Extracorporeal membrane oxygenation (ECMO) for severe acute respiratory distress syndrome (ARDS) has known a growing interest over the last decades with promising results during the 2009 A(H1N1) influenza epidemic. Targeting populations that can most benefit from this therapy is now of major importance.

Survival has steadily improved for a decade, reaching up to 65% at hospital discharge in the most recent cohorts. However, ECMO is still marred by frequent and significant complications such as bleeding and nosocomial infections. In addition, physiological and psychological symptoms are commonly described in long-term follow-up of ECMO-treated ARDS survivors. Because this therapy is costly and exposes patients to significant complications, seven prediction models have been developed recently to help clinicians identify patients most likely to survive once ECMO has been initiated and to facilitate appropriate comparison of risk-adjusted outcomes between centres and over time. Higher age, immunocompromised status, associated extra-pulmonary organ dysfunction, low respiratory compliance and non-influenzae diagnosis seem to be the main determinants of poorer outcome.