ABSTRACT

Objective

: This study aims to develop a nomogram model to predict the survival of refractory cardiogenic shock (RCS) patients that received veno-arterial extracorporeal membrane oxygenation (VA-ECMO).

Methods

A total of 235 and 209 RCS patients were supported with VA-ECMO from January 2018 to December 2019 in Guangdong Provincial People’s Hospital, and from January 2020 to December 2020 in four third-grade and class-A hospitals were a development cohort (DC) and validation cohort (VC), respectively. Finally, 137 and 98 patients were included in the DC and VC. Multivariate logistic regression analysis was used to identify variables, and only these independent risk factors were used to establish the nomogram model. The receiver operating characteristic curve (ROC), calibration plot, decision curve, and clinical impact curves were used to evaluate the nomogram’s discriminative ability, predictive accuracy, and clinical application value.

Results

Pre-ECMO cardiogenic arrest (pre-ECA), lactate (Lac), inotropic score (IS), and modified nutrition risk in the critically ill score (mNUTRIC score) were incorporated into the nomogram. This showed good discrimination in the DC, with an area under ROC (AUROC) and a 95% confidence interval (CI) of 0.959 (0.911–0.986). The AUROC (95% CI) of the VC was 0.928 (0.858–0.971). The calibration plots of the DC and VC presented good calibration results. The decision curve and clinical impact curve of the nomogram provided improved benefits for RCS patients.

Conclusions

This study established a prediction nomogram composed of pre-ECA, Lac, IS, and mNUTRIC scores that could help clinicians to predict the survival probability at hospital discharge precisely and rapidly for RCS patients that received VA-ECMO.

中文翻译：

---

列线图预测难治性心源性休克后静脉-动脉体外膜肺氧合患者的生存结果

摘要

客观的

: 本研究旨在开发列线图模型来预测接受静脉-动脉体外膜肺氧合 (VA-ECMO) 的难治性心源性休克 (RCS) 患者的生存率。

方法

2018年1月至2019年12月，广东省人民医院共有235名和209名RCS患者接受了VA-ECMO支持，2020年1月至2020年12月在四家三级甲等医院为发展队列（DC）和验证队列（VC），分别。最后，137 和 98 名患者被纳入 DC 和 VC。多变量逻辑回归分析用于识别变量，仅使用这些独立危险因素建立列线图模型。采用受试者工作特征曲线（ROC）、校准图、决策曲线和临床影响曲线评价列线图的判别能力、预测准确性和临床应用价值。

结果

ECMO前心源性骤停（pre-ECA）、乳酸（Lac）、正性肌力评分（IS）和危重病评分（mNUTRIC评分）中的改良营养风险被纳入列线图。这表明 DC 具有良好的辨别力，ROC 下的面积 (AUROC) 和 0.959 (0.911–0.986) 的 95% 置信区间 (CI)。VC 的 AUROC (95% CI) 为 0.928 (0.858–0.971)。DC 和 VC 的校准曲线呈现出良好的校准结果。列线图的决策曲线和临床影响曲线为 RCS 患者提供了更好的益处。

结论

本研究建立了一个由 pre-ECA、Lac、IS 和 mNUTRIC 评分组成的预测列线图，可以帮助临床医生准确快速地预测接受 VA-ECMO 的 RCS 患者出院时的生存概率。