Time for Clinicians to Embrace Their Inner Bayesian?Reanalysis of Results of a Clinical Trial of Extracorporeal Membrane Oxygenation

This issue of JAMA includes a Special Communication by Goligher et al¹ reporting a Bayesian reanalysis of the results from the recent Extracorporeal Membrane Oxygenation (ECMO) to Rescue Lung Injury in Severe Acute Respiratory Distress Syndrome (ARDS) (EOLIA) trial. This trial, which tested whether routine early ECMO reduced mortality for patients with severe ARDS, was stopped early for futility, and concluded that ECMO was not shown to reduce mortality.² In contrast, Goligher et al found it highly probable that ECMO lowers mortality, incorporating various assumptions, although it is unclear whether the benefit is as large as that assumed when the EOLIA trial was designed. How can the conclusions drawn from these 2 analyses of the same trial be so different?