ABSTRACT
PURPOSE: To develop a risk prediction model for postoperative sarcopenia in elderly patients with patellar fractures in China. PATIENTS AND METHODS: We conducted a community survey of patients aged ≥55 years who underwent surgery for patellar fractures between January 2013 and October 2018, through telephone interviews, community visits, and outpatient follow-up. We established a predictive model for assessing the risk of sarcopenia after patellar fractures. We developed the prediction model by combining multivariate logistic regression analysis with the least absolute shrinkage model and selection operator regression (lasso analysis) as well as the Support Vector Machine (SVM) algorithm. The predictive quality and clinical utility of the predictive model were determined using C-index, calibration plots, and decision curve analysis. We also conducted internal sampling methods for qualitative assessment. RESULT: We recruited 137 participants (53 male; mean age, 65.7 years). Various risk factors were assessed, and low body mass index and advanced age were identified as the most important risk factor (P < 0.05). The prediction rate of the model was good (C-index: 0.88; 95% CI [0.80552-0.95448]), with a satisfactory correction effect. The C index is 0.97 in the validation queue and 0.894 in the entire cohort. Decision curve analysis suggested good clinical practicability. CONCLUSION: Our prediction model shows promise as a cost-effective tool for predicting the risk of postoperative sarcopenia in elderly patients based on the following: advanced age, low body mass index, diabetes, less outdoor exercise, no postoperative rehabilitation, different surgical methods, diabetes, open fracture, and removal of internal fixation.
