Can a head injury cause dementia years later? What the long-run evidence shows

Short answer: Head injury is associated with a higher risk of dementia, and the association can still be detected decades later — but how much higher depends enormously on how bad the injury was and how many there have been. Pooling 32 studies, any head injury carried about a 66% higher risk. A single injury in one large 25-year cohort carried 25% higher risk; two or more carried 114%. And in the largest UK study, risk was highest in the first two and a half years after injury and no association remained beyond 40 years. If you are reading this because of one knock on the head a long time ago, that is the most relevant sentence on this page.

There is also a finding that almost no article on this subject mentions, and it changes what you should actually do. Read on.

This is educational content, not medical advice. It describes patterns in population research, not a prediction about you. Any new or worsening symptom after a head injury needs a clinician, not an article.

What the numbers actually say

The pooled estimate. A 2023 systematic review in the Journal of Neurotrauma combined 32 studies of dementia after traumatic brain injury of any severity. The pooled risk ratio was 1.66 (95% CI 1.42–1.93) — about 66% higher risk. Associations were stronger in younger people, in men, and in cohorts from Asia. Veterans, notably, were not at higher risk than civilians once exposure was accounted for.

Number of injuries matters more than the headline. A 25-year follow-up of 14,376 people in the American ARIC study, published in Alzheimer's & Dementia in 2021, found head injury associated with dementia at HR 1.44 (1.30–1.57) — with a clear dose-response:

• One head injury: HR 1.25 (1.13–1.39)
• Two or more: HR 2.14 (1.86–2.46)

Severity matters at least as much. A 2026 analysis of 397,581 UK Biobank participants followed for 12.5 years found head injury at 1.72 (1.50–1.97) and diagnosed TBI at 1.86 (1.46–2.38) — but recurrent head injuries at 4.05 (2.24–7.32) and severe TBI at 4.50 (3.18–6.37). People with repeated injuries also showed smaller hippocampal volumes and more white-matter damage on brain imaging.

(Some studies report much larger numbers than others. A 2026 study of 501,710 UK Biobank participants reported TBI associated with early-onset dementia at HR 4.06 and late-onset at 2.51 — but that is the same underlying cohort as the study above, analysed differently. Two papers from one biobank are not two independent confirmations, and the gap between 1.72 and 2.51 for essentially the same question is a good illustration of how much study design moves these numbers.)

How long does the risk last?

This is the question most people actually came for, and the honest answer has two halves that don't perfectly agree.

The signal persists for decades. The ARIC study's whole design was a 25-year follow-up, and the association was still there at the end of it.

But it fades. The UK Biobank analysis looked specifically at time since injury. Dementia risk was highest in the first 30 months after a head injury — HR 2.20 (1.66–2.92) — and declined from there, with no association remaining beyond 40 years post-injury.

That early spike deserves scepticism rather than alarm. Some of it is almost certainly the injury. Some of it is the reverse: the earliest, unnamed phase of dementia affects balance, attention and judgement years before anyone gives it a name, which makes a fall — and a head injury — more likely. When an association is strongest right before diagnosis and fades the further back you look, that pattern is a warning sign about the direction of cause. (It is one of seven ways a dementia statistic can mislead you.)

Is any of it genuinely causal? The same 2026 paper ran a Mendelian randomisation analysis — a method that uses inherited genetic variation, fixed at conception, as a natural experiment, so it cannot be distorted by lifestyle or reverse causation. It supported a causal link between TBI and Alzheimer's disease, at OR 1.17 (1.01–1.37). Note how modest that is, and that its lower bound sits just above no effect. The best available read: there is probably a real causal component, and it is smaller than the observational numbers suggest.

The finding nobody mentions: most head injuries happen at 71

Nearly every article about head injury and dementia is about sport — football, boxing, heading a ball. The best long-run data on ordinary people tells a very different story.

Researchers followed 10,333 participants in the Framingham Heart Study, across the original and offspring cohorts, from 1948 to 2022 — seven decades — identifying head injuries from comprehensive medical-record review rather than recall. What they found:

• Most TBIs occurred late in life. Mean age at injury: 74 in the original cohort, 71 in the offspring cohort.
• Falls were the most common cause — 65.4% and 82.8% of injuries respectively.
• TBI was associated with dementia-related death in a dose-dependent way: mild TBI HR 1.60 (1.31–1.97), moderate-to-severe HR 3.67 (2.31–5.80).
• It was associated with all-cause mortality (1.15, 1.06–1.26) — but not with death from non-dementia causes, and mild TBI alone was not significantly associated with all-cause mortality (1.06, 0.96–1.16). Almost all of the excess mortality ran through dementia.

The authors' own conclusion: "Preventing falls to reduce TBI could have important implications for dementia and mortality."

(A note on those two numbers, because the coverage garbled them: 1.60 and 3.67 are hazard ratios for dementia-related mortality, not for being diagnosed with dementia. Headlines rendering them as "60% higher dementia risk" and "267%" are describing a different outcome from the one the study measured.)

And the loop runs both ways

A 2023 analysis in Neurology, again using ARIC — 13,081 people, median 23 years of follow-up — asked the reverse question: after a head injury, what happens to your risk of falling?

People who had sustained a head injury had 2.01 times the risk (1.85–2.18) of a subsequent fall requiring hospital care. Accounting for the competing risk of dying first, still 1.69 (1.57–1.82). Dose-response again: more injuries and worse injuries, higher fall risk.

So: falls cause most head injuries, and head injuries roughly double the risk of future falls. That is a loop, and it is the single most useful thing on this page, because loops can be broken at any point.

What this means for what you should actually do

If you are 45 or 55 and worried about a rugby injury at 19, the evidence says: a single mild injury that long ago is a small and fading signal, and your attention is better spent elsewhere. If you have had repeated head injuries, or a severe one, the association is substantially larger — worth raising with a clinician, and worth taking the rest of your modifiable risk factors seriously.

But for almost everyone reading this, the head injury that matters most hasn't happened yet, and it will most likely be a fall, in your seventies. That is not a grim thought. It is the most encouraging sentence here, because fall risk is one of the most modifiable things in medicine, and every component of it is already something you might be working on:

1. Balance and lower-body strength. The core of every fall-prevention programme that works. Also good for your brain in its own right.

2. Your eyes. Poor vision is a leading cause of falls — and vision loss is itself on the Lancet Commission's list. Get the check, update the prescription, and be careful with the transition period after new varifocals.

3. Standing up. Feeling light-headed on standing is common, treatable, and a direct route to a fall. Blood pressure that is too low on standing matters here, which is a different conversation from midlife blood pressure control — worth raising both with your clinician at the same visit.

4. Your medicine cabinet. Sedatives, sleeping tablets and drugs with anticholinergic effects all increase fall risk, and several are on the list of medicines most often flagged for review in older adults. We have written separately about sleeping tablets and anticholinergic burden. Never stop a prescribed medicine on your own — ask for a review.

5. Hearing. Balance and hearing share an organ. Untreated hearing loss is on the Commission's list for dementia and is independently associated with falls.

6. The obvious ones. Helmets for cycling and skiing, seatbelts, handrails, decent lighting, and getting the loose rug out of the hallway. Unfashionable, and they work.

Keeping it in proportion

Head injury is one of fourteen modifiable factors on the Lancet Commission's list, and the share of all dementia cases attributed to it across a whole population is small. That is not the same as saying the risk to an individual who has had a serious head injury is small — a population share depends heavily on how many people are exposed, not just on how much harm the exposure does. The 2023 meta-analysis shows this neatly: the attributable risk among US veterans came out at twice the general population's, "largely due to the high prevalence of TBI exposure" — while veteran status itself was not associated with higher individual risk. More exposure, not more danger. Here is what all fourteen factors are, in plain English.

Two other details from the research are worth stating carefully. The ARIC study found stronger associations in women than men, and in White than Black participants. Findings like these usually say more about how injuries and dementia get recorded, and about who reaches care and when, than about anyone's biology — and the authors treat them as needing explanation rather than as settled facts.

How we research and review our content →

An independent physician audited the Solenna app. We published all nine findings →

Educational content only. It does not diagnose any condition and is not a substitute for professional medical advice. Seek medical attention for any head injury with loss of consciousness, persistent headache, vomiting, confusion or drowsiness. Do not start, stop or change any prescribed medication — including sedatives or sleep medication — without speaking to the clinician who prescribed it.