News · Nutrition & Diet
Mediterranean diet beat low-fat on 20-year heart risk
Researchers emulated a trial in 12,197 higher-risk US adults from two Harvard cohorts. Estimated 20-year cardiovascular risk was 28.2% on a Mediterranean diet against 35.9% on a low-fat one.
Based on a target trial emulation in 12,197 adults from the Nurses' Health Study and Health Professionals Follow-up Study, with 3,469 confirmed cardiovascular events
- Three named diets were compared head to head over 20 years in 12,197 higher-risk US adults.
- Estimated risk of heart attack or stroke: 35.9% low-fat, 31.2% AHA goals, 28.2% Mediterranean.
- That is roughly 8 fewer events per 100 people choosing Mediterranean over low-fat.
- In a general population the gap narrowed sharply, to about 10% lower risk.
- Diet was self-reported every four years, and nobody was actually assigned a diet.
Thirty-six against twenty-eight.
Those are the estimated chances, out of 100, that an older American adult already carrying heart risk factors has a heart attack, stroke or related event over twenty years, depending on whether they follow a low-fat diet or a Mediterranean one.
The gap is roughly eight people per hundred, and it comes from a study in the American Journal of Clinical Nutrition that set out to compare three named diets against each other rather than test one at a time.
What the three diets were
The comparison is the interesting part, because low-fat was not chosen as a straw man. It was official advice for decades, and a great many people over 60 built their habits around it.
Against it the researchers set the Mediterranean diet, and the food-based AHA-2020 dietary goals, AHA standing for the American Heart Association, which specify targets for fruit and vegetables, whole grains, fish, nuts, salt and sugary drinks rather than a fat percentage.
Three real strategies, all of which a doctor might reasonably recommend, run against each other over twenty years.
How the three diets were compared without a trial
Nobody was assigned a diet here, and the method used to work around that is worth understanding because it is now common.
The researchers emulated a target trial: they specified the randomized trial they wished existed, including who would be eligible and what each strategy involved, then analyzed long-running cohort data as though people had been assigned to those strategies.
The data came from 12,197 United States adults aged 55 to 80 y with diabetes or 3 major CVD risk factor or more, drawn from two long-running Harvard cohorts of nurses and other health professionals. Diet was assessed every 4 y via validated food frequency questionnaires, and cardiovascular events were confirmed by medical records rather than self-reported.
Over the twenty years, 3469 CVD cases occurred, CVD meaning cardiovascular disease: heart attack, stroke, an artery procedure, or death from any of them. That is a large number of real events, and it is what gives the estimates their precision.
What the Mediterranean diet estimate came to
Absolute risks, not relative ones, which is the more useful way round.
Pooled absolute risks of CVD were 35.9% under the low-fat diet, 28.2% under the Mediterranean diet, and 31.2% under the American Heart Association goals.
Put plainly: for every hundred people in this higher-risk group, an estimated 36 would have a cardiovascular event over twenty years eating low-fat, against 28 eating Mediterranean. The heart association’s goals landed between the two.
Both alternatives beat low-fat. The Mediterranean pattern beat it by more.
Why the Mediterranean advantage shrinks in healthier people
Here is the number that keeps the finding honest, and it is in the same paper.
The headline comparison was run in people who already had diabetes or a stack of risk factors, and who were aged 55 to 80. When the researchers re-ran it in a general population not selected for age, diabetes, or CVD risk factors, the Mediterranean advantage fell to roughly a tenth lower risk rather than a fifth.
That is not a contradiction. It is arithmetic: a strategy that removes a share of your risk removes more when you have more of it. The practical reading is that this evidence is strongest for exactly the group it studied, and thinner for a healthy 40-year-old.
What the diet questionnaires cannot fix
Two limits, and the first is unavoidable in any study of this kind.
Diet came from questionnaires filled in every four years. People misremember, and they under-report the things they know they should not eat. The method used here handles confounding better than a plain correlation does, and it cannot turn recalled eating into measured eating.
The second is who was in the cohorts. Nurses and other health professionals differ from the general population in education, income and health behavior, and the twenty-year risk estimates carry those differences with them.
What survives both caveats is the comparison itself. The three strategies were assessed the same way, in the same people, over the same period. If the measurement is imperfect, it is imperfect in the same direction for all three, and low-fat still came last.
People also ask
What is a target trial emulation?
A way of using observational data to answer the question a randomized trial would have answered. Researchers specify the trial they wish had been run, including who would be eligible and what the strategies are, then analyze existing cohort data as if participants had been assigned to those strategies. It removes several of the biases that make ordinary diet-and-disease correlations unreliable. It does not remove the fact that nobody was randomized.
What were the three diets?
A low-fat diet, a Mediterranean diet, and the food-based AHA-2020 dietary goals, which are the American Heart Association's targets covering fruit and vegetables, whole grains, fish, nuts, sodium and sugary drinks. The comparison matters because the low-fat pattern dominated official advice for decades.
What were the numbers?
Pooled absolute 20-year risks of cardiovascular disease were 35.9% (95% CI, 33.1%-38.5%) under the low-fat diet, 28.2% (95% CI, 25.8%-31.4%) under the Mediterranean diet, and 31.2% (95% CI, 29.1%-33.4%) under the AHA-2020 goals. Against the low-fat diet, risk ratios were 0.79 (95% CI, 0.70-0.91) for Mediterranean and 0.87 (95% CI, 0.79-0.96) for the AHA goals.
Who was studied?
12,197 United States adults aged 55 to 80 years with diabetes or at least three major cardiovascular risk factors, drawn from the Nurses' Health Study and the Health Professionals Follow-up Study. This is a higher-risk group by design, which is part of why the absolute numbers are large.
Does it work as well for everyone?
The advantage shrank considerably. In a general population not selected for age, diabetes, or cardiovascular risk factors, the risk ratios were 0.90 (95% CI, 0.84-0.95) for the Mediterranean diet and 0.94 (95% CI, 0.90-0.99) for the AHA goals. Same direction, much smaller effect, which is what you would expect when the baseline risk is lower.
What is the main weakness?
Diet was assessed every 4 years via validated food frequency questionnaires, so it rests on what people remembered and reported. And the cohorts are health professionals, who are not representative of the general US population in education, income or health behavior. The emulation method handles some confounding; it cannot manufacture a randomized comparison.
References
- Li, Y., Stern, D., Martinez-Gonzalez, M. A., et al. Cardiovascular disease risk under low-fat, Mediterranean, and American Heart Association diets: a target trial emulation in United States adults. American Journal of Clinical Nutrition, 2026.
- American Heart Association. The American Heart Association Diet and Lifestyle Recommendations.
- National Heart, Lung, and Blood Institute. DASH Eating Plan. US National Institutes of Health.