Multidimensional pre-season screening to predict musculoskeletal injuries in competitive swimmers: a prospective cohort study

Swimming is physically demanding, yet prospective evidence identifying independent injury predictors remains limited, particularly regarding the predictive value of multidimensional functional screening batteries. In this 9-month prospective cohort study, 170 competitive swimmers (aged 18–60 years) underwent baseline assessment of demographics, training load, injury history, and kinetic-chain function. Musculoskeletal injuries were monitored biweekly using the OSTRC Overuse Injury Questionnaire. To prevent overfitting, Least Absolute Shrinkage and Selection Operator (LASSO) penalized regression was used for variable selection, followed by a refitted multivariable logistic regression to identify independent predictors. Model performance was assessed using discrimination and calibration metrics. Fifty-two swimmers (30.6%) sustained at least one musculoskeletal injury (incidence: 2.29 per 1,000 training hours). The LASSO-selected 7-predictor model demonstrated excellent discrimination (AUC = 0.886, 95% CI 0.840–0.933) and good calibration. Independent risk factors included previous injury (adjusted OR = 5.45, 95% CI 2.27–14.00), greater weekly training time (OR = 1.46, 95% CI 1.18–1.85), and higher BMI (OR = 1.37, 95% CI 1.24–1.53). Significant protective factors included a structured warm-up (OR = 0.35, 95% CI 0.16–0.75), greater left-arm explosive power (OR = 0.97, 95% CI 0.96–0.98), and a higher trunk flexor-to-total endurance ratio (OR = 0.96, 95% CI 0.92–0.99). A multidimensional pre-season screening protocol integrating anthropometrics, training exposure, and kinetic-chain functional metrics may have potential as a tool for identifying swimmers at higher injury risk. Although the identified modifiable factors provide potential targets for future injury-prevention strategies, these findings should be interpreted cautiously. External validation in larger, stroke-stratified cohorts is needed before the model can be considered for clinical or applied use.