左髂总静脉受压继发髂股静脉血栓形成风险预测模型构建和评价  

Establishment and evaluation of a risk prediction model used for predicting iliofemoral venous thrombosis secondary to compression of the left common iliac ve

作　　者：朱桥华 杨轮 黄江远 刘勇刚 黄维 罗美华 陈德华 余启荣 ZHU Qiaohua;YANG Lun;HUANG Jiangyuan;LIU Yonggang;HUANG Wei;LUO Meihua;CHEN Dehua;YU Qirong(Department of Oncology and Vascular Intervention,Shunde Hospital,Southern Medical University,Foshan,Guangdong Province 528308,China)

机构地区：[1]南方医科大学顺德医院肿瘤和血管介入科,广东佛山528308 [2]珠海市慢性病防治中心 [3]南方医科大学顺德医院影像诊断科 [4]广州中医药大学顺德医院放射科

出　　处：《介入放射学杂志》2023年第4期371-375,共5页Journal of Interventional Radiology

基　　金：广东省医学科学技术研究基金(B2022125)。

摘　　要：目的 构建并评价左髂总静脉(LCIV)受压继发髂股静脉血栓形成的风险预测模型。方法 回顾性分析2012年11月至2019年12月在南方医科大学顺德医院接受血管内介入治疗的血栓性髂静脉受压综合征患者临床资料。根据纳入和排除标准选取治疗组和对照组,两组患者按年龄-性别1∶1进行匹配。根据CT横断面上所测LCIV和右髂总静脉最小短径计算LCIV受压比例。采用Padua量表评分评估静脉血栓栓塞症(VTE)危险因素,多因素logistic回归分析构建左髂股静脉血栓形成风险预测模型,受试者工作特征曲线(ROC)、Hosmer-Lemeshow拟合优度检验及k折交叉验证评价模型预测效能。结果 治疗组共纳入93例患者,其中男23例,女70例,年龄(57.1±16.9)岁(22~88岁)。治疗组、对照组LCIV受压比例分别为(77.10±12.88)%、(42.11±21.22)%(P<0.01),LCIV受压与髂股静脉血栓形成呈相关性,并与VTE危险因素间存在协同作用。所构建预测模型:P(髂股静脉血栓形成概率)=exp(logit P)/[1+exp(logit P)],logit P=-9.07+0.61×VTE危险因素评分+0.13×LCIV受压比例。logit P的ROC曲线下面积为0.940(0.908~0.971),Hosmer-Lemeshow拟合优度为P=0.563,k折交叉验证训练准确度为0.871±0.007,预测准确度为0.844±0.048。结论 基于LCIV受压比例和Padua危险因素评分所建左髂股静脉血栓形成风险模型具有良好的预测效能。Objective To establish and evaluate a risk prediction model used for predicting iliofemoral venous thrombosis secondary to compression of the left common iliac vein(LCIV). Methods The clinical data of patients with thrombotic iliac vein compression syndrome, who received endovascular intervention therapy at the Shunde Hospital of Southern Medical University of China between November 2012 and December2019, were retrospectively analyzed. According to the inclusion and exclusion criteria, patients of study group and control group were selected. The patients were matched by age-sex at 1:1 ratio in both groups. The LCIV compression ratio was calculated based on the minimum short diameters of LCIV and right common iliac vein,which were measured on CT transverse images. Padua scale score was used to assess the risk factors for venous thromboembolism(VTE). Multivariate logistic regression analysis was used to establish the risk prediction model for left iliofemoral vein thrombosis, the receiver operating characteristic(ROC) curve, Hosmer-Lemeshow goodnessof-fit test and κ-fold cross validation were used to evaluate the model performance. Results A total of 93 patients were included in the study group, including 23 males and 70 females, with a mean age of(57.1±16.9) years(range of 22-88 years). The compression ratio of LCIV in study group and control group was(77.10±12.88)%and(42.11±21.22)% respectively(P<0.01), the compression degree of LCIV was correlated with the iliofemoral vein thrombosis and had a synergistic interaction with the VTE risk factors. In the established predictive model, P(iliofemoral vein thrombosis probability)=exp(logit P)÷[1+exp(logit P)], logit P=-9.07+0.61×VTE risk factor score+0.13×LCIV compression ratio. The area under ROC curve of logit P was 0.940(0.908-0.971), HosmerLemeshow goodness of fit was P=0.563, κ-fold cross-validation training accuracy was(0.871±0.007), and prediction accuracy was(0.844±0.048). Conclusion The risk prediction model, which is established based on LCIV compres

关 键 词：髂静脉受压 髂股静脉血栓形成 风险预测模型 预测效能 

分 类 号：R654.4[医药卫生—外科学]