摘要
目的建立肝硬化患者伴门静脉血栓形成(portal vein thrombosis,PVT)的列线图预测模型并进行验证,以评价其临床应用价值。方法收集2015年1月~2022年10月笔者医院收治的464例肝硬化伴或不伴PVT患者的临床资料,应用单因素和多因素Logistic回归分析筛选肝硬化患者发生PVT的独立危险因素并建立列线图预测模型,并对列线图预测模型进行区分度、校准度和临床有效性验证。结果单因素及多因素Logistic回归分析结果显示,脾切除史、门静脉直径、D-二聚体及C反应蛋白与白蛋白比值(C-reactive protein-to-albumin ratio,CAR)是肝硬化伴PVT的独立危险因素(P<0.05),将上述独立危险因素纳入并成功建立列线图预测模型。分别在建模人群及验证人群中对列线图模型进行验证:其受试者工作特征(receiver operating characteristic,ROC)曲线下面积分别为0.846(95%CI:0.803~0.888)和0.815(95%CI:0.743~0.887),预测模型区分度良好;Hosmer-Lemeshow检验中P值分别为0.783和0.453,模型拟合良好;校准图显示,列线图模型对肝硬化伴PVT的预测概率显示出良好的校准;临床决策曲线分析(decision curve analysis,DCA)表明,该列线图具有良好的应用价值。结论通过建立并验证一种简单、实用的肝硬化患者伴发PVT风险的列线图预测模型,可以帮助临床医师早期识别及干预高危患者。
Objective To establish and verify nomogram prediction model with portal vein thrombosis in patients with cirrhosis,so as to evaluate its clinical application value.MethodsThe clinical data of 464 patients with cirrhosis with or without PVT admitted by our hospital from January 2015 to October 2022 were collected,univariate and multivariate Logistic regression analysis were used to screen the independent risk factors for PvT in patients with liver cirrhosis,and a nomogram prediction model was established,and the nomogram pre-diction model was verified for discrimination,calibration and clinical validity.Results Univariate and multivariate Logistic regression a-nalysis showed that the history of splenectomy,portal vein diameter,D-dimer and C-reactive protein-to-albumin ratio(CAR)were independent risk factors for cirrhosis with portal vein thrombosis(P<O.05),the above independent risk factors were incorporated and a nomogram prediction model was successfully established.The nomogram was verified in both the modeling and validation populations:the area under the receiver operating characteristic(R0C)curve were 0.846(95%CI:0.803-0.888)and 0.815(95%Cl:0.743-0.887),respectively,and the prediction model had good discrimination.The P-values in Hosmer-Lemeshow test were 0.783 and O.453,respectively,and the model fitted well.The calibration chart showed that the nomogram model showed good calibration for the pre-diction probability of liver cirrhosis with PVT.Decision curve analysis(DCA)showed that the nomogram had good application value.Conclusion By establishing and validating a simple and practical nomogram prediction model for the risk of PVT in patients with cirrho-sis,it can help clinicians identify and intervene high-risk patients early.
作者
朱梦幻
李多华
吴克俭
ZHU Menghuan;LI Duohua;WU Kejian(The First Clinical Medical Collge of Xuzhou Medical Universily,Jiangsu 221000,China)
出处
《医学研究杂志》
2024年第7期156-162,共7页
Journal of Medical Research
