Reply to Epelbaum: Standards and Stereotypes in COVID-19.

From the Authors:

We thank Dr. Epelbaum for the opportunity to further discuss our data. Dr. Epelbaum raises several important points for discussion.

Massachusetts General Hospital and Beth Israel Deaconess Medical Center are tertiary care hospitals at the epicenter of the coronavirus disease (COVID-19) response in Boston, Massachusetts. As of May 16, 2020, the hospitals have cared for more than 700 critically ill patients with COVID-19. Our motivation to publish our frontline systematic observations stemmed from concern that anecdotal reports, susceptible to cognitive biases (1), were gaining significant attention and impacting practice patterns for respiratory failure (2).

As of May 16, 2020, our cohort of 66 patients had a median follow-up of 53 days (range, 48–67 d) with 14 total deaths (21.2%) (3). Fifty-one patients (75.8%) had left the ICU, and 46 patients (70.0%) were discharged alive from the hospital. Six patients (9.1%) remained hospitalized. Fifteen of 66 patients (21.2%) received a tracheostomy; 8 patients (53%) were decannulated. We agree that further studies are needed about the timing and use of tracheostomy in COVID-19 respiratory failure. Dr. Epelbaum states that reports from New York and Italy are discordant with the mortality in our cohort; however, the mortality rates are in fact quite similar. The cited article from New York has been formally corrected to reflect a mortality of 24.5% among mechanically ventilated patients. In the cited manuscript from Italy, mortality at the time of censoring was 26% for ICU patients (4). All early reports should be interpreted in the context of limited follow-up and absence of risk adjustment. However, our original report described a follow-up period which, to our knowledge, was longer than any published series outside China. We would also emphasize that the 70% hospital discharge rate in our cohort does not support Dr. Epelbaum’s statement of “remarkable propensity for sudden death” following initial ICU discharge.

Dr. Epelbaum reports that the comparison of our cohort to the PROSEVA (Prone Positioning in Severe ARDS) (5) trial reveals “‘unARDS-like’ clinical behavior.” First, we caution against direct comparison of our entire cohort to PROSEVA. PROSEVA enrolled patients with severe acute respiratory distress syndrome (ARDS), whereas our cohort included all patients intubated for respiratory failure. If we examine comparable patients from our cohort—those with moderate to severe ARDS who were treated with prone positioning—we see similar physiologic responses. Patients in PROSEVA (Figure S2 from Reference 5) had a mean PaO:FiO ratio of approximately 110 when supine, prior to first proning session, with improvement to 190 at the end of the first prone session (difference, +80). Patients in our cohort improved from a median PaO:FiO ratio of 150 when supine to 232 during the first prone session (difference, +82).

Finally, we share Dr. Epelbaum’s concern about the use of off-label therapies for ARDS, including statins. Preclinical and observational data published prior to COVID-19 suggested a role for statins for the treatment of viral infections, including severe acute respiratory syndrome coronavirus (SARS-CoV), 2009 influenza A (H1N1), and Middle East respiratory syndrome (6, 7). Therefore, at the time of our study, hospital guidelines recommended initiating statins for accepted secondary indications frequently present in patients with COVID-19 and not for ARDS per se. We agree that statin therapy has not shown benefit in unselected patients with ARDS and requires further study. We also agree that off-label therapies should ideally be used in the context of clinical trials.

Frontline experience remains critical to inform real-time patient care and shape future research. We urge the continued rigorous study of these observations to accurately inform the management of critically ill patients with COVID-19.