HJNO Jul/Aug 2026

24 JUL / AUG 2026 I  U.S. HEALTHCARE JOURNALS Therefore, we also estimated the associa- tions of highest level of play with the out- comes excluding youth players. We con- ducted all linear regressions andANCOVAs adjusting for self-report history of alcohol and drug abuse as covariates. RESULTS The study sample included 3970 male former American football players (mean [SD] age, 55.93 [10.00] years; 471 [12.0%] AfricanAmerican or Black, 16 [0.4%] Asian, 30 [0.8%] NativeAmerican, 20 [0.5%] Pacific Islander, 3198 [81.0%] White, 122 [3.1%] mul- tiple races, and 67 [1.7%] other race) enrolled in HITSS. Two substudies were performed: (1) all 3970 football players and (2) 943 players and controls (mean [SD] age, 58.46 [10.37] years; 42 [4.5%] AfricanAmerican or Black, 7 [0.7%] Asian, 4 [0.4%] NativeAmeri- can, 858 [91.0%] White, 8 [2.8%] multiple races, and 4 [0.4%] other race), including 661 football players and 282 Brain Health Registry controls. Table 1 lists the charac- teristics of the football and control groups. Table 2 lists the characteristics of all the football players by highest level of play. eTable 3 stratifies Table 1 by level of play. Linear regression models with multiway cluster-robust SEs revealed that, compared with controls, football players had worse scores on PALFAMS (B=−0.64; 95% CI, −1.23 to 0.05; β =−0.15; P =.03), square root PALTEA (B=0.31; 95% CI, 0.07-0.54; β =0.18; P =.01), log ECog (B=0.11; 95% CI, 0.07-0.15; β =0.38; P <.001), and GDS-15 (B=0.62; 95% CI, 0.39-0.86; β =0.37; P <.001). In the cohort of 3970 football players, those who com- pleted PALwere older, had more education, and had a lower proportion of professional players (eTable 4 in Supplement). Level of Play ANCOVAmodels demonstrated a signifi- cant omnibus effect for highest level of play on PALFAMS (F=3.88, adjusted P =.03, par- tial η η 2 =0.02), square root PALTEA (F=5.47, adjusted P =.009, partial η 2 =0.02), log ECog (F=11.63, adjusted P < .001, partial η 2 =0.007), BRI (F=8.95, adjusted P < .001, partial η 2 =0.005), and GDS-15 (F=4.48, adjusted P =.01, partial η 2 =0.007). Post hoc Tukey- adjusted pairwise comparisons revealed professional players had worse scores com- pared with college and high school or youth, and college players had worse scores than high school or youth (Figure 2). Binary logistic regressionmodels demon- strated that, compared with youth or high school players, professionals had higher odds of clinically meaningful elevations on the following scales: ECog (odds ratio [OR], 1.36; 95% CI, 1.04-1.78; P =.02), BRI (OR, 1.61; 95% CI, 1.22-2.13; P <.001), and GDS-15 (OR, 2.21; 95% CI, 1.32-3.94; P = .004). College players had significantly higher odds of clinically meaningful elevations compared with youth or high school players on the fol- lowing scales: ECog (OR, 1.29; 95% CI, 1.08- 1.54; P =.006), BRI (OR, 1.29; 95% CI, 1.05- 1.58; P =.01), and GDS-15 (OR, 1.42; 95% CI, 1.05-1.93; P =.02) (eTable 5 in Supplement 1). Years of Play Multivariable linear regressions revealed significant associations between total years of play and higher BRI (B=0.27; 95%