【摘要】 目的 联合Caprini评分和凝血指标D-二聚体构建PICC相关性血栓可视化模型。方法 回顾性纳入了2018年9月—2022年9月在我院行PICC置管并维护的413例患者，收集患者的人口学资料、置管资料、Caprini评分以及D-二聚体等相关指标，通过多因素Logistic回归模型分析独立危险因素，MedCalc软件分别绘制Caprini评分、D-二聚体以及两者联合的受试者工作特征曲线（ROC曲线）并对比曲线下面积（AUC），采用R软件绘制列线图。结果 本研究共纳入413例患者，发生PICC相关性血栓患者为38例（9.2%）。多因素Logistic回归分析结果显示，年龄、Caprini评分、D-二聚体、BMI、既往病史、化疗、导管规格、肢体选择为发生PICC相关性血栓发生的独立危险因素。对比ROC曲线下面积（AUC）:Caprini评分单独预测为0.714,D-二聚体单独预测为0.636，两者联合预测为0.876,(95%CI:0.840-0.906)。通过对分析结果的可视化处理，显示联合预测模型的区分度较高，Brier评分和校正曲线都表现了较好的校准度。H-L检验（χ2=3.505,P>0.05）结果显示，该模型的拟合度较好。结论 Caprini评分联合凝血指标D-二聚体构建的PICC相关性血栓可视化模型具有较好的预测效果，可以有效地预测PICC相关性血栓的早期风险，为临床治疗提供有力的参考依据。更多还原

【Abstract】 Objective To construct a visualization model of PICC-catheter related thrombosis risk with Caprini score and D-dimer.Methods A total of 413 patients who received PICC catheterization and maintenance in our hospital from September 2018 to September 2022 were retrospectively included. Demographic data, catheterization data, Caprini score, D-dimer and other related indicators of patients were collected. Independent risk factors were analyzed by using multiple logistic regression. Caprini score, Ddimer and their combined receiver operating characteristic curve(ROC curve) were drawn respectively by MedCalc software, and the area under the curve(AUC) was compared. Results The incidence of PICC-associated thrombosis was 9.2%(38 patients). Multiple logistic regression showed that age, Caprini score, D-dimer, BMI, past medical history, chemotherapy, catheter size, and limb selection were independent risk factors for PICC-catheter related thrombosis. The AUC of Caprini score alone was 0.714; that of Ddimer alone 0.636, and that of the combined 0.876(95%CI: 0.840～0.906). Visualization processing showed that the joint prediction model had a high degree of differentiation, and the Brier score and correction curve both showed a good calibration degree. H-L inspection(χ2=3.505,P>0.05) showed good fitting of the model. Conclusion The visualization model of PICC-catheter related thrombosis constructed by Caprini score combined with D-dimer can effectively predict the early risk of PICC-catheter related thrombosis, and provide reference for clinical treatment.更多还原