摘要
目的 探讨影响维持性血液透析(maintenance hemodialysis,MHD)患者心血管并发症的危险因素,建立列线图预测模型并验证该模型的预测效果。方法 采用便利抽样法,选取2016年1月至2018年3月本院收治的437例MHD患者,采用随机数字表法按照7 3分为训练集(n=305)与测试集(n=132),单因素分析与多因素Cox比例风险回归模型筛选MHD患者心血管并发症的独立危险因素,并建立列线图模型,采用测试集对模型进行外部验证。通过受试者工作特征曲线(receiver operating characteristic curve,ROC)、校准曲线、决策曲线分析(decision curve analysis,DCA)以证实列线图模型的预测效能和临床应用价值。结果 多因素分析显示,年龄≥60岁(HR=2.047,95%CI:1.164~3.599)、透析龄≥24个月(HR=1.707,95%CI:1.013~2.877)、QT间期离散度≥63 ms(HR=1.773,95%CI:1.123~2.801)、C反应蛋白水平(HR=1.142,95%CI:1.054~1.238)、尿酸水平(HR=1.028,95%CI:1.006~1.052)是MHD患者心血管并发症的独立危险因素(均P<0.05)。基于上述因素建立列线图预测模型,内部验证模型校准曲线接近45°对角线,C-index为0.785,预测1年及3年心血管并发症风险的ROC曲线下面积分别为0.819(95%CI:0.772~0.861)、0.758(95%CI:0.706~0.805),预测临界值分别为262分、90分。测试集校准曲线接近45°对角线,C-index为0.724,预测1年及3年心血管并发症风险的ROC曲线下面积分别为0.805(95%CI:0.742~0.858)、0.728(95%CI:0.659~0.789),测试集和训练集预测效能比较,差异无统计学意义(P=0.246)。DCA显示训练集与测试集阈值概率范围分别为0.10~0.65与0.05~0.50,模型表现为正的临床净收益。结论 基于MHD患者心血管并发症危险因素建立的列线图预测模型具有良好的区分度、准确性与临床实用性,针对透析3个月以上的MHD患者,当1年得分≥262分及3年得分≥90分时应加强心血管并发症监测,制定预防性干预措施。
Objective To explore the risk factors of cardiovascular complications in patients under maintenance haemodialysis(MHD), establish a nomogram prediction model and verify the prediction effect of the model. Methods It was a retrospective study. The data of 437 MHD patients admitted to our hospital from January 2016 to March 2018. According to the ratio of 7 3, the patients were divided into training group(n=305) and test group(n=132) by random number table. Univariate and multivariate Cox proportional hazard regression analysis were applied to screen independent risk factors of cardiovascular complications in the MHD patients. A nomogram model was then established based on independent risk factors. The testing group was used for external validation of the model. The receiver operating characteristic curve (ROC),calibration curve and clinical decision curve analysis (DCA) were applied to verify the predictive performance of the nomogram model.Results Multivariate Cox analysis showed that age≥60 years (HR=2.047,95%CI:1.164-3.599),length of dialysis≥24 months (HR=1.707,95%CI:1.013-2.877),ECG QT-interval dispersion≥63 ms (HR=1.773,95%CI:1.123-2.801),C-reactive protein level (HR=1.142,95%CI:1.054-1.238),and uric acid level (HR=1.028,95%CI:1.006-1.052) were independent risk factors of the cardiovascular complications in MHD patients (all P<0.05).A nomogram prediction model was established based on above factors.In the internal verification,with the calibration curve close to 45°diagonal and the C-index at 0.785,the risk of cardiovascular complications in 1 year and 3 years were 0.819(95%CI:0.772-0.861) and 0.758(95%CI:0.706-0.805) respectively,according to the area under the ROC curve(AUC) for prediction,and with the corresponding prediction thresholds at 262 points and 90 points,respectively.In the external validation,the calibration curve was close to 45°diagonal,the C-index was 0.724,and the AUC for prediction of the risk of cardiovascular complications at 1 and 3 years were 0.805 (95%CI:0.742-0.858) and 0.728
作者
徐庆华
赵君花
甄长萍
梁新蕊
Xu Qinghua;Zhao Junhua;Zhen Changping;Liang Xinrui(Blood Purification Center,Department of Nephrology,the First Affiliated Hospital of Naval Military Medical University,Shanghai,200433,China)
出处
《现代临床护理》
2022年第6期6-14,共9页
Modern Clinical Nursing
关键词
维持性血液透析
心血管并发症
预测模型
maintenance haemodialysis
cardiovascular complications
predictive model
