An External Validation of Scoring Systems in Mortality Prediction in Veno-Venous Extracorporeal Membrane Oxygenation

Abstract

Veno-venous extracorporeal membrane oxygenation (VV ECMO) offers the last resort in the treatment of acute respiratory distress syndrome (ARDS). Various scoring systems have been established, yet external validation of these scoring systems in the Asian population remains scarce. We aim to identify factors associated with hospital mortality and to validate various scoring systems in the prediction of hospital mortality. A retrospective analysis of adults admitted to Pamela Youde Nethersole Eastern Hospital intensive care unit who received VV ECMO from January 1, 2010 to June 30, 2019 was performed. Demographics, ventilation strategies, rescue therapies, and clinical outcomes were compared. The primary outcome was hospital mortality and secondary outcomes were intensive care unit (ICU) mortality, ICU, and hospital length of stay. There were 122 VV ECMO performed for ARDS, of which 78 survived and 44 died. VV ECMO performed for viral pneumonitis was significantly associated with better survival (55.1% vs. 25%, p = 0.001) compared with other causes. As for prediction scores, the PREdiction of Survival on ECMO Therapy-Score had the highest area under receiver operator curve of 0.733 (95% confidence interval [CI]: 0.643-0.823), whereas that of PRedicting dEath for SEvere ARDS on VV ECMO score was 0.662 (95% CI: 0.561-0.764), Respiratory Extracorporeal Membrane Oxygenation Survival Prediction score was 0.657 (95% CI: 0.553-0.761), Sequential Organ Failure Assessment score was 0.652 (95% CI: 0.547-0.757), and VV ECMO mortality score was 0.637 (95% CI: 0.532-0.742). In our cohort, VV ECMO performed for viral pneumonitis was associated with a higher hospital survival. Prediction scores are helpful in our population and provide a useful reference to hospital mortality.