Improving Predictive Isometric Peak Torque Values In Division I Football Players

PURPOSE: To identify demographic values that may predict isometric knee extension peak torque and establish normative values in Division I collegiate football players. METHODS: 104 healthy Division 1 football players, 19.17 years (+/- 3.70), performed knee extension isometric peak force, as part of annual baseline testing. Peak torque knee extension (ft./lb) per percent body weight was recorded. A composite limb score was calculated by mean performance across limbs for each individual. Age, height, weight, and history of lower extremity surgery were recorded. Univariable and multivariable models were created for each PPT to determine the importance of recorded descriptive variables. Quantile regression was used to develop performance percentiles for each PPT. A secondary analysis of individual Chi-Square Automatic Interaction Detections (CHAID) was utilized to identify distinct sets of significant predictors for various subgroups or levels of data. RESULTS: In univariable regressions, weight was a significant predictor for isometric knee extension (p = .001, r2 = .318) with moderate effect size. Multivariable regression showed predictive patterns for weight that were identical to those shown in univariable analysis. Quantile regression performance percentile curves were successfully created for isometric peak torque as a factor of weight. However, CHAID analysis found that football position was the only significant predictor of isometric peak torque (p = .001, r2 = .355). CHAID model, performance subgrouping was created (Figure 1.). CONCLUSIONS: Weight and position can accurately predict performance on knee extension isometric peak torque in collegiate football players. The quantile regression performance percentile curves created could serve as a tool for clinicians, enhancing their ability to benchmark Division 1 football players’ performance. CHAID modeling allows for a robust prediction of performance based on football position.