尿素氮/白蛋白比值的列线图模型对于老年心力衰竭患者预后的预测能力研究OACSTPCD
Constructing and validating a nomogram incorporating blood urea nitrogen/albumin ratio to predict prognosis of heart failure in the elderly
背景 随着人口老龄化的加剧,心力衰竭(heart failure,HF)患病率呈上升趋势,老年心力衰竭的预警、风险分层研究对改善预后具有重要意义.目的 探讨血尿素氮-白蛋白比值(blood urea nitrogen to albumin ratio,BAR)对老年心力衰竭患者1年内死亡的预测价值;建立包含BAR在内的列线图,并验证其临床价值.方法 建模数据来自解放军总医院2016年1月-2021年10月诊断为心力衰竭的1 259例老年患者,验证队列来自2021年10月-2022年10月诊断为心力衰竭的314例老年患者.以1年病死率为结局指标,采用Logistic回归分析筛选预后因素,根据预后因素构建列线图;ROC曲线、校准曲线和临床决策曲线分析(decision curve analysis,DCA)用于评估模型的区分度、一致性和临床效益;通过计算ROC 曲线下面积和 DCA,比较列线图与 N 端脑钠肽前体(N-terminal pro-B-type natriureticpeptide,NT-proBNP)及 BIOSTAT-CHF预测模型的性能.结果 训练队列和验证队列中分别有27.2%(343/1 259)和25.2%(79/314)患者在1年内死亡.多因素Logistic回归结果显示,年龄、D-二聚体、NT-proBNP、BAR、血管紧张素转化酶抑制剂(angiotensin converting enzyme inhibitor,ACEI)/血管紧张素受体阻滞剂(angiotensin receptor blockers,ARB)使用与死亡结局独立关联(P<0.05).基于以上5个因素构建列线图,列线图展现出良好的区分度(AUC=0.838,P<0.001)、一致性(Hosmer-Lemeshowx2=4.251,P=0.834)和临床效益.列线图模型与NT-proBNP和欧洲BIOSTAT-CHF预测模型相比预测效能均更佳(0.838 vs 0.703和0.825,P<0.05).DCA结果提示,列线图相比NT-proBNP、BIOSTAT-CHF预测模型均增加了更多的净获益.结论BAR是老年心力衰竭患者1年内死亡的独立预测因子.包含BAR在内的列线图展现出优异的辨别力和校准能力,为预后评估提供有力支持.
Background With the aging of population,the prevalence of heart failure(HF)is on the rise,and exploring risk stratification markers and prognostic models for elderly patients with HF is important for improving prognosis.Objective To investigate association of blood urea nitrogen to albumin ratio(BAR)with death within 1 year in elderly patients with heart failure,and then propose a nomogram including BAR,verifying its predictive value.Methods Training set data were obtained from 1 259 elderly patients diagnosed with heart failure from January 2016 to October 2021 in Chinese PLA General Hospital,and the validation set data were obtained from 314 elderly patients diagnosed with heart failure from October 2021 to October 2022.Logistic regression analysis was used to screen risk factors and construct the nomogram based on risk factors;ROC curves,calibration curves,and clinical decision curve analysis(DCA)were used to assess model discrimination,consistency,and clinical benefit;The performance of the nomogram was compared with that of N-terminal brain natriuretic peptide precursor(NT-proBNP)and the BIOSTAT-CHF prediction model by calculating the area under the ROC curve and DCA.Results In the training cohort,27.2%(343 out of 1 259)patients died within 1 year,while in the validation cohort,25.2%(79 out of 314)patients died within the same timeframe.Multivariate Logistic regression showed that age,D-dimer,NT-proBNP,BAR,and ACEI/ARB use independently associated with mortality(P<0.05).The nomogram was constructed based on these five factors,which demonstrated good discrimination ability(AUC=0.838,P<0.001),consistency(Hosmer-Lemeshow x2=4.251,P=0.834),and clinical benefit.And the predictive efficacy of the nomogram was better compared with both NT-proBNP and the European BIOSTAT-CHF prediction model(0.838 vs 0.703 and 0.825,both P<0.05).The DCA results suggested that the nomogram added more net benefit than both the NT-proBNP and the BIOSTAT-CHF prediction model.Conclusion BAR is an independent predictor of death within 1 year in elderly patients with heart failure.The nomograms with BAR demonstrate excellent discrimination and calibration ability,providing strong support for prognostic assessment.
秦继;高维阳;黄淘克;张安迪;刘宏斌
解放军医学院,北京 100853解放军总医院第二医学中心心血管内科,国家老年疾病临床医学研究中心,北京 100853
临床医学
心力衰竭血清尿素氮-白蛋白比值列线图1年内死亡风险预测模型
heart failureserum urea nitrogen-albumin rationomogramrisk of death within 1 yearpredictive model
《解放军医学院学报》 2024 (005)
486-492,502 / 8
评论