慢性心力衰竭患者易损期非计划性再入院风险预测模型的构建及验证

Construction and Validation of Risk Prediction Model for Unplanned Readmission during Vulnerable Period in Patients with Chronic Heart Failure

目的构建慢性心力衰竭患者易损期非计划性再入院风险预测模型,并验证该模型的应用效果。方法采用便利抽样法,选取2019年1月至2020年4月在江苏大学附属医院心血管内科住院的慢性心力衰竭患者360例,按7∶3的比例随机分为建模组(n=252)和内部验证组(n=108),然后选取2020年6月至2021年1月在本院住院的慢性心力衰竭患者126例作为外部验证组。收集研究对象的一般资料、临床资料及实验室检查指标,通过病历系统和电话随访获取慢性心力衰竭患者出院后3个月非计划性再入院情况。基于建模组的数据资料,采用多因素Logistic回归分析方法建立预测模型,通过Hosmer-Lemeshow检验和ROC曲线评估模型的拟合效果及区分度,并建立线段式静态诺谟图和动态诺谟图。通过内部验证和外部验证评价模型的预测效果。结果多因素Logistic回归分析结果显示,心房颤动、营养不良、贫血、脑钠肽升高、左心室射血分数降低是慢性心力衰竭患者易损期非计划性再入院的影响因素(P<0.05)。预测模型公式为:P=1/{1+exp〔-(1.601×心房颤动+1.267×营养不良+1.756×贫血+0.989×脑钠肽升高+1.805×左心室射血分数降低-4.405)〕}。Hosmer-Lemeshow检验结果显示,P=0.776;ROC曲线分析结果显示,模型预测建模组患者易损期非计划性再入院的曲线下面积为0.886,约登指数最大值为0.618,最佳临界值为0.150,灵敏度为0.927,特异度为0.690。内部验证和外部验证的正确率分别为87.04%和86.51%。结论合并心房颤动、营养不良、贫血及脑钠肽升高、左心室射血分数降低是慢性心力衰竭患者易损期非计划性再入院的独立危险因素,本研究构建的模型预测慢性心力衰竭患者易损期非计划性再入院风险的效果良好,可为准确识别易损期非计划性再入院高危风险患者提供借鉴。

Objective To construct a risk prediction model for unplanned readmission during vulnerable period in patients with chronic heart failure and verify the application effect of this model.Methods A total of 360 patients with chronic heart failure hospitalized in the Department of Cardiovascular Medicine,Affiliated Hospital of Jiangsu University from January 2019 to April 2020 were selected by convenience sampling method.They were randomly divided into the modeling group(n=252)and the internal validation group(n=108)in a 7∶3 ratio.Then,126 patients with chronic heart failure hospitalized in the same hospital from June 2020 to January 2021 were selected as the external validation group.General data,clinical data,and laboratory indicators were collected.The incidence of unplanned readmission at 3 months after discharge in patients with chronic heart failure was obtained through the medical record system and telephone follow-up.Based on the data of the modeling group,the prediction model was established by multivariate Logistic regression analysis method.The goodness of fit and differentiation of the model were evaluated by Hosmer-Lemeshow test and ROC curve,and the static and dynamic nomograms were established.The prediction effect of the model was evaluated through internal validation and external validation.Results Multivariate Logistic regression analysis showed that atrial fibrillation,malnutrition,anemia,elevated brain natriuretic peptide,and decreased left ventricular ejection fraction were the influencing factors of unplanned readmission during vulnerable period in patients with chronic heart failure(P<0.05).The prediction model was as follows:P=1/{1+exp[-(1.601×atrial fibrillation+1.267×malnutr ition+1.756×anemia+0.989×elevated brain natriuretic peptide+1.805×decreased left ventricular ejection fraction-4.405)]}.Hosmer-Lemeshow test results showed that P=0.776,ROC curve analysis results showed that the area under curve of the model for predicting the risk of unplanned readmission during vulnerable period in patients with chronic heart failure was 0.886,the biggest Yuden index was 0.618,the optimal critical value was 0.150,the sensitivity was 0.927,and the specificity was 0.690.The validity of internal and external validation was 87.04%and 86.51%,respectively.Conclusion Atrial fibrillation,malnutrition,anemia,elevated brain natriuretic peptide,and decreased left ventricular ejection fraction are independent risk factors of unplanned readmission during vulnerable period in patients with chronic heart failure.The model constructed in this study is effective in predicting the risk of unplanned readmission during vulnerable period in patients with chronic heart failure,which can provide a reference for accurately identifying patients at high risk of unplanned readmission during vulnerable period.