心力衰竭病人非计划再入院的风险因素分析及构建预测模型

黄荔荔; 徐琴; 李雯; 陈培培

心力衰竭病人非计划再入院的风险因素分析及构建预测模型

Analysis of the risk factors for unplanned readmission in patients with heart failure and construction of predictive models

摘要

摘要:
目的： 开发预测心力衰竭（HF）病人非计划再入院风险动态列线图。
方法： 选取252例HF病人，根据是否发生非计划再入院分为再入院与非再入院组，收集2组病人临床资料，采用多因素logistic回归分析影响HF病人非计划再入院独立危险因素并构建列线图。利用校准曲线、决策曲线分析和临床影响曲线对列线图预测效能进行评估。利用DynNom等包将列线图发布至网络中开发预测非计划再入院风险的动态列线图。
结果： 252例HF病人中，102例（40.48%）出现非计划再入院。多因素logistic回归分析显示年龄、糖尿病、慢性肾脏病（CKD）和N末端脑钠肽前体（NT–proBNP）均是HF病人发生非计划再入院风险独立危险因素（P ＜ 0.05），白蛋白（ALB）是独立保护因素（P ＜ 0.01）。校准曲线分析结果显示列线图预测结果与实际观察结果高度一致，列线图校准度较好；决策曲线分析结果显示，在大多数阈值下该列线图都能提供显著意义的临床净收益；临床影响曲线分析结果显示随着阈值增大该列线图模型判定高风险再入院人群与实际发生人群高度匹配。动态列线图交互界面见https://inynomoasd.shinyapps.io/muadongtailiexiantu/。
结论： 基于年龄、糖尿病、CKD、ALB和NT proBNP构建预测HF病人非计划再入院风险的动态列线图能有效识别高风险再入院人群，为及时调整治疗方案、优化病人管理策略从而降低再住院率提供依据。

Abstract:
Objective To develop a dynamic column chart for predicting the risk of unplanned readmission in patients with heart failure (HF).
Methods A total of 252 HF patients were selected, and divided into the readmitted group and non-readmitted group according to whether unplanned readmission occurred. The clinical data of two groups were collected. Multivariate logistic regression was used to analyze the independent risk factors influencing unplanned readmission in HF patients, and a nomogram was constructed. The predictive efficacy of the nomogram was evaluated by using calibration curves, decision curve analysis and clinical impact curves. A dynamic nomogram for predicting the risk of unplanned readmission was developed by publishing the nomogram to the network using packages such as DynNom.
Results Among 252 HF patients, 102 cases (40.48%) occurred the unplanned readmission. The results of Multivariate logistic regression analysis showed that the age, diabetes, chronic kidney disease (CKD) and N-terminal pro-brain natriuretic peptide (NT-proBNP) were the independent risk factors of unplanned readmission in HF patients (P ＜ 0.05), and the albumin (ALB) was an independent protective factor (P ＜ 0.01). The analysis results of the calibration curve showed that the prediction results of the nomogram were highly consistent with the actual observation results, and the calibration degree of nomogram was good. The results of the decision curve analysis showed that at most thresholds, this nomogram could provide significant clinical net benefits. The results of the clinical impact curve analysis showed that as the threshold increasing, the nomogram model highly matched the high-risk readmission population with actual occurrence population. Dynamic nomogram interface saw https://inynomoasd.shinyapps.io/muadongtailiexiantu/.
Conclusions The construction of a dynamic nomogram for predicting the risk of unplanned readmission in HF patients based on age, diabetes, CKD, ALB and NT-probNP can effectively identify the high-risk readmission populations, and provide a basis for timely adjustment of treatment plans and optimization of patient management strategies to reduce the readmission rate

HTML全文

参考文献(22)

施引文献

资源附件(0)