A multicenter study on the risk assessment model of fracture nonunion after intramedullary nailing operation for subtrochanteric fracture of femur

Objective To study the influencing factors of fracture nonunion after intramedullary nailing for subtrochanteric fracture and construct a risk assessment model. Methods A retrospective analysis was performed on 251 patients with intramedullary nail fractures of the femoral subtrochanteric fracture from February 2006 to January 2018. According to the different treatment time, the 251 patients included in this study were divided into the modeling group and the verification group. In the modeling group, postoperative fracture nonunion rate, general data, fracture related factors, surgical reduction related factors, mechanical and biological factors were calculated, and the influencing factors of fracture nonunion were screened by univariate analysis. Indicators with statistical differences in univariate analysis were analyzed using Logistic regression model for multivariate analysis to build the risk assessment model. The influencing factors were re-evaluated through the verification group, and the differentiation and calibration of the model were evaluated. Results Fracture nonunion occurred in 34 of 149 patients in the modeling group. Among the 13 potential influencing factors, univariate analysis and logistic regression analysis showed that postoperative hip varus, intramedullary nail fixation failure and complete open reduction were the risk factors of fracture nonunion. Postoperative reduction of medial cortex was a protective factor for fracture nonunion, and a regression equation was established. Based on the logistic regression model, the Nomogram diagram was drawn. In the verification group, fracture nonunion occurred in 24 of 149 patients. The area under the ROC curve was AUC=0.883>0.7, indicating that there was a moderate differentiation to evaluate the occurrence of fracture nonunion after operation. The goodness of fit test: the H-L test (χ2=2.921, P=0.712) showed that the model had a good calibration. Conclusion The risk factors of fracture nonunion were hip varus, failure of intramedullary nail fixation and complete open reduction after intramedullary nailing of subtrochanteric fracture, and postoperative reduction of medial cortex was the protective factor. The risk assessment model has moderate differentiation and good calibration, which can provide reference for the risk assessment of fracture nonunion after subtrochanteric fracture operation. Key words: Femoral fracture; Fracture, ununited; Forecasting; Calibration