Development and validation of a model for predicting refracture in elderly patients with hip fracture

Abstract

Abstract Purpose The goal of our study was to determine the incidence of refractures within two years after hip fracture surgery in elderly patients. LASSO Cox was used to gather independent indicators. The nomogram model was developed to predict refractures. Methods In this retrospective analysis, 1254 participants were recruited between 2019 and 2021 at two general hospitals in China. Baseline variables were collected, including demographics, comorbidities, and living conditions. The primary outcome was refractures within two years. This model was developed using LASSO Cox and validated using bootstrapping methods. The calibration and discriminative ability were evaluated using C-index values, AUC values, calibration plots, and decision curves. Results A total of 119 patients with refractures were identified via telephone interview. Seven independent indicators were predictors for refractures, including age, Harris, anemia, neurologic disease, visual impairment, anti-osteoporosis, and exercise. Based on bootstrap resampling, the calibration curve value was 0.869 (95% CI: 0.795–0.987), and the AUC was 0.793 (95% CI 0.751–0.834). Conclusion This model shows potential as a prediction instrument that permits clinicians to precisely detect the high risk of refractures and further help patients make decisions.