Altamens

Altamens Research · Evidence review

What Actually Supports Cognitive Health as We Age?

An evidence map of medical risk management, movement, hearing, social connection and cognitive training — plus what brain games can and cannot claim.

Altamens Research12 min read

Abstract

This review asks which modifiable interventions have the strongest evidence for supporting cognitive health in older adults, and what can responsibly be said about cognitive training. Rather than a single headline claim, we organise the evidence into a ladder: clinician-managed medical risks, foundational everyday behaviours, social and cognitively meaningful activity, and targeted cognitive practice.

Across sources, the most defensible position is not that brain games prevent dementia. It is that cognitive health is influenced by many factors, that a subset of medical risks carries genuinely strong randomized evidence, and that cognitive training can improve practiced or closely related abilities by small amounts. Transfer to everyday function is limited and inconsistent. We label each item by evidence type — guideline, randomized trial, meta-analysis or observational — and keep improving a trained skill separate from preventing disease.

Key findings

  • Cognitive health is shaped by vascular, metabolic, sensory, behavioural, environmental and social factors — not by any single product or activity.
  • The WHO's 2026 guideline attributes up to 45% of dementia risk at the population level to modifiable factors; this is not a promise for any individual.
  • Intensive blood-pressure control has unusually strong randomized evidence for reducing mild cognitive impairment (SPRINT MIND).
  • Cognitive training produces small average gains on cognitive tests; transfer to everyday function varies and durability is not established.
  • Commercial brain-training apps have not been shown to reproduce the outcomes of the structured ACTIVE-trial protocols.

1The research question

Older adults, and the people who care for them, are surrounded by claims about how to keep the mind sharp. Many of those claims collapse the distinction between two very different ideas: getting better at a task you practice, and lowering your risk of dementia. This review asks a narrower, answerable question — which modifiable interventions have the strongest evidence for supporting cognitive health in older adults, and what can responsibly be said about cognitive training?

The most defensible public resource is not a claim that brain games prevent dementia. It is an evidence ladder showing that cognitive health is shaped by vascular, metabolic, sensory, behavioural, environmental and social factors. Cognitive training can improve practiced or closely related abilities by small amounts, but transfer to everyday function varies, and commercial games have not been shown to reproduce the outcomes of structured clinical protocols.1

2How to read this evidence

Not all evidence is equal, and honest communication depends on saying which kind you are relying on. Throughout, each intervention is labelled by evidence type:

  • Guideline — a formal recommendation from a body such as the WHO, synthesising many studies.
  • Randomized trial — participants are randomly assigned, the strongest design for causal claims.
  • Meta-analysis — a statistical pooling of many studies, only as strong as the trials inside it.
  • Observational evidence — associations seen in populations, which cannot by themselves prove cause.

3Tier A — Clinician-managed medical risks

The interventions with the strongest evidence are largely medical, and they are decisions to make with a clinician rather than to self-administer. Blood pressure is the clearest example.

Blood-pressure treatment has unusually strong randomized evidence. In SPRINT MIND, an intensive systolic target below 120 mmHg versus below 140 mmHg reduced mild cognitive impairment and the combined outcome of mild cognitive impairment or probable dementia. The probable-dementia endpoint alone was not statistically significant.7

Figure 1SPRINT MIND: intensive vs standard blood-pressure control
Hazard ratio · null at 1.0 · hover a row for focus
Probable dementia (alone)0.83 (0.67–1.04)· ns
Mild cognitive impairment0.81 (0.69–0.95)
Combined MCI / probable dementia0.85 (0.74–0.97)
0.630.740.850.971.08
← Favours intensive controlFavours standard →

Hazard ratios below 1.0 favour intensive control. The mild cognitive impairment and combined outcomes reached significance; probable dementia alone did not, because its confidence interval crosses 1.0.

Diabetes, high cholesterol and depression sit in the same tier: they appear as modifiable risk factors in the guideline evidence, and they are conditions managed with, and by, clinicians. The practical takeaway is not a supplement or a gadget but attending to these conditions with professional care.

Sensory health belongs here too, and the hearing evidence is genuinely nuanced. ACHIEVE found no significant difference in three-year cognitive decline in the full randomized cohort. A prespecified analysis suggested different effects in a higher-risk sub-population, so the result supports risk-based investigation rather than a universal claim that hearing aids prevent decline.8

4Tier B — Foundational behaviours

Alongside medical care, national guidance converges on a familiar set of behaviours. The NIA advises managing chronic conditions, treating hearing and vision loss, sleeping seven to nine hours, avoiding smoking, limiting alcohol, eating a balanced diet, staying physically active, and remaining socially and cognitively engaged.1

  • Physical activity — consistently associated with slower cognitive decline, though much of this evidence is observational.
  • Sleep — seven to nine hours is the commonly advised range for older adults.
  • Not smoking and limiting alcohol — both feature as modifiable risk factors across the guideline and Commission evidence.
  • A balanced diet — dietary patterns are linked to cardiometabolic health, which in turn relates to cognitive health.

5Tier C — Social and cognitively meaningful activity

Social isolation is one of the modifiable risk factors in the life-course model, and the WHO's 2026 guidance includes social engagement, cognitive stimulation and cognitive training among its risk-reduction recommendations.2,3

The most conservative reading is that staying socially connected and cognitively engaged is a reasonable, low-risk part of healthy ageing that appears in formal guidance — while remembering that much of the underlying data is observational and cannot, on its own, establish cause.

6Tier D — Targeted cognitive practice

Computerized cognitive training produces modest average gains on cognitive tests. A meta-analysis of 52 randomized trials in 4,885 healthy older adults found a small overall effect, with variation by domain and programme design; durability was not established.5 A larger 215-study meta-analysis reported an overall effect, with significant near transfer and smaller far transfer, and found that multicomponent programmes were more consistently associated with transfer than narrow single-task practice.6

A 20-year follow-up of the ACTIVE trial added a further caution about long-term claims. Speed-of-processing training with booster sessions was associated with fewer claims-based dementia diagnoses over 20 years (HR 0.75, 95% CI 0.59–0.95); no benefit appeared without boosters, and memory or reasoning training showed no main effect.9 A subsequent reanalysis questioned the result after correcting for multiple comparisons, so it remains promising rather than proof that commercial brain games prevent dementia.10

Figure 2Cognitive-training effect sizes (Hedges' g)
Overall · Lampit 2014g = 0.22 (0.15–0.29)
Overall · Basak 2020g = 0.28
Near transfer · Basak 2020g = 0.37
Far transfer · Basak 2020g = 0.22
0.000.200.350.50

By convention ≈ 0.2 is a small effect, ≈ 0.5 medium, ≈ 0.8 large. Wide intervals mean greater uncertainty.

Effect sizes are small. These come from different meta-analyses with different populations and methods and should not be read as directly interchangeable. 'Near transfer' means gains on closely related tasks; 'far transfer' means gains on more distant abilities.

7The 14-factor life-course model

The 2024 Lancet Commission identifies 14 potentially modifiable factors across the life course. High LDL cholesterol and untreated vision loss were newly added in 2024. The Commission is explicit that these estimates describe the population and do not guarantee that any individual will avoid dementia.4

Figure 3Fourteen modifiable factors across the life course

Early life · 1

Midlife · 10

Later life · 3

Population-attributable risk. Population-attributable risk: the share of cases that might be avoided if a factor were removed across a whole population — not the risk change for one person.

Grouped by the life stage at which each factor is emphasised. Two factors — high LDL cholesterol and untreated vision loss — were new in the 2024 update. These are population-attributable estimates, not individual predictions.

8The cognitive health evidence ladder

Pulling the tiers together, the table below is the practical asset: each action labelled by evidence type, measured outcome, direction, confidence and the caveat that keeps it honest. Read top to bottom as strongest-evidence medical care first, then behaviours, then social and cognitive activity.

Figure 4The cognitive health evidence ladder for older adults

Tier A · Clinician-managed medical risks

Blood-pressure controlModerate–high
Randomized trialMCI; combined MCI / probable dementiaReduced risk

Probable-dementia endpoint alone not significant; targets are clinical decisions.

Diabetes, cholesterol & depression careModerate
GuidelineDementia risk factorsManaged risk

Condition management is individualised medical care.

Hearing interventionMixed
Randomized trial3-year cognitive declineNo overall effect; possible benefit in higher-risk

Prespecified subgroup, not a universal claim.

Vision careModerate
ObservationalDementia risk (new 2024 factor)Associated lower risk

Observational; treat vision concerns with a clinician.

Tier B · Foundational behaviours

Physical activityModerate
Guideline / observationalCognitive declineAssociated lower risk

Largely observational; assumes partial causality.

Sleep, not smoking, limited alcohol, balanced dietModerate
GuidelineOverall brain & cardiometabolic healthSupportive

Foundational behaviours; not a targeted treatment.

Tier C · Social & cognitive activity

Social & cognitively engaging activityModerate
Guideline / observationalRisk reduction (WHO recommendation)Supportive

Recommended, but underlying data is largely observational.

Tier D · Targeted cognitive practice

Structured cognitive trainingModerate
Meta-analysisTrained & related cognitive testsSmall gain

Transfer varies; durability not established.

Commercial brain-training appsLow
Guideline cautionEveryday cognitionNot established

Not shown to match ACTIVE-trial outcomes (NIA).

Confidence reflects the strength and consistency of the evidence for the stated outcome, not a promise of benefit for any individual.

9Limitations

  • Many risk-factor estimates are observational and assume at least partial causality.
  • The evidence base is weighted toward high-income settings and may not generalise everywhere.
  • Cognitive outcomes differ across studies, and effect sizes from different meta-analyses are not directly interchangeable.
  • Population-level risk reduction cannot predict any individual's outcome.

For symptoms, diagnoses, hearing or vision concerns, and any treatment targets, speak to a qualified clinician. This review is an educational map of the evidence, not a substitute for individual medical advice.

References

  1. 1.National Institute on Aging (NIA) Cognitive Health and Older Adults. U.S. National Institute on Aging, 2026. https://www.nia.nih.gov/health/brain-health/cognitive-health-and-older-adults
  2. 2.World Health Organization Risk reduction of cognitive decline and dementia: WHO guidelines (second edition). World Health Organization, 2026. https://www.who.int/publications/i/item/9789240123557
  3. 3.World Health Organization New WHO guidelines: up to 45% of dementia risk could be prevented or delayed. World Health Organization (news), 2026. https://www.who.int/news/item/15-07-2026-new-who-guidelines--up-to-45--of-dementia-risk-could-be-prevented-or-delayed
  4. 4.Livingston G, et al. (Lancet Commission) Dementia prevention, intervention, and care: 2024 report of the Lancet standing Commission. The Lancet (Lancet standing Commission), 2024. https://doi.org/10.1016/S0140-6736(24)01296-0
  5. 5.Lampit A, Hallock H, Valenzuela M Computerized cognitive training in cognitively healthy older adults: a systematic review and meta-analysis of effect modifiers. PLoS Medicine (via Europe PMC), 2014. 52 randomized trials, 4,885 healthy older adults. https://europepmc.org/articles/PMC4236015
  6. 6.Basak C, et al. Meta-analysis of computerized cognitive training in older adults: near- and far-transfer effects. Europe PMC, 2020. 215-study meta-analysis. https://europepmc.org/articles/PMC7050567
  7. 7.SPRINT MIND Investigators (Williamson JD, et al.) Effect of intensive vs standard blood pressure control on probable dementia and mild cognitive impairment. JAMA (via Europe PMC), 2019. SPRINT MIND randomized trial. https://europepmc.org/articles/PMC6439590
  8. 8.Lin FR, et al. (ACHIEVE Collaborative) Hearing intervention versus health education control to reduce cognitive decline (ACHIEVE): a randomised controlled trial. The Lancet (via Europe PMC), 2023. ACHIEVE randomized trial. https://europepmc.org/articles/PMC10529382
  9. 9.Coe NB, et al. Impact of cognitive training on claims-based diagnosed dementia over 20 years: evidence from the ACTIVE study. Alzheimer's & Dementia: Translational Research & Clinical Interventions, 2026. ACTIVE 20-year follow-up; speed training with boosters HR 0.75 (0.59–0.95). DOI: 10.1002/trc2.70197. https://europepmc.org/articles/PMC12884427
  10. 10.Dickens WT Reanalyzing the impact of cognitive training over 20 years in the ACTIVE study: losing the forest in the trees. Alzheimer's & Dementia: Translational Research & Clinical Interventions, 2026. Reanalysis after correcting for multiple comparisons. DOI: 10.1002/trc2.70280. https://europepmc.org/articles/PMC13284741