Analysis of influencing factors of early femoral head necrosis after femoral neck fracture and construction of nomogram prediction model

Abstract

To explore the risk factors of early femoral head necrosis in patients with femoral neck fracture after operation, and to establish a nomogram prediction model. A total of 167 patients with femoral neck fracture from January 2020 to April 2022 were selected and divided into necrosis group and non-necrosis group according to whether femoral head necrosis occurred in the early postoperative period. There were 21 males and 17 females in the necrosis group, aged from 33 to 72 years old, with an average of (53.49±10.96) years old, and the time from injury to operation ranged from 40 to 67 hours, with average time of(53.46±7.23) hours. There were 72 males and 57 females in the non-necrosis group, aged from 18 to 83 years, with an average of (52.78±12.55) years old, and the time from injury to operation was 18 to 65 hours, with an average time of(39.88±7.79) hours. The potential influencing factors, including patient gender, diabetes mellitus, hypertension, chronic liver disease, posterior inclination angle of the femoral head, operation mode, fracture displacement, fracture line location, preoperative braking traction, screw arrangement mode, reduction quality, age, body mass index(BMI), and injury to operation time were subjected to single factor analysis. Logistic multivariate regression analysis was conducted for factors with a significance level of P<0.05. The incidence of femoral head necrosis in 167 patients with femoral neck fracture was 22.76%. The following factors were identified as independent risk factors for early postoperative femoral head necrosis in patients with femoral neck fractures:coexisting diabetes[OR=5.139, 95%CI(1.405, 18.793), P=0.013], displaced fracture [OR=3.723, 95%CI(1.105, 12.541), P=0.034], preoperative immobilization[OR=3.444, 95%CI(1.038, 11.427), P=0.043], quality of reduction [OR=3.524, 95%CI(1.676, 7.411), P=0.001], and time from injury to surgery[OR=1.270, 95%CI(1.154, 1.399), P=0.000]. The Hosmer-Lemeshow goodness-of-fit test(χ2=3.951, P=0.862), the area under the receiver operator characteristic(ROC) curve was 0.944[P<0.001, 95%CI(0.903, 0.987)], with a sensitivity of 89.50%, the specificity was 88.40%, the maximum Youden index was 0.779, and the overall trend of the model correction curve was close to the ideal curve. Model regression equation was Z=1.637 × diabetes + 1.314× fracture displacement+1.237 × preoperative braking traction+1.260 × reduction quality + 0.239×injury to operation time-18.310. The occurrence of early femoral head necrosis in patients with femoral neck fracture postoperatively is affected by multiple factors. The risk early warning model established according to the factors has good predictive efficacy.