Accurately selecting patients who are likely to benefit from venovenous extracorporeal membrane oxygenation (VV-ECMO) for severe respiratory failure is crucial given its low availability, substantial invasiveness, associated complications, and high financial costs. Existing selection criteria are influenced by multidimensional factors, including local logistics and geographical variation in health-care provision, expert opinion, and data from observational studies and a single randomised controlled trial of ECMO for severe acute respiratory distress syndrome.

Evolving outcomes of patients on VV-ECMO during the COVID-19 pandemic showed that current selection criteria might not effectively guide decision making under unprecedented levels of health-care system strain, and accentuated the need for reliable prognostic models and better understanding of both short-term and long-term outcomes.

In this context, we read with interest the two studies by Roberto Lorusso and colleaguesand by Matthieu Schmidt and colleagues, published in The Lancet Respiratory Medicine.

In their prospective, multicentre observational cohort study, Lorusso and colleagues sought to identify associations between clinical characteristics at the time of VV-ECMO cannulation and in-hospital mortality, and to report 6-month functional outcomes in patients supported with VV-ECMO during the first wave of the COVID-19 pandemic (March–September, 2020). Data from 1215 patients were analysed using mixed Cox proportional hazards models. In-hospital mortality was 50% (602/1215 patients died), and 95% (549/577) of those with available follow-up data who survived to hospital discharge were still alive at 6 months. Factors associated with in-hospital mortality included being age 60 years or older, use of inotropes or vasopressors before initiation of VV-ECMO, and time from intubation to ECMO of 4 days or longer.