Rapid Response: Why CTE Prevalence Estimates in Former NFL Players May Be Inflated
Rapid Response
Why CTE Prevalence Estimates in Former NFL Players May Be Inflated
BMJ, August 31, 2026
Dear Editor:
We read with great interest the paper by Daneshvar et al.[1] While we commend them for their work on this important topic, the indicated range for the prevalence of CTE among National Football League (NFL) players is inflated. The ranges indicated are correct for the deaths they analyzed, but extrapolating this to all players is problematic.
Most NFL players are still living. According to Pro Football Reference (PFR), 19,012 NFL players satisfying the authors’ inclusion criteria were alive through 2007. The deaths analyzed represent 9.0% of that population. To be representative of all NFL players those 9% must be a random sample, which it is demonstrably not.
These players are aging and more deaths will occur over time. The increase from 834 deaths between 2008-2015 (8 years) to 878 between 2016-2021 (6 years) suggests this [1] and is seen in National Death Index (NDI) data on NFL players. However, the authors show a CTE-dementia association, and cite evidence indicating CTE is associated with adverse outcomes that lead to earlier death. Thus, earlier deaths seen in this population will be enriched for CTE pathology, and not representative, because it is likely that many players have not died yet precisely because they do not have CTE-associated conditions.
There is a related problem of age-dependent brain donation. The percentage of donated brains among younger players (e.g. 20-29 years: 34.6%) is greater than among older players (e.g. 80-89 years: 13.1%) (Figure 1)[1]. Since virtually all donated brains showed CTE, the estimated CTE lower bound (CTElb) prevalence is effectively the weighted average of the proportion donated at the different ages. However, living players will skew to older ages at death precisely because they have not died yet. This is seen as increasing proportions of older deaths over time in NDI or PFR data. Notably, brain donors’ average age at death does not really rise from 2008-2021 (Supplemental Table 3)[1] showing that donors skew young, which inflates CTElb. To make a statement about all NFL players, one needs to consider all 19,012 players who could have died during the time period analyzed. That gives a CTElb of 315/19,012=1.7% (1.2% for 2016-2021, 1 in 83). This is likely low as more CTE will certainly be found, but it is the correct lower bound for the full 19,012 NFL players, whereas the 18.4% reported is correct only for the non-representative deaths that have occurred in that group to date.
These points are in addition to the selection bias raised in the accompanying commentary[2] from restricting to donated brains. While the authors attempt to mitigate this using inverse probability weighting, if a strong contributor to brain donation is premortem behavioral issues (e.g., memory loss), then the weighting misses a critical predictor of donation. The authors’ use of cause of death in the weighting model doesn’t fix this because, as they demonstrated, death certificates don’t capture dementia well. This selection would lead to the non-donor CTE percentage being less than that in donors, making the true prevalence (potentially substantially) less than the calculated CTE upper bound.
None of these concerns argue against CTE being a serious issue with potentially dramatic consequences affecting players and their families. Indeed, the first concerns we raise are only concerns if in fact CTE is associated with earlier death. However, these point to the prevalence estimates in all NFL players being lower than what is reported. At this time, drawing conclusions based on non-random donation of brains to estimate true prevalence remains challenging. To prevent misinterpretation, a future paper should explain these issues with data and consider approaches more suited to incorporating postmortem data with the living players.
References:
1. Daneshvar D H NCJ, Abdolmohammadi B, Luster C B, Uretsky M, Martin B M et al. . Prevalence of chronic traumatic encephalopathy at death in National Football League players: retrospective population based cohort study, 2008-21. BMJ 2026;394:e100418.
2. Kurth T, Riley RD. Chronic traumatic encephalopathy in former NFL players. BMJ 2026;394:e100526. doi: 10.1136/bmj-2026-100526 [published Online First: 2026/08/25 23:33]
Competing interests: All authors work with the Football Players Health Study at Harvard that is funded by the NFLPA.
AI use: None declared
31 August 2026
Marc G Weisskopf
Professor
Michael Leung, Niki Konstantinides, Rachel Grashow
Harvard TH Chan School of Public Health
Boston, MA, USA
(Originally posted on the BMJ website)
(Note: September 10, 2026)
While there may be differences in methodological approaches used to estimate the lower bound prevalence of CTE among former athletes, there is no disagreement regarding the central conclusion that CTE is a real neuropathologic condition associated with significant clinical consequences for affected individuals and their families. Moreover, the association between the most severe forms of neurodegenerative pathologic change and the development of dementia appears to be compelling. We believe that ongoing discussion of these complex issues is best advanced through rigorous scientific discourse in forums that promote open collegial scientific debate, critical evaluation, and continual refinement of knowledge. Clinical science has always progressed through such dynamic evolution of evidence and ideas. Throughout these discussions, however, we must all remain mindful of the voices of former players and their families, placing their experiences at the center of our efforts while pursuing strategies that impact disease, enhance resilience, preserve function, and improve quality of life.