Duration of Noninvasive Respiratory Support and Extracorporeal Membrane Oxygenation Outcomes: Connecting the Dots.

To the Editor: We thank Lo Presti et al.1 for their letter related to our examination of the impact of duration of noninvasive respiratory support (RS) in patients with COVID-19 supported with extracorporeal membrane oxygenation (ECMO). The points raised highlight the challenges faced in the retrospective analysis of a complex disease process. The inability to account for baseline triage characteristics, lung compliance, or reason for RS failure are valid critiques of our analysis. Without reservation, our study should be viewed in the context of rapidly emerging evidence related to the support of patients with COVID-19. Given the absence of prospective trials, clinicians are left to weigh available indirect evidence that the duration of RS may be associated with worse outcomes in this context. Since authoring our manuscript, several additional studies have further corroborated our group’s findings. We wish to draw attention to these works, discuss the relevance to our findings, and emphasize the need for further investigation in this domain. First, Barbaro et al.2 presented registry data of patients with COVID-19 supported with ECMO. These investigators identified an increase in mortality over the course of the pandemic. What changed during this period to account for this observation? The authors postulated several reasons, of which, changes in pre-ECMO treatment and more frequent employment of noninvasive ventilation relate specifically to our findings. Similarly, Braaten et al.3 analyzed temporal trends in ECMO outcomes and found worse mortality in patients supported with ECMO later in the pandemic, and further, that longer time from admission to ECMO cannulation was associated with worse ECMO outcomes. In our cohort, a longer duration of RS (days) was significantly associated with a diminished rate of ECMO decannulation (sHR: 0.93; 95% CI, 0.87–0.98), whereas the duration of invasive mechanical ventilation (IMV) (sHR: 0.98; 95% CI, 0.88–1.09) was not. Moreover, the duration of RS remained significantly associated with worse ECMO outcomes even when adjusted for the duration of IMV.4 Though we dichotomize the duration of RS by median duration in our article, the precise threshold duration of RS associated with the poorest outcomes remains to be elucidated. The finding that duration of IMV is not related to ECMO outcomes in COVID-19 has been replicated by other investigators.5,6 Notably, Riera et al.6 reported days from symptoms to cannulation (possibly reflective of prolonged RS), but not IMV days, were independently associated with mortality. ECMO is a tool to decrease the physiologic insult of IMV by facilitating less damaging native lung ventilation strategies. It reasons that violation of lung-protective goals either during IMV or before tracheal intubation would similarly result in lung damage and worse overall outcomes. Clinician awareness and implementation of lung-protective ventilation in patients receiving IMV have likely improved over time and may explain the absence of impact of IMV duration on ECMO outcomes seen in patients with COVID-19. Paradoxically, crisis care during the current pandemic may have resulted in more frequent violation of lung-protective metrics via deleterious RS settings as a strategy to avoid IMV. Direct evidence implicating the duration of RS to worse ECMO outcomes in patients with COVID-19 is limited but growing. Connecting the dots of available evidence supports our preliminary findings, and we, like Lo Presti et al.5, support a prospective study of this hypothesis.