What WHO actually recommends, with the strength attached
The World Health Organization published the second edition of Risk reduction of cognitive decline and dementia on 10 July 2026, replacing the 2019 first edition. It is a GRADE-rated guideline, meaning every recommendation carries both a strength (strong or conditional) and a certainty rating for the underlying evidence. Reproducing only the recommendation and dropping the two labels is how a cautious document becomes a confident listicle, so both are here.
| What WHO addresses | Recommendation | Strength | Certainty |
|---|---|---|---|
| Physical activity, adults with normal cognition | Should be recommended | Strong | Moderate |
| Tobacco cessation | Should be offered to adults who use tobacco | Strong | Low |
| Vitamin B, vitamin E, omega-3 PUFA, multivitamins and minerals | Not recommended | Strong, against | Moderate |
| Healthy balanced dietary pattern | May be recommended | Conditional | Moderate |
| Reducing hazardous and harmful drinking | May be offered | Conditional | Moderate |
| Cognitive training | May be offered to older adults | Conditional | Low |
| Managing midlife overweight or obesity | May be offered | Conditional | Low to moderate |
| Managing hypertension | May be offered to adults with hypertension | Conditional | Very low |
| Managing diabetes | May be offered to adults with diabetes | Conditional | Very low |
| Hearing aids for hearing loss | May be offered | Conditional | Low |
| Social activity, cognitive stimulation | May be recommended or encouraged | Conditional | Very low |
| Menopausal hormone therapy, women 65 and older | Not recommended for this purpose | Conditional, against | Very low |
Read down the strength column. One item is strong on moderate certainty for people with normal cognition, and it is physical activity. Tobacco cessation is also strong, on low certainty. Everything else pointing towards an action is conditional, several on very low certainty. That is a much more modest picture than the usual twelve-point checklist, and it comes from the body that reviewed the evidence.
The half most articles delete
WHO also publishes what it could not recommend, under its own heading of insufficient evidence for recommendation. For the specific purpose of reducing the risk of cognitive decline or dementia, that list includes management of depression, pharmacological interventions typically used to reduce the risk of recurrent stroke, specific interventions after traumatic brain injury, vision impairment interventions, and interventions to improve sleep quality or duration as well as those to treat sleep-wake disorders.
Every one of those appears on popular prevention lists. Naming them is not pessimism; it is the difference between a guideline and a marketing page. Sleep in particular is worth calling out, since it is heavily promoted in this context and WHO issued two separate insufficient-evidence statements about it.
What the randomized trials found
Four trials do most of the work in this field, and the honest summary of them is mixed.
FINGER randomized 60 to 77 year olds at elevated risk to a two-year multidomain program of diet, exercise, cognitive training and vascular risk monitoring. Its primary outcome was change in a neuropsychological test battery Z score, not dementia. The between-group difference in change per year was 0.022 (95% CI 0.002 to 0.042, p=0.030): statistically significant and small. The trial calls itself a proof-of-concept randomized controlled trial and concludes the intervention could improve or maintain cognitive functioning. Adverse events occurred in 7 percent of the intervention group against 1 percent of controls, mostly musculoskeletal pain.
US POINTER, published in JAMA in August 2025, is the large American follow-up, and its design detail is decisive. Both arms received a lifestyle intervention, structured or self-guided; there was no do-nothing control. The structured group's global cognitive composite improved faster by 0.029 SD per year (95% CI 0.008 to 0.050, p=0.008). That tells you structure beat self-direction. It cannot tell you how much either beat doing nothing, and the authors themselves note that further investigation will help address clinical relevance.
SPRINT MIND is the trial that matters most here, because dementia was its primary outcome, which is rare. Over a median 5.11 years, treating blood pressure to a systolic target below 120 rather than below 140 produced 7.2 versus 8.6 cases of adjudicated probable dementia per 1,000 person-years, a hazard ratio of 0.83 (95% CI 0.67 to 1.04). That confidence interval crosses 1, so the primary result was null, and the investigators note the trial may have been underpowered after early termination. But intensive control significantly reduced mild cognitive impairment (HR 0.81, 95% CI 0.69 to 0.95) and the combined outcome of MCI or probable dementia (HR 0.85, 95% CI 0.74 to 0.97).
ACHIEVE tested hearing aids, the risk factor with the largest attributed share in the popular version of the story. In 977 adults aged 70 to 84, three-year cognitive change did not differ between the hearing intervention and health education control (difference 0.002, 95% CI -0.077 to 0.081, p=0.96). A prespecified sensitivity analysis found a significant difference between the two enrolled cohorts (p for interaction 0.010), suggesting a hearing intervention might reduce cognitive change in populations at increased risk but not in those at decreased risk. The primary result was null, and the subgroup finding is a sensitivity analysis, not a headline.
The MIND diet trial was also null. Over three years in 604 participants, global cognition improved in both arms with a mean difference of 0.035 standardized units (95% CI -0.022 to 0.092, p=0.23), and MRI measures were similar. Its comparator was a control diet with mild caloric restriction, and all participants got dietary counselling and weight-loss support, so this was not diet against nothing.
The 45 percent figure, handled properly
WHO restates it in its own introduction: according to a 2024 Lancet Commission report, up to 45 percent of dementia risk is attributed to 14 modifiable risk factors. That sentence is worth reading precisely, because it is routinely rewritten into something it does not say.
45 percent is a modeled population attributable fraction computed from observational data. It answers what share of cases might be associated with a set of factors if those factors were eliminated from an entire population. It is not a trial result, and it is not an estimate of what any intervention has achieved or what any individual can achieve. "Up to 45 percent of dementia risk is attributed to modifiable factors" is accurate. "You can prevent 45 percent of dementia" is not, and no trial on this page supports it.
What this leaves you with
A short list, which is the point.
- Move. The only strong recommendation on moderate certainty for people with normal cognition. Our guide to strength training after 40 and the one on walking and blood pressure cover the doses with evidence behind them.
- Stop smoking. Strong recommendation, and the benefits extend well beyond cognition.
- Treat blood pressure. Weakly recommended for this specific purpose by WHO, and supported by the best randomized signal in the field, on MCI.
- Do not buy supplements for this. WHO recommends against vitamin B, vitamin E, omega-3 and multivitamins for dementia risk reduction, strongly, on moderate-certainty evidence, in people without established deficiencies. Our guides on multivitamins and NMN and NR reach the same place from the trial evidence.
- Treat the rest on their own merits. Hearing loss, diabetes, alcohol, weight and social connection are worth addressing for reasons that do not depend on the dementia case, which is fortunate, because the dementia case for most of them is conditional and low certainty.
One boundary is worth stating because it is regularly blurred. Preventing dementia and treating it are different questions. DAPA randomized 494 people who already had mild to moderate dementia to a structured aerobic and strength program and found a small statistically significant negative effect on its primary cognitive outcome (mean difference -1.4, 95% CI -2.62 to -0.17), concluding the program produced no clinically meaningful benefit. Exercise's case here is about risk in people who do not have dementia.
Frequently asked questions
What is the single best thing I can do to reduce dementia risk?
On the current evidence, physical activity. It is the only intervention WHO's 2026 guideline recommends strongly on moderate-certainty evidence for adults with normal cognition. It is worth being clear about what that means: the evidence is about reducing the risk of cognitive decline, not a guarantee against dementia.
Do brain training games work?
WHO conditionally recommends cognitive training for older adults with normal cognition or mild cognitive impairment, on low-certainty evidence, and defines it as targeted, repetitive exercise stimulating specific cognitive domains. Cognitive stimulation, meaning structured group activities and everyday activities like reading and games, is conditional on very low certainty. Neither is a strong recommendation.
Should I take omega-3 or a B-complex for brain health?
WHO recommends against vitamin B, vitamin E, omega-3 polyunsaturated fatty acids and multivitamins or minerals for reducing the risk of cognitive decline and dementia, in people without established deficiencies. That is a strong recommendation on moderate-certainty evidence, which is a firmer conclusion than most of the positive recommendations in the same guideline.
Will hearing aids protect my memory?
ACHIEVE tested this directly and found no difference in three-year cognitive decline in its primary analysis. A prespecified sensitivity analysis suggested a possible benefit in a higher-risk subgroup. Hearing aids are worth having for hearing; the dementia claim outran the trial.
Is it too late in my sixties?
Nothing in these documents supports a cutoff. WHO's physical activity recommendation applies to adults with normal cognition and extends, conditionally, to those with mild cognitive impairment. SPRINT MIND enrolled adults with hypertension at a median age well into later life and found its MCI benefit there.
When to talk with a clinician
New or worsening memory problems, getting lost in familiar places, difficulty with words, or changes noticed by people close to you are reasons to be assessed rather than to start a prevention program. Assessment matters because several treatable conditions imitate early dementia, including thyroid disease, B12 deficiency, depression, sleep apnea and medication effects. Our guides to B12 levels and a raised TSH cover two of them.
