Preventing Cognitive Decline

Dementia is not ordinary forgetfulness and it is not an inevitable part of ageing. It is an umbrella term for impairment severe enough to interfere with daily life, caused by several diseases and injuries; mixed causes are common.¹

“Prevention” sounds binary. The evidence supports a narrower promise: across populations, changing some exposures and treating some conditions may reduce average risk or delay onset. Age, genes, chance, biology, opportunity, environment and access to care still matter. A diagnosis is never proof of personal failure.

The World Health Organization’s 2026 guideline recommends action across health behaviours, medical conditions, social and cognitive activity, hearing and environmental exposure, while repeatedly grading uncertainty and distinguishing strong from conditional advice.² The 2024 Lancet Commission estimated that 14 potentially modifiable factors collectively account for about 45% of dementia cases worldwide.³

Normal ageing can include slower recall. Persistent change that disrupts finances, medication, navigation, language, safety or familiar tasks deserves assessment. Sudden confusion is urgent: infection, stroke, medication effects and other treatable problems can resemble decline. Evaluation can identify contributors and open support and planning.⁴

Think in life courses, not miracle ingredients: learning, vascular health, sensory access, relationships, environment and injury protection interact.

The stages are not rigid. WHO’s evidence review also shows why a neat risk map is not a map of equal proof: recommendations range from strong to conditional, several with low or very-low certainty.⁵

No one can optimise away deprivation, unsafe work or inaccessible care. When recommendations compete for limited resources, prioritise the clearest immediate benefit and clinical need.

The brain depends on circulation, metabolism, sensation and safety. Start with established needs, not a proprietary “neuroprotective” stack.

Newer genetic and longitudinal analyses offer no sound reason to prescribe alcohol for cognitive health. Apparent benefits of light drinking can arise when former drinkers are grouped with abstainers or when early illness causes people to drink less; harmful use clearly carries risks.⁶

In ACHIEVE, a hearing intervention did not slow three-year cognitive decline across all 977 adults. A prespecified higher-risk cardiovascular subgroup showed 48% less decline; healthier volunteers did not.⁷ That subgroup result is encouraging, not universal. Treat hearing loss because communication and participation matter now; regard risk reduction as possible.

Physical activity improves health on many fronts. Its dementia-specific evidence is less precise than headlines imply.

For general health, WHO advises 150–300 minutes of moderate aerobic activity a week, or 75–150 vigorous, or an equivalent mix, plus strength work on at least two days. Older adults should add varied balance-and-strength activity on at least three days when able. Any activity is better than none.⁸ These are general-health targets, not a proven dementia-prevention dose.

A 2024 analysis of 104 longitudinal studies found only a very small association between activity and later impairment after bias correction, no dose–response, and no significant association beyond ten years.⁹ A 2023 umbrella review of randomised trials found that apparent cognitive benefits became negligible after adjusting for active controls, baseline differences and publication bias.¹⁰ Other syntheses find modest domain gains: “may help” is reasonable; “reverses decline” is not.

EXERT randomised 296 sedentary adults with amnestic MCI to aerobic exercise or low-intensity stretching, balance and mobility—not an aerobic-plus-resistance package. Cognition was stable in both groups at 12 months, without a significant difference. A nonrandomised outside comparison cannot prove prevention.¹²

Movement still has a compelling safety and function story. A Cochrane review of 108 trials found exercise reduced falls in community-dwelling older people, particularly through balance and functional programmes.¹³ Start below your ceiling and seek guidance for unstable symptoms, significant pain, falls or complex conditions.

Plant-forward, minimally processed food supports general health. Mediterranean and MIND patterns are templates, not proven dementia shields.

A 2025 meta-analysis of 23 observational studies associated higher Mediterranean-diet adherence with lower risks of cognitive impairment, dementia and Alzheimer’s disease.¹⁴ Adherents may also differ in education, income, activity, smoking, care and early disease. Adjustment cannot remove every difference.

In the largest dedicated MIND trial, 604 at-risk adults received MIND plus mild calorie restriction or an equally intensive control diet. After three years, cognition and MRI outcomes did not differ significantly.¹⁵ A PREDIMED substudy found suggestive cognitive benefits from Mediterranean diets plus olive oil or nuts, but involved 447 people at one site and did not show fewer dementia cases.¹⁶

Capsules do not inherit the associations of a whole diet. In VITAL-Cog, 1 gram a day of marine omega-3 did not improve cognition over two to three years.¹⁷ COSMOS found a small average multivitamin advantage—about 0.07 standard deviations for global cognition across three substudies from one parent trial—but no demonstrated reduction in MCI or dementia. The popular “two years younger” description is a model-based comparison with age-related test differences, not rejuvenation or extra disease-free years.¹⁸

Correcting a documented deficiency is medical care, not a prevention stack. Supplements can interact or harm. Ask whether the exact product changed a meaningful clinical outcome and was independently replicated.

Training can improve the ability practised. Transfer to everyday cognition or dementia risk is less certain.

For people with MCI, the American Academy of Neurology concluded that six months of exercise is likely to improve cognitive measures and cognitive training may improve them; it found no high-quality evidence supporting drug treatment at the time of its guideline.¹⁹ WHO 2026 similarly makes conditional recommendations for cognitive training and stimulation, with low or very-low certainty and no established optimal format or dose.

ACTIVE enrolled 2,832 independent older adults. Ten sessions produced gains mainly in the abilities trained; reasoning and speed effects remained at ten years, while memory-test effects did not. Training groups reported less difficulty with instrumental daily activities, although attrition and self-report complicate interpretation.²⁰

A 2026 follow-up linked Medicare claims for 2,021 participants. No randomised arm had a significant main effect on diagnosed dementia. Speed trainees who completed a later booster had fewer diagnoses, but booster completion was post-randomisation and may mark healthier, more engaged people.²¹ This is a signal for one speed protocol, not proof for brain games generally.

Broad reviews find reliable improvement on practised tasks but weak or inconsistent “far transfer” to unrelated abilities and daily life.²² Commercial claims can outrun even that evidence: the US Federal Trade Commission settled charges that Lumosity had made unsupported claims about reducing or delaying age-related cognitive impairment and other conditions.²³

Sleep, mood, stress and connection affect thinking today. Their relationship with later dementia is complex and may run both ways.

In the Whitehall II cohort, persistent sleep of six hours or less in midlife was associated with later dementia.²⁴ This is observational, not proof that longer sleep prevents disease. Clearance is a plausible research pathway; “eight hours clears amyloid” is not a human treatment claim.

Loud snoring, witnessed pauses or gasping, marked daytime sleepiness and morning headaches can signal obstructive sleep apnoea and deserve assessment.²⁵ Treat apnoea for established health reasons; prevention remains uncertain. Review pain, medicines, alcohol, shift work, caregiving and environment rather than blaming discipline.

Mindfulness, relaxation, time outdoors or biofeedback may help some people with stress. A meta-analysis in subjective cognitive decline or MCI found short-term improvements in anxiety, perceived stress and quality of life, but not stress biomarkers—and it did not establish dementia prevention.²⁶ Claims that mindfulness protects dendritic spines by lowering cortisol turn a possible mechanism into a clinical certainty.

Combining sensible targets is pragmatic. Bundled trials cannot reveal which ingredient mattered or prove synergy without a factorial design.

A 2021 synthesis of multidomain trials found a small cognitive-test advantage but no significant reduction in incident dementia.³² FINGER follow-up reported a randomised difference through year seven; later engagement comparisons were observational and vulnerable to healthy-adherer and attrition bias.³³ As of this review, these programmes have not established that they prevent dementia.

A mechanism justifies study; it does not establish a durable human benefit.

Watch verbs cross the laboratory-to-headline gap: “may influence” becomes “boosts,” and association becomes protection. Ask: In whom? Compared with what? Measured how? For how long?

Make the next useful action safe and repeatable, then add support where barriers are greatest.

1. Clarify the starting pointList current conditions, medicines, symptoms, falls, sensory needs, food access, movement capacity, sleep and what already brings meaning.
2. Choose one anchorSelect the change with the clearest immediate value: a health appointment, tobacco support, a walk, a meal pattern, hearing care or a class.
3. Add structurePut it in the calendar, remove one obstacle, invite a partner, arrange transport or professional guidance, and define the smallest viable version.
4. Review and extendAfter several weeks, assess safety, consistency and daily benefit. Keep, adapt or replace it; then add the next domain if capacity allows.

A five-minute start may suit someone rebuilding after illness. Malnutrition may require energy and protein, not restriction; loneliness may require transport or hearing support before a group. Personalisation changes the route.

Before changing behaviour or spending money, identify the design, comparator, outcome and absolute difference.

Ask who was missing. Results from resourced volunteers may not transfer to people facing frailty, poverty, disability or caregiving. Access is part of the intervention.

There is no neuroplasticity master switch or recipe that makes dementia impossible. Protect vascular and metabolic health, preserve movement, eat nourishing food, avoid tobacco and harmful alcohol, address hearing and other needs, stay engaged, and build environments that make those actions possible.

Name the outcome. Healthy actions improve present wellbeing; some improve specific tests; multidomain programmes yield small composite-score gains in selected groups. None proves rejuvenation, guaranteed independence or dementia prevention.

A useful plan begins with the most important treatable need, makes one sustainable change, adds support and measures what matters in life. It leaves room for disability, culture, money, caregiving, enjoyment and uncertainty. It also leaves blame behind.

1. WHO dementia fact sheet.
2. Risk reduction of cognitive decline and dementia: WHO guidelines, second edition.
3. Dementia prevention, intervention, and care: 2024 Lancet Commission.
4. What is dementia? Symptoms, types and diagnosis.
5. WHO recommendations and supporting evidence.
6. Alcohol use and dementia risk: cohort and genetic analyses.
7. ACHIEVE hearing intervention randomised trial.
8. WHO guidelines on physical activity and sedentary behaviour.
9. Physical activity and cognitive decline: systematic review and meta-analysis.
10. Exercise and cognition: umbrella review of randomised trials.
11. Aerobic exercise and hippocampal volume: meta-analysis.
12. EXERT randomised trial in adults with MCI.
13. Exercise for preventing falls in community-dwelling older people.
14. Mediterranean diet and age-related cognitive disorders: meta-analysis.
15. Trial of the MIND diet for prevention of cognitive decline.
16. Mediterranean diet and cognitive function: PREDIMED substudy.
17. VITAL-Cog omega-3 randomised trial.
18. COSMOS multivitamin supplementation and cognition.
19. Practice guideline update: mild cognitive impairment.
20. Ten-year effects of the ACTIVE cognitive training trial.
21. ACTIVE claims-based dementia outcomes over 20 years.
22. Do “brain-training” programmes work? Review.
23. FTC action on unsupported Lumosity claims.
24. Sleep duration in midlife and incident dementia.
25. Sleep apnoea: symptoms and diagnosis.
26. Mindfulness in subjective cognitive decline or MCI: meta-analysis.
27. FINGER multidomain intervention randomised trial.
28. US POINTER randomised clinical trial.
29. LatAm-FINGERS randomised clinical trial.
30. MAPT multidomain and omega-3 trial.
31. preDIVA vascular-care cluster randomised trial.
32. Multidomain interventions for dementia prevention: systematic review.
33. FINGER cognition and engagement through long-term follow-up.