Analysis of high-risk factors for preoperative DVT in elderly patients with simple hip fractures and construction of a nomogram prediction model

Abstract

Abstract Background Hip fractures are anatomically classified in relation to femoral neck, intertrochanteric or subtrochanteric fractures. Simple hip fractures discussed in this study are femoral neck fractures or intertrochanteric fractures, which are the most common types of hip fractures. Controversy remains regarding the value of biochemical indices of thrombosis in elderly patients with fractures. A retrospective study was conducted to investigate the index admission data in blood draws of elderly patients with hip fractures and their high-risk factors for deep venous thrombosis (DVT). A nomogram prediction model for DVT was established to facilitate a rapid, accurate, and effective prediction based on the results. Methods The data were based on 562 elderly patients undergoing hip fracture surgery, from whom 274 patients were selected for enrollment. The 274 patients were divided into two groups using preoperative vascular color Doppler ultrasonography. Chi-square tests, t-tests, and U tests were conducted, and logistic regression analysis was conducted showing different factors between the two groups. Independent risk factors with statistical significance (P < 0.05) were obtained, and the logistic regression equation and the new variable prediction probability_1 (PRE_1) were constructed. The receiver operating characteristic (ROC) curve of risk factors and PRE_1 was drawn to obtain the area under the curve (AUC) and truncation value of each risk factor. Finally, a nomogram prediction model was constructed using the R programming language to calculate the concordance index (C-index). Results Time from injury to hospitalization, platelet (PLT) count, D-dimer level, fibrinogen (FIB) level, and systemic immune-inflammatory index (SII) score were independent risk factors for preoperative DVT in elderly patients with hip fractures. The logistic regression equation and PRE_1 were constructed by combining the above factors. ROC analysis showed that the area under the curve for PRE_1 (AUC = 0.808) was greater than that of the other factors. The sensitivity of PRE_1 (sensitivity = 0.756) was also higher than that of the other factors, and the specificity of PRE_1 (specificity = 0.756) was higher than that of two other factors. Moreover, a predictive nomogram was established, and the results showed a high consistency between the actual probability and the predicted probability (C-index = 0.808), indicating a high predictive value in fractures accompanied by DVT. Conclusions This study confirmed that SII score could be used as a risk factor in the prediction of DVT occurrence. A nomogram prediction model was constructed by combining 5 independent risk factors: time from injury to admission, PLT count, D-dimer level, FIB level, and SII score, which had high predictive values for fractures accompanied by DVT. This model use is limited to simple hip fracture.