Brain Health

The 2024 Lancet Commission puts that figure on 14 modifiable risk factors.^[1] The list is dominated by cardiovascular and sensory items, which points at the most useful reframing available here: the brain is downstream of cardiovascular and metabolic health, not a separate organ. What's good for the heart is overwhelmingly good for the brain.

The evidence has converged on a small set of high-leverage actions — treat hearing loss early, treat blood pressure and cholesterol in midlife, correct vision, defend metabolic health, stay socially connected, keep learning hard things. One caveat runs through all of it: no trial has yet shown that any of these prevents dementia as an endpoint. What exists is a strong risk-factor map, several trials that move cognition, and a couple of natural experiments that move diagnoses. The things that don't work — brain-training apps, isolated cognitive supplements, blue-light glasses — are mapped just as clearly.

What the evidence actually supports

The 14 modifiable risk factors — Moderate

Moderate, not Strong, and it matters why: the Commission is a careful synthesis of observational evidence, its shares assume the associations are causal, and none of it is trial-derived.^[2] The "share of cases" column is the population attributable fraction — the proportion of dementia that would not occur if that factor were eliminated across the whole population. It is not a personal risk estimate.

Ordered by that share, the 14 are hearing loss and high LDL ("bad") cholesterol at 7% each; less education and social isolation at 5%; depression, traumatic brain injury and air pollution at 3%; physical inactivity, smoking, type 2 diabetes, hypertension and untreated visual loss at 2%; and obesity and excessive alcohol at 1%. Cholesterol and vision loss are the two added in 2024. Dementia prevention carries the full table, with each factor's risk multiplier and how common it is.

The ranking is driven by how common a factor is, not how dangerous it is. Hearing loss and high cholesterol tie at the top on ordinary risk multipliers — 1.4× and 1.3× — because 59% and 77% of the relevant population have them. Depression carries by far the largest individual risk of the 14 (2.2×) but affects 7% of people, so it contributes 3%. Hypertension's small share is often misread as "blood pressure doesn't matter much for the brain"; in fact its individual dementia effect is one of the weakest measured (1.2×) despite being present in nearly a third of adults. The case for treating it hard is cardiovascular, plus a real effect on milder cognitive impairment.

Ten of the 14 are midlife factors, one is early-life (education) and only three are late-life (social isolation, air pollution, uncorrected vision loss). The window that matters most is roughly ages 45 to 65 — which is also when most people assume there is nothing to do yet.

The estimates may be generous. Re-tested inside a single cohort of 5,526 Americans, only three of the 14 factors survived — and midlife hypertension came out protective, which is itself a reason to distrust the re-analysis — leading those authors to warn that meta-analytic estimates may be inflated by publication bias. That work is a conference abstract, not a peer-reviewed paper: a caution flag, not a refutation.^[3] Dementia prevention carries the full table of risk multipliers and prevalences.

Strong — including strong evidence of no effect

• Cataract surgery is a genuine intervention, not a cosmetic one. In a randomised trial, first-eye surgery in women over 70 cut the rate of falling by about a third; notably, the companion trial of second-eye surgery did not significantly reduce falls.^[4]
• Noise damage is dose-dependent, and the scale is logarithmic. Safe weekly exposure runs from about 40 hours at 80 decibels (busy traffic, a hairdryer) to roughly 20 minutes at 100 decibels (a loud concert, a leaf blower). Filtered earplugs are the cheapest brain-health purchase available.
• Blue-light-blocking glasses do nothing measurable. A Cochrane review of 17 randomised trials found little or no effect on visual fatigue, visual acuity or sleep, and no randomised evidence at all on macular health.^[5]
• Lutein and zeaxanthin supplements do not improve cognition. The decisive test randomised more than 3,000 older adults and was null, as was its omega-3 arm.^[6] Leafy greens remain worth eating for the eye; the pill is not worth taking for the brain.

Moderate

• Treating hearing loss. One null randomised trial with a promising subgroup, plus a consistent observational signal with a clean dose-response. Enough to act on; not proof that hearing aids prevent dementia.
• Intensive blood-pressure control, for milder cognitive impairment. SPRINT-MIND tested a target below 120 mmHg systolic against a standard target below 140, and cut mild cognitive impairment by about a fifth and the combined impairment-or-dementia endpoint by about 15%; the dementia-alone endpoint moved the right way but remained compatible with no effect, and the trial was probably underpowered for it.^[7] See Blood pressure.
• Education and cognitive reserve. The one cognitive item the Commission actually lists. A meta-analysis of 27 longitudinal studies puts early-life education at about 18% lower dementia risk, midlife occupational complexity at about 9%, and late-life engagement at about 19% — meaning what a 60-year-old does with their time has effect sizes comparable to whether they went to college.^[8] About 28% of education's protective effect runs through the complexity of the job it makes possible.^[9]
• Social connection. Within that same late-life window it is the strongest single reserve marker measured — roughly 30% lower dementia risk, against about 9% for cognitive activity.^[10] A meta-analysis of 15 randomised trials of social interventions in 1,785 older adults found improved global cognition, though no effect on executive function specifically.^[11]
• Sleep duration and regularity, and treating sleep apnea. Deep (slow-wave) sleep is when the brain's waste-clearance system appears most active at removing beta-amyloid, and a Framingham follow-up found each percentage-point yearly decline in deep sleep tracked about 27% higher dementia risk.^[12] Both claims are weaker than they sound. The human causal evidence is a single 17-person crossover study that moved beta-amyloid but not tau, the other core Alzheimer's protein,^[13] and carrying the APOE ε4 risk gene predicted faster loss of deep sleep — so early disease may be eroding the sleep rather than the reverse. Total sleep duration is the better-supported target than deep sleep specifically. See Sleep architecture.
• The shingles vaccine. The most surprising recent finding in the field. A natural experiment in Wales, where a birthdate cut-off made the vaccine abruptly available to people born on or after 2 September 1933 but not to those born a week earlier, found new dementia diagnoses over seven years fell by 3.5 percentage points in absolute terms — about a fifth in relative terms.^[14] The birthdate design mimics randomisation and sidesteps the healthy-vaccinee bias that plagues ordinary observational work, and an independent Australian analysis found the same direction.^[15] The evidence rests on the older live-attenuated vaccine; data for the now-standard recombinant one are still accruing.

Weak / preliminary

• MIND-diet adherence for dementia specifically. The observational signal is large — roughly half the Alzheimer's incidence in the top third of adherence, in a cohort of 923 older adults^[16] — but the only randomised trial of the pattern, in 604 at-risk older adults over three years, was a clean null on cognition and brain imaging.^[17] Follow the pattern on cardiovascular grounds; the dementia-specific claim is not established.
• Ultra-processed food and the brain. A 2026 systematic review in BMJ Nutrition, Prevention & Health found three of five major cohorts linked ultra-processed-food intake to accelerated cognitive decline and higher dementia risk even after adjusting for adherence to an otherwise healthy dietary pattern — a hint of a processing-specific effect on top of nutrient quality, from observational data only.^[18] See Ultra-processed food.
• Discretionary cognitive activity — apps, puzzles, taking up a language late. Small, inconsistent, badly confounded.
• Cognitive supplements — omega-3, B vitamins, ginkgo, curcumin. Null or inconsistent on cognitive endpoints; ginkgo is negative in two large prevention trials. See Dementia prevention.

One confound governs almost all the observational evidence above, and is worth naming once. The silent early phase of dementia depresses engagement for years before diagnosis. In 28-year follow-up of 10,308 British civil servants, physical activity began declining up to nine years before diagnosis, and across the full follow-up there was no association between activity and dementia at all — the authors concluded that earlier positive findings likely reflected reverse causation rather than protection.^[19] Any cohort followed for less than that silent window is exposed to the same problem.

Why the brain is downstream of the heart

The single most consequential idea here is that the brain is not a separate organ. The same vascular and metabolic biology that drives heart attacks, strokes and type 2 diabetes drives the slow grind of cognitive decline. The cochlea is one of the most metabolically demanding microvascular structures in the body — which is why hearing loss tracks the same biology as cardiovascular disease and now reads as a biomarker of cochlear blood-supply health as much as an isolated sensory deficit.^[20] The retina is the only place in the body where blood vessels can be photographed without cutting anyone open, which is why a routine photograph of the back of the eye is turning into a research-grade systemic-aging read-out.

The implication is that the highest-leverage brain interventions are mostly not brain-specific. Treating blood pressure and cholesterol, defending cardiorespiratory fitness, sleeping regularly, eating a vegetable-and-fish-heavy pattern and not smoking dominate any cognitive supplement on offer. The brain-specific actions that do matter — treating hearing loss, correcting visual loss, building reserve through demanding new learning — are not substitutes for cardiovascular care; they stack on top of it. The integrated case is collected under Dementia prevention.

Hearing: the largest single modifiable midlife lever — Moderate

Hearing loss has been quietly reclassified over the last decade. It is no longer an isolated sensory deficit — it is a systemic biomarker of cochlear blood-supply health, a measurable driver of accelerated brain atrophy, and, tied with high LDL cholesterol at a rounded 7% share, the largest modifiable midlife dementia risk factor.^[21]

The interventional evidence is more equivocal than the headlines suggest. ACHIEVE, the only randomised trial, enrolled 977 older adults from two very different cohorts — an at-risk group drawn from the Atherosclerosis Risk in Communities (ARIC) study and healthier de-novo volunteers — and missed its primary endpoint: hearing intervention did not slow cognitive decline overall. The much-quoted 48% reduction is a subgroup finding confined to the higher-risk ARIC participants, who were older and declining faster — hypothesis-generating, not established.^[22] The 2026 ASPREE analysis is a target-trial emulation — an observational design built to imitate a trial, not a trial — in 2,777 adults aged 70 and over with moderate hearing loss. Over seven years, dementia occurred in 5.0% of those prescribed hearing aids against 7.5% of those not: about a third lower in relative terms, or roughly 25 fewer dementia diagnoses per 1,000 people over seven years, with a clean dose-response by how often the aids were actually worn.^[23]

So: a null trial with a promising subgroup, plus a consistent observational signal. Good reason to treat hearing loss — worth doing on its own merits anyway — but not proof that hearing aids prevent dementia. Two caveats travel with the ASPREE result: the benefit shows up as crossing the diagnostic threshold rather than as sharper day-to-day thinking, and the device that lives in a drawer is not the intervention.

The systemic signal is broader than dementia. A 2025 meta-analysis of cohort studies found adults with hearing loss die at a modestly higher rate — about 21% higher all-cause mortality (hazard ratio 1.21, 95% confidence interval 1.13–1.31), roughly 22% higher cardiovascular and 11% higher cancer mortality. That bracketed range is where the true effect most plausibly lies; when it sits entirely above 1.0, as here, chance alone is an unlikely explanation. Loss measured on an audiogram carries a stronger signal (about 28% higher) than self-reported loss — the test picks up physiology people don't notice.^[24] Hearing-aid use is itself independently associated with lower long-term mortality in 9,885 US adults followed a median 10.4 years.^[25]

The diagnostic shift matters too. Standard pure-tone audiometry caps at 8 kHz, but extended high-frequency audiometry, up to 16 kHz, catches damage years earlier. The 2026 STOP (Swedish Tinnitus Outreach Project) analysis attributed roughly 64% of the age-related variance in real-world speech-in-noise understanding to extended high-frequency thresholds, against 16% for the standard test — striking, but from a single 263-adult analysis that is still a non-peer-reviewed preprint, so treat the numbers as provisional.^[26] The advice survives the caveat: if you struggle in restaurants but pass a standard audiogram, that is the test to ask for, and the 2026 American Academy of Otolaryngology–Head and Neck Surgery Foundation guideline now recommends the workup.^[27]

Hearing covers the ACHIEVE and ASPREE designs in detail, the extended high-frequency case, the noise × air-pollution synergy, the diabetes-and-hearing crosswalk nobody screens for, ototoxic chemicals, the cochlear biology, and why gene therapy will not reach age-related hearing loss any time soon.

Vision: correct the deficit first, admire the frontier second — Moderate

Untreated visual impairment joined the modifiable list in 2024, as one of the smaller levers at about 2% of cases.^[28] The dementia link appears to run mostly through functional vision loss rather than shared pathology — blindness substantially raises Alzheimer's risk, while non-exudative (dry) age-related macular degeneration without severe impairment does not — though causality remains unresolved, with shared neuropathology, sensory deprivation and reverse causation all still plausible.^[29] The practical implication is identical to hearing: treat the correctable deficit early.

The unglamorous conditions carry the strongest evidence. Cataract is the leading cause of blindness worldwide, and its surgery is one of the highest-value operations in medicine: first-eye surgery in women over 70 cut the falling rate by about a third in a randomised trial.^[30] Observational data also link cataract extraction to roughly 29% lower dementia risk over the following decade — with glaucoma surgery, which does not restore vision, showing no such association, an internal comparison that supports a vision-specific pathway, though the finding stays open to healthy-patient selection bias.^[31] Glaucoma is silent until it is advanced and the vision it takes does not come back; the LiGHT trial randomised more than 700 people and established an outpatient laser as a reasonable first-line alternative to daily drops, with about 70% of laser-first eyes still drop-free at six years.^[32] And for dry eye, the most popular remedy fails: the large DREAM trial found 3,000 mg/day of fish oil no better than placebo.^[33] None of this needs a frontier scan — it needs a comprehensive dilated exam on a sensible cadence.

The frontier is genuinely interesting and genuinely not ready (Weak / preliminary). Deep-learning models read photographs of the back of the eye to produce a Retinal Age Gap — the difference between AI-predicted retinal age and chronological age. In a UK Biobank analysis of 35,913 adults, each additional year of gap was associated with about 2% higher all-cause mortality, with steeper increases at the upper quartiles for causes other than cardiovascular disease and cancer.^[34] A related model, RetiAGE, predicts ten-year cause-specific mortality from the same images: people in the top quarter of the gap had roughly two and a half times the cardiovascular mortality and about 1.6 times the cancer mortality of those in the bottom quarter.^[35] In the studies published so far these models have predicted mortality about as well as some DNA-methylation clocks, from a 30-second clinic photo — but head-to-head validation is limited, no professional body uses them for risk stratification, and no eye exam will print you one.^[36]

Two other things belong on the practical list. Macular pigment — lutein and zeaxanthin, concentrated at the centre of the retina — filters short-wavelength light and improves visual performance, and leafy-green-heavy diets raise it; the food route has no downside, while the supplement failed the decisive cognitive trial.^[37] And the 670 nm deep-red light finding is thinner than its coverage: brief morning exposure improved a colour-contrast surrogate in small pilots, but a 12-month controlled study found no structural benefit at all — no change in drusen or photoreceptor-layer thickness — so confidence should stay low regardless of what device you buy.^[38]

Vision covers cataract, glaucoma, dry eye and myopia in full, the Retinal Age Gap models and their limits, macular pigment and the AREDS2 evidence, the 670 nm data, the macular-degeneration-and-dementia mediation, and why blue-light glasses move no trial endpoint.

Cognitive engagement: build reserve, and don't write off people — Moderate for reserve, Weak for the products

The mechanism behind the education signal is the cognitive-reserve hypothesis: autopsy studies routinely find brains with extensive Alzheimer's-type pathology in people who showed no symptoms in life, and minimal pathology in people who clearly had dementia. The brain holds redundant capacity between damage and observable decline, and lifelong intellectual challenge builds it. A useful shorthand for what that protects is brainspan — the years during which neural networks stay efficient enough to support agency and coherent self-regulation. It is a recent coinage rather than settled vocabulary, but it names a real gap: a body can keep functioning while the brain has lost the bandwidth that defines a person.^[39]

Reserve buys time, not immunity. Once compensatory networks finally fail, decline tends to be faster in people who had more reserve, because more damage had to accumulate before symptoms broke through. The net effect on quality of life is still strongly positive — more years of full function — but it is a delay, not a cure.

The bad news side is firm. In 2016 the US Federal Trade Commission fined the makers of Lumosity $2 million for deceptive advertising, finding the company lacked competent evidence for its claims.^[40] More than 70 cognitive scientists convened by the Stanford Center on Longevity and the Max Planck Institute concluded there is little evidence that brain games improve broad cognitive abilities or help anyone navigate everyday life better.^[41] The field is not unanimous — a rival letter drew 133 signatures, several from scientists with commercial ties — but the independent adjudication and Cochrane both favour the sceptical reading: computerised training produces small immediate benefits of uncertain clinical importance, with no evidence the effect on global cognition persists at twelve months.^[42] The underlying problem is the failure of "far transfer" — people reliably get better at the game they play, and the gains stop there.

The most-quoted exception is narrower than advertised. The ACTIVE trial randomised 2,832 cognitively healthy adults aged 65 and over to memory, reasoning, visual speed-of-processing training, or a no-contact control; booster sessions were later offered to a randomly chosen 60% of trained participants.^[43] Twenty years on, the speed-of-processing arm showed about 25% lower dementia incidence — but only among those who also did the boosters. Without boosters there was no benefit, and none of the pre-specified whole-arm comparisons against control reached statistical significance.^[44] At ten years, the durable gains were confined to the specific ability each arm had trained.^[45] The trial also carries the pillar's only harm signal for a cognitive intervention: among participants with 12 years of education or less, memory and speed training were associated with higher 20-year mortality.^[46]

Don't over-correct into "socialising doesn't count." One short trial found that learning demanding new skills improved episodic memory where socialising about familiar topics did not — but across the wider literature social connection is the strongest late-life reserve marker there is, well ahead of cognitive activity, and randomised social interventions improve global cognition.^[47] The two are not substitutes. The option with support from both directions is combining them: group classes, ensemble music, collaborative crafts, conversation circles.

Two popular claims deserve less confidence than they get. Crosswords beat a purpose-built computerised trainer in one head-to-head trial — but the 107 participants already had mild cognitive impairment, both arms received an active intervention so there was no do-nothing comparison, and the primary brain-imaging outcome was null.^[48] And the familiar "bilingualism delays dementia by four to five years" figure traces to a proponents' commentary: the best prospective meta-analysis found a 4% difference well inside the range expected from chance,^[49] and a 2026 cohort found no advantage in any cognitive domain.^[50] Learn a language for its own sake, not as insurance.

Cognitive engagement covers the reserve mechanism and its life-course staging, the ACTIVE follow-ups in full, the brain-training evidence and the fight around it, occupational complexity, retirement as a cognitive risk transition, and the dose-and-novelty question.

Dementia prevention: the integrative pillar — Moderate for the bundle

Putting the 14 factors into practice is the job of Dementia prevention. The strongest multi-domain evidence comes from FINGER, a Finnish trial that combined diet, exercise, cognitive training and vascular-risk monitoring in 1,260 at-risk older adults and improved cognition over two years — though the between-group difference was small.^[51] Its American successor U.S. POINTER randomised 2,111 adults to a structured versus a self-guided version of the same programme: the structured arm did better on global cognition by about 0.03 standard deviations per year, both arms improved, the advantage was confined to executive function, and memory — the domain that matters most for dementia — showed essentially nothing. There was no do-nothing arm, and the trial was funded and staffed by the Alzheimer's Association, with the structured arm using a commercial brain-training product.^[52] The benefit was statistically indistinguishable between carriers and non-carriers of the APOE ε4 risk gene — which is the practical answer to a family history: the levers look the same either way.

Two things people ask about belong here rather than on any checklist. Anti-amyloid antibodies (lecanemab, donanemab) produce modest disease-slowing in early symptomatic Alzheimer's and are not a prevention strategy for healthy adults: over 18 months lecanemab reduced decline by 0.45 points on an 18-point clinical rating scale — about 27% relative slowing on a score that worsened 1.66 points on placebo — with brain swelling or microbleeds in about 13% of lecanemab and 24% of donanemab recipients (Caution for the harm profile).^[53] And purpose is a real observational association with a specific window: in a 28-year prospective study of 4,632 adults, purpose measured at 52 predicted better cognition but not lower dementia, while purpose maintained through ages 63 to 70 predicted lower dementia at 80, with modest effect sizes — and a steeper decline between 52 and 70 went the other way.^[54] Late midlife — retirement, empty nest, role loss, early widowhood — is where it is won or lost. See Purpose.

The integrated message: cardiovascular risk reduction is the engine of cognitive protection, sensory and social levers stack on top, and the actions are concrete enough to put on a list.

Practical brain-health checklist

1. Treat hearing loss. Baseline audiometry between 40 and 50 per the 2026 guideline, sooner with any concern. If you struggle in noisy rooms but pass a standard audiogram, ask for extended high-frequency audiometry. Wear the aids you're prescribed. See Hearing.
2. Get a comprehensive dilated eye exam by midlife and treat any correctable loss — cataract surgery, glasses, low-vision rehabilitation, macular-degeneration management. Eat leafy greens daily for macular lutein and zeaxanthin: the pattern, not the pill. See Vision.
3. Treat blood pressure and cholesterol in midlife. The Commission's target is systolic below 130 mmHg from age 40 if tolerated — stricter than routine practice, looser than the below-120 that SPRINT-MIND tested, and justified by cardiovascular gains plus the cognitive-impairment signal rather than a proven dementia effect. High LDL cholesterol ties hearing loss for the largest share of cases. See Blood pressure and Lipid management.
4. Stay metabolically healthy. Avoid type 2 diabetes and treat insulin resistance; well-controlled diabetes is a different thing from uncontrolled for cognitive outcomes. Track HbA1c (the three-month blood-sugar average), a lipid panel and waist circumference in midlife rather than waiting.
5. Train both ways, and don't sit through the day. Aerobic exercise is the modality with a signal on the volume of the hippocampus, the brain's main memory structure — from a single small trial, so read it as suggestive;^[55] for resistance training, one 12-month trial in older women improved a single executive-function measure.^[56] Prolonged sitting tracks brain atrophy independent of whether you meet the exercise guidelines, and about 7,000 steps a day is associated with 38% lower dementia incidence than 2,000.^[57] See Exercise and Sitting.
6. Sleep 7–8 hours on a regular schedule, and screen for sleep apnea. Treating obstructive sleep apnea improves cognitive trajectory in some studies, and it is heavily underdiagnosed. See Sleep.
7. Eat MIND-pattern. Leafy greens daily, berries twice a week or more, fish, nuts, olive oil, whole grains, beans, poultry. Limit red and processed meat, butter, fried foods, sweets and ultra-processed food. Do it for the cardiovascular evidence; the dementia-specific claim rests on cohorts, not the trial. See Dietary patterns.
8. Stay socially connected. The strongest late-life reserve marker measured, ahead of cognitive activity — real relationships, regular contact, groups you actually turn up to.
9. Keep learning things that are genuinely hard. Languages, instruments, demanding crafts, complex bodies of knowledge — ideally with other people, which combines the two strongest signals. Rotate as a skill becomes automatic. See Cognitive engagement.
10. Treat depression, at any age. It carries the highest individual risk of the 14 factors, and it is both a risk factor and an early symptom — a reason to treat it, not to dismiss it.
11. Defend purpose through the late-midlife transition. Plan the next role before the current one ends. See Purpose.
12. Don't smoke; drink minimally; reduce air-pollution exposure where feasible — filtration in homes near major roads is a small, plausible win. See Environmental toxins.
13. Wear seatbelts and helmets; avoid repeated head impacts from contact sports.
14. Ask about the shingles vaccine at the age it's offered. Quasi-experimental evidence, not a randomised trial — but the cleanest natural experiment in the field.

What's overrated

• Commercial "brain training" apps as a substitute for demanding real-world learning. Small immediate gains, gone within twelve months, no far transfer.
• Cognitive supplements as standalone interventions — omega-3, B vitamins, ginkgo, curcumin. Mostly null or inconsistent on cognitive endpoints. The DO-HEALTH trial is sometimes cited the other way, which gets it backwards: that trial missed all six of its pre-specified primary endpoints, cognition among them; what moved was a post-hoc analysis of methylation aging clocks in 777 of its 2,157 participants — surrogate markers, not cognition and not dementia.^[58] See Omega-3.
• Blue-light-blocking glasses for sleep or eye strain. No movement on any relevant trial endpoint.
• Crosswords and bilingualism as prevention. Both are fine things to do. Neither has the evidence its reputation implies.
• Anti-amyloid antibodies as prevention. A treatment for early symptomatic disease with a real harm profile, not something a healthy adult should be seeking.
• Treating cognitive decline as inevitable past a certain age. The late-life share of cognitive reserve is as large as the early-life share.
• Single-supplement "anti-Alzheimer's" stacks. No supplement combination has any dementia-endpoint evidence at all; the risk-factor levers on this page at least have a coherent map behind them.

---

For the full ordered action list drawn from every pillar, see the healthspan long list.