The Mediterranean diet: what survived the trials and what did not

The Mediterranean diet is not a single thing. It describes an eating pattern documented in southern Europe in the mid-twentieth century: abundant vegetables, fruit, legumes, wholegrains and nuts; olive oil as the principal fat; moderate fish; modest dairy, mostly fermented; little red or processed meat; and wine with meals in some populations.

Its reputation comes largely from one trial, and the story of that trial is worth knowing.

PREDIMED

PREDIMED randomised around 7,400 Spanish adults at high cardiovascular risk to a Mediterranean diet supplemented with extra-virgin olive oil, the same diet supplemented with mixed nuts, or a control low-fat diet. Published in 2013, it reported roughly a 30 per cent reduction in major cardiovascular events in the Mediterranean arms.

This was a landmark: a hard-outcome randomised trial of a whole dietary pattern, which almost never happens because such trials are extraordinarily difficult and expensive.

In 2018 the paper was retracted and republished. Investigation had found that a proportion of participants were not properly individually randomised — some were enrolled as households, and one site randomised by clinic rather than by person. The authors reanalysed excluding and adjusting for the affected participants.

The revised results were substantially similar: the effect estimate held. But the trial can no longer be described as cleanly randomised throughout, and its evidential weight is somewhat reduced. Both facts are usually omitted from articles enthusing about olive oil.

Where the claims outran the data

Cognition and dementia. Observational associations are consistent and encouraging. The randomised evidence is weaker. The MIND diet trial, testing a Mediterranean-DASH hybrid explicitly designed for brain health, found no significant cognitive advantage over a mild calorie-restriction control over three years. A PREDIMED substudy suggested cognitive benefit, but it was a secondary analysis in a subset.

Cancer. A PREDIMED secondary analysis reported reduced breast cancer incidence in the olive oil arm, based on a small number of events. It is hypothesis-generating rather than conclusive.

Longevity. Observational cohorts consistently associate adherence with lower all-cause mortality. Randomised evidence for mortality specifically is not strong.

What is doing the work

Nobody knows precisely, and it is likely to be several things acting together rather than one active ingredient — which is exactly why the reductionist approach of isolating a component into a supplement has repeatedly failed.

Plausible contributors: replacing saturated with monounsaturated and polyunsaturated fat; high fibre intake from legumes, vegetables and wholegrains; polyphenols in extra-virgin olive oil, which are largely absent from refined olive oil; nuts, which have unusually consistent observational and trial data; oily fish; and the displacement of red meat, processed meat and refined carbohydrate.

The displacement point deserves emphasis. Much of the benefit of any healthy dietary pattern comes from what it crowds out, and studies that add Mediterranean foods to an otherwise unchanged diet find much smaller effects.

Adopting it outside the Mediterranean

The principles transfer; the specific foods need not. What matters is the structure:

• Legumes several times a week. The most consistently overlooked element in northern European adaptations, and probably one of the more important.
• Extra-virgin olive oil as the main fat — the extra-virgin qualifier matters, as the polyphenols are removed by refining. PREDIMED participants consumed around 50 ml a day, which is a lot more than most people use.
• A handful of nuts daily. Roughly 30 g.
• Vegetables at both main meals, in quantity.
• Fish twice a week, including oily fish.
• Wholegrains rather than refined.
• Red meat occasionally, processed meat rarely.

Local equivalents are fine. Rapeseed oil is a reasonable second choice where olive oil is unaffordable. Tinned sardines and mackerel are cheap and count. Frozen vegetables count. The pattern does not require a Tuscan grocer.

The wine question

Traditional descriptions include moderate wine with meals, and this is the component with the weakest justification. Given the evidence discussed elsewhere on alcohol — carcinogenicity from low doses, and Mendelian randomisation studies that do not support a cardiovascular benefit — recommending wine as part of a health-promoting diet is not defensible.

The rest of the pattern stands without it.

The honest summary

The Mediterranean pattern is the best-evidenced diet available for cardiovascular health, it is palatable, it is culturally flexible, and it is not a miracle. If the goal is fewer heart attacks and better metabolic markers, it earns its reputation. If the goal is preventing dementia, the evidence is not there yet, whatever the headlines have said.