Context:Neurocognitive reaction time has been associated with musculoskeletal injury risk, but visuomotor reaction time (VMRT) derived from tests that present greater challenges to visual stimulus detection and motor response execution may have a stronger association.Objective:To assess VMRT as a predictor of injury and the extent to which improvement may result from VMRT training.Design:Cohort study.Setting:University athletic performance center.Participants:76 National Collegiate Athletic Association Division-I FCS football players (19.5 ± 1.4 y, 1.85 ± 0.06 m, 102.98 ± 19.06 kg).Interventions:Preparticipation and postseason assessments. A subset of players who exhibited slowest VMRT in relation to the cohort’s postseason median value participated in a 6-wk training program.Main Outcome Measures:Injury occurrence was related to preparticipation VMRT, which was represented by both number of target hits in 60 s and average elapsed time between hits (ms). Receiver operating characteristic analysis identified the optimum cut point for a binary injury risk classification. A nonparametric repeated-measures analysis of ranks procedure was used to compare posttraining VMRT values for slow players who completed at least half of the training sessions (n = 15) with those for untrained fast players (n = 27).Results:A preparticipation cut point of ≤85 hits (≥705 ms) discriminated injured from noninjured players with odds ratio = 2.30 (90% confidence interval, 1.05–5.06). Slow players who completed the training exhibited significant improvement in visuomotor performance compared with baseline (standardized response mean = 2.53), whereas untrained players exhibited a small performance decrement (group × trial interaction effect, L2 = 28.74; P < .001).Conclusions:Slow VMRT appears to be an important and modifiable injury risk factor for college football players. More research is needed to refine visuomotor reaction-time screening and training methods and to determine the extent to which improved performance values can reduce injury incidence.
Efficient maximum likelihood estimation in semiparametric mixture models
