A Prediction Score for In-Hospital Mortality for Patients With Refractory Ischemic Cardiogenic Shock After VA-ECMO: From SMART-RESCUE Registry

초록

Background Systematic prediction model applicable for refractory ischemic cardiogenic shock (RICS) requiring extracorporeal membrane oxygenation (ECMO) has not been developed to date. We developed a prediction model that assesses in-hospital mortality of RICS with ECMO insertion. Methods SMART-RESCUE is a nationwide registry including 12 major centers and enrolled 1,247 patients with cardiogenic shock (CS) from 2014 to 2018: 954 retrospective data and 293 prospective data. Predictors of in-hospital mortality were extracted by logistic regression analysis, and prediction equation was created with variable transformation. External validation was done with prospective data. Results Among 954 retrospective data, 389 patients underwent ECMO insertion, and 290 patients had RICS; 105 patients (36.2%) survived. Eleven predictors were selected for prediction of in-hospital mortality: age, diastolic blood pressure, left ventricular ejection fraction after ECMO insertion, lactic acid after ECMO insertion, peak creatine kinase (CK)-MB, inotropic score, shock to ECMO insertion time, history of extracorporeal cardiopulmonary resuscitation, use of continuous renal replacement therapy, use of mechanical ventilator, and successful coronary revascularization. SMART-RESCUE score showed competent sensitivity (area under the curve = 0.8664). Internal and external validation also revealed qualified performance. (C-statistic = 0.878, 0.8969, respectively) Download : Download high-res image (213KB)Download : Download full-size image Conclusion SMART-RESCUE score is a first predictive model focusing on patients with RICS requiring ECMO insertion, which is highly predictive. Categories CORONARY: Hemodynamic Support and Cardiogenic Shock