Have you long assumed that a diagnosis of chronic traumatic encephalopathy tells you almost everything you need to know about a patient’s future? A major new study published in The BMJ suggests otherwise. It is not the presence of CTE that predicts dementia. It is the severity.
For those of us working with patients who carry a history of repetitive head impact, whether from contact sport, occupational exposure, or blast injury, this distinction matters. It changes the question we should be asking at the bedside, and it changes what we tell patients and families about risk.
What the study did
Daneshvar and colleagues, working with a large multidisciplinary team that includes some of the most recognised names in CTE neuropathology, tackled a problem that has quietly undermined this field for years: brain donation selection bias.
Almost everything we thought we knew about CTE prevalence has come from brain banks. Families of players who showed cognitive or behavioural decline before death have been more likely to donate. That means a brain bank full of CTE-positive donors tells us almost nothing about how common CTE is in the wider population of players who die, most of whose brains are never studied at all.
This study took a different approach. The researchers identified every one of the 1,712 former NFL players who died between 2008 and 2021, not just the 338 whose brains were examined. Of those 338 donors, 315 (93.2%) had confirmed CTE. Rather than reporting that headline figure as a population estimate, which would be badly misleading, they calculated the full range of possible prevalence across all 1,712 deaths.
For the period 2016 to 2021, when donation rates were highest, that range was 24.5% to 97.7%. The lower bound assumes every undonated brain was CTE-negative, which is about as conservative an assumption as you can make. Even so, at minimum, one in four former NFL players who died in that period had CTE at death.
The finding that matters more than the headline number
The prevalence range will draw the attention. The more clinically useful finding sits underneath it.
Among brain donors, dementia was present in 90.4% of those with stage IV CTE, the most severe grade, compared with 39.1% of those with no CTE at all. After adjustment for selection bias using inverse probability weighting, stage IV CTE remained significantly associated with dementia (risk ratio 1.44, 95% CI 1.16 to 1.78).
Here is the part worth sitting with: the presence of CTE at any stage was not significantly associated with dementia in the corresponding analysis. Only severe, stage IV disease carried that signal.
This is not a small technical footnote. It means a positive CTE finding, on its own, does not tell you whether a person experienced clinically meaningful cognitive decline. Burden and stage appear to matter far more than a binary yes or no. The scientific and clinical question shifts from “does this person have CTE” to “how much, and why did it progress this far in this person and not another.”
Exposure, severity, and the question we keep coming back to
Duration of NFL play was not associated with simply having CTE. It was associated with having stage IV CTE (risk ratio 1.07 per additional year). One reading of this is that in an already extremely high-exposure population, many individuals may have already crossed some threshold for CTE to develop at all, while continued exposure drives how far the disease progresses.
That leaves an old question standing squarely in the room. Why do some heavily exposed individuals develop severe pathology while others, with what looks like comparable exposure, do not? The authors point to genetic factors, vascular health, age, co-pathology, and other biological modifiers as the likely next frontier. I would add: we should not assume the answer lies solely in the head impacts themselves. The patient in front of us is never just their exposure history.
CTE rarely travels alone
The paper is admirably careful on this point. These brains frequently showed multiple pathologies together, including Lewy body disease, Alzheimer’s disease, frontotemporal lobar degeneration, TDP-43 pathology, and vascular disease. Among donors with dementia, coexisting neurodegenerative pathology was common.
So finding dementia in someone with stage IV CTE does not mean CTE caused it, or caused it alone. Interestingly, among donors with CTE and no other diagnosed neurodegenerative disease, roughly 53% still had dementia. That relationship deserves serious, sustained investigation rather than being either overclaimed or dismissed.
An uncomfortable finding about death certificates
One result deserves more attention than it will probably get. Among donors clinically adjudicated as having dementia, only 40.6% had a neurodegenerative disease listed as a primary or secondary cause of death.
This is a reminder that applies well beyond CTE and well beyond American football. If we rely on death certificates, or on any single administrative record, to judge how common neurodegenerative disease is in an exposed population, we will underestimate it substantially. The absence of a diagnosis on paper is not evidence of the absence of disease in life.
Limitations worth keeping in view
The authors are transparent about the boundaries of what this study can tell us, and it is worth naming them plainly for anyone using this paper to guide their own thinking:
- The true population prevalence remains unknown. This study narrows the floor; it does not settle the number.
- Brain donation remains non-random. Donors differed from non-donors in career length, Pro Bowl selections, and Hall of Fame status, all factors the inverse probability weighting tries, but cannot fully guarantee, to correct for.
- These findings describe NFL players, one of the most heavily exposed populations of contact-sport athletes anywhere. The authors explicitly caution against generalising even to the broader American football community, let alone to single-concussion presentations, amateur athletes, or other sports.
- CTE remains a postmortem diagnosis. Nothing here changes our inability to diagnose it definitively in a living patient.
- Collider bias, arising from conditioning the analysis on both death and brain donation, is openly discussed by the authors, who argue it likely biases their findings toward underestimating the CTE-dementia association rather than overstating it. That is a reasonable argument, though inherently difficult to verify.
Why this matters for how we assess patients
This paper is not about vestibular or balance assessment directly. But it reinforces something I return to constantly in clinical practice: a diagnosis or pathology label does not tell you what a person is experiencing, or what they will experience. Severity, trajectory, and individual variability matter enormously, and treating a diagnostic category as a fixed predictor of outcome does a disservice to the patient in front of us.
For any of us working with patients carrying a history of repetitive head trauma, whether that is a single significant concussion, occupational blast exposure, or years of contact sport, the lesson is not “check for CTE.” It is to keep asking why some patients with similar histories do so much better, or so much worse, than others. That question does not have a neuropathology answer yet. It may have a clinical one, in how we assess, monitor, and individualise care over time.
A recent review of this paper by Dr Ioannis Mavroudis, consultant neurologist, put it well: the debate has moved past simply asking whether CTE occurs in people exposed to repetitive head impacts. The harder and more useful questions are about dose-response, individual susceptibility, and how to identify risk during life rather than only confirming it after death.
Those are the questions worth building our future research agenda around.