Evaluation of outcome scoring systems for patients on extracorporeal membrane oxygenation

Background. Extracorporeal membrane oxygenation (ECMO) has been used in critical conditions such as life-threatening respiratory failure or postcardiotomy cardiogenic shock. This investigation compares the predictive value of Acute Physiology, Age and Chronic Health Evaluation IV (APACHE IV), earlier APACHE models, Sequential Organ Failure Assessment (SOFA), and the risk of renal failure, injury to the kidney, failure of kidney function, loss of kidney function, and end-stage renal failure (RIFLE) classification obtained on the first day of ECMO support for hospital mortality in critically ill patients. Methods. We reviewed the medical records of 78 critically ill patients on ECMO support at the specialized intensive care unit in a tertiary care university hospital from March 2002 to October 2005. Demographic, clinical, and laboratory variables and five scoring systems were retrospectively gathered as predicators of survival on ECMO day 1. Results. The overall mortality rate was 60.3%. The most common condition requiring ECMO was cardiogenic shock. Goodness-of-fit was good for APACHE IV but not the APACHE III model. The APACHE IV and APACHE III scoring systems displayed excellent areas under the receiver operating characteristic curve (0.922 +/- 0.030 and 0.907 +/- 0.038, respectively). Furthermore, APACHE IV correlated significantly with APACHE III scores in individual patients (r(2) = 0.902; p < 0.001). Finally, cumulative survival rates at 6-month follow-up after hospital discharge differed significantly (p < 0.001 for APACHE IV <= 49% versus APACHE IV > 49%). Conclusions. This study confirms the grave prognosis of critically ill patients receiving ECMO support. The APACHE IV proved to be a reproducible evaluation tool with excellent prognostic abilities in these patients.