Dementia Risk Factors, and How Much of It You Can Actually Change

"I forgot the word for colander last week. My father had Alzheimer's. I need to know if I'm starting." That is the quietest fear in primary care and the one patients apologize for raising. It deserves better than reassurance, and better than the tidy listicle about crosswords and blueberries.

What follows is the long-horizon question: not "is this dementia," but how much of the risk is actually yours to change. If your worry is the word-finding and fog of the menopause transition specifically, that is a different and far more reassuring topic, covered in our article on menopause brain fog. The good news here is real, and mostly not the news you have been sold.

What the 2024 Lancet Commission actually found

The reference point here is the Lancet standing Commission on dementia prevention, intervention and care, led by Gill Livingston, whose 2024 report is the current standard. It identifies 14 modifiable risk factors across the life course and estimates that eliminating all of them would theoretically prevent or delay around 45% of dementia cases worldwide.

The 14, grouped by when they matter most:

• Early life: lower levels of education.
• Midlife: hearing impairment, high LDL cholesterol, hypertension, obesity, excessive alcohol, traumatic brain injury, physical inactivity, diabetes, smoking, depression.
• Later life: social isolation, air pollution, untreated vision loss.

Two are new. The 2024 update added high LDL cholesterol in midlife from around age 40 and untreated vision loss in later life, together about 9% of cases (7% and 2%). The 2020 report listed 12 factors linked to 40%. The largest individual shares belong to hearing impairment and high LDL cholesterol at 7% each, then less education in early life and social isolation in later life at 5% each.

That last point is the most useful thing in the report. The single largest modifiable contributor to dementia in the world is hearing loss, and the second is a cholesterol number your primary care doctor already has on file.

The four with the most behind them

Hearing. The ACHIEVE trial, published in The Lancet in 2023, randomized 977 adults aged 70 to 84 with untreated hearing loss to hearing aids with audiological counselling or a health education control, and followed cognition for three years. In the full cohort the difference was essentially zero. But a prespecified sensitivity analysis found the effect differed sharply by subgroup: among participants recruited from the long-running ARIC cardiovascular cohort, who were older and carried more risk factors, hearing intervention slowed three-year cognitive decline by 48%. A 2025 secondary analysis found decline about 62% slower among the quartile at highest predicted risk. The reasonable reading is that hearing aids are not a general cognitive tonic for healthy 70-year-olds and are probably meaningful for people already vulnerable. Treating hearing loss has enough other benefits that this is an easy call regardless.

Blood pressure. SPRINT MIND, published in JAMA in 2019, randomized 9,361 adults with hypertension to a systolic target below 120 or below 140. Intensive control reduced mild cognitive impairment (hazard ratio 0.81) and the combined outcome of MCI or probable dementia (hazard ratio 0.85). For probable dementia alone the hazard ratio was 0.83, but the confidence interval crossed 1.00 and the trial stopped early for cardiovascular benefit, leaving it underpowered there. So the best randomized evidence shows blood pressure control moves cognitive outcomes and stops just short of proving it prevents dementia. It is still the strongest lever most people have.

Cholesterol and vascular risk. High LDL from around age 40 now sits on the list at 7% of cases, which puts lipid management inside brain health rather than only heart health. Our guide to cardiovascular risk factors covers how we assess and treat it. Diabetes and smoking belong in the same bucket: what damages small vessels damages the brain. Vision is the other 2024 addition: cataract surgery and an updated glasses prescription are the least glamorous and most available interventions here.

Where the list gets weaker, and why that matters

Most of the 14 come from observational studies. We know people with the risk factor develop more dementia; we do not know it caused it. Two problems recur.

The first is ordinary confounding: people who exercise, stay connected and read a great deal also tend to be wealthier, better educated, less depressed and better doctored. Statistics adjust for some of that, never all of it.

The second is reverse causation, the more interesting problem. Dementia begins in the brain years before anyone notices a symptom. During that silent phase people withdraw socially, stop exercising, and become depressed. Measure inactivity and isolation, follow those people forward, and both will predict dementia. The arrow may point the other way.

This is not speculation. Sabia and colleagues, in the BMJ in 2017, followed 10,308 participants in the Whitehall II cohort for an average of 27 years. Midlife physical activity showed no association with later dementia (hazard ratio 1.00). What they found instead was that physical activity in people who developed dementia began to decline up to nine years before diagnosis, and they concluded that earlier findings of lower dementia in active people were likely reverse causation. Depression and social isolation are open to the same critique, and the Commission says so.

Then there is US POINTER, published in JAMA in 2025, the largest American test of a structured multidomain lifestyle program. It randomized 2,111 adults aged 60 to 79 at elevated risk to a structured two-year program or a self-guided version. Both groups improved. The structured arm improved more, by 0.029 standard deviations per year: statistically significant, and small. The imaging study found no group differences in amyloid, tau or hippocampal volume.

Sleep belongs here for the same reason. It is not among the Commission's 14, because the relationship runs both ways and the evidence is not strong enough to call it causal. Poor sleep is still worth fixing on its own terms, covered in our piece on menopause and sleep problems.

The questions patients actually ask

Should I get tested for APOE4? This deserves a real conversation rather than a checkbox. APOE4 is the common genetic variant most strongly associated with late-onset Alzheimer's disease. A 2024 Nature Medicine study by Fortea and colleagues, analyzing thousands of individuals, argued that people carrying two copies should be considered to have a genetically determined form of Alzheimer's disease rather than merely a risk factor: nearly all had abnormal amyloid biomarkers by age 65, and symptom onset was more predictable than in the general population.

That is important science. It does not yet change management. No treatment starts on a positive result, and the standard advice for a carrier, control blood pressure and cholesterol, treat hearing loss, stay active, is what we would tell you anyway. What a result can change is planning, long-term care and finances, prevention trial eligibility, and how you feel every time you forget a word, permanently, because you cannot un-know it. APOE4 status also carries implications for anti-amyloid drug eligibility and side effect risk. We are glad to order it, and we would rather talk it through first. Genetic panels are ordered and interpreted here; the laboratory work is done outside the office.

What about brain training and memory supplements? Short version below, and it is not generous.

How this works at Framework Health

1. We go through the 14 with you, individually. An hour is long enough to work out which ones actually apply to you.
2. We treat the vascular ones properly. Blood pressure to a real target, lipids assessed and managed, diabetes and smoking addressed. These are the levers with trial evidence behind them.
3. We check the senses nobody checks. Hearing and vision get asked about and referred, not assumed. This is the highest-yield underused move in the field.
4. We follow up and re-measure. If cognition is the concern, we set a baseline, recheck at a sensible interval, and work up anything that changes rather than calling it normal aging by default.

Questions we hear about this every week

The bottom line

Roughly half of dementia risk appears to be modifiable at a population level, and the levers that matter most are unglamorous: hearing aids, blood pressure, cholesterol, glasses, not smoking, staying connected. Some of the list is association rather than proven cause, and being honest about that is what separates useful advice from marketing. Nothing you can buy in a bottle is on it. If this is a worry you have carried privately, bring it to a visit long enough to work through it, because the fear is usually heavier than the plan.