Efficacy of outcome prediction of the respiratory ECMO survival prediction score and the predicting death for severe ARDS on VV-ECMO score for patients with acute respiratory distress syndrome on extracorporeal membrane oxygenation

Background
Extracorporeal Membrane Oxygenation (ECMO) therapy for respiratory failure is an increasingly popular modality of support. Patient selection is an important aspect of outcome success. This review assesses the efficacy of the popular prognostic tools Respiratory ECMO Survival Prediction Score (RESP) and Predicting Death for Severe ARDS on VV-ECMO score (PRESERVE) for ECMO patient selection.

Methods
A literature search was performed. Six publications were found to match the specified selection criteria. These publications were assessed and compared using the area under the receiver operating characteristic (AUROC) curve statistical method to ascertain the discriminatory ability of the models to predict treatment outcome.

Results
Six articles were included in this review from 306 screened, of which all were retrospective cohort studies. Data was generated over a period of 3–9 years from 13 referring hospitals. Studies consisted of 467 male and 221 female (30 unknown) participants in total with a high heterogeneity. The PRESERVE prognostic model was found to have a higher AUROC score than the RESP model, however both models were found to be sub-optimal in their discriminatory ability. A high chance of bias was seen across all included studies.

Conclusion
It was the findings of this review, indicated by analysis using the AUROC measures, that the prognostic model PRESERVE performed better than RESP for predicting post ECMO therapy outcomes, for patients presenting with Acute Respiratory Distress Syndrome within their respective validated time frames, i.e., RESP at Intensive care unit (ICU) discharge and PRESERVE at 6 months post ICU discharge. However, It was recognized that comparator groups were small thereby introducing bias into the study. Further prospective, randomized studies would be necessary to effectively assess the utility of these predictive survival scores.