Has saturated fat been exonerated?

No, and the claim rests on a specific and identifiable analytical error. Meta-analyses showing that reducing saturated fat produces no clear benefit are real, but they combine trials in which saturated fat was replaced by quite different things. When it is replaced by polyunsaturated fat, cardiovascular events fall. When it is replaced by refined carbohydrate, they do not. Since total calories are broadly fixed, nothing can be removed from a diet without something taking its place, so asking whether saturated fat is harmful without specifying the replacement is not an answerable question. Separately, saturated fat raises LDL cholesterol in controlled feeding studies, and LDL is causally related to heart disease through drug trials, genetic studies and dose-response evidence. Any exoneration argument has to explain that away, and none does.

Why do the studies seem to contradict each other?

Mostly because they are asking different questions and being reported as though they were asking the same one. Substitution analyses ask what happens when saturated fat is swapped for a specified alternative and give consistent answers. Reduction analyses ask what happens when saturated fat goes down without specifying what replaces it, and give mixed answers because the replacements differ. Observational studies of high-fat versus low-fat populations may compare well-nourished with under-nourished groups, as in the widely publicised PURE study, where the lowest-fat groups were in low-income settings and low fat intake reflected poverty rather than dietary choice. Once the substitution framework is applied, most of the apparent contradiction in this literature disappears.

What about butter, cheese and full-fat dairy?

This is a genuine and interesting complication rather than a rhetorical concession. Cohort studies consistently find that fermented dairy — particularly yoghurt and cheese — is not associated with the cardiovascular harm that its saturated fat content would predict, and may be neutral or slightly favourable. Explanations under investigation include the physical structure of the food matrix altering fat absorption, calcium binding fatty acids in the gut so they are excreted rather than absorbed, bioactive peptides produced by fermentation, and the particular fatty acid profile of dairy fat. Butter behaves less favourably than cheese in these comparisons. None of this overturns the LDL relationship; it demonstrates that whole foods are not always predictable from a single nutrient.

Is coconut oil healthy?

No, and the argument made for it involves a straightforward category error. The favourable metabolic research concerns medium-chain triglycerides, meaning fatty acids of six to ten carbons, which are absorbed directly into the portal circulation and metabolised differently from ordinary fats. Coconut oil is predominantly lauric acid, which has twelve carbons and behaves largely like a conventional long-chain saturated fat: it is packaged into chylomicrons and it raises LDL cholesterol. A systematic review of clinical trials found that coconut oil raises LDL substantially more than unsaturated plant oils, with no demonstrated benefit to offset this. It is a pleasant cooking fat with a distinctive flavour and there is no need to avoid it entirely, but it should be treated as what it is — a saturated fat.

So what should I actually eat?

The advice is about replacement rather than restriction, which is both more precise and more useful than the low-fat message of the past. Replace butter, lard, tallow, coconut oil and fatty processed meats with olive oil, rapeseed and other seed oils, nuts, seeds, oily fish, avocado and legumes. Do not replace them with low-fat products in which the fat has been substituted by sugar or refined starch, which was the mistake of the low-fat era and which plausibly did harm. Fermented dairy such as yoghurt and cheese appears to behave better than its fat content predicts and does not require the same caution. WHO recommends keeping saturated fat below 10% of total energy, explicitly specifying what should take its place — the specification is the important part.

How does GMJ News cover saturated fat?

GMJ News covers the LDL relationship and its causal evidence base including familial hypercholesterolaemia and PCSK9 natural experiments, substitution analysis as the resolution of the apparent contradictions, the Keys and Hegsted equations, the low-fat era and why it damaged trust, the PURE study and its methodological problems, the dairy matrix effect as a genuine unexplained finding, heterogeneity between individual saturated fatty acids including stearic acid, the coconut oil and medium-chain triglyceride category error, and what WHO guidance actually specifies about replacement. Part of PHIG's contested science and cardiovascular prevention programme.

Home › Topics › Saturated Fat

Saturated Fat

GMJ News knowledge hub · last reviewed September 2026 · Georgian Medical Journal

Summary Statistics Glossary GMJ news FAQ Documents Organizations Research

The claim that saturated fat has been “exonerated” is among the most consequential misreadings of nutritional evidence in circulation, and it rests on a specific analytical error: meta-analyses showing that reducing saturated fat has no clear effect on mortality are frequently cited without noting that the outcome depends entirely on what replaces it (WHO). Replacing saturated fat with polyunsaturated fat lowers cardiovascular events in randomised trials; replacing it with refined carbohydrate does not; replacing it with monounsaturated fat or whole grains is beneficial. Saturated fat unambiguously raises LDL cholesterol, and LDL is causally related to atherosclerosis through trial, genetic and dose-response evidence — so any argument that saturated fat is harmless must explain away the entire LDL causality literature, which it cannot. The honest complications worth stating are that different saturated fatty acids behave differently, that dairy matrix effects appear not to follow the pattern predicted by fat content alone, and that coconut oil raises LDL and has no demonstrated compensating benefit.

Key messages

SETTLED: saturated fat raises LDL cholesterol, and LDL causes atherosclerosis

Both halves are established. Controlled feeding studies show a predictable, dose-dependent rise in LDL cholesterol with increased saturated fat intake. LDL causality is supported by randomised trials of multiple drug classes acting through different mechanisms, by Mendelian randomisation across dozens of genetic variants, and by consistent dose-response. Any claim that saturated fat is harmless must therefore either dispute LDL causality — which requires dismissing one of the strongest evidence bases in medicine — or argue that saturated fat somehow raises LDL without consequence, for which no mechanism has been offered.

THE CENTRAL ERROR: reduction studies answer a different question from substitution studies

This is the analytical mistake on which the entire "saturated fat exonerated" argument rests. Meta-analyses showing that reducing saturated fat produces no clear mortality benefit are real, but they pool trials in which saturated fat was replaced by different things. Replacement with polyunsaturated fat reduces cardiovascular events; replacement with refined carbohydrate does not; replacement with whole grains or unsaturated fat does. Because energy intake is broadly fixed, nothing can be reduced without something replacing it, so "is saturated fat harmful?" is not a well-formed question. "Compared with what?" is the only version that can be answered.

SETTLED: the dairy matrix behaves differently from its fat content alone

One of the genuine complications, and it is real rather than a concession. Cohort evidence consistently finds that fermented dairy — yoghurt and cheese — is not associated with the cardiovascular harm predicted from its saturated fat content, and may be neutral or modestly beneficial. Proposed explanations include the physical food matrix affecting fat absorption, calcium binding fatty acids in the gut, fermentation-derived bioactive peptides, and the specific fatty acid profile of dairy fat. This does not overturn the LDL relationship; it indicates that whole foods cannot always be predicted from single nutrients.

UNSUPPORTED: that coconut oil is beneficial because of medium-chain triglycerides

A specific and widespread error worth correcting directly. The metabolic research on medium-chain triglycerides concerns C6 to C10 fatty acids, which are absorbed via the portal circulation and handled differently from long-chain fats. Coconut oil is predominantly lauric acid (C12), which is largely packaged into chylomicrons like a conventional long-chain saturated fat and raises LDL cholesterol accordingly. Systematic review of clinical trials confirms that coconut oil raises LDL substantially more than unsaturated plant oils, with no compensating outcome benefit demonstrated. The marketing rests on applying research about one class of molecule to a product mostly composed of another.

GENUINELY OPEN: whether all saturated fatty acids behave identically

The category "saturated fat" contains chemically distinct molecules with different metabolic handling. Lauric, myristic and palmitic acids raise LDL; stearic acid appears broadly neutral for LDL, though it may affect other pathways. Odd-chain saturated fatty acids from dairy show inverse associations with diabetes in cohort studies. This heterogeneity is scientifically real and is used legitimately in the literature and illegitimately in popular argument, where the existence of one neutral fatty acid is presented as exonerating the whole class. Guidelines address the aggregate because that is what people eat.

WHAT FOLLOWS: the advice is about replacement, not restriction

The practical conclusion is more precise and more actionable than "eat less fat", which was the flawed advice of the low-fat era and which in practice led to substitution by refined carbohydrate — plausibly making things worse and contributing directly to the current distrust. The evidence supports replacing butter, tallow, coconut oil and fatty processed meat with olive oil, seed oils, nuts, oily fish and legumes. It does not support replacing them with low-fat products loaded with sugar. Acknowledging that the earlier guidance was wrong in this specific way is necessary for the current advice to be believed.

Key statistics

Dose-dependent

saturated fat raises LDL cholesterol predictably in controlled feeding studies

Keys/Hegsted equations

Replacement matters

SFA replaced by PUFA lowers events; replaced by refined carbohydrate does not

Cochrane/Circulation

Dairy matrix

fermented dairy is not associated with predicted harm — a genuine unexplained finding

Cohort meta-analyses

Coconut oil

raises LDL substantially more than unsaturated plant oils, with no outcome benefit shown

Circulation 2020

Stearic acid

appears LDL-neutral unlike lauric, myristic and palmitic — real heterogeneity within the class

AJCN

<10% energy

WHO recommendation for saturated fat intake, with replacement by unsaturated fat

WHO 2023

Saturated fat — where the disagreement actually lies

Source: Bars show strength of supporting evidence. The substitution framework resolves most of the apparent contradiction in this literature.

Glossary of key terms

The Keys and Hegsted equations
Nutritional biochemistryPredictive equations derived from controlled feeding studies in the 1950s and 1960s that quantify the change in serum cholesterol produced by changing dietary saturated fat, polyunsaturated fat and cholesterol. They remain remarkably accurate when tested against modern metabolic ward data, and they are the quantitative foundation of dietary fat guidance. Their persistence is significant in a field where much has been revised: the relationship between dietary saturated fat and blood LDL is one of the most reproducible findings in nutrition, established under strictly controlled conditions long before the epidemiological literature that is now disputed. The controversy concerns what LDL change means for outcomes, not whether the LDL change occurs.

The low-fat era and why trust was lost
History/PolicyFrom the late 1970s, dietary guidance in several countries emphasised reducing total fat, which the food industry met with low-fat products in which fat was replaced largely by refined carbohydrate and sugar. The population consequence was an increase in refined carbohydrate consumption during a period of rising obesity and type 2 diabetes. Whether the guidance caused this is contested and probably overstated, but the substitution was certainly not what the underlying evidence supported, which even then pointed to replacing saturated with unsaturated fat rather than with anything at all. This history matters because it is the strongest legitimate argument in the sceptics' case, and because refusing to acknowledge it makes current guidance less credible rather than more.

The PURE study controversy
EvidenceThe Prospective Urban Rural Epidemiology study, published in the Lancet in 2017, reported that higher saturated fat intake was associated with lower mortality and no increase in cardiovascular disease across 18 countries, and was widely reported as overturning dietary fat guidance. The methodological objections are substantial: the lowest fat intake groups were in low-income countries where very low fat intake reflects overall dietary inadequacy and poverty rather than a chosen low-fat diet, so the comparison is between nutritional deprivation and adequacy rather than between dietary patterns; food frequency questionnaires were not validated identically across settings; and the substitution question was not addressed. It is a genuine dataset that does not support the conclusion drawn from it in popular coverage.

Individual saturated fatty acids
BiochemistryLauric acid (C12) raises total cholesterol most but also raises HDL, complicating interpretation; myristic (C14) is the most potent LDL-raiser per gram; palmitic (C16) is the most abundant in the diet and a substantial LDL-raiser; stearic (C18) is largely LDL-neutral, apparently because it is rapidly desaturated to oleic acid in vivo. Chocolate and beef contain substantial stearic acid, which is one reason their lipid effects are less adverse than total saturated fat content suggests. Odd-chain fatty acids C15 and C17, largely from dairy, are inversely associated with type 2 diabetes in cohorts, though whether as cause or as intake biomarker is unresolved. Guidelines target the aggregate because dietary sources always deliver mixtures.

Familial hypercholesterolaemia as a natural experiment
GeneticsThe strongest single argument against dismissing LDL, and one that requires no dietary epidemiology at all. People with heterozygous familial hypercholesterolaemia have a lifelong LDL elevation from a single gene defect, with normal diet, weight and lifestyle, and their untreated risk of premature coronary disease is increased roughly twenty-fold. Conversely, people with loss-of-function PCSK9 variants have lifelong low LDL and markedly reduced coronary risk. Neither group differs systematically in saturated fat intake. These natural experiments isolate LDL from every dietary and lifestyle confounder, and they are the reason LDL causality is regarded as settled rather than merely well-supported.

What WHO and national guidelines actually recommend
PolicyWHO's 2023 guideline recommends reducing saturated fatty acid intake to 10% of total energy and trans fatty acids to 1%, with the specific instruction to replace them with polyunsaturated fatty acids, monounsaturated fatty acids from plant sources, or carbohydrate from foods containing naturally occurring dietary fibre such as whole grains, vegetables, fruits and pulses. The explicit specification of the replacement is the substantive change from earlier guidance and directly addresses the low-fat era error. National guidelines in Europe, North America and Australasia are aligned. The recommendation is not "avoid fat" and has not been for many years, though it is still widely reported and remembered as though it were.