危重症患者下肢深静脉血栓风险预测模型的构建及评价

目的建立危重症患者下肢深静脉血栓风险预测模型,并评价预测效能。方法选取重症医学科患者420例,分为建模组300例和验证组120例。应用Logistic回归分析筛选下肢深静脉血栓发生的独立危险因素,建立下肢深静脉血栓风险预测模型。应用Bootstrap法进行模型内部验证,利用验证组数据对模型进行外部验证。结果Logistic回归分析显示,血浆D-二聚体、机械通气、静脉血栓史、血管升压素和糖尿病是危重症患者发生下肢深静脉血栓的独立危险因素。建模组风险预测模型的曲线下面积为0.935,验证组曲线下面积为0.925。Hosmer-Lemeshow检验结果P=0.901。结论本研究建立的列线图预测效能较好,可操作性强,有利于下肢深静脉血栓早筛查、早诊断。

Development and evaluation of a prediction model for lower limb deep venous thrombosis in critically ill patients

Objective To develop a prediction model for lower limb deep venous thrombosis (DVT) in critically ill patients and to evaluate the predictive validity. Methods A total of 420 critically ill patients were divided into two parts:300 cases for model development and 120 cases for verification. Logistic regression analysis was performed to identify risk factors and formulate prediction model. Bootstrap method was used for internal validation of the model, and 120 cases for external validation of the model. Results Logistic regression analysis showed that plasma D-dimer, mechanical ventilation, venous thromboembolism history, vasopressor use and diabetes were independent risk factors for lower limb DVT in critically ill patients. The area under ROC curve was 0.935 in internal validation and 0.925 in external validation. Hosmer-Lemeshow test showed P=0.901. Conclusion The nomogram established in this study for prediction of lower limb DVT in critically ill patient population has good predictive performance and strong operability, which is conducive to early screening and diagnosis.