Cholesterol, LDL and seed oils · both sides of the dispute, with the numbers
Plate 47
Lipid record
What a cohort measured
48
Sources on the page
10
Questions in people
On this page. Counts, not a grade and not a ranking.
No goal on this page meets the rule for a stronger record. The rule is printed above the goals, and the count it read is printed under each one.
Read this first
Educational use only. This page is not medical advice and not a reason to start, stop or change a statin or any prescription. Doses and diets here are what a study ran. Familial hypercholesterolemia, established heart disease, diabetes and pregnancy are questions for your own doctor. Aaron Cuha is a coach, not a doctor.
Educational use only. This database summarizes published research and is not medical advice. Where a dose appears it is a record of what a trial or clinic ran, or what people describe doing, with its source attached. Nothing here is a recommendation for personal use.
Cholesterol is a waxy fat every cell needs: it builds cell membranes and is the raw material for vitamin D, bile and the steroid hormones. It travels in the blood inside particles. LDL, low-density lipoprotein, carries it out to the tissues and is the number most guidelines target. HDL, high-density lipoprotein, carries it back and is often called the good cholesterol. Total cholesterol is the sum.
Seed oils are oils pressed from seeds such as soybean, corn, safflower, sunflower and cottonseed. Their main fat is linoleic acid, an omega-6 polyunsaturated fat. For decades the advice was to swap saturated animal fat for these oils to lower cholesterol and prevent heart disease. The 2024 book Dark Calories argues the opposite: that seed oils high in linoleic acid drive inflammation and chronic disease. This page tests both the old advice and the challenge against the trials and cohorts.
The dispute is really several questions with different answers: whether high LDL harms arteries, whether low cholesterol is dangerous in older age, whether statins help healthy older people live longer, whether eggs and saturated fat matter, and whether seed oils help or hurt. Each gets its own card below, graded by the same printed rule as every other page on this site.
Each side has a strong story and real evidence. Neither is the whole picture. The three cards below are the strongest points against the advice of the last half century; the amber box under them is the strongest case for it.
The advice was shaped by industry money
Internal Sugar Research Foundation documents show it funded a 1965 New England Journal of Medicine review that singled out fat and cholesterol as the cause of heart disease and downplayed sugar. It set the review's objective and saw drafts, and the funding was never disclosed.
The buried seed oil trials
The Sydney Diet Heart Study swapped animal fat for safflower oil in 458 men after a heart event, from 1966 to 1973. Its full data were recovered and published only in 2013: 17.6 percent of the seed oil group died, against 11.8 percent of controls.
Seed oil now fills our body fat
Linoleic acid stored in Americans' body fat rose 136 percent over the last half century, tracking the shift to soybean and other seed oils. No earlier generation has carried this much of it.
The case for the old advice
People born with PCSK9 gene variants have lower LDL from birth. In one cohort a 28 percent lower LDL went with 88 percent less coronary heart disease. Nobody chooses their genes, so illness and lifestyle cannot explain it.
Across 28 randomised statin trials and 186,854 people, each 1.0 mmol/L lower LDL cut major vascular events by 21 percent, at every age studied, though the coronary benefit shrank with age.
Even in the Japanese cohort where low LDL went with more deaths in people over 65, it went with less coronary artery disease (HR 0.56). The extra deaths came from somewhere other than the arteries.
Every number here is what a study measured or a trial ran, named beside it. None is a target for a reader. Lipid results are reported in mg/dL in the US and mmol/L almost everywhere else, so the same study can look different depending on the unit.
Each row is a record from a paper. It is not a plan for you.
| Goal | Who | What the trial ran | Source |
|---|---|---|---|
| Statins after 70 | 9971 Australians aged 70 and over without heart disease, diabetes or dementia | Atorvastatin 40 mg once a day or identical placebo, median 5.9 years. | PMID 42670961 |
| Seed oil swap, Sydney | 458 men aged 30 to 59 after a heart event | Safflower oil and safflower margarine in place of animal fats and common margarines, from 1966 to 1973. | PMID 23386268 |
| Seed oil swap, Minnesota | 4393 men and 4664 women in state hospitals and a nursing home | A diet with 9 percent of calories from saturated fat and 15 percent from polyunsaturated fat, against 18 and 5 percent. | PMID 2643423 |
| Cutting seed oils, migraine | 182 adults with migraine | 16 weeks with two thirds of food provided: omega-3 raised to 1.5 g a day and linoleic acid cut to 1.8 percent of calories or less, against omega-3 alone or control. | PMID 34526307 |
| Cutting seed oils, head injury | 122 people with persistent post-traumatic headache | 12 weeks of a high omega-3, low omega-6 diet against control. | PMID 40600308 |
| High LDL on keto | 80 lean keto dieters with LDL of 190 mg/dL or more | No intervention: a mean 4.7 years on their own ketogenic diets, scanned once and compared with matched controls. | PMID 39372369 |
Stronger records first. Nulls and marketing-ahead goals stay on the same page. Every card is people only; animal work is in its own section below. Nothing here is ranked against the others, and no clinic or device is recommended.
How the tier is decided, so you can check it against the counts printed under each one. Stronger: a clinical guideline or a medical society's indication list recommends it in a named setting, or a pooled analysis of randomised trials favours the change named on the card and the source rated the certainty moderate. Thin: the pooled result was null or went against the change named on the card for the goal as named, the source rated it very low, or nothing was pooled at all. Mixed: everything else, including a result that moved one measure and not another.
Mixed, local, or early
5 observational cohorts pooled, 2,317 people; certainty not graded by the source; result favours the change named on the card
Who was studied
Older adults followed over years. A pooled analysis of 5 cohorts and 2,317 people with a median age of 85; a 2016 review of 19 cohort studies and 68,094 people over 60; and single cohorts of 1,391,616 Koreans, 283,289 Japanese, 211,290 Chinese, 12,334 healthy older adults in ASPREE, three Danish cohorts, and 999 Italians. All observational. Some authors declare drug company consultancies.
Higher LDL went with lower death in cohorts covering 92 percent of the people in the 2016 review, and with lower death per 1 mmol/L higher LDL in the very old (HR 0.87). In Japan, LDL under 80 mg/dL went with more deaths (HR 1.57) and more disability (HR 1.31). In China, LDL under 100 went with more deaths, and the authors proposed 100 to 159 as optimal after 65. In ASPREE the lowest death was at 3.3 to 3.4 mmol/L, and in Italy a total cholesterol under 200 had the highest death rate.
Setting in the evidence
Community-dwelling older adults, mostly not on cholesterol drugs.
What this does not show
That raising LDL makes anyone live longer. Illness, frailty and poor nutrition lower cholesterol, so low cholesterol may be a sign of disease rather than its cause. In the same cohorts, low LDL still went with fewer heart attacks, and in ASPREE higher LDL meant more cardiovascular deaths.
Mixed, local, or early
28 randomised trials pooled, 186,854 people; certainty not graded by the source; result mixed
Who was studied
STAREE, a randomised double-blind trial of 9971 Australians aged 70 and over without heart disease, diabetes or dementia, government funded. And an analysis of 28 randomised statin trials, 186,854 people, of whom 14,483 were over 75.
In STAREE, major cardiovascular events fell from 412 to 297 people (HR 0.70). Death, dementia or lasting disability: 637 against 676 (HR 0.94, not significant). Serious adverse events were 2.7 percent in each group, with more muscle, liver and diabetes events on atorvastatin. Across 28 trials, each 1.0 mmol/L lower LDL cut major vascular events 21 percent, with a smaller coronary benefit at older ages.
Setting in the evidence
Atorvastatin 40 mg a day in STAREE; various statins in the pooled trials.
What this does not show
Longer disability-free life from a statin in healthy people over 70, over six years. It also does not apply to people who already have heart disease.
Mixed, local, or early
17 randomised trials pooled, 66,337 people; certainty mixed; result mixed
Who was studied
A 2026 risk-stratified review of 17 randomised trials and 66,337 people, low to moderate certainty. And the Minnesota Coronary Experiment, 4393 men and 4664 women, first reported in 1989 and reanalysed from recovered data in 2016.
Cutting saturated fat gave all-cause death RR 0.96 and non-fatal heart attack 0.86, neither significant. For low-risk people the absolute effects were below the authors' importance thresholds; for high-risk people above them, more so when the replacement was polyunsaturated fat. In Minnesota, halving saturated fat from 18 to 9 percent of calories lowered cholesterol but did not change heart events or deaths.
Setting in the evidence
Diet advice and supplied diets, mostly decades-old trials.
What this does not show
That saturated fat is harmless for everyone. The high-risk subgroup may benefit, and the review's certainty is low to moderate.
Mixed, local, or early
31 randomised trials pooled; certainty not graded by the source; result mixed
Who was studied
Randomised: 182 adults with migraine for 16 weeks and 122 people with post-traumatic headache for 12 weeks, NIH funded; 31 trials of lowering the omega-6 to omega-3 ratio; and 30 trials of adding linoleic acid. Observational and genetic: 2802 and 12,401 people.
In migraine, adding omega-3 and cutting linoleic acid cut headache days by 4.0 a month against control, twice the 2.0 from omega-3 alone. In post-traumatic headache, the same diet cut headache days by 2.1 a month and pain intensity. Across 31 trials, a lower omega-6 ratio reduced TNF-alpha and IL-6 but not CRP. Across 30 trials, adding linoleic acid did not raise inflammation markers overall, though larger increases may raise CRP. The genetic study found no simple pro-inflammatory rule for omega-6.
Setting in the evidence
Supplied or guided diets low in linoleic acid and high in fish omega-3.
What this does not show
That every kind of inflammation responds, or that cutting seed oils alone, without raising omega-3, does the same. The headache impact scores did not change significantly in either headache trial.
Mixed, local, or early
cohorts, genetic studies and trials pooled, count not printed in the abstract; certainty not graded by the source; result favours the change named on the card
Who was studied
A genetic study inside a 15-year cohort of 3363 Black and 9524 white Americans, and the European Atherosclerosis Society consensus pooling over 200 cohorts, genetic studies and trials with more than 2 million people. Critics answer in a 2018 review.
PCSK9 variants that lowered LDL 28 percent went with 88 percent less coronary heart disease (HR 0.11), and a milder variant lowering LDL 15 percent with 47 percent less (HR 0.50). The consensus concluded LDL causes atherosclerotic disease, rising with dose and years of exposure. The 2018 critics' review argues the consensus rests on misleading statistics and excluded unsuccessful trials.
Setting in the evidence
Natural genetic variation, and drug trials.
What this does not show
What happens when LDL is driven very low late in life, or that the same applies in older age, where the cohorts above point the other way on overall death.
Thin, null, or marketing ahead
nothing pooled; certainty not graded by the source; no pooled result
Who was studied
3392 Dutch adults aged 70 to 78 followed a median 6.7 years, observational. And 473 patients with heart disease kept on evolocumab for a median 5.1 more years, at a median LDL of 35 mg/dL.
A one standard deviation higher LDL to HDL ratio went with 19 percent lower dementia risk, independent of statins and APOE genotype. In the evolocumab group, executive function did not change over long follow-up.
Setting in the evidence
Community-dwelling older adults; patients with heart disease on drug therapy.
What this does not show
That raising cholesterol prevents dementia, or that very low LDL is safe for the brain in people already at high dementia risk. Neither question has been tested in a trial.
Thin, null, or marketing ahead
33 cohort risk estimates pooled, 1,720,108 people; certainty not graded by the source; result null
Who was studied
Observational: a 2020 meta-analysis of 33 risk estimates and 1,720,108 people, three large US cohorts, and a 2019 pooled analysis of 29,615 US adults from 6 cohorts. Plus the American Heart Association's 2020 science advisory.
One more egg a day was not linked to cardiovascular disease (RR 0.98), and in the US cohorts an egg a day or more was not either (HR 0.93). The AHA advisory says observational studies generally show no significant link, and guidelines dropped the daily cholesterol cap. The 2019 analysis disagreed: each additional 300 mg of dietary cholesterol a day went with more heart disease (HR 1.17).
Setting in the evidence
Usual diets, reported by questionnaire.
What this does not show
That eggs lower risk, or what happens at very high intakes in people whose LDL rises sharply with dietary cholesterol.
Thin, null, or marketing ahead
19 randomised trials pooled, 6,461 people; certainty low; result null
Who was studied
Randomised: 19 trials and 6461 people in Cochrane; the Sydney Diet Heart Study, 458 men; the Minnesota Coronary Experiment, 9423 people; and a 2017 analysis splitting trials by how well they were controlled. Not randomised: the Finnish Mental Hospital Study. Observational: blood and fat tissue linoleic acid in 68,659 people across 30 cohorts and in 811,069 people with diet data, one pool with Unilever support.
Cochrane found more omega-6 made no difference to deaths (RR 1.00). In Sydney, deaths were 17.6 against 11.8 percent (HR 1.62). In Minnesota, cholesterol fell 13.8 percent with no fewer deaths, and each 30 mg/dL fall went with 22 percent more. Adequately controlled trials only: no effect on heart deaths (1.13). The non-randomised Finnish study found about half as much coronary disease on the high polyunsaturated diet, and the observational pools found higher linoleic acid went with less heart disease and lower death (RR 0.87). A Danish fat-biopsy study found no link either way.
Setting in the evidence
Safflower, corn and soybean oils and margarines in place of animal fats.
What this does not show
A clean verdict. The randomised record shows no lives saved; the observational record shows people with more linoleic acid in their blood doing better. Those answer different questions, and the trials were mostly decades old and in men.
Thin, null, or marketing ahead
nothing pooled; certainty not graded by the source; no pooled result
Who was studied
52,268 men and 64,240 women in two Copenhagen cohorts. Observational.
Death followed a U shape. The lowest was at about 73 mg/dL for men and 93 for women. Men at 116 or more had about twice the death rate (HR 2.06); women at 135 or more had HR 1.68.
Setting in the evidence
General population, natural HDL levels.
What this does not show
Why. The authors say the finding needs confirmation in other studies.
Thin, null, or marketing ahead
nothing pooled; certainty not graded by the source; no pooled result
Who was studied
80 lean keto dieters with LDL averaging 272 mg/dL, scanned once against matched controls, funded by the Citizen Science Foundation; a one-year follow-up of 100 such dieters, now retracted; a web survey of 548 low-carbohydrate dieters; and one case report.
In the 2024 study, plaque and calcium scores did not differ from controls with lower LDL, and LDL did not track plaque. The 2025 follow-up scanned 100 people with median LDL of 237 mg/dL a year apart. Read from its own numbers: noncalcified plaque volume rose by a median of 18.9 cubic millimetres, and even the lower quartile rose 9.3, so at least three quarters of participants added plaque in a year. Within this group, ApoB and years of high LDL did not predict how much; baseline plaque did. Because everyone in it had very high LDL, it could not compare high LDL with normal LDL. The paper was retracted in 2026 at the request of its authors and the editors, who cited methodology errors. The survey found leaner people had bigger LDL rises, which fell when some carbohydrate returned. One man held an LDL of 574 for nearly seven years with no plaque.
Setting in the evidence
Self-chosen ketogenic diets.
What this does not show
That high LDL on keto is safe over decades, or that it is the cause of the plaque growth: the follow-up had no comparison group with normal LDL. Long-term outcomes have not been studied.
One person's own account, shown on its own. Not blinded, no comparison group, and never added into any grade on this page.
Aaron Cuha, owner of this archive
Eliminated seed oils from his diet and reports that his inflammation is gone. One person, with no comparison group and no blinding, recorded here as his own account. The randomised trial closest to what he did is the migraine trial on the inflammation card, where cutting linoleic acid on top of omega-3 doubled the drop in headache days.
This page is about people. Animal cholesterol and seed oil studies were not used, so nothing here rests on a mouse. The section stays on the page so that is visible.
Two kinds of evidence are kept apart on every card. Randomised trials show what happens when one thing is changed. Cohorts show who does well in real life, but cannot fully separate cause from sign. Both are kept, and neither is thrown out for being mainstream or for not being mainstream.
The older-adult evidence is the strongest support for the view that low cholesterol is dangerous, and it is all observational. Illness lowers cholesterol, which is the main reason the cohorts may overstate the risk; the same cohorts still show low LDL meaning fewer heart attacks.
The seed oil evidence is the reverse. The randomised trials, including two recovered decades late, never showed the swap saving lives, and Sydney showed harm. The observational blood studies point the other way, and one pool had support from a margarine maker.
Money runs in every direction on this topic. Drug companies fund many cholesterol researchers; a margarine maker supported one seed oil analysis; the sugar industry secretly funded the review that blamed fat; a keto advocacy foundation funded the lean mass hyper-responder scans. Each source below names its own.
The blocks above are the studies. The block below is a judgment about what to do with them. It has to say where it goes past the measurements.
Aaron Cuha’s position
This is opinion, not a finding. Everything else on this page is the evidence.
Cholesterol is essential, and in older adults low cholesterol is a warning sign that most large studies tie to dying sooner. The randomised trials never showed that swapping animal fat for seed oils saves lives, and cutting seed oils has real trial support for pain and inflammation. What the record does not show is that high LDL is harmless for arteries in general: lifelong genetic evidence and statin trials point the other way, and the keto study most cited for that claim was retracted.
The reasoning
Where this goes beyond the evidence
This position goes beyond the evidence in two places. It treats the older-adult cohorts as a warning sign rather than proof that higher cholesterol protects, which is a judgment about reverse causation the cohorts cannot settle. And it treats the seed oil trials as stronger than the blood-level cohorts because they were randomised, though those trials were decades old, small by modern standards and mostly in men. A modern randomised trial of cutting seed oils, measuring death and heart disease, would settle what neither side can now.
Who this is not for
Anyone with familial hypercholesterolemia, a prior heart attack or stroke, or diabetes, whose treatment belongs with their own doctor; anyone considering stopping a prescribed statin because of this page; and anyone who wants this page to replace a clinician.
Aaron Cuha, who built this archive and reads the studies behind the pages. He is a coach, not a doctor, and no medical credential is claimed for him anywhere on this site. More about who writes this
Anyone with familial hypercholesterolemia
A genetic condition of very high LDL from birth. The genetic evidence on this page bears on it directly, and the older-adult cohorts studied general populations, not this condition.
Anyone after a heart attack or stroke
Those trials are a different question. The STAREE result is for people over 70 without heart disease.
Anyone stopping a statin on their own
STAREE found fewer cardiovascular events on atorvastatin (HR 0.70). Changing a prescription belongs with the doctor who wrote it.
Anyone wanting a target number
The ranges here are where death was lowest in observational cohorts. They are not treatment targets for any one person.
In 19 blinded trials and 123,940 people, muscle pain or weakness was 27.1 percent on a statin and 26.6 percent on placebo. In the first year only about one in 15 complaints was caused by the statin; high-intensity doses caused more.
Low or moderate intensity statins raised new diabetes diagnoses by 10 percent, and high intensity by 36 percent, in blinded trials.
Driving LDL below 55 mg/dL in 8 trials and 122,802 patients did not raise bleeding strokes (OR 1.05). Genetically lower LDL went with fewer bacterial infections and less sepsis but more viral infections and HIV/AIDS (RR 1.64).
Low cholesterol correlates with cancer, depression, bleeding strokes and infection, and in older people often signals illness. A 2026 review advises individual assessment for the very elderly and frail rather than chasing a number.
In observational studies, the highest fried food eaters had 28 percent more major cardiovascular events and 37 percent more heart failure, though not more deaths.
This site teaches a blood panel first. A lipid panel, with LDL, HDL, triglycerides and ideally ApoB, is the starting point; the marker pages explain each number. Nothing here replaces reading your own results with your own doctor.
Disclosed both ways: the owner of this archive eliminated seed oils and reports his inflammation is gone. That is recorded as a first-hand report in its own section, not counted in any grade. His full protocol is at /my-protocols.
Nothing on this page is for sale. No seller wrote it, and no drug, diet brand or oil is recommended or ranked.
Four approaches people take to cholesterol and fats. None is ranked, and none is a prescription.
A daily drug that lowers LDL by blocking the liver's cholesterol production.
Evidence match. STAREE in healthy people over 70, and 28 randomised trials pooled.
Mismatch. In STAREE it cut cardiovascular events but did not extend disability-free life; muscle and diabetes effects are real at higher doses.
A diet that in lean people can raise LDL sharply, the lean mass hyper-responder pattern.
Evidence match. The 2024 cross-sectional scan of 80 people found no extra plaque.
Mismatch. The 2025 follow-up was retracted, and no long-term outcome study exists.
Removing soybean, corn, safflower, sunflower and similar oils, as Dark Calories argues.
Evidence match. The Sydney and Minnesota trials, and the migraine and head injury trials of cutting linoleic acid.
Mismatch. Observational blood studies tie higher linoleic acid to less heart disease.
Olive oil, fish, vegetables, nuts and legumes, a whole-diet approach.
Evidence match. The Lyon Diet Heart Study after a first heart attack, 14 events against 44.
Mismatch. It tests a whole diet, so no single fat can be credited.
Floor habits and the neighbouring pages. None of these is a cart.
Marker
What the number on your panel means.
Marker
A count of the particles, not the cholesterol in them.
Floor
A marker, or a panel you still need. Not a diagnosis.
Floor
Eighteen patterns, and what held up at equal calories.
Disclosure
Where cutting seed oils sits in a personal plan, labelled as one.
Order
Measure first. Floor habits before gadgets.
Map
Every pillar, this one among them, held to the same rule.
Knowledge stays free. No seller wrote this page. Every source below opens the record that was checked before the sentence was written, with what it found and who funded it.
Historical analysis of internal Sugar Research Foundation documents. The foundation funded a 1965 New England Journal of Medicine review that singled out fat and cholesterol and downplayed sugar; it set the review's objective, supplied articles and saw drafts, and its funding was not disclosed. NIH funded; no disclosures.
Randomised trial run from 1966 to 1973, its full data recovered decades later. 458 men aged 30 to 59 after a heart event; 221 replaced animal fats with safflower oil, 237 did not. Deaths: 17.6 against 11.8 percent (HR 1.62); cardiovascular deaths HR 1.70; coronary deaths HR 1.74. Supported by the Life Insurance Medical Research Fund of Australia and New Zealand and the NIH alcohol institute; no financial relationships declared.
Systematic review of US studies measuring linoleic acid in body fat. It rose 136 percent over the last half century, tracking the rise in dietary linoleic acid as fat sources shifted to seed oils such as soybean oil. NIH funded; no conflicts.
Genetic study within a 15-year cohort. In 3363 Black participants, PCSK9 variants carried by 2.6 percent lowered LDL 28 percent and coronary heart disease 88 percent (HR 0.11). In 9524 white participants, a variant in 3.2 percent lowered LDL 15 percent and coronary disease 47 percent (HR 0.50). NIH funded.
28 randomised trials, 186,854 people, 14,483 of them over 75. Each 1.0 mmol/L lower LDL cut major vascular events by 21 percent (RR 0.79); the benefit for coronary events shrank with age. Funded by the British Heart Foundation, the Medical Research Council and the Wellcome Trust.
Observational, 283,289 Japanese residents aged 65 to 90, mean follow-up 6.3 years. LDL under 80 mg/dL went with more functional disability (HR 1.31) and more deaths (HR 1.57), and with less coronary artery disease (HR 0.56). The authors declare no conflicts.
Randomised double-blind trial, 9971 Australians aged 70 and over without heart disease, diabetes or dementia, atorvastatin 40 mg or placebo for a median 5.9 years. Major cardiovascular events: 297 against 412 people (HR 0.70). Death, dementia or lasting disability: 637 against 676 (HR 0.94, not significant). Serious adverse events 2.7 percent in each group. Funded by the National Health and Medical Research Council and the National Heart Foundation of Australia.
The original report of the same trial, 4393 men and 4664 women. Saturated fat 9 against 18 percent of calories; cholesterol fell from 207 to 175 mg/dL against 203. No differences in heart events, heart deaths or total deaths. Public Health Service funded.
Randomised trial, 182 adults with migraine, 88 percent women, 16 weeks, two thirds of food provided. Against control, raising omega-3 and cutting linoleic acid to 1.8 percent of calories or less cut headache days by 4.0 a month; raising omega-3 alone, by 2.0. The headache impact score did not change significantly. NIH funded; the NIH has claimed intellectual property on stable analogs of oxidized lipid mediators naming two coauthors as inventors.
Randomised trial, 122 people with persistent post-traumatic headache, 12 weeks. The high omega-3, low omega-6 diet cut headache days by 2.1 a month and pain intensity by 0.9 points against control; the headache impact score did not change significantly. NIH intramural funding.
Cross-sectional, 80 lean keto dieters with diet-induced LDL of 190 mg/dL or more, mean 272, for a mean 4.7 years, matched to controls with lower LDL. Coronary calcium and plaque scores did not differ, and LDL did not correlate with plaque. Funded by the Citizen Science Foundation; the authors report no relevant relationships.
Individual data from 5 observational cohorts, 2,317 people with a median age of 85. Each 1 mmol/L higher LDL went with lower 5-year death (HR 0.87), most clearly in the least fit (HR 0.85) and not in the fittest (HR 0.98). Funded by UK research councils, the British Heart Foundation and the Department of Health; nothing to disclose.
Systematic review of 19 cohort studies, 30 cohorts and 68,094 people aged 60 and over. Higher LDL went with lower all-cause death in 16 cohorts, 14 of them significantly, covering 92 percent of the people; the rest found no association. No funding or conflict statement in the PubMed record.
Observational, 211,290 adults aged 65 and over, median follow-up 3.08 years, 5333 deaths. Against LDL of 100 to 129 mg/dL, LDL under 100 went with higher all-cause death, and cardiovascular death was higher both under 70 (HR 1.338) and from 160 up to 190 (HR 1.437). The authors propose 100 to 159 mg/dL as the optimal range at this age. Funded by Chinese national and Guangdong provincial science grants.
Observational, 1,391,616 Koreans aged 65 and over without heart disease, 233,963 deaths over a mean 8.50 years. Lower LDL went with fewer heart attacks and strokes; all-cause death followed an inverted J shape, but in statin users lower LDL went with fewer deaths. PubMed tags it as non-government funded; no conflict statement in the record.
Observational analysis of 12,334 healthy adults averaging 75.2, not on cholesterol drugs, 1250 deaths over a median 7 years. U-shaped: lowest all-cause death at LDL of 3.3 to 3.4 mmol/L. Each 1 mmol/L higher LDL went with 9 percent fewer deaths overall and 16 percent fewer cancer deaths, but 19 percent more cardiovascular deaths. NIH funded; authors declare research support or honoraria from Amgen, Boehringer-Ingelheim, Merck, Pfizer and Sanofi, and one is the lead investigator of STAREE.
Observational, Danish adults not on cholesterol drugs in cohorts of 9486, 56,796 and 48,105. The LDL with the lowest death rate fell from 4.0 mmol/L (155 mg/dL) in 1991 to 1994 to 3.5 mmol/L (135 mg/dL) in 2010 to 2018. Declared interests: consultancies or talks for AstraZeneca, Sanofi, Amgen, Novartis, Novo Nordisk, Abbott, Kowa and others; one author is employed by Novo Nordisk.
Observational, 999 Italians aged 65 and over, 6 years. Death risk rose sharply below a total cholesterol of about 200 mg/dL. Against under 200, 200 to 239 had an HR of 0.72; the pattern was clearer in women and held after excluding statin users. The authors suggest low cholesterol may signal malnutrition or inflammation. No conflicts declared.
Observational analysis of 3392 Dutch adults aged 70 to 78, median 6.7 years, 231 dementia diagnoses. A one standard deviation higher LDL to HDL ratio went with 19 percent lower dementia risk, independent of statins and APOE genotype. PubMed tags it as non-government funded; no conflicts declared.
473 patients with heart disease followed a median 5.1 more years on evolocumab, median LDL 35 mg/dL. Executive function did not change. Mostly white men with a median age of 62. No funding or conflict statement in the PubMed record.
Registry record for STAREE, sponsored by Monash University, a university, not a drug company.
Observational. In three US cohorts, at least one egg a day was not linked to heart disease (HR 0.93). The updated meta-analysis of 33 risk estimates and 1,720,108 people found one more egg a day was not linked to cardiovascular disease (RR 0.98). Funded by the NIH and the Canadian Institutes of Health Research.
Science advisory. Observational studies generally do not show a significant link between dietary cholesterol and heart disease, guidelines dropped a specific dietary cholesterol target, and most intervention studies link intakes above current averages to higher LDL. No funding or conflict statement in the PubMed record.
Observational, 29,615 US adults from 6 cohorts, median 17.5 years. Each additional 300 mg of dietary cholesterol a day went with more heart disease (HR 1.17) and more deaths (HR 1.18); each additional half egg a day with HR 1.06. NIH funded.
17 randomised trials, 66,337 people, low to moderate certainty. Cutting saturated fat: all-cause death RR 0.96, cardiovascular death 0.93, non-fatal heart attack 0.86, none significant. For people at low heart risk the absolute effects were below the authors' importance thresholds; for high risk, above them, more so when replaced with polyunsaturated fat. No funding or conflict statement in the PubMed record.
Randomised double-blind trial in a nursing home and six state mental hospitals, 9423 people aged 20 to 97, data recovered and analysed in 2016. Corn oil in place of saturated fat lowered cholesterol 13.8 percent against 1.0 percent, with no fewer deaths; each 30 mg/dL fall went with 22 percent more deaths (HR 1.22). No benefit at autopsy. A meta-analysis of 5 such trials, 10,808 people, found no benefit on coronary death (1.13) or all deaths (1.07). Funded by the NIH; no financial relationships declared.
19 randomised trials, 6461 people, followed one to eight years. More omega-6 made little or no difference to all-cause death (RR 1.00) or cardiovascular events (0.97), low quality; heart attacks RR 0.88, not significant. The World Health Organization funded the review update.
Diet-heart trials split by how well they were controlled. Adequately controlled trials only: no effect on major heart events (RR 1.06), heart deaths (1.13) or total deaths (1.07). Adding the inadequately controlled trials produced a fall in total heart events (RR 0.80). No funding or conflict statement in the PubMed record.
Not randomised: two Helsinki mental hospitals, 1959 to 1971, one on a cholesterol-lowering diet high in polyunsaturated fat, then the diets reversed after six years. Coronary heart disease was about half as common during the high polyunsaturated years. Public Health Service funded.
Observational, 30 cohort studies in 13 countries, 68,659 people, 15,198 cardiovascular events. Higher linoleic acid in blood and fat tissue went with lower total cardiovascular disease and cardiovascular death. Two authors report research support from Unilever, a margarine and spreads maker, for this work; NIH and UK funders.
Observational, 38 studies and 44 cohorts, 811,069 people with diet data. Highest against lowest linoleic acid intake: total death RR 0.87, cardiovascular death 0.87, cancer death 0.89. NIH funded.
Observational, from 57,053 Danes, fat biopsies from 3167 random participants and 2819 heart attack cases over 16 years. Linoleic acid in body fat was not linked to heart attacks (HR 1.03, top against bottom fifth). No funding or conflict statement in the PubMed record.
31 randomised trials. A lower omega-6 to omega-3 ratio reduced TNF-alpha (SMD -0.270) and IL-6 (-0.153), and did not change CRP. No funding or conflict statement in the PubMed record.
30 randomised trials, 1377 people. Adding linoleic acid did not significantly change TNF, IL-6 or CRP overall; subgroup analysis suggested larger increases may raise CRP. No funding or conflict statement in the PubMed record.
Cohort and genetic analyses in 2802 and 12,401 people. A higher omega-6 to omega-3 ratio went with higher inflammation markers in the cohorts, but the genetic analysis found no simple rule that omega-6 inflames and omega-3 calms. Funded by the UK Medical Research Council and the Wellcome Trust.
Observational, 52,268 men and 64,240 women in Copenhagen. U-shaped: lowest death at HDL of about 73 mg/dL in men and 93 in women. Men at 116 or more: HR 2.06; women at 135 or more: HR 1.68. No funding or conflict statement in the PubMed record.
RETRACTED in 2026, at the request of its authors and the editors. The study's own data: 100 lean keto dieters with median LDL of 237 mg/dL and ApoB of 178 mg/dL, scanned a year apart. Noncalcified plaque volume rose by a median of 18.9 cubic millimetres (interquartile range 9.3 to 47.0) and percent atheroma volume by 0.8 percent. Neither ApoB, its change, nor total LDL exposure was associated with how much plaque grew; baseline plaque was. Funded by the Citizen Science Foundation; one author co-wrote a low-carbohydrate cookbook and donates the royalties, and another receives membership contributions.
Retraction notice. Retracted at the request of the authors and the editors after concerns about methodology; the errors were judged too great to correct with a corrigendum. No funding or conflict statement in the PubMed record.
Web survey of 548 adults on low-carbohydrate diets and five clinical cases. Leaner people with low triglyceride to HDL ratios had larger LDL rises; 100 met the lean mass hyper-responder pattern. In the cases, reintroducing some carbohydrate lowered LDL markedly. No funding or conflict statement in the PubMed record.
One man on a ketogenic diet for ulcerative colitis. LDL rose from 95 to 574 mg/dL and total cholesterol to 705, held for nearly seven years, with zero coronary plaque on CT angiography. One person. Declared interest: one author is a partner in a lab testing company.
Consensus statement pooling over 200 prospective cohorts, genetic studies and randomised trials with more than 2 million people and over 150,000 cardiovascular events. Concludes the link between LDL and atherosclerotic disease is causal and grows with the dose and the years of exposure. No funding or conflict statement in the PubMed record.
Review by critics of the cholesterol hypothesis. Argues three major consensus reviews rest on misleading statistics, exclusion of unsuccessful trials and contradictory observations, and that the hypothesis fails the Bradford Hill criteria. No funding or conflict statement in the PubMed record.
19 blinded placebo-controlled trials, 123,940 people. Muscle pain or weakness: 27.1 against 26.6 percent (RR 1.03). In year 1 an excess of 11 per 1000 person-years, so only about one in 15 complaints on a statin was caused by it; more with high-intensity doses. Funded by the British Heart Foundation and the Medical Research Council; some authors' institutions hold grants from Novartis, Abbott, Amgen and others.
Blinded statin trials, 123,940 people against placebo. Low or moderate intensity statins raised new-onset diabetes by 10 percent (RR 1.10) and high intensity by 36 percent (RR 1.36). Funded by the Wellcome Trust; most authors' department accepts no personal honoraria.
8 randomised trials, 122,802 high-risk patients. Driving LDL below 55 mg/dL did not increase bleeding strokes (OR 1.05). No funding or conflict statement in the PubMed record.
Observational and genetic analyses in Copenhagen, UK Biobank and FinnGen. Genetically 1 mmol/L lower LDL went with fewer bacterial infections (RR 0.91) and less sepsis (0.91), but more viral infections (1.15) and HIV/AIDS (1.64). No conflicts declared.
Narrative review. Low cholesterol correlates with cancer, depression, hemorrhagic stroke and infection, but these associations are highly susceptible to reverse causation, and trials and genetic studies have not shown that drug-lowered LDL causes them. Advises individual assessment for the very elderly and frail. No funding or conflict statement in the PubMed record.
Observational studies. Highest against lowest fried food intake: major cardiovascular events RR 1.28, heart failure 1.37, cardiovascular death 1.02 and all-cause death 1.03. High heterogeneity and recall bias noted. PubMed tags it as non-government funded; no competing interests declared.
Educational use only. This page is not medical advice and not a reason to start, stop or change a statin or any prescription. Doses and diets here are what a study ran. Familial hypercholesterolemia, established heart disease, diabetes and pregnancy are questions for your own doctor. Aaron Cuha is a coach, not a doctor.
Educational use only. This database summarizes published research and is not medical advice. Where a dose appears it is a record of what a trial or clinic ran, or what people describe doing, with its source attached. Nothing here is a recommendation for personal use.