High-flow oxygen does not reduce mortality in COVID-19 respiratory failure

The use of high-flow nasal cannula oxygen, as opposed to standard oxygen, in the first-line setting, did not reduce mortality rates in patients admitted to intensive care units (ICUs) for respiratory failure due to COVID-19, results of the SOHO-COVID trial showed.

The study population comprised 711 adults (mean age 61 years, 30 percent female) with respiratory failure due to COVID-19 and PaO₂/FiO₂* ≤200 mmHg (mean 130 mmHg) while breathing oxygen at ≥10 L/min for ≥15 minutes who were admitted to 34 ICUs in France. They were randomized (median 2.6 hours post-ICU admission) 1:1 to receive high-flow oxygen (flow rate of ≥50 L/min delivered via a heated humidifier [mean 51 L/min]; n=357) or standard oxygen (flow rate of ≥10 L/min delivered via a non-rebreathing mask [mean 13 L/min]; n=354). Fractions were adjusted to maintain oxygen saturation at 92–96 percent in both groups. High-flow oxygen therapy was carried out for ≥48 hours and was switched to standard oxygen upon achievement of oxygen saturation of ≥92 percent and a respiratory rate ≤25/min and FiO₂ ≤40 percent. At baseline, mean respiratory rate was 29 breaths/min.

Mortality at day 28 did not significantly differ between patients assigned to high-flow and standard oxygen (10 percent vs 11 percent; absolute difference, –1.2 percent, 95 percent confidence interval [CI], –5.8 to 3.4 percent; p=0.60; hazard ratio [HR], 0.88, 95 percent CI, 0.56–1.38; p=0.57 [high-flow vs standard oxygen]). [JAMA 2022;328:1212-1222]

There was a significant reduction in intubation rate with high-flow vs standard oxygen (45 percent vs 53 percent; absolute difference, –7.7 percent, 95 percent CI, –14.9 to –0.4 percent; p=0.04; HR, 0.77, 95 percent CI, 0.63–0.96; p=0.03). The main reasons for intubation were severe respiratory failure (84 percent) and life-threatening hypoxaemia (31 percent) which were similar between groups.

“[T]he decreased intubation rate in the high-flow oxygen group cannot be explained by any inconsistency in intubation indications,” the investigators said.

In-ICU mortality did not significantly differ between patients in the high-flow and standard oxygen groups (12 percent vs 15 percent; absolute difference, –2.9 percent; p=0.25), nor did in-hospital mortality (13 percent vs 15 percent; absolute difference, –2.1 percent; p=0.42), mortality at 90 days (13 percent vs 15 percent; absolute difference, –1.5 percent; p=0.56), and length of stay in the ICU (median 8 vs 9 days; p=0.91) or hospital (median 16 days in each group; p=0.54).

At day 28, there was no significant difference in the number of ventilator-free days between the high-flow and standard oxygen groups (median 28 vs 23 days; absolute difference, 0.5 days; p=0.07).

One hour after initiating treatment, mean PaO₂ was lower in the high-flow vs standard oxygen group (75 vs 80 mmHg; absolute difference, –5.0 mmHg; p=0.006), while dyspnoea score was improved (44 percent vs 28 percent with slight or marked improvement; p<0.001).

The most frequently occurring adverse event was ventilator-associated pneumonia (58 percent vs 53 percent [high-flow vs standard oxygen]). Two patients in the high-flow oxygen and five in the standard oxygen group experienced cardiac arrest that required intubation.  

The overall mortality rate was lower than expected, potentially due to “improved management of patients with COVID-19,” remarked the investigators.

Despite the lack of mortality difference between groups, “the decreased risk of intubation and need for invasive mechanical ventilation may be considered an important outcome for patients with acute respiratory failure,” they continued. “[I]t may also help avoid the use of ICU ventilators in resource-constrained settings during a pandemic.”