作者：王云涛，李少珂，张娜娜，刘怡雪，王立成

单位：

1.453000河南省新乡市，新乡医学院 2.450000河南省郑州市第七人民医院心血管外二科 通信作者：王立成，E-mail：man.58@163.com

Units：

1.Xinxiang Medical College, Xinxiang 453000, China 2.The Second Department of Cardiovascular Surgery, Zhengzhou No.7 People's Hospital, Zhengzhou 450000, China Corresponding author: WANG Licheng, E-mail: man.58@163.com

关键词：

心房颤动; 术后心房颤动; 冠状动脉旁路移植术,非体外循环; 影响因素分析; 逻辑回归; K近邻;

Keywords：

Atrial fibrillation; Postoperative atrial fibrillation; Coronary artery bypass, off-pump; Root cause analysis;Logistic regression; K-nearest neighbors; Follow-up studies

CLC：

DOI：

10.12114/j.issn.1008-5971.2022.00.112

Funds：

河南省医学科技攻关计划项目（LHGJ20200734）

摘要：

目的 探讨行非体外循环冠状动脉旁路移植术(OPCABG)患者发生术后心房颤动(POAF)的影响因素,构建分类器模型,并分析随访12个月心脏彩色超声检查结果。方法 选择2016年4月至2018年4月在郑州市第七人民医院行OPCABG的患者202例,按照3∶1的比例随机分割为训练集(n=151)和测试集(n=51)。收集患者围术期临床资料。采用多因素Logistic回归模型分析行OPCABG患者发生POAF的影响因素,基于逻辑回归(LR)和K近邻算法(KNN)构建分类器模型。随访两组患者术后12个月心脏彩色超声检查结果及主要不良心脑血管事件(MACCE)发生情况。结果 训练集患者根据是否发生POAF分为POAF组和非POAF组。多因素Logistic回归分析结果显示,年龄、合并高血压、合并急性心肌梗死、合并右冠状动脉狭窄、合并频发房性期前收缩或房性心动过速、合并肾功能不全、术前服用β-受体阻滞剂、术前左心房前后径(LAAP)、术后左心室射血分数(LVEF)是行OPCABG患者发生POAF的影响因素(P<0.05)。LR分类器拟合的回归方程为:ln〔PPOAF/(1-PPOAF)〕=0.107×年龄+0.092×合并高血压+0.052×合并急性心肌梗死+0.057×合并右冠状动脉狭窄+0.104×合并频发房性期前收缩或房性心动过速+0.084×合并肾功能不全-0.097×术前服用β-受体阻滞剂+0.053×术前LAAP-0.056×术后LVEF-0.318。测试集数据构建混淆矩阵,两种模型的准确度分别是0.922和0.882。非POAF组和POAF组术后3、6、12个月LVEF分别较本组术前升高,左心室收缩末期内径(LVESD)、左心室舒张末期内径(LVEDD)和LAAP分别较本组术前降低(P<0.05)。术后12个月POAF组心房颤动复发8例,复发率为22.2%。术后12个月POAF组MACCE发生率高于非POAF组(χ2=5.043,P=0.025)。结论 年龄、合并高血压、合并急性心肌梗死、合并右冠状动脉狭窄、合并频发房性期前收缩或房性心动过速、合并肾功能不全、术前服用β-受体阻滞剂、术前LAAP、术后LVEF是行OPCABG患者发生POAF的影响因素。两种分类器模型预测效能较好。随访期间患者心脏结构及功能均有改善,但是发生POAF者MACCE发生率较高。

Abstract：

【Abstract】　Objective To investigate the influencing factors of postoperative atrial fibrillation (POAF) in patientsundergoing off-pump coronary artery bypass grafting (OPCABG) ,construct the classifier models, and analyze the results of cardiaccolor Doppler ultrasound after 12 months of follow-up.MethodsA total of 202 patients who received OPCABG in the ZhengzhouNo.7 People's Hospital from April 2016 to April 2018 were selected, they were randomly divided into training set (n=151) andtest set (n=51) according to the ratio of 3∶1. The perioperative clinical data of the patients were collected. Multivariate Logisticregression model was used to analyze the influencing factors of POAF in patients undergoing OPCABG, and two classifier modelswere constructed based on Logistic regression (LR) and K-nearest neighbors (KNN) . The results of cardiac color Dopplerultrasound and the incidence of major adverse cardiac and cerebrovascular event (MACCE) at 12 months after OPCABG werefollowed up. Results The patients in the training set were divided into POAF group and non-POAF group according to whetherPOAF occurred. Multivariate Logistic regression analysis showed that age, hypertension, acute myocardial infarction (AMI) , rightcoronary artery (RCA) stenosis, frequent atrial premature contraction or atrial tachycardia, renal insufficiency and preoperative useof β-receptor blockers, preoperative left anterior and posterior atrial diameter (LAAP) and postoperative left ventricular ejectionfraction (LVEF) were the influencing factors of POAF in patients undergoing OPCABG (P < 0.05) . The regression equation was:ln [PPOAF/ (1-PPOAF) ] = 0.107 × age + 0.092 × hypertension + 0.052 × AMI + 0.057 × RCA stenosis + 0.104 × frequent atrialpremature contraction or atrial tachycardia + 0.084 × renal insufficiency - 0.097 × preoperative use of β-receptor blockers+ 0.053 × preoperative LAAP - 0.056 × postoperative LVEF - 0.318. The confusion matrix of the test set data was constructedaccording to the regression equation, and its accuracy was 0.922. At 3, 6 and 12 months after operation, respectively LVEF washigher than that before operation, and LVESD, LVEDD and LAAP were lower than those before operation in the non-POAFgroup and POAF group (P < 0.05) . At 12 months after operation, recurrence of atrial fibrillation in 8 cases in POAF group, andthe recurrence rate was 22.2%. The incidence of MACCE in POAF group was higher than that in non-POAF group (χ2 =5.043,P=0.025) . Conclusion Age, hypertension, AMI, RCA stenosis, frequent atrial premature contraction or atrial tachycardia, renalinsufficiency and preoperative use of β-receptor blockers, preoperative LAAP and postoperative LVEF are the influencing factorsof POAF in patients undergoing OPCABG. The prediction efficiency of the two classifier models are good. The cardiac structureand function of patients are improved during follow-up, but the incidence of MACCE is higher in patients with POAF.

ReferenceList：