軍事士官候補生および海軍兵学校生徒における膝の使いすぎによる傷害の予測

Musculoskeletal injuries are prevalent during military training, with overuse knee injuries representing a major source of medical attention and time-loss. The early transition into military academy life is marked by considerable physical and psychological stressors, creating a high-risk window for injury development-particularly among individuals with an injury history. Thus, the aim of this study was to develop and internally validate a multivariable prediction model for overuse knee injuries among first-year military cadets with a history of knee injury.

This was a prospective cohort study, which included 1,265 newly matriculated cadets and midshipmen with a recent history of knee injury from the U.S. Air Force, Army, and Naval Academies. Participants completed standardized baseline testing, including sport and physical training history, lower-extremity isometric strength, and jump-landing biomechanical assessments. Incident overuse knee injuries were prospectively tracked over a 9-month period using medical record review. A multivariable logistic regression was used to develop a prediction model and dynamic nomogram for real-world use. Decision curve analysis was completed to evaluate clinical utility.

Among our sample, 389 (30.8%) trainees sustained at least 1 overuse knee injury within their first academic year. The internally validated prediction model demonstrated moderate discrimination (area under the receiver operator characteristic [AUC] = 0.66; 95% CI, 0.65, 0.67) and stable calibration (0.79; 95% CI, 0.77, 0.81). Decision curve analysis indicated that our final prediction model would correctly identify 29 additional participants out of every 100 as high risk for injury compared with not using a model at all.

This study presents a novel, internally validated prediction model for overuse knee injuries in a high-risk trainee population with prior knee injury. Subgrouping by prior injury status performed better than applying the model to the entire cohort, highlighting the potential efficiency of anatomically specific injury history as a first-level filter for development of injury prediction models. Although this specific study's model performance was moderate, the decision curve analysis supports its potential clinical utility for guiding targeted prevention efforts in military trainees.