What actually protects memory: separating the evidence from the marketing

There is a reliable pattern in brain-health research. An observational study finds that people who do X have less dementia. Headlines follow. A supplement appears. Then, some years later, a randomised trial gives X to one group and a placebo to another, and the effect disappears.

This has now happened to ginkgo biloba, to vitamin E, to B vitamins in unselected populations, and to a long list of nutraceuticals with confident packaging. It is not that the original observations were fabricated. It is that people who take ginkgo are different from people who do not in a hundred other ways, and observational data cannot separate the herb from the person.

So what survives the trials?

The Lancet Commission's list

The most useful summary available is the Lancet Commission on dementia prevention, intervention and care, which has been updated several times since 2017. Its central estimate is that a substantial share of dementia cases worldwide — the 2020 update said around 40 per cent — are associated with modifiable risk factors. That figure is a population-attributable fraction, not a personal guarantee, and it assumes causality that is not fully proven for every factor. It is still the best map we have.

The factors it identifies cluster by life stage. In midlife: hearing loss, high blood pressure, obesity, excessive alcohol and head injury. In later life: smoking, depression, social isolation, physical inactivity, air pollution and diabetes. Less education in early life is the remaining one.

Two things stand out from that list. First, almost all of it is vascular or sensory rather than specifically neurological. Second, hearing loss is at the top of the midlife group, which surprises most people.

Hearing aids, blood pressure, and the unglamorous middle

The hearing association is strong and consistent across cohorts. The mechanism is debated — it may be that straining to hear consumes cognitive resources, that sensory deprivation accelerates atrophy in auditory cortex, that hearing loss drives social withdrawal, or all three. The ACHIEVE trial, published in 2023, gave hearing aids to older adults and did not find a cognitive benefit in the overall population. But in the subgroup at higher risk of decline, the hearing intervention slowed cognitive change by about half over three years. That is a subgroup finding and should be treated with appropriate caution, but it is the most encouraging result in this field for some time.

Blood pressure is on firmer ground. The SPRINT MIND trial randomised older adults to intensive blood pressure control (systolic under 120) versus standard (under 140), and found a significant reduction in mild cognitive impairment. The effect on probable dementia itself did not reach statistical significance, but the trial was stopped early for cardiovascular benefit, which reduced its power to detect the slower-developing outcome.

Treating hypertension in midlife is, on current evidence, one of the more defensible things a person can do for their future cognition. It also happens to be the thing most people are already being nagged about and quietly ignoring.

Exercise: real, but oversold

Physical activity has the most enthusiastic press and a more mixed trial record than the enthusiasm suggests. Observational data is strongly positive. Randomised trials in people who already have cognitive impairment have been largely disappointing — the DAPA trial of exercise in mild-to-moderate dementia found no cognitive benefit, and a slight worsening on one measure.

The reasonable interpretation is that exercise is prevention rather than treatment, and that trials of a year or two in already-affected people are testing the wrong thing at the wrong time. Exercise also improves the vascular risk factors that clearly do matter, so the indirect route is well supported even where the direct evidence is thin. Aim for it. Do not expect it to reverse anything.

Diet, and the MIND diet specifically

The MIND diet — a hybrid of Mediterranean and DASH eating patterns emphasising leafy greens, berries, nuts, olive oil, fish and beans — produced striking observational results, with high adherence associated with cognitive function equivalent to being several years younger.

The randomised trial, published in the New England Journal of Medicine in 2023, compared the MIND diet to a mild caloric restriction control over three years. Both groups improved cognitively. There was no significant difference between them. This is the same pattern again, and it should temper the claims considerably.

It does not mean the diet is worthless. It is a good diet for cardiovascular health, which is on the causal pathway, and three years is short for an outcome that develops over decades. But "eat berries to prevent Alzheimer's" is not something the trial evidence currently supports.

Cognitive training and the crossword question

Brain-training programmes reliably make you better at the trained task. Transfer to untrained abilities or to everyday function is weak and contested; the ACTIVE trial found some durable effects, particularly for reasoning and speed of processing, but the field has a history of overclaiming that resulted in a Federal Trade Commission settlement against one prominent company.

Learning something genuinely new and effortful — a language, an instrument, a craft — is a better bet than a puzzle app, if only because it usually comes with social contact and sustained challenge rather than a leaderboard.

What to tell your parents

Get the hearing checked. Get the blood pressure treated. Keep moving, keep seeing people, and be sceptical of anything sold in a bottle with a picture of a brain on it. It is a duller answer than the advertising offers, and it is the one the trials support.