Dementia is the outcome people fear most and the one they are told least about preventing — usually because the honest answer used to be that nothing was known. That is no longer the position, and the most useful part of what changed is that most of the modifiable risk is vascular, which means the rest of this topic has been about it all along.
1. What the 45% figure is, and what it is not
Read this before quoting the number
- It is a population attributable fraction. It answers: if every one of these fourteen exposures were removed from the whole population, how much dementia would not occur? That is a public health question, not a personal prognosis.
- It assumes the associations are causal and that complete elimination is possible. Neither is true. Nobody eliminates their hearing loss, deprivation and air pollution to zero.
- The factors overlap heavily. Hypertension, diabetes, obesity, inactivity and LDL travel together in the same people, so their individual percentages cannot simply be added.
- Most of it is not individual behaviour. Education, air pollution, deprivation and access to hearing aids are structural. A page like this one addresses the smaller, personal share.
- What it does establish is direction and priority, and that is genuinely new. Twenty years ago dementia was widely presented as bad luck. It is substantially not.
2. The fourteen, and the four that matter most
| Life stage | Factors |
|---|---|
| Early life | Less education |
| Midlife, from about 45 | Hearing loss · Raised LDL cholesterol · High blood pressure · Obesity · Excess alcohol · Traumatic brain injury |
| Later life | Smoking · Depression · Social isolation · Physical inactivity · Diabetes · Air pollution · Untreated vision loss |
Two were added in 2024 and both are worth pausing on. Raised LDL in midlife is estimated to account for about 7% of cases, on evidence including a meta-analysis of three UK cohorts in which each 1 mmol/L higher LDL was associated with about 8% higher all-cause dementia incidence. Untreated vision loss in later life accounts for about 2%.
Which means this site has been writing about dementia prevention for some time without labelling it. LDL and apoB, blood pressure, physical activity, alcohol, sleep, air pollution and head injury all have chapters already. The dementia argument does not add a new set of instructions; it adds a second reason for the existing ones — and for many people a more motivating one than heart attack risk.
3. Hearing, which is the largest single factor and the least acted on
- Hearing loss is the biggest contributor in the Commission's global estimates, and hearing aids are free on the NHS. The gap between those two facts is the point of this section.
- The mechanism is plausible in more than one way — reduced cognitive stimulation, the cognitive cost of straining to hear, and social withdrawal, which is itself a separate factor on the list.
- The trial evidence is real but narrower than the headlines. The ACHIEVE randomised trial of hearing intervention found no significant slowing of cognitive decline in the overall population, but a substantial effect in the pre-specified subgroup at higher cardiovascular risk. That is a positive signal in the group most likely to decline, not a demonstrated benefit for everyone.
- The practical position is still straightforward. Untreated hearing loss has costs that have nothing to do with dementia — isolation, low mood, safety — and the intervention is free, reversible and low-risk. Get a hearing test at 60 and act on it, and treat the average delay of a decade between noticing and doing something as the problem it is.
4. Vision, which is easier still
Untreated vision loss was added in 2024, and the largest treatable cause in this age group is cataract — where the intervention is a routine day-case operation with a very high success rate. Sight tests are free on the NHS from 60, and vision loss is frequently unnoticed because it arrives gradually. As with hearing, the argument does not rest on the dementia link alone: falls, driving and independence all sit on the same finding.
5. The shingles vaccine result, which is unusually well identified
Over seven years, the eligible group had roughly a fifth fewer new dementia diagnoses. Published in Nature in 2025.
Why this is stronger than the usual observational study — and still not proof
- It sidesteps the bias that ruins vaccine research. People who get vaccinated are healthier, better organised and more engaged with healthcare than people who do not — which makes ordinary comparisons useless. A birth-date cutoff removes that entirely.
- The negative controls behave correctly. The vaccine affected shingles and dementia and not other causes of death or illness, and did not lead to more uptake of other vaccines or preventive care. That is exactly what you would want to see, and it is what a confounded result usually fails.
- It has replicated. Australia produced the same finding by the same method, with concordant signals reported from England, New Zealand and Canada, and a separate US analysis of a vaccine switch pointed the same way.
- The caveats are real. It is still not a randomised trial; the benefit appeared largely in women in both natural experiments and nobody yet knows why; the mechanism is unsettled; and the studied product was Zostavax, the older live-attenuated vaccine, while the UK programme now uses the recombinant Shingrix — the take-the-vaccine-you-are-offered conclusion carries across, but the evidence was generated on a different product. A randomised trial is being sought.
- What follows practically: Moderate — and shingles vaccination is offered on the NHS on its own merits, at ages set by the national programme. If you are eligible, this is a reason to take it up rather than defer it. It is not a reason to seek out a discontinued vaccine, and it is not yet a dementia treatment.
6. What the evidence does not support
| Claim | Position |
|---|---|
| Brain-training apps and puzzles | You get better at the task and it does not generalise. Cognitively demanding activity — learning something genuinely new, work, and above all social interaction — has better support than any commercial programme |
| Supplements for brain health | Nothing has convincing outcome evidence in people who are not deficient. Ginkgo, omega-3 and multivitamins have all been trialled for cognitive decline without a persuasive result. See supplements for the general argument |
| Coconut oil, ketone products, "type 3 diabetes" | Mechanistic claims sold well ahead of the evidence |
| Aluminium in cookware and antiperspirants | Investigated for decades and not supported. A durable myth rather than a live concern |
| Blood biomarker tests sold direct to consumers | Blood tests for Alzheimer's pathology are advancing genuinely fast and are changing specialist practice. Buying one privately without a clinical pathway attached tells you something you may not be able to act on, and the counselling matters as much as the result |
7. The new drugs, and where the UK stands
Lecanemab and donanemab are monoclonal antibodies that clear amyloid from the brain in early Alzheimer's disease. They are the first treatments to modify the disease process rather than the symptoms, which is a genuine scientific milestone.
- The effect size is modest — a slowing of decline over 18 months, of a magnitude that is statistically clear and clinically debated.
- They carry real risks, principally brain swelling and small bleeds — amyloid-related imaging abnormalities — which require monitoring by MRI and are commoner in people carrying two copies of APOE4.
- They are given by infusion, need specialist diagnosis and repeated scanning, and apply only to early disease.
- Licensed here, and not funded — but that position is moving. The MHRA licensed lecanemab in August 2024 and donanemab in October 2024, in both cases excluding people who carry two copies of APOE4, in whom the imaging abnormalities are commonest. NICE's final guidance in June 2025 did not recommend either for NHS use on cost-effectiveness grounds. An appeal was then partly upheld, and in March 2026 NICE announced it is reconsidering both drugs; revised guidance is awaited. Checked August 2026 — if you are reading this later, check the current NICE position rather than trusting this paragraph.
- The distinction worth holding on to is that this is a value-for-money judgement rather than a safety one, and the two are constantly conflated in reporting.
What that means for a reader today: these drugs are not the answer to the question this chapter is about. Everything above is available now, free, and applies to the 45% that is modifiable — while the drugs address a fraction of established disease at the margins. That ordering is not a criticism of the science; it is where the available benefit currently sits.
8. What to actually do
Treat the vascular risk factors, and treat them from midlife
Blood pressure, LDL, diabetes and smoking. The timing matters more here than for the heart — the Commission's LDL and blood pressure findings are specifically about midlife exposure, decades before any symptom. The rest of this topic is the detail.
Get your hearing tested, and wear the aids
The single largest factor, free on the NHS, and typically delayed by about a decade.
Get your eyes tested, and have the cataract done
Free from 60. A routine operation.
Stay physically active and socially connected
Two separate factors that support each other, and the reason group activity beats solitary exercise for this outcome specifically.
Protect your head, and treat sleep and mood as medical
Helmets and fall prevention; and depression and poor sleep are on the list in their own right. See sleep.
Take the vaccines you are offered
Shingles for the reason in section 5, and the rest on their own merits.
And do not treat a family history as a verdict
Genetic risk and modifiable risk are largely independent, and the Commission's evidence indicates that addressing these factors matters even in people at high genetic risk. A parent with dementia is a reason to act earlier, not a reason to assume it is decided.