Tip · Nutrition

Mediterranean diet

Olive oil, nuts, vegetables, legumes, fish. In PREDIMED with 7,447 people at high risk, heart attack, stroke or cardiac death occurred about 30 percent less often under a Mediterranean diet with olive oil or nuts. The study was retracted in 2018 because of errors in random allocation and reanalyzed; the result held.

Olive oil, nuts, vegetables, legumes and fish: the Mediterranean diet is one of the few dietary patterns that have been tested in large randomized trials with heart attack and stroke as endpoints. Its most important study was retracted and republished. This page explains why the result nevertheless stands, and where the evidence ends.

In short

In PREDIMED, 7,447 people in Spain at high cardiovascular risk, heart attack, stroke or cardiac death occurred about 30 percent less often under a Mediterranean diet with extra virgin olive oil or nuts than under a low-fat diet. In 2018, the authors retracted the study because, among other things, randomization had not been carried out correctly at two of eleven centers, and published a reanalysis with an almost identical result. CORDIOPREV confirmed the benefit in 1,002 heart patients. All-cause mortality did not demonstrably fall in PREDIMED, and for the brain there is no evidence from randomized trials.

What the Mediterranean diet is

The impetus came in the 1960s from the Seven Countries Study: in the Mediterranean regions, markedly fewer people died of coronary heart disease, presumably because of the different diet. Since then, the pattern has been defined, not a single food. The Cochrane review requires two core features: a high ratio of monounsaturated to saturated fat, that is, olive oil as the main cooking fat or nuts, and plenty of plant foods such as fruit, vegetables and legumes. Added to these are whole grains, more fish, little meat and meat products, dairy products in moderation and, in the classic description, low to moderate amounts of red wine.

There are separate pages on individual components: Omega-3 and Hydroxytyrosol and olive polyphenols.

The story of PREDIMED: retraction and republication

PREDIMED ran at eleven Spanish centers. 7,447 people aged 55 to 80, 57 percent women, at high risk but without cardiovascular disease, were assigned to three diets: Mediterranean with extra virgin olive oil, Mediterranean with nuts, or, as a control, the advice to eat less fat. Olive oil and nuts were provided free of charge, along with quarterly training sessions. After a mean of 4.8 years, the study was stopped following a prespecified interim analysis and published in 2013 in the New England Journal of Medicine.

Later, the authors found deviations from the protocol: household members of participants had been enrolled in the same group without their own random allocation, at one of the eleven centers some participants were assigned entirely without randomization, and at another the allocation tables were apparently used inconsistently. According to a later paper, this affected 866 participants, 11.6 percent. In June 2018, the authors retracted the 2013 publication and published, in the same issue, a reanalysis that no longer relies solely on the assumption that all participants were randomly allocated.

The result barely changed. In 2013, the hazard ratio was 0.70 with olive oil and 0.72 with nuts; in 2018, 0.69 and 0.72. Even without the 1,588 participants whose allocation was known or presumed to have deviated from the protocol, the result remained similar. The price of the correction: since then, PREDIMED is no longer a pure randomized trial, but one with statistical adjustment.

What is well supported

What is supported is protection against cardiovascular events, in two large randomized trials and in two situations. Before disease, in PREDIMED, there were 96 events under olive oil, 83 under nuts and 109 in the control group; the decrease was clearest for strokes. After disease, in CORDIOPREV, there were 87 versus 111 events compared with a low-fat diet. This fits with large cohorts such as the Women’s Health Study with 25,315 women and a mortality that was 23 percent lower with high adherence to the diet.

What the studies show

PREDIMED, New England Journal of Medicine 2018

The reanalysis after the retraction: 7,447 participants, 288 major cardiovascular events over a median of 4.8 years. 3.8 percent in the olive oil group, 3.4 percent in the nut group, 4.4 percent in the control group. Hazard ratio 0.69 for olive oil and 0.72 for nuts, similar also after propensity score adjustment and without the participants who deviated from the protocol.

Cochrane review 2019

30 randomized trials with 12,461 participants. For PREDIMED, Cochrane rates the evidence for fewer strokes as moderate, from 24 to 14 per 1,000 people. For cardiovascular mortality (hazard ratio 0.81) and all-cause mortality (1.0), little to no effect was seen, with low certainty. Among the risk factors, systolic blood pressure fell by about 3 mmHg, LDL cholesterol only slightly. Cochrane excluded two further studies on secondary prevention from the main analysis because of doubts about data quality.

CORDIOPREV, Lancet 2022

1,002 patients with coronary heart disease in Córdoba, randomized to a Mediterranean or a low-fat diet, seven years of follow-up, endpoints assessed blinded. 198 major events, 87 under the Mediterranean and 111 under the low-fat diet, a hazard ratio of 0.72 to 0.75 depending on the model. In the 827 men, the difference was clear, 16.2 versus 22.8 percent; in the 175 women, none was found.

Lyon Diet Heart Study, Lancet 1994 and Circulation 1999

605 people after a first heart attack, a Mediterranean diet rich in alpha-linolenic acid versus the heart diet customary at the time. After 27 months, 3 versus 16 cardiac deaths and 5 versus 17 non-fatal heart attacks; after 46 months, 14 versus 44 events of this kind. Instead of olive oil, a rapeseed oil margarine was used here. Cochrane rates the evidence as low.

Women’s Health Study, JAMA Network Open 2024

25,315 initially healthy women, followed for a mean of 24.7 years, 3,879 deaths. With high adherence to the Mediterranean diet, mortality was 23 percent lower than with low adherence, and still 11 percent lower after adjustment for further lifestyle factors. Small metabolites explained the most, at 14.8 percent, and inflammatory markers 13.0 percent; LDL cholesterol contributed less than 3 percent. This is an observation, not an experiment.

Weight in PREDIMED, Lancet Diabetes & Endocrinology 2019

Without a calorie target and without an exercise program, no group gained more weight over five years than the low-fat control group: minus 0.4 kilograms with olive oil, practically zero with nuts. Waist circumference grew by just under one centimeter less in the nut group. The first version from 2016 was retracted because of the same randomization errors and reanalyzed.

MIND trial, New England Journal of Medicine 2023

The MIND diet is a mix of the Mediterranean and the blood-pressure-lowering DASH diet with a focus on the brain. 604 older adults without cognitive impairment but with overweight and a family history, both groups with mild calorie reduction, three years. Cognitive performance improved in both groups; no difference between them was found, not even on MRI.

Where the data stop

Since 2018, PREDIMED has been a corrected study: the reanalysis uses statistical methods that no longer rest entirely on random allocation. The control group only received the advice to eat less fat, and PREDIMED tested a Mediterranean diet plus free olive oil or nuts, not the pattern as such. Among the funders of the PREDIMED substudies and of CORDIOPREV were organizations from the olive oil and nut industries, alongside public funds.

Fewer cardiovascular events does not yet mean living longer: all-cause mortality did not demonstrably fall in PREDIMED, and Cochrane rates the evidence for hard endpoints as low to moderate. In CORDIOPREV, only the men benefited; the women’s group was small, with 175 participants. For the brain, there is no randomized evidence; the MIND trial showed no difference. And both large randomized trials took place in Spain, with middle-aged and older people at increased risk or with existing heart disease.

How to do it

Make olive oil your main fat; in PREDIMED it was extra virgin olive oil, used generously. Eat nuts daily, plenty of vegetables, fruit and legumes, along with whole grains, fish more often, less meat and meat products, and dairy products in moderation. That is the combination that Cochrane defines as Mediterranean and that was tested in the studies.

You do not have to count calories for this: PREDIMED worked without a calorie target, and despite all the fat, no group gained more weight over five years than the low-fat control group. Red wine is part of the classic description, but what was tested was the diet with olive oil or nuts, not the alcohol; a benefit of its own for wine has therefore not been shown, more on this on the page Cutting down on alcohol. If you already have coronary heart disease, the diet does not replace treatment but complements it, and for exactly this situation there is a large randomized trial, CORDIOPREV.

Safety

The Mediterranean diet consists of ordinary foods, and in the studies that recorded adverse events, these were absent or minor according to Cochrane; the first PREDIMED publication reported no side effects from the diet. The obvious concern that plenty of fat makes you fat was not confirmed by PREDIMED over five years. Anyone who does not tolerate nuts has an alternative in the olive oil variant, which performed just as well in PREDIMED. Alcohol carries its own risks, which should not be credited to the diet. Anyone taking medication for diabetes or high blood pressure who changes their diet substantially is best accompanied by a physician.

BK-Score Well supported

Human evidence		8
Mechanism		6
Safety data		8
Hype gap		6
Track record of use		10

Evidence 8: Two large randomized trials with hard endpoints, PREDIMED with 7,447 people without heart disease (hazard ratio 0.69 with olive oil and 0.72 with nuts) and CORDIOPREV with 1,002 heart patients (0.72 to 0.75), plus the Lyon study after heart attack and cohorts such as the Women’s Health Study with 25,315 women over almost 25 years; points deducted because PREDIMED 2013 was retracted and reanalyzed in 2018 without the assumption of complete randomization, and Cochrane rates the evidence for hard endpoints as low to moderate. Mechanism 6: Systolic blood pressure falls by about 3 mmHg, LDL and triglycerides only slightly, and in the Women’s Health Study, metabolites, inflammatory markers and lipid values each explained only 10 to 15 percent of the association. Safety 8: These are ordinary foods; adverse events were absent or minor in the studies that recorded them, and despite plenty of fat there was no greater weight gain over five years; side effects, however, were systematically recorded in only a few studies. Hype 6: The cardiovascular effect holds up, but all-cause mortality did not demonstrably fall in PREDIMED, the effect in CORDIOPREV was seen only in men, and for the brain the MIND trial with 604 older adults showed no difference. Use 10: A dietary pattern lived for generations, scientifically accompanied since the Seven Countries Study of the 1960s. Direction positive.

The score rates the state of knowledge, not the effect. “Safety data 9” means well studied – not harmless.
Subjective assessment by Biohacking Kompakt based on published scoring rules – not a scientific rating and not a medical recommendation. Rules and all ratings (German)

Frequently asked questions about the Mediterranean diet

What is part of the Mediterranean diet?

Olive oil as the main cooking fat or nuts, and plenty of fruit, vegetables and legumes; these are the two core features according to Cochrane. Added to these are whole grains, more fish, little meat and meat products, dairy products in moderation and, in the classic description, low to moderate amounts of red wine.

Why was the PREDIMED study retracted?

Because of errors in random allocation. Household members were enrolled without their own randomization, at one of the eleven centers some participants were assigned entirely without randomization, and at another the allocation tables were used inconsistently. The authors retracted the 2013 publication in June 2018 and published a reanalysis.

Does the result still hold?

Yes, with one caveat. The reanalysis arrived at a hazard ratio of 0.69 with olive oil and 0.72 with nuts, practically as in 2013, even without the 1,588 participants with deviating allocation. However, it no longer relies solely on randomization, and Cochrane rates the evidence for hard endpoints as low to moderate. CORDIOPREV independently confirmed the benefit in heart patients.

Don’t you gain weight with so much olive oil and nuts?

Not in PREDIMED. Without a calorie target, none of the Mediterranean groups gained more weight over five years than the low-fat control group; the olive oil group even gained 0.4 kilograms less, narrowly not statistically significant. Waist circumference grew by just under one centimeter less in the nut group.

Do I have to drink red wine for it?

No. Wine is part of the classic description, but PREDIMED tested the diet with olive oil or nuts, not the alcohol. A benefit of its own for wine has not been shown by these studies.

Does the Mediterranean diet also protect the brain?

That is not supported. In the MIND trial with 604 older adults, cognitive performance under a Mediterranean-DASH hybrid diet did not improve more over three years than in a control group that also reduced calories slightly, and MRI showed no difference either.

Sources

Open in the database – with all tips, filters and BK-Score (German app)

Information only, not medical advice and not a usage recommendation. With pre-existing conditions and before major changes, consult a physician. Last updated: 2026-11-07.