The risk assessment model of fracture nonunion after intramedullary nailing for subtrochanteric femur fracture.

To investigate the influencing factors of fracture nonunion after intramedullary nailing for subtrochanteric fractures and to construct a risk assessment model.Based on the multicenter retrospective analysis of 251 patients, all patients were divided into modeling group and 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. Logistic regression model was used for multifactor analysis to construct the risk assessment model. Based on the logistic regression model, the risk prediction model was constructed by drawing the Nomogram diagram. Through the verification group, the influencing factors were evaluated again, and the differentiation and calibration of the model were evaluated. The calibration degree was evaluated by Hosmer-Lemeshow test, goodness of fit test, and calibration curve. The discriminant degree was evaluated by the receiver operating characteristic curve.Fracture nonunion occurred in 34 of 149 patients in the modeling group. Among the 14 potential influencing factors, univariate analysis and logistic regression analysis showed that postoperative hip varus, intramedullary nail fixation failure, and reduction of fracture with large incision were the risk factors of fracture nonunion. The medial cortex fracture was seen reduced on X-Ray was a protective factor for fracture nonunion, and a regression equation was established. Based on the logistic regression model, the Nomogram diagram is drawn. Twenty-four cases of fracture nonunion occurred in the verification group. The area under the receiver operating characteristic curve was area under curve =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 Hosmers-Lemeshow test (X2 = 2.921, P = .712 > .05) showed that the model had a good calibration.After intramedullary nailing of subtrochanteric fracture, hip varus, failure of intramedullary nail fixation and wide surgical dissection are the risk factors of fracture nonunion, and the postoperative reduction of medial cortex fracture is protective factor.National key research and development projects: 2016YFC0105806.