缩窄性心包炎超声特征的logistic回归分析  

Echocardiographic features of constrictive pericarditis:a logistic regression analysis

作　　者：李国英 徐明[2] 刘玲玲 黄泳淇 LI Guo-ying;XUMing;LIU Ling-ling(Department of Ultrasound,the First Affiliated Hospital of Guangzhou Medical University,Guangzhou 510000,China)

机构地区：[1]广州医科大学附属第一医院超声科,广东广州510000 [2]广州医科大学附属第二医院超声科,广东广州510163

出　　处：《放射学实践》2023年第8期1065-1069,共5页Radiologic Practice

摘　　要：目的:利用logistic回归分析探讨超声诊断缩窄性心包炎(CP)的特征性指标。方法:回顾性分析116例疑似CP患者的常规二维、脉冲多普勒超声(PW)以及组织多普勒成像技术(TDI)超声心动图声像特征。根据手术、心导管检测或心内膜活检结果,CP组78例(其病因结核60例,占77%,特发或手术后8例,占10%,不明原因10例,占13%),非CP组38例。根据CP组与非CP组间比较结果,我们将心包增厚、室间隔抖动征、心室变形、吸气后二尖瓣E峰峰值流速变化率(△Emv)>25%、二尖瓣环隔侧舒张期组织位移速度(隔e')>8 cm/s以及隔e'>二尖瓣环外侧舒张期组织位移速度(侧e')纳入多因素logistic回归分析,确定超声特征诊断CP的独立影响因子并构建联合诊断模型。采用受试者工作特征(ROC)曲线评价各项超声特征及联合诊断模型对CP的诊断效能。结果:多因素logistic回归分析结果显示室间隔抖动征(X1)、隔e'>8 cm/s(X2)和吸气后△Emv>25%(X3)为CP的独立影响因素(P均<0.05);建立的预测CP的联合诊断模型为Logit(P)=-3.166+2.758X1+2.832X2+2.255X3。此模型预测CP的符合率为89.8%,AUC为0.931。结论:利用多因素logistic回归分析建立超声特征联合诊断模型有利于提高对CP的诊断效能。Objective:To investigate the characteristics of constrictive pericarditis(CP)by use of echocardiography.Methods:The routine two-dimensional,pulsed doppler ultrasound(PW),and tissue doppler imaging techniques(TDI)echocardiographic sonographic features of 116 suspected CP patients were retrospectively analyzed.According to the results of surgery and cardiac catheterization or endocardial biopsy,78 were divided in the CP group and 38 in the non-CP group.There were 60 cases of tuberculosis of its etiology,accounting for 77%;8 cases of idiopathic or postoperative disease,accounting for 10%;And 10 cases of unknown cause,accounting for 13%.According to the results of comparison between CP group and non-CP group,we divided pericardial thickening,ventricular septal shift,ventricular deformation,the rate of change of mitral valve E peak velocity after inhalation(△Emv)>25%,septal diastolic tissue displacement velocity of mitral annulus(septal e')>8cm/s and septum e'>the lateral diastolic tissue displacement velocity of the mitral annulus(lateral e')were included in the multivariate logistic regression model for analysis,and the independent influencing factors of ultrasound features in the diagnosis of CP were finally established.The receiver characteristic(ROC)curve was used to evaluate the efficacy of multiple ultrasound features in diagnosing CP alone and in combination,respectively.Results:Ventricular septal shift(X1),septal e'>8cm/s(X2)and△Emv rate of change after inspiration≥25%(X3),these three variables were independent influencing factors of CP(all P<0.05).The CP diagnostic model was Logit(P)=-3.166+2.758X1+2.832X2+2.255X3,and the accuracy of this model forecast was 89.8%,and the AUC was 0.931.Conclusion:The establishment of multi-factor logistic regression model for ultrasound features is conducive to improve the diagnostic efficiency of CP.

关 键 词：超声心动图 缩窄性心包炎 LOGISTIC回归分析 诊断模型 诊断效能 

分 类 号：R445.1[医药卫生—影像医学与核医学] R542.1[医药卫生—诊断学]