充血性心力衰竭合并心源性休克患者的院内死亡预测模型构建及临床效能分析

Construction of in-hospital death prediction model for patients with congestive heart failure complicated with cardiogenic shock and clinical efficacy analysis

目的基于重症监护医学信息数据库(MIMIC)构建一个预测充血性心力衰竭合并心源性休克(CS)患者院内死亡的模型并评价其临床效能。方法本研究为回顾性研究,纳入MIMIC-Ⅳ中2008—2019年诊断为充血性心力衰竭合并CS的患者2090例,根据是否发生院内死亡,将所有患者分为非死亡组(1434例)及死亡组(656例)。收集患者的临床特征、生命体征、实验室检查结果及系统评分等数据。通过Lasso回归筛选相关变量,并采用多因素Logistic回归分析独立预测因素,构建院内死亡的列线图预测模型。使用Bootstrap法进行内部验证,并通过受试者工作特征(ROC)曲线分析和决策曲线分析(DCA)对预测模型进行评价。结果多因素Logistic回归分析结果显示,末梢血氧饱和度[比值比=0.968,95%置信区间(CI):0.949~0.987,P=0.001]和血白蛋白(比值比=0.764,95%CI:0.626~0.932,P=0.008)为院内死亡的独立保护因素;年龄(比值比=1.043,95%CI:1.034~1.051,P<0.001)、女性(比值比=1.304,95%CI:1.052~1.615,P=0.015)、体温<36℃(比值比=1.720,95%CI:1.284~2.304,P<0.001)、慢性阻塞性肺疾病(比值比=1.404,95%CI:1.131~1.744,P=0.002)、血液透析(比值比=2.210,95%CI:1.710~2.856,P<0.001)、血乳酸(比值比=1.149,95%CI:1.100~1.200,P<0.001)、序贯器官衰竭评估评分(比值比=1.113,95%CI:1.080~1.146,P<0.001)为院内死亡的独立危险因素。基于多因素Logistic回归分析构建充血性心力衰竭合并CS患者院内死亡的列线图风险模型,ROC曲线下面积为0.766,敏感度为72.6%,特异度为66.9%。内部验证的重抽样校准曲线表明模型理想曲线和实际曲线拟合良好。DCA表明该模型具有较高的临床净获益。结论本研究基于MIMIC-Ⅳ,通过年龄、性别、慢性阻塞性肺疾病、末梢血氧饱和度、体温<36℃、血液透析、血白蛋白、血乳酸、序贯器官衰竭评估评分构建了充血性心力衰竭合并CS患者的院内死亡列线图预测模型。该模型具有较高区分度、校准度、预测效能以及临床净获益,能帮助早期识别高风险患者,并优化治疗决策。

Objective To develop a model based on the Medical Information Mart for Intensive Care(MIMIC)database to predict in-hospital mortality in patients with congestive heart failure and cardiogenic shock(CS),and evaluate its clinical efficacy.Methods This retrospective study included 2090 patients diagnosed as congestive heart failure complicated with CS from 2008 to 2019 from the MIMIC-Ⅳdatabase.All patients were divided into non-death group(1434 cases)and death group(656 cases)according to whether there was in-hospital death.Data on clinical characteristics,vital signs,laboratory results,and system scores were collected.Lasso regression was used to select relevant variables,and multivariate logistic regression analysis was employed to identify independent predictors and construct a nomogram model for predicting in-hospital mortality.Internal validation was performed using the Bootstrap method.The model was evaluated using the receiver operating characteristic(ROC)curve and decision curve analysis(DCA).Results Multivariate Logistic regression analysis showed that peripheral oxygen saturation[odds radio(OR)=0.968,95%confidence interval(CI):0.949-0.987,P=0.001]and serum albumin(OR=0.764,95%CI:0.626-0.932,P=0.008)were independent protection factors for patients in-hospital death,and age(OR=1.043,95%CI:1.034-1.051,P<0.001),female(OR=1.304,95%CI:1.052-1.615,P=0.015),body temperature<36℃(OR=1.720,95%CI:1.284-2.304,P<0.001),chronic obstructive pulmonary disease(OR=1.404,95%CI:1.131-1.744,P=0.002),hemodialysis(OR=2.210,95%CI:1.710-2.856,P<0.001),serum lactate levels(OR=1.149,95%CI:1.100-1.200,P<0.001)and sequential organs failure assessment(SOFA)score(OR=1.113,95%CI:1.080-1.146,P<0.001)were independent risk factors.Based on multivariate Logistic regression analysis,a nomogram risk model of nosocomial death in patients with congestive heart failure complicated with CS was constructed.The nomogram model demonstrated an area under the ROC curve of 0.766,with a sensitivity of 72.6%and specificity of 66.9%.Internal validation showed good agreement between the predicted and actual values on the calibration curve.Decision curve analysis indicated that the model provided significant clinical net benefit.Conclusions Based on the MIMIC-Ⅳdatabase,this study developed a nomogram model incorporating age,gender,chronic obstructive pulmonary disease,peripheral oxygen saturation,body temperature<36℃,hemodialysis,serum albumin,serum lactate,and SOFA score to predict in-hospital mortality in patients with CS complicated by congestive heart failure cardiogenic shock.The model demonstrated high discriminative ability,calibration,predictive performance,and clinical net benefit,enabling early identification of high-risk patients and optimization of treatment decisions.