Patients supported by extracorporeal membrane oxygenation and acute dialysis: acute physiology and chronic health evaluation score in predicting hospital mortality.

Extracorporeal membrane oxygenation (ECMO) can provide short-term cardiopulmonary support to critically ill patients. Among ECMO patients, acute renal failure requiring dialysis has an ominous prognosis. However, a prognostic scoring system and risk factors adjustment for hospital mortality in these patients have not been elucidated previously. A multicenter observational cohort study was conducted from January 2002 to December 2006. Information obtained included demographics, biochemical variables, Acute Physiology and Chronic Health Evaluation (APACHE) II, III, and IV scores at ICU admission and initial acute dialysis, and hospital mortality in 102 non-coronary artery bypass graft (CABG) patients receiving ECMO support with acute dialysis. This retrospective cohort study included 70 men and 32 women with a mean age of 47.9 ± 15.7 years. Seventy-two patients (70.6%) had hospital mortality. The area under the receiver operating characteristic curve showed APACHE IV (0.653) had a better discriminative power to predict hospital mortality than APACHE II (0.584) and APACHE III (0.634) at initializing dialysis. Hosmer-Lemeshow statistics showed good calibration for all three scores to predict hospital mortality at initializing dialysis (APACHE IV, P = 0.392; APACHE III, P = 0.094; and APACHE II, P = 0.673). Independent predictors for hospital mortality by multivariate logistic regression analysis were higher central venous pressure (odds ratio [OR], 1.11; confidence interval [CI] 95%, 1.02-1.20; P = 0.016), higher APACHE IV score at initializing dialysis (OR, 1.03; CI 95%, 1.01-1.05; P = 0.003), and latency from hospital admission to dialysis (OR, 1.04; CI 95%, 1.00-1.08; P = 0.033). High mortality rate was noted in non-CABG patients receiving ECMO and acute dialysis. Predialysis APACHE IV scores had good calibration and moderate discrimination in predicting hospital mortality in these patients. Because ECMO support could stabilize cardiopulmonary status, APACHE IV scores would likewise underestimate disease severity at lower score ranges in these patients.