Rationale, feasibility, and utility of RCTs of ECMO vs MV
As a rescue in the emergency setting, an RCT of ECMO versus non-ECMO in severe hypoxemia is ethically unacceptable and comparable to one in situations such as CPR. Instead, comparing ECMO to non-ECMO to decrease ventilator-induced lung injury in stable ARDS requires first a realistic estimate of the mortality attributable to mechanical ventilation. EOLIA was planned assuming that protective mechanical ventilation per se has a 20% mortality, while ECMO itself has a zero risk of mortality. Negative trials do not indicate whether a given intervention is useless, but simply fail to confirm the hypothesis [6]. Indeed, a reduction of 20% in absolute mortality is, in our opinion, an unrealistic goal, whereas the observed recruitment rate of 0.06 patients/unit/month represents real world conditions. Calculations based on the characteristics of the patients in EOLIA show that 624 patients would be required for a study with sufficient power to detect a significant 11% absolute mortality reduction in ECMO from 46% mortality of non-ECMO patients. With the enrolment rate of the CESAR trial (0.03 patients/unit/month) or the EOLIA trial (0.058 patients/unit/month) and 100 participating units, such a study would take 17 or 9 years, respectively.
And finally, will this study change clinical practice or the incidence of ECMO use? Looking back on 50 years of ARDS and 40 years of ECMO history, this seems rather unlikely, since this study yet again provides data for the weal and woe of ECMO just as the previous studies did. Statistics are an operational tool and not a religion; the knowledge, skill, and common sense of physicians are the values in the balance with “0.05”.