Purpose
To develop a model to predict re-injury after being cleared to return to full duty from an initial injury.
Methods
This was a prediction model derivation cohort study. Military service members cleared for unrestricted full duty after sustaining a musculoskeletal injury were enrolled from 3 large military hospitals. Medical history, demographics, psychological profile, physical performance (Y-Balance TestTM, Functional Movement ScreenTM, Selective Functional Movement Assessment, triple hop, closed chain ankle dorsiflexion, 2-mile run, 75% bodyweight carry time), and past injury history were assessed. Monthly text messages, medical records and limited duty databases were used to identify injuries resulting in time lost from work in the following year.
Results
450 participants (65 females), ages 18-45 were analyzed. 15 variables were included in the final model. The Area Under the Curve was 0.74 (95CI 0.69, 0.80), indicating good performance. The calibration score of the model was 1.05 (95CI 0.80, 1.30) indicating very good performance. With an injury incidence in our cohort of 38.0%, the treat all net benefit was 0.000, and the net benefit of our predictive model was 0.251. This means 25 additional soldiers out of every 100 were correctly identified as high risk for injury compared to not using a prediction model at all.
Conclusions
This multivariable model accurately predicted injury risk after returning for full duty and was better than not using a prediction model at all (an additional 25 out of every 100 tactical athletes were correctly identified). This model provides guidance for proper decision-making about when these individuals are not ready to return to full duty, with higher risk of a subsequent injury.
Click to expand...