The MIND diet and dementia risk: the trial versus the cohorts

People who scored highest on the MIND diet developed dementia less often than people who scored lowest.² Then researchers handed the diet to one group of older adults for three years and a control diet to another. Cognition improved about the same in both.³ Both results are real; they answer different questions.

What the cohort studies actually found

The MIND diet combines the Mediterranean diet with the DASH diet.¹ It then leans further into foods that population studies had already linked to lower dementia risk: leafy greens, berries, nuts, olive oil, fish. Most of the early evidence for it came from exactly that kind of population study. Researchers hand out a questionnaire. They score how closely each person's usual eating matches the diet. Then they watch who develops dementia over the following years.

A pooled analysis of eleven such cohort studies covered more than 224,000 people and over 5,000 cases of dementia.² People in the highest third of MIND diet scores had a lower rate of dementia than people in the lowest third.²

A broader review of the literature on the MIND diet and cognition reported the same pattern repeating. Ten of eleven studies that looked at dementia or Alzheimer's risk found a positive association with the diet.¹ A separate review of risk factors for early-onset dementia listed higher MIND diet adherence, alongside more education and injury-free exercise, among the factors that looked protective.⁴

That is a consistent signal across a lot of data. It is also, by its nature, a signal that can be confounded. People who eat this way tend to differ from people who do not in ways a questionnaire cannot fully capture. Income matters. So do other health habits. So does the possibility that early, undiagnosed cognitive change had already nudged some people's eating in one direction before anyone noticed. Cohort studies can adjust for the factors researchers think to measure. They cannot adjust for the ones they miss.

What the trial found instead

Only one study has tested the MIND diet the way a drug would be tested. Researchers assigned people at random to follow it or not, then watched what happened. That trial ran three years, in 604 older adults who had no cognitive impairment at the start but did have a family history of dementia. Half were coached to follow the MIND diet with mild calorie restriction. Half followed a control diet with the same calorie restriction.³

Cognition improved in both groups over the three years. That is a reminder that practice effects and general health coaching can move test scores. The MIND-diet group's global cognition score rose by 0.205 standardized units, the control group's by 0.170. The gap between them, 0.035, was not statistically distinguishable from no difference at all.³

MIND diet vs. control diet, change in global cognition over three years0 = no effect

Difference between groups0.035 (-0.022 to 0.092)

Brain scans told the same story. The two groups looked similar on MRI measures of white-matter changes and brain volume by the end.³

That is a genuinely different result from the cohort data, not a smaller version of the same one. The trial is not saying the diet has a small effect. It is saying that over three years, in this group of people, the diet could not be told apart from a comparably healthy control diet.

Why the two kinds of study disagree

Both designs looked at the same diet, in populations that overlap in age and health status, and still landed in different places. That gap is worth sitting with rather than resolving too quickly in either direction, because the instinct to pick a favorite number is exactly what turns a mixed literature into a false certainty. Part of the gap is what each design can catch. A cohort study following people for a decade or more can pick up a slow-moving effect that a three-year trial simply has not had time to show. Part of it is what "the control diet" means inside a trial. Participants who agree to join a dementia-prevention study, and who then get regular counseling, rarely eat badly even in the comparison arm. That narrows the gap between groups from the start, before the diet itself does anything.

Part of it may also be the questionnaire problem working in reverse. A cohort study measures whatever people were already doing, including whatever led them to eat that way in the first place. A trial measures something narrower: what happens when people are handed a plan on day one and asked to follow it from there.

Neither design is wrong. They are answering different versions of the same question. The cohort data says that people who happen to eat this way, for whatever reason, get dementia at a lower rate over the years that follow. The trial says that assigning the diet on purpose did not clearly change cognition or brain measures over three years, at least in a population that started out relatively healthy for its age.

What remains unknown

The eleven-cohort estimate and the trial result cannot both be given equal weight without saying which kind of evidence answers which question.^2,3 Three years is also not long by the standard of a disease that develops over decades. A trial that length may simply be too short to catch an effect that a ten-year cohort study can see.³ Whether a longer intervention, an earlier start, or a different population would close the gap between the cohort finding and the trial finding is still an open question. No single record here answers it, and none should be read as if it did.