Abstract: BACKGROUND: Intertrochanteric femoral fractures are the main type of fragility fracture in the elderly, and proximal femoral nail antirotation is the preferred surgical option, but the factors associated with postoperative internal fixation failure are controversial.
OBJECTIVE: A new “three-column” classification of intertrochanteric femoral fractures was proposed by evaluating patients’ imaging data preoperatively and analyzing its interaction with postoperative internal fixation failure. A risk prediction model was developed and validated by using numerical algorithms, which facilitates clinicians to identify and intervene in high-risk patients preoperatively. 
METHODS: Patients with intertrochanteric femur fractures in Sanshui Branch of Foshan Hospital of Traditional Chinese Medicine between June 2012 and June 2022 were selected. The patients were divided into the internal fixation failure group and the internal fixation maintenance group according to whether they had internal fixation failure after surgery. According to the preoperative radiographs, the proximal femur was divided into three columns: the medial column, the lateral column, and the middle column. Each column had different subgroups. The relationship between the morphological characteristics of the “three columns” and the failure of proximal femoral nail antirotation internal fixation was analyzed, and the independent risk factors for internal fixation failure were screened out by single and then multifactorial logistic regression analyses. A risk prediction model was constructed according to the independent risk factors using R language software. The Bootstrap method was used to resample 1 000 times. The area under the curve, calibration curve, and clinical decision curve were used to evaluate the differentiation, calibration ability, and clinical application value of the model. The Youden index was used to determine the optimal risk cut-off value of the prediction model, according to which the patients were divided into high and low risk groups. The stability and extensibility of the model were evaluated according to the accuracy of its risk prediction ability.
RESULTS AND CONCLUSION: (1) The four independent risk factors for postoperative internal fixation failure after surgery were predicted using the “three-column” typing system: medial column (comminuted fracture of the lesser trochanter and femoral talar) [odds ratio=5.385, 95%CI(1.961, 14.782), P=0.001], medial column (chimney type) [odds ratio=2.893, 95%CI(1.167, 7.173), P=0.022], lateral column (lateral wall thickness < 20.5 mm) [odds ratio=2.804, 95%CI(1.078, 7.297), P=0.035] and lateral column (lateral wall fracture) [odds ratio=4.278, 95%CI(1.670, 10.959), P=0.012]. (2) The constructed risk prediction model showed good discrimination and accuracy [area under the receiver operating characteristic curve=0.852, 95%CI(0.837, 0.922)]. The calibration curve showed good agreement between the model-predicted risk and the actual risk of occurrence. (3) The clinical decision curve suggested that the model had good clinical applicability when the risk threshold probability was in the range of 0.2-0.82. The risk probability of 28% was the optimal threshold for risk stratification of the model, and the predictive performance of the model was better in patients with different risk groups. (4) The “three-column” typing system constructs a predictive model to calculate the risk probability of postoperative internal fixation failure in patients with intertrochanteric femoral fractures. This method is accurate, simple, and easy to apply clinically, and can be used as a digital tool to guide personalized clinical treatment.

Key words: intertrochanteric femoral fracture, proximal femoral nail antirotation, independent risk factor, internal fixation failure, typing system, risk prediction model, orthopedic implants