G. Peigh1, H. T. Pitcher1, N. Cavarocchi1, H. Hirose1 1Thomas Jefferson University,Philadelphia, PA, USA
Introduction: While the use of cardiac ECMO is increasing in adult patients, an analysis of risk factors is still in its infancy. Even though standard intensive care unit (ICU) risk scores such as Simplified Acute Physiology Score II (SAPS II), Sequential Organ Failure Assessment (SOFA) and Acute Physiology And Chronic Health Evaluation II (APACHE II), or disease specific scores such as Model for End-stage Liver Disease (MELD), Kidney Risk, Injury, Failure, Loss of Function, ESRD (RIFLE), Predicting death for Severe ARDS On VV ECMO (PRESERVE) and ECMOnet scores exist, they may not apply to cardiac ECMO patients as their risk factors differ from the variables these common scores use.
Methods: Between 2010 and 2014, 107 ECMO procedures were performed at our hospital. 73 procedures were to support patients on cardiac ECMO. Patient demographics, preoperative conditions, and survival were retrospectively analyzed with IRB approval.
Results: Cardiac ECMO was performed on 73 patients (47 males and 26 females) with a mean age of 48 ± 14 years. The most common etiologies (#) for ECMO were acute myocardial infraction (19), acute on chronic heart failure (14), post-cardiotomy failure (13), malignant arrhythmia (11), and others (16). The mean duration of ECMO support was 9.2 ± 6.1 days. 47/73 patients (64%) survived ECMO. 27/47 patients (57%) were discharged from the hospital. The risk factors independently associated with death on ECMO were high lactate levels (p=0.02) and post-cardiotomy failure (p=0.03). Pre-ECMO SAPS II, SOFA, APACHE II, MELD, RIFLE, ECMOnet and PRESERVE scores were not correlated with survival (Table). Analysis of pre-ECMO risk factors indicated that elevated lactate (>2.0 mmol/dl), metabolic acidosis (HCO3<20 mEq/L), renal dysfunction (RIFLE score of ‘injury’ or above), and having a post-cardiotomy failure predicted death. Applying this data into a new Simplified Cardiac ECMO Score (min 0, max 4) predicted survival (survivors 1.5 ± 1.1; non-survivors 2.4 ± 0.8; p=0.0006). The area under the curve (AUC) was 0.55 for SOFA, 0.60 for APACHE II, 0.54 for SAPS II, 0.57 for MELD, 0.49 for PRESERVE, 0.62 for ECMOnet, 0.60 for RIFLE, and 0.72 for the Simplified Cardiac ECMO score. Although our score has a higher AUC, a small sample size led to an underpowered study. Despite the fact that there is clinical significance, there was not statistical significance between the various AUC curves.
Conclusion: Common ICU or disease specific risk scores calculated for cardiac ECMO patients prior to the initiation of ECMO did not correlate with ECMO survival. Although the Simplified Cardiac ECMO Score needs to be further investigated, it helps predict futile efforts in high-risk populations.
