Fats — the messiest story

TL;DR. For fifty years, dietary fat was the villain and margarine was the hero. Both were wrong. The picture that survives careful trials: saturated fat from real food is roughly neutral on heart disease; trans fats from partial hydrogenation were the actually-dangerous category we were told to eat instead; replacing saturated fat with olive oil and nuts (Lyon Diet Heart, PREDIMED) cuts cardiac events 30–70 percent; replacing it with refined carbohydrate does not; and the industrial diet has collapsed the omega-6 to omega-3 ratio from around 3:1 to between 10:1 and 25:1, with consequences that show up in cell membranes, inflammation, and mood. Seed oils are not poison in moderate amounts — the ratio and the source are the levers. This is the messiest story in nutrition because the science actually changed, and the public messaging never caught up.

What you'll learn

• The four fat families — saturated, monounsaturated, polyunsaturated, trans — and how their structure dictates how they behave in food and in your body.
• The rise and quiet collapse of the lipid hypothesis, from Ancel Keys to the 2016 recovery of the Minnesota Coronary Experiment.
• Why replacing saturated fat with refined carbohydrate produced no health gain, while replacing it with unsaturated fat from olive oil and nuts cut cardiac events 30–70 percent.
• The omega-6 to omega-3 collapse and what it means for inflammation and mood.
• An honest read on the seed-oil debate: not poison, not health food, ratio-and-context dependent.
• Why trans fats were the one fat the campaign actually got right — eventually.

The fat-types crash course

Dietary fat is not one substance. It is a family of fatty-acid chains that differ in length, in the number and position of their double bonds, and — for unsaturated fats — in the geometry of those bonds. Those structural details determine how a fat behaves at room temperature, how it oxidizes when heated, what your body does with it, and what it does to your cell membranes.

Saturated fatty acids (SFA) have no double bonds. They pack tightly, so they are solid at room temperature — butter, lard, beef tallow, coconut oil, the fat in dairy and red meat. Chemically stable and heat-tolerant.

Monounsaturated fatty acids (MUFA) have one double bond. Olive oil, avocado oil, and most of the fat in nuts. Liquid at room temperature, reasonably stable to heat.

Polyunsaturated fatty acids (PUFA) have two or more double bonds. Liquid even in the fridge, the most prone to oxidation. PUFAs split into two essential families your body cannot make from scratch: omega-6 (linoleic acid, mostly from seed oils and grains) and omega-3 (alpha-linolenic acid from leaves and flax, EPA and DHA from cold-water fish). Both are required for life; the ratio between them is one of the most consequential numbers in modern nutrition.

Trans fatty acids (TFA) have at least one double bond in the trans configuration — straight, instead of the natural bent cis shape. Trace amounts occur naturally in dairy. The dangerous version is industrial: partial hydrogenation blasts liquid vegetable oil with hydrogen to flip cis bonds to trans and create a solid, shelf-stable fat — margarine, Crisco, the frying oil that fast food was built on.

Each family does different things inside the body. SFA and MUFA are stable membrane components and metabolic fuel. Omega-6 PUFAs are precursors to pro-inflammatory eicosanoids; omega-3 PUFAs are precursors to anti-inflammatory ones. Trans fats lodge into membranes where cis fats belong, distort them, raise LDL, lower HDL, raise triglycerides, and promote clotting. Of the four, only trans fats have no biological role and no safe intake threshold.

The lipid hypothesis melt

The story most Americans absorbed — saturated fat causes heart disease, so eat margarine — came from a single research program. Ancel Keys, a Minnesota physiologist, proposed in the 1950s that dietary saturated fat raises blood cholesterol and that blood cholesterol drives coronary heart disease. His Seven Countries Study plotted SFA intake against heart-disease mortality across seven national diets and found a striking line. The American Heart Association endorsed the framework in 1961.

The political consolidation came in 1977. Senator George McGovern's Select Committee on Nutrition drafted Dietary Goals telling Americans to eat less meat; the cattle and dairy lobbies attacked, and the recommendation was rewritten to talk about saturated fat instead. With that linguistic move, U.S. dietary advice switched from naming foods to naming nutrients. Butter was replaced with margarine. Whole milk with skim. Lard with partially hydrogenated shortening.

Then the trials started reporting back.

Walter Willett and Frank Hu's 2001 review in the Journal of the American College of Nutrition — "Types of Dietary Fat and Risk of Coronary Heart Disease" — read every prospective cohort then available. Only two studies showed a significant positive link between SFA and CHD. Replacing SFA with carbohydrate did nothing. Replacing it with PUFA helped modestly. Replacing it with trans fat made things worse.

The Women's Health Initiative, a $415 million randomized trial of 48,000 postmenopausal women on a low-fat diet, reported in 2006 no significant reduction in heart disease, breast cancer, or colorectal cancer. The PURE study (Dehghan et al., Lancet 2017) followed 135,000 people across 18 countries for 7.4 years; higher SFA intake was associated with lower total mortality and no significant cardiovascular effect, while the recommended high-carb replacement was associated with higher mortality.

Two trials that should have settled the matter in the 1970s were sat on. The Sydney Diet Heart Study (1966–1973) randomized 458 post-MI men to replace SFA with safflower oil. LDL fell. Mortality went up. The Minnesota Coronary Experiment (1968–1973), a 9,423-person inpatient trial led by Keys himself, did the same with corn oil. LDL fell. Mortality did not improve, and in older participants it worsened. Both datasets were buried for forty years. Christopher Ramsden's NIH team recovered and re-analyzed them in 2013 and 2016 BMJ papers. The cleanest tests of the "lower SFA, lower LDL, save lives" causal chain failed at the third link.

The lipid hypothesis was a hypothesis, adopted as policy before it was settled. The one fat strongly linked to heart disease in the data was the one the campaign told Americans to eat instead of butter.

What replaces matters

The careful trials do not support "saturated fat is fine, eat what you want." They support a different claim: what you replace a fat with matters more than the fat you cut.

The Lyon Diet Heart Study (de Lorgeril, Circulation 1999) randomized 605 French heart-attack survivors to either standard low-fat advice or a Mediterranean-pattern diet rich in olive oil, canola-based margarine high in alpha-linolenic acid, vegetables, fruits, fish, and nuts. After roughly four years the Mediterranean group had about 70 percent fewer second heart attacks and 56 percent lower all-cause mortality. The trial was stopped early because the gap was too large to ethically continue. Total cholesterol barely differed between groups — the benefit was not running through the LDL number.

PREDIMED (Estruch et al., NEJM 2013) was the larger replication: 7,447 Spanish adults at high cardiovascular risk, randomized to a Mediterranean diet with extra-virgin olive oil, a Mediterranean diet with mixed nuts, or a low-fat control. Both Mediterranean arms cut major cardiovascular events by about 30 percent versus the low-fat group, even though they ate substantially more total fat. Stopped early on efficacy.

The shared pattern: SFA → unsaturated fat from whole-food sources cuts events sharply. SFA → refined carbohydrate is a wash or worse. This is what makes the "calorie is a calorie" framing fail. A calorie from olive oil and a calorie from corn syrup do different things in the body, and the matched-calorie trials show it.

The omega-6 to omega-3 collapse

A second story runs underneath the first. Omega-3 fats are made in the chloroplasts of green leaves — alpha-linolenic acid is among the most common molecules in growing photosynthetic tissue. Omega-6 fats are concentrated in seeds. For most of human evolution we ate green things, animals that had eaten green things, and the occasional seed; the estimated pre-industrial omega-6 to omega-3 ratio is around 3:1.

Industrial agriculture put both humans and our food animals on a seed diet. Cattle finished on corn. Chickens and pigs fed corn and soy. Cooking fats moved from butter, lard, and olive oil to soybean, corn, cottonseed, and sunflower. About 9 percent of all American calories now come from one fatty acid — linoleic acid, the omega-6 in soybean oil. The modern ratio sits between 10:1 and 25:1, depending on methodology. U.S. adipose-tissue omega-6 content rose from 9 percent in 1959 to 21 percent in 2008.

The ratio matters because omega-6 and omega-3 fats compete for the same enzymes and end up in the same cell membranes. NIH lipid researcher Joseph Hibbeln has linked rising population-level omega-6 intake to rising cardiovascular mortality, depression, suicide, and homicide across countries. Casey Means notes in Good Energy that membrane omega-6:omega-3 ratios shift within three days of a dietary change, and that distorted membranes affect insulin receptor function, mitochondrial dynamics, and inflammatory signaling. Clinical evidence for omega-3 supplements is mixed; the population-level ratio collapse and its membrane consequences are consistent.

Seed oils — the honest take

The seed-oil discourse has gotten polarized in a way the science does not warrant. Two extreme positions are wrong: "seed oils are toxic, avoid all of them" and "seed oils are health food, the more the better."

The mechanistic concerns are real. Modern industrial seed oils are produced by an RBD process — refined, bleached, deodorized — that uses high heat and solvents (hexane) to extract yield from low-value seeds (soy, cottonseed, sunflower, safflower, canola, corn). The polyunsaturated chains oxidize easily, especially when reheated at high temperature in commercial deep fryers. Oxidized lipids generate aldehydes (4-hydroxynonenal, malondialdehyde) that damage proteins and DNA. Heating extra-virgin olive oil past about 160°C also generates trans bonds; this is not a seed-oil-only phenomenon.

The mechanistic comforts are also real. Linoleic acid is an essential fatty acid. Replacing SFA with PUFA in trials consistently lowers LDL. A tablespoon of canola oil in a home-cooked stir-fry is not the population at risk; the population at risk is one drawing a quarter of its calories from soybean oil in restaurant frying and packaged snacks.

The defensible position: not poison in moderate culinary amounts, not health food. The problem is industrial dominance — the volume of omega-6 in the American food supply, the repeatedly heated fryer environments that oxidize it, and the displacement of better fats (olive, avocado, butter, fish) by cheap, ubiquitous alternatives. The lever is the ratio and the source.

Olive oil and the Mediterranean evidence

Olive oil is the most-tested cooking fat in the human literature. PREDIMED, Lyon, and a generation of cohort studies converge on the same pattern: 2–4 tablespoons of extra-virgin olive oil a day, inside a diet rich in vegetables, legumes, fish, and nuts, is associated with lower cardiovascular events, lower cognitive decline, and lower depression rates.

The "why" is more than the fatty-acid profile. Extra-virgin olive oil contains roughly 30 polyphenols, including oleocanthal, which acts as a natural ibuprofen-like cyclooxygenase inhibitor. Refined olive oil ("light" or "pure") loses most of these. The Mediterranean trial signal is specifically for the unrefined, high-polyphenol product used as the primary cooking and dressing fat.

There is no magic to the Mediterranean diet beyond the combination — generous unsaturated fat from real-food sources, plants doing most of the rest, occasional fish and dairy, very little processed food. It is the most-tested dietary pattern in human history and the cleanest answer to "what replaces saturated fat well."

Trans fats — the actual win

The campaign that mostly went wrong got one thing right, late. Trans fats from partial hydrogenation are unambiguously harmful: they raise LDL, lower HDL, raise triglycerides, promote clotting, drive systemic inflammation, and damage endothelial function. Lipid biochemist Mary Enig was warning about them in the 1970s and was ignored by an industry that needed margarine to be the answer.

The FDA finally listened. A 2003 labeling rule required trans-fat disclosure; food companies reformulated rather than print the number. In 2015 the FDA ruled that partially hydrogenated oils were no longer Generally Recognized as Safe, with full phase-out by 2018. Estimated prevented heart attacks: 72,000 to 220,000 per year in the United States.

This is what nutrition science looks like when it works: a falsifiable claim, replicated measurements, a regulatory response, a population outcome that moved in the predicted direction. The science worked here because the question was narrow, the mechanism was specific, and there was nothing the food industry wanted to defend. The lesson is not that experts are always right; it is that the method can produce a real answer when it is allowed to.

Frequently asked questions

Butter or margarine?

Butter, in normal amounts, in the context of an otherwise reasonable diet. Modern soft-spread margarines without partial hydrogenation are not dangerous, but the trans-fat-era stick margarine you may remember from the 1990s killed people. Real butter from grass-fed cows has the additional bonus of more omega-3 and conjugated linoleic acid than feedlot dairy.

Is saturated fat okay?

From real-food sources, in normal culinary amounts, in someone without a specific genetic lipid disorder: probably yes. The 2017 PURE study of 135,000 people and a 635,000-person meta-analysis on butter both failed to find the harm signal that fifty years of advice predicted. People with familial hypercholesterolemia and people whose LDL responds strongly to dietary fat are exceptions worth talking to a doctor about.

What about coconut oil?

Coconut oil is mostly medium-chain saturated fat (lauric acid). It raises LDL more than olive oil does. It also raises HDL. The trial evidence for distinct health benefits beyond "tropical cooking fat that doesn't oxidize when heated" is thin. Useful in cooking; not a superfood.

Are seed oils evil?

No. Industrial omega-6 dominance is a problem at the population level; a tablespoon of canola oil in a home-cooked meal is not. The lever is total volume, repeated high-heat frying, and what better fats are being displaced.

What about cholesterol on the egg?

Dietary cholesterol is largely uncoupled from blood cholesterol for most people. The liver produces most of your cholesterol regardless of intake. A daily egg shows no heart-disease link in healthy people in cohort data; diabetics are an exception worth tracking.

Does saturated fat in cheese and yogurt act differently from SFA in red meat?

Probably yes. Fermented dairy carries a different fatty-acid mix (odd-chain SFA appear cardioprotective), live microbes, and a matrix that changes absorption. Meta-analyses find neutral or modestly favorable cardiovascular signals for cheese and yogurt that do not appear for processed red meat.

Is air-frying healthier than deep-frying?

For oil exposure, yes — less total fat, much less oxidized oil, no reused fryer oil. Air-frying still generates acrylamide from carbohydrate browning, same as oven-roasting. The win is avoiding repeatedly heated industrial oils.

Sources

1. Willett, W. Eat, Drink, and Be Healthy. Free Press, 2017. The "what replaces" framework; Lyon and PREDIMED; the case against trans fats.
2. Spector, T. Spoon-Fed. Jonathan Cape, 2020. PURE, the butter meta-analysis, the durable saturated-fat myth.
3. Pollan, M. In Defense of Food. Penguin Press, 2008. The leaves-to-seeds shift, the omega-6 surplus, the political history of the lipid hypothesis.
4. Lustig, R. H. Metabolical. Harper Wave, 2021. The Sydney Diet Heart and Minnesota Coronary recoveries; membrane integrity as a subcellular pathology.
5. Means, C. Good Energy. Avery, 2024. Omega-6:omega-3 ratio and cell-membrane structure; the seven mitochondrial insults.
6. Hu, F. B., Manson, J. E., & Willett, W. C. (2001). "Types of Dietary Fat and Risk of Coronary Heart Disease: A Critical Review." JACN, 20(1), 5–19.
7. de Lorgeril, M., et al. (1999). "Mediterranean Diet, Traditional Risk Factors, and the Rate of Cardiovascular Complications After Myocardial Infarction." Circulation, 99(6), 779–785.
8. Estruch, R., et al. (2013/2018). "Primary Prevention of Cardiovascular Disease with a Mediterranean Diet." NEJM, 368, 1279–1290.
9. Ramsden, C. E., et al. (2013). "Use of dietary linoleic acid for secondary prevention of coronary heart disease and death: evaluation of recovered data from the Sydney Diet Heart Study." BMJ, 346, e8707.
10. Ramsden, C. E., et al. (2016). "Re-evaluation of the traditional diet-heart hypothesis: analysis of recovered data from Minnesota Coronary Experiment (1968–73)." BMJ, 353, i1246.
11. Dehghan, M., et al. (2017). "Associations of fats and carbohydrate intake with cardiovascular disease and mortality in 18 countries from five continents (PURE)." Lancet, 390(10107), 2050–2062.
12. Hibbeln, J. R., et al. (2006). "Healthy intakes of n-3 and n-6 fatty acids: estimations considering worldwide diversity." American Journal of Clinical Nutrition, 83(6 Suppl), 1483S–1493S.

Related modules

• C2: Macronutrients — the boring map that still works
• C4: Sugar — the clearest case in nutrition →
• C9: Inside your cells (mitochondria, ATP, insulin) →

Related glossary terms

• Saturated fatty acid
• Monounsaturated fatty acid
• Polyunsaturated fatty acid
• Omega-3
• Omega-6
• Trans fatty acid
• Essential fatty acid
• Lipoprotein
• LDL
• HDL