For a decade the NFL had one reliable defense against the brain-bank numbers, and it was a good one: families donate brains when something has already gone wrong, so a study of donated brains tells you about the donors and nothing about the league.
A paper published Tuesday in the BMJ takes that defense away, and it does it by conceding the point.
The researchers assumed the worst-case version of the objection. Then they asked what the lowest possible prevalence of chronic traumatic encephalopathy among deceased NFL players could be if every single non-donor had been perfectly healthy. The answer they got is one in four.
What the Study Did Differently
The team behind it comes from Mass General Brigham, the Boston University CTE Center and the Concussion and CTE Foundation, with Jesse Mez and Daniel Daneshvar as the lead names. They started from the roster of former NFL players who died over the study window and worked with tissue from three separate brain banks, at Boston University, Mount Sinai and the UCSF Alzheimer’s Disease Research Center, rather than the single collection that produced the widely quoted 2023 figures.
The raw numbers, as the BMJ set them out:
- 1,712 former NFL players in the death cohort
- 338 brains donated and examined
- 315 of those, or 93.2 percent, showed CTE
- 202 donors, just under 60 percent, had a clinician’s diagnosis of dementia
- 104 donors, about 31 percent, had stage IV CTE, the most advanced form
The 93.2 percent is the number that will get quoted, and it is the one that means the least. It describes a self-selected group. The number that matters is the range the researchers built on top of it: among players who died between 2016 and 2021, the true prevalence sits somewhere between 24.5 percent and 97.7 percent. Across the longer window the floor drops to 18.5 percent.
That range looks uselessly wide until you notice what the bottom of it costs the league. NPR framed the finding as a minimum rather than an estimate, which is the right reading. Twenty-four and a half percent is the number you get if you hand the NFL every benefit of every doubt.
Why a Floor Beats an Estimate
Public-health arguments usually fail on uncertainty. An industry facing an inconvenient study reaches for the confidence interval, points at the width of it, and says the science is not settled. It works because it is often technically true.
A bounded floor is a different kind of object. It does not ask you to trust the sample. It says: pick any assumption you like about the players nobody studied, including the most generous one available, and the number still does not go below a quarter. That is the analytical move here, and it is why this paper is harder to argue with than the 2023 result that put CTE in nearly 92 percent of examined brains. That study produced a striking figure and an obvious rebuttal. This one produces a smaller figure and no rebuttal.
Mez put the implication in fairly restrained terms, saying that if at least 25 percent of NFL players who died had these changes in their brain, it reads as a call to action for more discussion of risk, more research, and more support for affected families. Daneshvar was blunter, describing the result as a wake-up call for the NFL and the scientific community. Boston University’s own summary of the work carries both.
The Dementia Link Is the Part Families Will Feel
The prevalence argument is the one the league will litigate. The dementia finding is the one that lands in living rooms.
Nearly 60 percent of the donors had been diagnosed with dementia by a clinician, and the association tracked with severity: donors with stage IV CTE showed markedly higher dementia rates than those with milder pathology. CNN’s write-up makes the practical point that this is what moves CTE from a postmortem curiosity to a predictor of how someone’s last years actually go, which is the thing wives and children and adult sons have been describing to reporters since Mike Webster.
It also reframes what the sport is being asked to fix. NBC News noted the researchers’ emphasis on cumulative repetitive head impacts rather than diagnosed concussions. That distinction is the whole ballgame. Concussion protocols address the visible events. If the driver is the thousands of sub-concussive hits a lineman absorbs in practice across fifteen years, then the protocol is treating the symptom and the exposure is unchanged.
The League Has a Business Answer and a Health Problem
The NFL Players Association emailed its membership after the study dropped, as Pro Football Talk reported, which tells you the union read it as significant even before the league said anything.
What the league says next will probably sound like what it said in 2023, and there is a structural reason for that beyond stubbornness. The NFL is a media business whose product happens to be a collision sport. Its energy this year has gone into the things that move revenue, and its loudest ongoing fight has been with Nielsen over how its audience gets counted, not over how its workforce ages. That is not hypocrisy exactly. It is what happens when the incentives sit on one side of the ledger and the cost sits on the other, twenty years later, in a house in Florida.
CBS News captured the researchers’ own conclusion, which is aimed less at the NFL than at everyone below it: reduce repetitive head impacts at every level of the sport. Youth football, high school, college. The professional players in this study are the ones with a union, a pension and a settlement fund. The fourteen-year-olds do not have any of that.
What Would Actually Count as a Response
Watch for three things, none of which is a statement.
The first is practice structure, because that is where the sub-concussive load lives and where a rule change costs the league nothing in broadcast inventory. The second is whether the NFL funds prevalence research it does not control, which is the only way its own numbers ever become credible. The third is what happens to the disability and dementia benefits in the collective bargaining agreement, because a study establishing a floor of one in four is also a study establishing that the current provisions were designed around a much smaller number.
A quarter is the most forgiving reading available. Nobody involved in this research thinks it is the right one.