关键词: 股骨颈骨折, 股骨头坏死, 内固定, 骨折并发症, 预测模型, 骨科植入物

Abstract: BACKGROUND: The most common complication of traumatic femoral neck fractures after internal fixation is femoral head necrosis. Currently, many studies have reported on the risk factors that affect the occurrence and development of postoperative femoral head necrosis, but there is still a lack of tools to predict the risk of femoral head necrosis after internal fixation of femoral neck fractures.
OBJECTIVE: To develop a predictive model that estimates the risk of femoral head necrosis shortly after patients with femoral neck fractures receive cannulated screw internal fixation.
METHODS: A retrospective analysis reviewed clinical records of 172 patients who underwent cannulated screw internal fixation for femoral neck fractures at Department of Orthopedics of Mianyang Central Hospital from January 2013 to June 2023. Patients were categorized into two groups based on the presence or absence of femoral head necrosis within one year post-operation: the necrosis group and the non-necrosis group. Univariate analysis, Lasso regression, and multivariate Logistic regression techniques were employed to identify the determinants of femoral head necrosis. A nomogram prediction model was constructed using R language's "rms" package, version 4.0. The receiver operating characteristic curve was used to evaluate the discriminatory ability of the model. The Hosmer-Lemeshow test was used to evaluate the goodness of fit of the model, and the decision curve analysis was used to determine its clinical application benefits. Internal validation of the study was conducted using the Bootstrap method, involving 1 000 repeated samplings. To delve deeper into the primary factors influencing femoral head necrosis post-internal fixation of the femoral neck, this paper employed the SHAP method for data set analysis. 
RESULTS AND CONCLUSION: (1) The risk factors leading to femoral head necrosis in the short term after cannulated screw fixation of femoral neck fractures include: smoking, diabetes, Garden classification, fracture line location, reduction quality, age, and operation time. (2) The prediction model demonstrated robust performance, evidenced by an area under the curve of 0.940 (95% Confidence Interval: 0.903 to 0.977), indicating a high level of prediction accuracy. The model achieved a sensitivity of 90.2% and a specificity of 87.6%, indicating that its diagnostic performance was stable. The Hosmer-Lemeshow goodness-of-fit test yielded a chi-square value of 6.593 with a P-value of 0.581, confirming that the model’s predictions closely align with the observed outcomes. (3) The calibration curve of the model also performed well, and its overall trend was very close to the ideal curve, further proving the high accuracy of the model. (4) The internal validation was carried out by the Bootstrap method with 1 000 repeated samplings, and the area under the curve of the model internal validation was still as high as 0.939, proving that the model had good stability. (5) Through the decision curve, it is found that within the probability threshold range of 1% to 92%, the model can obtain the maximum net benefit value. (6) The SHAP analysis results show that among the risk factors analyzed in this study, the location of the fracture line serves as the most significant predictor of femoral head necrosis following internal fixation with cannulated screws in femoral neck fractures, and subcapital fractures are extremely prone to femoral head necrosis after surgery. (7) It is concluded that the validated prediction model demonstrates strong discriminative power and reliability, offering practical clinical utility. It serves as a useful reference tool for short-term risk assessment of femoral head necrosis following internal fixation of femoral neck fractures.

Key words: femoral neck fracture, femoral head necrosis, internal fixation, fracture complication, prediction model, orthopedic implant