The Diet That Worked Until It Was Tested

An observational study linked the MIND diet to 53% less Alzheimer's. A three-year randomized trial found no difference.

Does the MIND diet prevent Alzheimer's disease?

The honest answer is that it is associated with a lower rate of Alzheimer's diagnosis in observational studies, and that the one randomized trial built to prove it did not.

The MIND diet — Mediterranean-DASH Intervention for Neurodegenerative Delay — was assembled at Rush University by the late Martha Clare Morris, a nutritional epidemiologist. It is a hybrid of two diets already known to help blood pressure and heart disease, adjusted toward the foods that showed up in brain research: green leafy vegetables and berries in particular.

In 2015 Morris and colleagues published the study everyone still quotes. They followed 923 people, ages 58 to 98, in the Rush Memory and Aging Project for an average of 4.5 years, scoring their diets from a food questionnaire. There were 144 new Alzheimer's diagnoses. After adjusting for age, sex, education, APOE ε4 status, physical activity, cognitive activity and total calories, the top third of MIND scores had a hazard ratio of 0.47 (95% CI 0.29–0.76) — a 53% lower rate of Alzheimer's than the bottom third. The middle third had a 35% lower rate.

Two details from that paper deserve more attention than they usually get.

The first is a sensitivity analysis the headlines skipped. The obvious objection to a study like this is reverse causation: maybe people who were already developing Alzheimer's had started eating worse, and the diet score was measuring early disease rather than predicting it. The authors tested for exactly that. They reran the analysis after deleting the 33 cases diagnosed within two years, then after deleting the 60 diagnosed within three years. The top-tertile hazard ratio barely moved — 0.53 both times. That is a real answer to a real objection, and it makes the association more credible than most.

The second is the caveat the authors put in their own discussion: an observational design precludes the interpretation of the findings as cause and effect. Their words. They said a randomized trial would be needed.

What did the randomized trial actually find?

No significant difference in cognition after three years between the MIND diet and a control diet.

The trial was published in the New England Journal of Medicine in July 2023, led by Lisa Barnes at Rush, funded by a $14.5 million NIH grant, and run at Rush and the Harvard School of Public Health.

It enrolled 604 adults — 301 assigned to the MIND diet, 303 to a control diet. Everyone was 65 or older, overweight, eating a poor diet at baseline, had a family history of dementia, and had no cognitive impairment when they started. They were followed three years, with 93.4% completing.

Global cognition improved in both groups. The MIND group gained 0.205 standardized units; the control group gained 0.170. The mean difference was 0.035 standardized units (95% CI −0.022 to 0.092; P=0.23). Brain imaging outcomes at three years did not differ either.

That confidence interval is worth reading rather than skipping. It runs from a small harm to a small benefit. The trial did not prove the MIND diet does nothing; it failed to detect a difference, and the range of differences it could not rule out is narrow. Those are different statements, and the second is the accurate one.

Why did the two studies disagree?

Largely because the control group in the trial was not a bad diet. It was a coached, calorie-reduced diet with a dietitian attached, and it worked.

Both arms received individualized guidance from registered dietitians, regular phone and in-person consultations, and group sessions across three years. Both arms were asked to cut 250 calories a day. Both lost roughly 5 kilograms — about 5.5% of starting body weight, above the 3% threshold generally considered clinically meaningful.

By the end, the average MIND score was 11.0 in the intervention group and 8.3 in the control group, on a scale where the 2015 observational data put anything above roughly 8.5 in the protective range. The control group had drifted into the therapeutic zone. Lisa Barnes, the lead author, noted that both groups lost about 5 kg, "suggesting that it could have been weight loss that benefited cognition in this trial."

There is one more piece. A post-hoc analysis of 551 trial participants, presented at the 2024 Alzheimer's Association International Conference, ignored which arm people were assigned to and looked instead at how closely they actually ate the diet. A one-point higher MIND score was associated with 0.027 standardized units more cognitive improvement over three years (95% CI 0.004–0.051), and the top third of adherence beat the bottom third by 0.140 units (95% CI 0.033–0.247).

That is suggestive, and it is also the weakest design of the three — people who follow a diet closely differ from people who do not in a hundred ways a statistical adjustment cannot fully capture. It is a conference abstract, not a finished paper. Hold it lightly.

What is actually on the MIND diet?

Two vegetable servings a day, berries twice a week, fish once a week — considerably less demanding than the Mediterranean or DASH diets it came from.

The 2015 score had 15 components: ten to eat and five to limit.

Eat:

• Green leafy vegetables — 6 or more servings a week
• Other vegetables — 1 or more a day
• Berries — 2 or more a week
• Nuts — 5 or more a week
• Beans — more than 3 a week
• Whole grains — 3 or more a day
• Fish — 1 or more a week
• Poultry — 2 or more a week
• Olive oil as the main cooking oil at home
• Wine — up to one glass a day

Limit:

• Red meat — fewer than 4 servings a week
• Butter and stick margarine — less than 1 tablespoon a day
• Cheese — less than 1 serving a week
• Pastries and sweets — fewer than 5 a week
• Fried and fast food — less than 1 a week

The thresholds are deliberately low. The DASH and Mediterranean scoring systems ask for three to four daily servings each of fruit and vegetables, and the Mediterranean asks for six fish meals a week. The MIND diet asks for two vegetable servings a day and one fish meal a week. The authors' point was that modest change might be enough.

One honest note on the wine component: it earns a point in the scoring system, and this site is not going to recommend that anyone start drinking for their brain. Alcohol carries its own risks, and it interacts badly with a long list of medications older adults take.

The questionnaire had two limitations. Berry intake was captured by a single item about strawberries, with a top answer of "2 or more times per week." Olive oil was a single question about the oil usually used at home. Neither measure could see the upper end of real consumption — which, as the authors point out, would bias the result toward finding no effect rather than away from it.

Will this help the person I already care for?

There is no evidence that it will, and the calorie restriction built into the trial would be actively wrong for many people with dementia.

Every participant in both studies was free of dementia at the start. Neither study asked whether changing the diet of someone who already has Alzheimer's slows anything down. Nobody has answered that question, and it would be dishonest to imply otherwise.

What is known is that the arithmetic reverses at that point.

Unintentional weight loss is common in dementia and it is not benign — appetite changes, taste changes, forgetting to eat, difficulty using utensils, and swallowing problems all pull the same direction. The concern in mid and late dementia is usually getting enough food in, not getting less.

The Academy of Nutrition and Dietetics, in its 2018 position paper on nutrition for older adults in long-term and post-acute care, is direct about this: for frail older adults, overall health goals may not warrant a therapeutic diet at all, because of its effect on quality of life. A diet that is not palatable leads to poor intake, which leads to weight loss and undernutrition, which leads somewhere worse. The Academy's recommendation is the least restrictive diet the person's medical situation allows.

So a caregiver who reads about the MIND diet and starts removing the cheese, the butter and the pastries from a 78-year-old with moderate dementia who is already losing weight has followed reasonable-sounding advice into a real harm. If the person still eats well and enjoys food, adding leafy greens and berries is harmless and possibly good. Taking food away is a decision that should go through the person's doctor or a dietitian, not through an article about prevention.

One more finding, reported honestly. In the 2015 data, the MIND diet appeared less protective in people carrying APOE ε4, the main genetic risk variant. The statistical test for that interaction came out at P = .06 — marginal, not conclusive, and the authors flagged it as such. It is not a reason to skip vegetables. It is a reason to be careful about promising a genetically at-risk family that diet is their lever.

So what is worth doing?

Eat the diet if you want to. Expect it to help your heart and your blood pressure, where the evidence is strong, and treat the brain effect as plausible and unproven.

• The foods are good for you regardless. Leafy greens, berries, nuts, beans, fish and olive oil have solid cardiovascular evidence behind them. Nothing about the 2023 trial changes that.
• Vascular health is the strongest lever anyone has. Blood pressure, diabetes and cholesterol are managed with tools that have been proven in trials. Diet supports those. So does treating hearing loss, which currently has better dementia evidence behind it than any food.
• Do not buy anything. There is no MIND diet supplement, pill or subscription with evidence behind it. Anyone selling one is trading on a study that did not find what they are implying it found.
• If you are the caregiver, feed yourself too. The person who skips meals standing at the counter is usually the one doing the caring. That is the diet in this house most likely to be failing.
• Do not remove food from someone who is losing weight. Ask a dietitian or the person's doctor first. Weight loss in dementia is a symptom, not a success.