Preventing Dementia and Alzheimer's: 14 Modifiable Risk Factors, and What Is Sold as a Trend but Lacks Evidence

How much of dementia can be prevented? Many people over 40 ask this after watching a parent or someone close slowly start to forget. The best answer today comes from the Lancet Commission, which pooled evidence from around the world and concluded that about 45 percent of cases may be prevented or delayed.

This article separates the factors with strong evidence that you can act on from things sold as trends, such as brain-training games and certain supplements, where the evidence does not yet hold up. It is for anyone who wants to start caring for the brain today.

Population attributable fraction (PAF) is the share of cases that would theoretically disappear if a risk factor were eliminated across the whole population. It is a population-level figure, not the amount any single person can lower their own risk by.

A Three-Line Summary

1. The Lancet Commission 2024 names 14 modifiable risk factors that together are linked to about 45 percent of dementia cases, up from 12 factors in 2020. The two new ones are untreated vision loss and high LDL cholesterol in midlife.
2. The factors with strong evidence that you can act on are caring for hearing, exercise and multidomain programs, controlling blood pressure and diabetes, quitting smoking, reducing air pollution, and avoiding social isolation.
3. Most supplements for dementia prevention show zero evidence, such as ginkgo, or weak evidence, such as fish oil and vitamin E. Brain-training games on their own are unproven, and for the MIND diet the observational studies and the randomized trial conflict.

The 14 Modifiable Risk Factors and the 45 Percent Figure

The Lancet Commission report (Livingston et al.), 2024 edition, updates the 2020 edition by pooling evidence worldwide and naming 14 modifiable risk factors across the life course.

Life stage	Risk factors
Childhood	Less education
Midlife, ages 40 to 65	High LDL cholesterol*, high blood pressure, smoking, obesity, depression, heavy drinking, hearing loss
Later life, age 65 and over	Physical inactivity, diabetes, social isolation, air pollution, brain injury, untreated vision loss*

(* are the 2 new factors added in 2024.)

The untreated vision loss estimate comes from a meta-analysis of 14 cohorts, over 6 million people, and 171,888 dementia cases, finding a relative risk of about 1.47. High LDL in midlife accounts for about 7 percent of cases, with mechanisms tied to amyloid plaques and cerebrovascular risk.

It must be stressed that 45 percent is a theoretical population-level figure, assuming all factors could be eliminated entirely. It is not the amount any single person can lower their own risk by.

Factors With Strong Evidence That You Can Act On

Hearing

Hearing loss has the highest PAF, about 8 percent, among the modifiable factors. The ACHIEVE trial in 2023, with about 977 participants aged 70 to 84 at high cardiovascular risk, compared hearing aids against health education.

The point to read in full is this: across the whole cohort there was no difference in cognitive decline, but in a high-risk subgroup of 238, hearing aids slowed decline by about 48 percent. That suggests benefit for a high-risk group, not for everyone, and the subgroup result has not been confirmed in the main trial.

Exercise and Multidomain Programs

The FINGER trial, a randomized study of 1,260 high-risk participants aged 60 to 77, tested a multidomain program covering diet, exercise, cognitive training, and vascular risk control against general health advice. It was the first large trial to show that a multidomain intervention preserves brain function and lowers risk. A limitation is that participants were volunteers who were fairly educated and healthy, and may not match every population.

In 2025 the US POINTER trial in an American population, with 2,111 participants averaging 68 years, compared a structured multidomain program against a self-guided one. Both groups improved in cognition, and the structured group improved slightly more in a statistically significant way, with equal benefit whether or not people carried the APOE4 gene. ⚠️ Caveat: the comparison group was also doing something, not sitting idle, so the effect of actually taking action may be larger than the gap seen between groups.

As for exercise on its own, a meta-analysis of 16 randomized trials with 945 participants found a moderate effect on cognition, with sessions 3 to 4 times a week, 30 to 45 minutes each, sustained beyond 12 weeks.

Blood Pressure, Diabetes, Smoking, and Social Life

• Intensive blood pressure control: SPRINT MIND, with 9,361 participants, found that intensive control with a target below 120 reduced mild cognitive impairment (MCI) by about 19 percent (HR 0.81) and reduced the composite of MCI or dementia (HR 0.85). Full-blown dementia on its own did not reach significance.
• Diabetes raises dementia risk by about 1.5 to 2.5 times, and low blood sugar itself raises risk by about 49 percent.
• Quitting smoking in midlife slows decline. A Lancet Healthy Longevity 2025 study of 9,436 people across 12 countries found risk approaching that of never-smokers within about 10 years of quitting. It is never too late to quit.
• Social isolation raises risk by about 26 percent and loneliness by about 32 percent, on par with major vascular risk factors.
• Air pollution (PM2.5) shows a dose-response relationship; a 10 percent reduction in PM2.5 was associated with about 14 to 26 percent lower dementia risk in cohort studies.

Sold as a Trend, but the Evidence Is Not There

Supplements

Supplement	Evidence
Ginkgo biloba	The GEM study, 3,069 participants followed for 6.1 years, found no preventive effect on dementia at all, one of the most solid negative results.
Fish oil and omega-3	Observational studies suggest benefit, but randomized trials are mixed and inconclusive. Eating fish may differ from taking capsules.
Vitamin E	Randomized evidence is thin, high doses raise bleeding risk, and it is better obtained from food.
Multivitamins	The COSMOS-Mind study, 2,262 participants over 3 years, found a small benefit for memory, but did not prove it prevents dementia.

Brain-Training Games

A 2019 Cochrane review concluded that the evidence for brain-training games preventing dementia is insufficient. Studies are very mixed, and where there is benefit, it is usually as part of a multidomain program like FINGER, not games on their own. For commercial games advertised as dementia prevention, the evidence does not back the claim.

Points to Watch: Conflicting Claims That Must Not Be Buried

The MIND diet: observational studies and the randomized trial conflict Morris’s observational work at Rush University found that following the MIND diet was associated with a 35 to 53 percent lower risk of Alzheimer’s. But a 3-year randomized trial in NEJM in 2023, with 604 participants, found no difference in cognitive scores or MRI imaging between the MIND group and the control group. The mechanism sounds reasonable, but it has not been proven in a prevention trial.

Anti-amyloid drugs slow decline a little but carry risk, and are not preventive Lecanemab slows decline by about 27 percent over 18 months, and donanemab by about 35 percent in early Alzheimer’s. But they carry a risk of brain swelling and microbleeds (ARIA); the donanemab trial had 3 deaths from ARIA. People with two copies of the ApoE4 gene are at high ARIA risk, and there is no evidence that these drugs prevent disease in people without symptoms. They remain drugs that slow decline, not prevent it, and their clinical meaningfulness is still debated.

The Lesné scandal and the amyloid theory A 2006 paper by Lesné, claiming the protein Aβ56 caused dementia in mice, was found to contain doctored images and was retracted in 2024. This means that particular Aβ56 cannot be used as a drug target, but it does not overturn the whole amyloid theory, because amyloid buildup in Alzheimer’s is confirmed by several independent lines of evidence, including cerebrospinal fluid, PET, and autopsy.

Alcohol: the J-shaped curve people once believed has been challenged The old belief that light drinking prevents dementia has been challenged by newer evidence using a genetic analysis method (Mendelian randomization from UK Biobank, about 313,958 participants), which finds that alcohol at every level raises risk in a dose-dependent way. The earlier protective effect may have come from confounding.

What Is Still Unknown

• Does treating depression genuinely lower dementia risk? Depression is a risk factor, raising risk by about 70 percent, but evidence that treating it reduces risk is still lacking.
• Would controlling blood pressure before age 50 prevent dementia better than SPRINT MIND did in those over 50? Still awaiting an answer.
• Do anti-amyloid drugs help people who have amyloid but no symptoms yet? No answer yet.
• Risk-factor data specific to the Thai population, such as the heavy PM2.5 air pollution Thailand faces, should be a topic for further study.

A Small Step You Can Take

The strongest evidence today says that taking action on several fronts at once works better than relying on any single thing. Start by checking your hearing if sounds are unclear, moving your body regularly, controlling blood pressure and blood sugar as your doctor advises, quitting smoking, and keeping up relationships with the people around you. As for supplements and brain-training games advertised to prevent dementia, there is not enough evidence to pin your hopes on them.

This information is for understanding, not medical advice. Before starting any medication or changing treatment, always consult a doctor.