An audit of mortality by using ECMO specific scores and APACHE II scoring system in patients receiving extracorporeal membrane oxygenation in a tertiary intensive care unit in Hong Kong

Abstract

Outcomes of patients receiving extracorporeal membrane oxygenation (ECMO) therapies risk-adjusted by ECMO specific scores have rarely been reported. Our primary aim was to determine the risk adjusted outcome of these patients by the use of Respiratory Extracorporeal Membrane Oxygenation Survival Prediction (RESP), Survival After Veno-Arterial-ECMO (SAVE) and APACHE II scores. The differences in predicted mortality between these scoring systems were analyzed. This is a single-center retrospective study reviewing 62 patients who received venous-venous ECMO (VV-ECMO) and venous-arterial ECMO (VA-ECMO) between 2009 and 2017 in a tertiary ICU. Demographic, clinical, laboratory, imaging data were analyzed. Primary outcome measures were the crude mortality and standardized mortality ratio (SMR), expressed as observed mortality divided by expected mortality predicted from RESP, SAVE and APACHE II scores, respectively. We also examined the difference in predicted mortality derived from RESP, SAVE, and APACHE II by using Bland Altman plots. Twenty and Forty-two patients received VV-ECMO (20 runs) and VA-ECMO (43 runs) respectively. For VV-ECMO, the mean RESP was 0.6 [standard deviation (SD) 4.86] with RESP-SMR of 1.16 [95% confidence interval (95% CI) =0.44-1.88]. For VA-ECMO, the mean SAVE score was -7.8 (SD 5.6), with SAVE-SMR of 0.69 (95% CI =0.39-0.98). In the Bland Altman plot for VA-ECMO, mean difference in predicted mortality between SAVE and APACHE II was 17.6% (95% CI: 7.6% to 27.6%, P<0.0001), with 95% limit of agreement of -30.2% to 65.5% and beta coefficient of -1.04. APACHE II predicted mortality for VA-ECMO was lower than that for SAVE until it crossed SAVE prediction at about 80% mortality. After this point, it becomes progressively higher than that for SAVE. The mortality outcome of our patients on VV-ECMO and VA-ECMO compares favorably with predicted mortality based on RESP and SAVE, respectively. In our cohort of patients receiving VA-ECMO, APACHE II tends to underestimate mortality in lower risk patients, and overestimate the mortality in patients at high risk of death.