By Betty Tran, MD, MSc

Associate Professor of Medicine, Division of Pulmonary and Critical Care Medicine, Northwestern University Feinberg School of Medicine, Chicago

SYNOPSIS: In adult patients hospitalized with COVID-19 for one month, using a noninvasive respiratory protocol that encouraged high-flow nasal cannula, noninvasive mechanical ventilation, and self-proning did not result in any significant increase in mortality.

SOURCE: Soares WE 3rd, Schoenfeld EM, Visintainer P, et al. Safety assessment of a noninvasive respiratory protocol for adults with COVID-19. J Hosp Med 2020;15:734-738.

This was a retrospective chart review of 469 consecutive adult patients admitted to any of four hospitals in the Baystate Health system with a positive reverse transcriptase-polymerase chain reaction (RT-PCR) test for SARS-CoV-2 between March 15, 2020, and April 15, 2020. A noninvasive COVID-19 respiratory protocol (NCRP) encouraging early use of high-flow nasal cannula (HFNC), noninvasive ventilation (NIV), and self-proning was developed and implemented on April 3, 2020. The primary outcome was mortality during the post-NCRP implementation period compared to the pre-implementation period. Rates of proning, HFNC, NIV, and intubation before and after protocol implementation were analyzed. Secondary outcomes included unexpected cardiac arrests, ICU transfers and consultations, and rapid response team (RRT) activations in the pre-implementation vs. postimplementation period.

After protocol implementation, there was an increase in HFNC use (5.5% to 24.7%) and self-proning (7.5% to 22.8%), while intubation rates decreased from 25.2% to 10.7% (P < 0.01). The median time to mechanical ventilation increased from 0.66 days (interquartile range [IQR], 0.23-1.69 days) to 1.4 days (IQR, 0.21-2.9 days) in the pre-implementation vs. postimplementation period. Overall mortality was 26.2% during the study period. During the pre-implementation period, 61 of 254 patients died vs. 62 of 215 in the postimplementation period (P = 0.14). After excluding patients with established (prior to admission) do not resuscitate/do not intubate (DNR/DNI) orders, the mortality rate was comparable pre-implementation vs. postimplementation (21.8% vs. 21.9%). In terms of secondary outcomes, there was no increase in RRT activations or ICU consults in the postimplementation period. ICU transfers decreased in the postimplementation period. There was one unexpected cardiac arrest in the postimplementation period compared to none before the protocol.


Our management of COVID-19 patients has evolved rapidly. In the first months of the pandemic, COVID-19 patients were intubated and placed on mechanical ventilation early in their course, often after failing up to 6 L/min of nasal cannula. This strategy was based on perceptions and reports of rapid clinical decline, ineffectiveness of noninvasive methods, and incomplete knowledge of viral transmissibility with a focus on healthcare provider safety. In their review, Soares et al reported 24% of patients died in the pre-implementation period compared to 28.8% in the postimplementation period (P = 0.14), although this finding was not statistically significant. The comparison became more equal after excluding patients with an established DNR/DNI order before admission. Considering the study’s retrospective nature, we cannot evaluate the effectiveness of the noninvasive respiratory protocol itself because of the potential for selection bias and unadjusted confounders. In other words, if more patients died while on an early noninvasive respiratory therapy, does it mean this approach was harmful (in terms of delaying the need for invasive mechanical ventilation), or just that sicker patients will require invasive mechanical ventilation and be more likely to die regardless of the interventions they received?

Overall, the findings suggest an early, noninvasive respiratory approach can reduce the rate of mechanical ventilation while not significantly affecting overall mortality, although the latter conclusion is limited by retrospective data. Soares et al noted the failure to find a reduced mortality after protocol implementation could be related to other factors, namely because of an increased proportion of established DNR/DNI patients, many coming from skilled nursing facilities and nursing homes, admitted after the protocol was implemented compared to prior. Considering the one-month period of this chart review, there were no novel medications, and it was unlikely providers changed their interventions or treatment plans significantly during this time, except for perhaps encouraging more patient self-proning.

Even if there is no significant reduction in mortality with an early, noninvasive approach, the reduction in rates of invasive mechanical ventilation may produce other benefits the authors did not explore here, such as shorter length of stay, fewer hospital-related comorbidities, decreased rates of PTSD and functional impairment, and lower healthcare costs, while freeing resources, such as ventilators and ICU beds.