Noninvasive Ventilation Alone Cannot Solve COVID-19 Respiratory Failure

The respiratory management of severe coronavirus disease (COVID-19) acute hypoxemic respiratory failure has vacillated since the onset of the pandemic. Initially the consensus was to favor early intubation and initiation of invasive mechanical ventilation. Noninvasive ventilation modalities, which include continuous positive airway pressure (CPAP) and bilevel positive airway pressure (BPAP), and humidified high flow oxygen were avoided. This was done to reduce aerosol generation risking infection in others and unrestricted tidal volumes causing injurious transpulmonary pressures.

Over time, as personal protective equipment (PPE) became more available and with the recognition of the high mortality associated with early invasive mechanical ventilation in severe COVID-19, management moved from early intubation and toward initially managing patients on noninvasive ventilation or high flow oxygen. The Helmet Noninvasive Ventilation Versus High-Flow Oxygen Therapy in Acute Hypoxemic Respiratory Failure (HENIVOT) and Randomized Evaluation of COVID-19 Therapy–Respiratory Support (RECOVERY-RS) studies demonstrated favorable outcomes when using noninvasive ventilation.^{1 , 2} In HENIVOT, noninvasive ventilation was associated with reduced intubation compared with high flow oxygen, with an intubation rate of 30% versus 51% (P=.03). In RECOVERY-RS, a similar result was seen, with a 36.3% intubation rate in the CPAP arm versus 44.4% in the conventional oxygen arm (P=.03); there was no statistically significant difference seen in the high flow oxygen versus conventional oxygen comparison.

At face value, it may be tempting to say noninvasive ventilation modalities are superior to high flow oxygen. However, in both HENIVOT and RECOVERY-RS, there was no statistical difference in mortality when comparing high flow oxygen, noninvasive ventilation, or even conventional oxygenation strategies. Although intubation is a patient-centered outcome and has due importance, mortality is a graver outcome. Both trials were also complicated by unblinded designs. RECOVERY-RS also had significant contamination across the treatment arms, with 23% of patients in the conventional oxygen arm receiving high flow oxygen or CPAP; in the CPAP arm, 15.3% received high flow oxygen. In RECOVERY-RS, high flow oxygen was not directly compared with CPAP.

A reduced need for intubation but a similar rate of mortality would suggest that the use of noninvasive ventilation may be differentiating 2 different patient populations. The first is a population that might be able to recover without invasive mechanical ventilation if noninvasive ventilation is initiated early. The second is a population in which noninvasive ventilation will not be effective and the need for invasive mechanical ventilation is inevitable. What is not clear from these data is whether delaying an inevitable intubation has an impact on mortality or not. Data from the Large Observational Study to Understand the Global Impact of Severe Acute Respiratory Failure registry showed noninvasive ventilation failure requiring intubation was associated with increased ICU mortality, with a hazard ratio of 1.446 (95% confidence interval 1.159 to 1.805).³ Spontaneous positive pressure breathing, and the consequent high tidal volumes, may be the drivers of deterioration in COVID-19 acute hypoxemic respiratory failure through self-induced lung injury.^{4 , 5} At the physiologic level, positive pressure ventilation from any device—whether it is an noninvasive ventilation through a face or head appliance, or a ventilator through an endotracheal tube—affects the lung in a similar fashion.⁶

High flow oxygen has advantages over noninvasive ventilation. It is recognized to be more comfortable and with convincing evidence of benefit in non-COVID-19 acute respiratory distress syndrome at reducing intubation and death.⁷ In the HiFlo-covid study, high flow oxygen was associated with less intubation and quicker clinical recovery when compared with conventional oxygen in patients with COVID-19 acute hypoxemic respiratory failure.⁸ Similar findings were seen in the small study of Nair et al,⁹ with high flow oxygen trending toward less mortality and less need for intubation when compared with noninvasive ventilation. Unfortunately, both studies are limited by relatively small enrollments, limiting their interpretation.

How should we interpret these nuanced results, and what could be driving these findings? It may come down to the devices used. In a recent meta-analysis examining noninvasive ventilation versus high flow oxygen for COVID-19 acute hypoxemic respiratory failure across 23 studies, high flow oxygen was associated with reduced mortality compared with noninvasive ventilation.¹⁰ However, when high flow oxygen is compared with either CPAP or noninvasive ventilation with a helmet device, the mortality benefit is no longer found. This suggests BPAP and facemask devices are driving the harm. The reasons for increased mortality associated with BPAP may reflect the self-induced lung injury phenomenon and delayed necessary intubation. The helmet device may also be beneficial because of the increased comfort reported with these devices in comparison to facemasks with less aspiration risk. In the United States, helmet devices have been slow in widespread adoption, limiting their current utility when managing patients with COVID-19 acute hypoxemic respiratory failure.

Given the theoretical and observed concerns with the use of noninvasive ventilation, especially with facemask devices, we should be cautious in adopting noninvasive ventilation over high flow oxygen as our primary oxygen therapy in severe COVID-19 acute hypoxemic respiratory failure refractory to conventional oxygen therapy. High flow oxygen has a long track record of being comfortable, safe, and effective at reducing intubation and mortality in acute hypoxemic respiratory failure. We should continue to use high flow oxygen, with the acknowledgment that noninvasive ventilation may be beneficial in limited circumstances when limited to specific modalities and devices.