HJNO Sep/Oct 2023

level and NFT burden (Supplementary Table 4, Table 6). These analyses restricted to brain donors with a single known position at the highest level of play. All models were adjusted for age at death. Duration and cumulative exposure measures associated with CTE status and severity There was a significant association between duration of play, as well as the cumulative exposure measures, and CTE status (Table 4), or severity (Table 5), adjusted for age at death. Each addi- tional year of play was associated with 15% increased odds of being diagnosed with CTE and, amongst those with CTE, 14% increased odds of being diagnosed with severe CTE. Every additional estimated 1,000 head impacts was associated with 21% increased odds of being diagnosed with CTE and, among those with CTE, 13% increased odds of being diagnosed with severe CTE. Every additional estimated 10,000 g cumulative linear acceleration to the head was asso- ciated with 20% increased odds of being diagnosed with CTE and, among those with CTE, 19% increased odds of being diagnosed with severe CTE. Every additional estimated 1,000,000 rad/sec 2 cumulative rotational acceleration to the head was associated with 22% increased odds of being diagnosed with CTE and, among those with CTE, 20% increased odds of being diagnosed with severe CTE. CHII-G and CHII-R classified CTE status better than duration of play or CHII Based on the Bayesian information crite- rion (BIC), there was very strong evidence for improved model fit for models estimat- ing the relationship between CTE status and either duration of play, CHII-G, or CHII-R, compared with the model using CHII. Addi- tionally, there was very strong evidence for improved model fit for models using either CHII-G and CHII-R over duration of play alone. There was no evidence of differences in model fit for the model incorporating CHII-G compared to CHII-R. ROC analyses (Fig. 2A) indicated significant improvement in classifying CTE status for models using CHII-G (area under curve or CHII-R, but not CHII compared to models using duration of play alone. There were similar significant improvements in classification using CHII-G and CHII-R compared to CHII. There were no significant differences between CHII-G and CHII-R in classifying CTE status. Similarly, 10-fold cross validation analyses found lowest Fig. 1 | Athlete CTE Status by Years of American Football Played. Histograms (a) and percent distributions (b) for the study sample by duration of play. Of note: among the 69 athletes who played ≤ 5 years of football, 35 played another contact sport (13 ofwhomhadCTE), 21 were in themilitary (7 of whom had CTE), and 9 were exposed to combat as part of military service (4 of whom had CTE). 47 of the 69 athletes had one or more of these exposures (16 of whom had CTE). Of the 69 athletes who played ≤ 5 years of football, 9 had CTE and none of these exposures. Please see Supplementary Fig. 1 for additional details. Source data are provided as a Source Data file. CTE chronic traumatic encephalopathy.