Read Past the Headline: BU's CTE Paper Has Three Numbers, 93.5%, 74.5% and a 63.7% Floor. Here's What Each One Means

Everybody saw the headline Wednesday: about three in four NFL players who died between 2008 and 2023 had CTE. I went and read the actual paper in JAMA Network Open, because the headline number is the middle of three numbers, and the other two tell you more about what Boston University's researchers actually found.

Number one: 93.5 percent. That's the raw count. Of 387 former NFL players whose brains were donated to the UNITE brain bank and evaluated, 362 had CTE, according to the paper. That's not the estimate for all players, and the researchers say so. Donated brains skew toward people whose families saw something wrong. The study describes this as registry bias, and points out that a 2017 study found CTE in 110 of 111 former NFL players, a figure that "does not represent the prevalence in the entire NFL population."

Number two: 74.5 percent. This is the headline. The team, led by Julia Gallini, a biostatistics Ph.D. candidate, and senior author Yorghos Tripodis of the BU School of Public Health, took every former NFL player who died from January 2008 through December 2023, 2,362 men, using the National Death Index. Of those, 396 were brain donors and 1,966 were not. They then estimated how much more likely a player with clear cognitive symptoms was to have his brain donated. The answer was about five times as likely, a selection ratio of 4.97. Correcting the 93.5 percent for that gets you to 74.5 percent, with a 95 percent confidence range of 69.6 to 78.3 percent.

Number three: 63.7 percent. This is the floor. The researchers ran nine alternate versions of the math, and the most conservative one, which they describe as a "high-bias, lower-bound" estimate, still came out at 63.7 percent. The other versions ranged from 70.3 percent to 86.1 percent. The paper's conclusion is blunt: even if players with cognitive symptoms were 10 times more likely to have their brains donated, the estimated prevalence "would still exceed 60%."

The donor-versus-nondonor table is worth a look too. Donors died younger on average, 67.0 years versus 74.7, played longer in the league, 7.1 seasons versus 4.7, made more Pro Bowls and were more likely to be in the Hall of Fame, 5.5 percent versus 2.7 percent. Each extra year in the NFL raised the likelihood of donation by about 13 percent in the model. Position, Pro Bowls and Hall of Fame status were not significant predictors once everything else was accounted for. The study also found 41.1 percent of the full group of 2,362 were 80 or older when they died.

Then there's dementia. Among donors with available records, 64.9 percent had a history of dementia in life. Adjusted the same way, the estimate for all 2,362 players is 27.2 percent. The authors say the true lifetime figure is probably higher than that, because nondonors died older on average.

What the paper does not say matters just as much, and the researchers list it themselves. The estimate applies only to players who have died. It does not tell you how many living former players, or active players, have CTE. It counts all stages of CTE, and the authors note "many cases may have been early stage at the time of death and possibly symptom free." Their proxy for who donates relies on death certificates, which undercount dementia, and on survey-based estimates of mild cognitive impairment. BU's release says the same thing in plainer English: the estimate "only applies to former NFL players after death."

For scale, BU's release says CTE has been found in less than 1 percent of thousands of brain donors to community brain banks in the U.S., Australia, Austria and Brazil. "An estimated 75 percent prevalence among NFL players who died over a 16-year period is concerning," Gallini said in the release. Former quarterback Matt Hasselbeck, an ambassador for BU's DIAGNOSE CTE study, said he was "looking at this like getting a scouting report."

Disclosure note, because it's in the paper: the funders were a National Institute on Aging grant to BU's Alzheimer's Disease Research Center, the National Institute of Neurological Disorders and Stroke and the Concussion & CTE Foundation, and the authors list various outside relationships, including co-author Christopher Nowinski's reported travel support from the NFL and the NFL Players Association. The journal says the sponsors had no role in the analysis.

Dez's take: The lazy reading of this is "three out of four players have CTE." That's not what it says. The honest reading is tougher: BU went out of its way to knock down its own number, assumed the worst about who donates, and still couldn't get it under about 64 percent. When the floor is that high, the argument isn't really about the math anymore.

What we still don't know: how common CTE is in living former players, and in college and high school players, which the authors say will be much harder to estimate because the death records and symptom data they used for the NFL don't exist at those levels.

Former players looking for help can reach the NFL Life Line at 800-453-0078, which BU's release lists along with resources from the NFL Players Association and the Concussion & CTE Foundation HelpLine.