Improving the Numbers, But Harming the Patient? The Authors Respond.

Abstract

To the Editor: We thank da Costa et al.1 for their letter which makes some interesting observations. We agree that our study alone cannot be used to justify the use of extracorporeal membrane oxygenation (ECMO) as a strategy to protect the lungs in patients with moderate acute respiratory failure (ARF)—nor do we make a claim that it should. Severity of respiratory failure is not just defined by a single PaO2/FiO2 ratio value, but on a more complex clinical evaluation which must also take into account the duration of hypoxemia. We take this opportunity to stress that all patients in our study2 had ARF severe enough to receive ECMO based on the indications established by our national service. The “less severe” group had a median PaO2/FiO2 ratio of 105 mm Hg despite median FiO2 of 0.85, but fulfilled other criteria for ECMO. As this cohort was matched for PaCO2 with the “very severe” group, it is likely the price the patient has to pay for the PaO2/FiO2 ratio is injurious mechanical ventilation. Therefore, we would suggest that many of these patients would already meet criteria for ECMO in many systems, and that we are not advancing a “new” indication. There is now substantial evidence for the association between increased mechanical power, ventilator-induced lung injury and mortality in ARF,3 and in this context the use of ECMO allows the least damaging lung ventilation strategy4 and better outcomes.5 We would suggest, intuitively, that increased duration of high mechanical power to the lungs will thus increase mortality even if PaO2/FiO2 ratio is not yet “severe.” High mechanical power can—in the short term—increase PaO2/FiO2 masking the severity of disease, but ultimately increasing mortality. This was exactly what happened in the OSCILLATE Trial, where high-flow oscillatory ventilation resulted in better oxygenation but worse patient outcomes.6 The results of more recent studies are consistent with our findings: a large multinational cohort study of coronavirus patients suggested a survival benefit in those with PaO2/FiO2 ratio of 120–150 mm Hg.7 Da Costa et al. are correct also to identify the subset of ARF patients who experience acute right ventricular dysfunction, which is often exacerbated by “lung-protective” ventilation measures.8 Whether, or at what stage, such patients may benefit from ECMO support, is not a question our study can answer. The task then is how best to determine the answer to these questions? Do patients with “moderate-severe” respiratory failure benefit from ECMO? What about those with acute right ventricle (RV) dysfunction? Ultimately, we agree with da Costa et al. that in the absence of randomized controlled trials, the solution is in high-quality observational registries of ARF, not just of therapies such as ECMO, with advanced causal inference methods. For the individual patient, we suggest that the optimal strategy remains to identify the point at which the benefits of ECMO in preventing hypoxia and avoiding injurious lung ventilation outweigh its risks.