In August 2024 the Lancet standing Commission on dementia prevention, intervention and care published an updated report estimating that around 45% of dementia cases worldwide are potentially preventable by addressing 14 modifiable risk factors across the life course. The 2020 report had put the figure near 40% across 12 factors; the update added untreated vision loss and elevated LDL cholesterol. It is among the most consequential findings in modern brain health, and also among the most frequently overstated.

What Are the 14 Risk Factors?

The Commission’s list spans the life course. In early life: low education. In midlife: hearing loss, high LDL cholesterol, depression, traumatic brain injury, physical inactivity, diabetes, smoking, hypertension, obesity and excessive alcohol consumption. In later life: social isolation, air pollution and untreated vision loss.

Two things stand out. First, most are ordinary cardiovascular and metabolic health factors — what protects the heart largely protects the brain. Second, several are sensory. Hearing loss and untreated vision loss are both on the list, and both are highly treatable.

That sensory finding is arguably the most actionable item here. Uncorrected hearing loss is common, frequently untreated for years, and its inclusion suggests that addressing it is a genuine intervention rather than a quality-of-life nicety.

What Does "45% Preventable" Actually Mean?

This is where the finding is most often misread, so it deserves precision.

It is a population attributable fraction. It estimates the proportion of dementia cases that would not occur if these risk factors were entirely eliminated across a whole population. It does not mean any individual has a 45% chance of preventing their own dementia, and it does not mean that 45% of cases are somebody’s fault.

Complete elimination of these factors is also not achievable in practice. No population removes hypertension, hearing loss, air pollution and social isolation entirely. The realistic effect of improving them is smaller than the headline number.

And genuinely non-modifiable factors remain — age above all, plus genetics. Someone can do everything the evidence supports and still develop dementia. Framing prevention as a matter of individual discipline is both inaccurate and cruel to families who have watched it happen despite every effort.

Which Interventions Have the Best Evidence?

Exercise is the strongest, and it is one of the few with direct experimental evidence rather than observational association. In a randomized trial of 120 older adults averaging about 66 years, one year of aerobic exercise increased anterior hippocampal volume by roughly 2%, while controls declined 1.4% — a change the authors described as adding back 1 to 2 years of volume.

That distinction matters. Most risk factor evidence is observational, meaning it can be confounded: people who exercise differ in many ways from people who do not. The exercise trial randomized participants, which is a substantially stronger design.

Cardiovascular and metabolic management — blood pressure, cholesterol, diabetes — has strong supporting evidence and is straightforwardly actionable through routine medical care. Hearing correction is well supported and underused.

Sleep sits slightly outside the Commission’s list but is highly relevant, since glymphatic clearance activity increases markedly during sleep. Our article on why sleep is the brain’s maintenance shift covers that mechanism.

Do Brain Training Games Work?

Mostly not in the way they are sold. The consistent finding across this literature is that people get better at the trained task, with limited transfer to general cognitive ability or everyday function.

This should not be surprising given how plasticity works: the brain reorganizes the specific circuit being exercised rather than a general capacity. Practising a puzzle improves that puzzle.

What has better support is genuine novel learning — acquiring a language, an instrument, a complex skill. These are demanding, sustained and varied in a way a repeated game is not, and they overlap with the education and cognitive engagement factors on the Commission’s list. They also tend to involve social contact, which is independently protective.

The general principle: if it is easy, comfortable and repetitive, it is probably not driving much change. Challenge and novelty are the active ingredients.

Why Is Social Isolation on the List?

Because the evidence for social connection as a health factor is considerably stronger than most people assume. The US Surgeon General’s 2023 advisory documented that social isolation is associated with a 29% increased risk of premature mortality, alongside increased risk of dementia specifically.

The mechanisms are plausibly multiple. Social interaction is cognitively demanding — tracking conversation, reading intention, adjusting in real time is more complex than most puzzles. Connection also regulates the stress response, and chronic stress is associated with hippocampal atrophy.

There is a reverse-causation caveat worth being honest about: early dementia can cause social withdrawal, so some of the association may run backwards. The Commission’s inclusion reflects a judgement that the evidence supports a causal contribution, not that the direction is fully settled.

Our article on how friendship protects the brain covers this in more detail.

What Should You Actually Do?

Start with the medical basics, because they are the highest-yield and most neglected. Blood pressure, cholesterol, blood sugar, hearing and vision — all checkable through routine care, all on the Commission’s list, all treatable.

Then aerobic exercise, consistently, at whatever level you can sustain over years rather than weeks. The trial that produced measurable hippocampal growth used walking sustained over a year.

Then sleep, social connection, and genuine cognitive challenge — with the emphasis on genuine. And treat depression if present, since it appears on the risk factor list; see depression treatment. Our June article on the aging brain and lifelong learning covers the cognitive engagement side.

Finally: new or progressive memory change deserves medical evaluation rather than self-management. Some causes of cognitive symptoms — thyroid dysfunction, B12 deficiency, medication effects, sleep apnea, depression — are treatable, and distinguishing them from progressive decline requires proper assessment.