That's absence of evidence for a difference, not proof there's none.

Practice

Connections

Back. Lesson 1 introduced "instead of what?" with Li's table and promised Jakobsen beside it, and said feeding trials rarely reach disease: Minnesota fed people in hospitals, and only about a quarter of them stayed a year or longer. Lesson 2's WHI cut fat of every kind and left untested the swap of saturated fat for unsaturated; Minnesota and Sydney are two tests of it. Lesson 5 noted fat had an energy range but no requirement, and a 30 percent limit set before there were data; this is what the evidence on fat quality shows now.

Forward. Lesson 7 takes salt and sugar, two disputes with the same shape: a well-shown effect on an intermediate measure, an argued-over low end, and a method that can manufacture a curve. Lesson 8 returns to the value layer in Steen's thresholds, and lesson 9 reads the 2026 US guidelines.

Go deeper

• Hooper and colleagues, 2020, the Cochrane review, free at PubMed Central. Read here: the abstract, and from the full text the plain language summary, the Summary of Findings table with its footnotes, and the funding and interest statements.
• Ramsden and colleagues, 2016, free at PubMed Central. Read here: the abstract, and from the full text the design, diets, "What this study adds" and Strengths and limitations. Minnesota, with its authors' own list of its limits.
• Ference and colleagues, 2017, the EAS consensus, free at PubMed Central. Read here: the abstract, and from the full text the sections on Mendelian randomisation and statin trials, and funding and conflicts. Why LDL is treated as a cause.
• The Scientific Foundation for the Dietary Guidelines for Americans, 2025–2030, free. Read here: the front matter and Chapter 5, "Fats and Oils", in full. The dissent's case on the trials, in one chapter.

Sources

1. C. E. Ramsden and colleagues, "Re-evaluation of the traditional diet-heart hypothesis", BMJ, 2016, doi 10.1136/bmj.i1246. Read: the abstract, and from the full text the design, diets, "What this study adds" and Strengths and limitations. That much of it was unpublished is the abstract's "previously unpublished data". A 2024 erratum was not read.
2. C. E. Ramsden and colleagues, "Use of dietary linoleic acid for secondary prevention of coronary heart disease and death", BMJ, 2013, doi 10.1136/bmj.e8707. Read: the abstract.
3. L. Hooper and colleagues, "Reduction in saturated fat intake for cardiovascular disease", Cochrane Database of Systematic Reviews, 2020, doi 10.1002/14651858.CD011737.pub3. Read: the abstract, and from the full text the plain language summary, Summary of Findings table 1 with its footnotes, "What's new", "History", Implications for practice, Sources of support and Declarations of interest. The abstract and the full-text table differ slightly on some mortality figures and trial counts; this lesson quotes the abstract.
4. World Health Organization, "Trans fat", fact sheet, 24 January 2024. Read: the full page.
5. US Food and Drug Administration, "Final Determination Regarding Partially Hydrogenated Oils", Federal Register 80(116), 17 June 2015. Read: the Federal Register record (abstract, action and dates) only.
6. D. Mozaffarian and colleagues, "Trans fatty acids and cardiovascular disease", New England Journal of Medicine, 2006, doi 10.1056/NEJMra054035. Read: the opening paragraph only.
7. R. Chowdhury and colleagues, "Association of dietary, circulating, and supplement fatty acids with coronary risk", Annals of Internal Medicine, 2014, doi 10.7326/M13-1788. Read: the abstract as currently indexed. The paper was later corrected; the erratum was not read. Walter Willett's published criticism was read as reproduced on Harvard's Nutrition Source.
8. B. A. Ference and colleagues, "Low-density lipoproteins cause atherosclerotic cardiovascular disease", European Atherosclerosis Society consensus statement, European Heart Journal, 2017, doi 10.1093/eurheartj/ehx144. Read: the abstract, and from the full text the sections on Mendelian randomisation and statin trials, and funding and conflicts.
9. U. Ravnskov and colleagues, "Lack of an association or an inverse association between low-density-lipoprotein cholesterol and mortality in the elderly", BMJ Open, 2016, doi 10.1136/bmjopen-2015-010401. Read: the abstract. Competing interests not checked.
10. U. Ravnskov and colleagues, "LDL-C does not cause cardiovascular disease", Expert Review of Clinical Pharmacology, 2018, doi 10.1080/17512433.2018.1519391. Read: the abstract. Competing interests not checked.
11. F. M. Sacks and colleagues, "Dietary Fats and Cardiovascular Disease", AHA presidential advisory, Circulation, 2017, doi 10.1161/CIR.0000000000000510. Read: the abstract only.
12. World Health Organization, Saturated fatty acid and trans-fatty acid intake for adults and children, guideline summary, 2023, landing page. Read: the guideline summary in full; the 134-page full guideline was not opened.
13. J. P. Steen and colleagues, "Effect of Interventions Aimed at Reducing or Modifying Saturated Fat Intake", Annals of Internal Medicine, 2026 (online 16 December 2025), doi 10.7326/ANNALS-25-02229. Read: the abstract. Its primary funding source is stated as "None".
14. A. Astrup and colleagues, "Saturated Fats and Health", Journal of the American College of Cardiology, 2020, doi 10.1016/j.jacc.2020.05.077. Read: the abstract, and from a full-text PDF hosted on a dairy-promotion site (the journal's site was blocked) the highlights, conclusions, and funding and disclosures. Other authors' disclosures were not verified.
15. US Departments of Health and Human Services and of Agriculture, The Scientific Foundation for the Dietary Guidelines for Americans, 2025–2030, January 2026, PDF. Read: the front matter and Chapter 5, "Fats and Oils", in full.
16. Q. Sun and F. B. Hu, "Fatty Acids, Their Food Sources, and Cardiovascular Disease", Journal of Nutrition, 2026, doi 10.1016/j.tjnut.2026.101669. Read: the abstract.
17. J. J. Christensen and colleagues, "Dietary fat quality, plasma atherogenic lipoproteins, and ASCVD", Atherosclerosis, 2024, doi 10.1016/j.atherosclerosis.2023.117433. Read: the abstract.
18. P. W. Siri-Tarino, Q. Sun, F. B. Hu and R. M. Krauss, "Meta-analysis of prospective cohort studies evaluating the association of saturated fat with cardiovascular disease", American Journal of Clinical Nutrition, 2010, doi 10.3945/ajcn.2009.27725. Read: the abstract, and from the full text the methods, funnel plot results, discussion and funding footnote.
19. S. Yamada and colleagues, "Saturated Fat Restriction for Cardiovascular Disease Prevention", JMA Journal, 2025, doi 10.31662/jmaj.2024-0324. Read: the abstract.
20. US Departments of Health and Human Services and of Agriculture, 2015-2020 Dietary Guidelines for Americans, December 2015. Read: the Chapter 1 text around page 32.
21. M. U. Jakobsen and colleagues, "Major types of dietary fat and risk of coronary heart disease", American Journal of Clinical Nutrition, 2009, doi 10.3945/ajcn.2008.27124. Read: the abstract.
22. Y. Li and colleagues, "Saturated Fats Compared With Unsaturated Fats and Sources of Carbohydrates", Journal of the American College of Cardiology, 2015, doi 10.1016/j.jacc.2015.07.055. Read: the abstract.
23. T. K. Thorning and colleagues, "Whole dairy matrix or single nutrients", American Journal of Clinical Nutrition, 2017, doi 10.3945/ajcn.116.151548. Read: the abstract. Funding not checked.
24. K. Trieu and colleagues, "Biomarkers of dairy fat intake", PLoS Medicine, 2021, doi 10.1371/journal.pmed.1003763. Read: the abstract.
25. V. W. Zhong and colleagues, "Associations of Dietary Cholesterol or Egg Consumption", JAMA, 2019, doi 10.1001/jama.2019.1572. Read: the abstract and the full text's limitations section.
26. J.-P. Drouin-Chartier and colleagues, "Egg consumption and risk of cardiovascular disease", BMJ, 2020, doi 10.1136/bmj.m513. Read: the abstract.
27. Seven Countries Study, sevencountriesstudy.com, archived January 2025. Read: the home page and five named pages on the study, its cohorts, its sponsors and its saturated fat findings. Run by people connected to the study.
28. C. E. Kearns, L. A. Schmidt and S. A. Glantz, "Sugar Industry and Coronary Heart Disease Research", JAMA Internal Medicine, 2016, doi 10.1001/jamainternmed.2016.5394. Read: the abstract, and from the full text the methods, payment passage, limitations, conclusions and disclosures.
29. D. M. Johns and G. M. Oppenheimer, "Was there ever really a 'sugar conspiracy'?", Science, 2018, doi 10.1126/science.aaq1618. Not read. Quotations and summary are from Columbia University's press release as republished by ScienceDaily, 15 February 2018.
30. US Departments of Health and Human Services and of Agriculture, Dietary Guidelines for Americans, 2025–2030, January 2026, PDF. Read: the whole document. Supports the dated line only; lesson 9 carries the rest.
31. J. A. S. Carson and colleagues, "Dietary Cholesterol and Cardiovascular Risk", AHA science advisory, Circulation, 2020, doi 10.1161/CIR.0000000000000743. Read: the abstract only.