RESUMO
OBJECTIVE: To determine whether high-flow oxygen therapy by nasal cannulae (HFNC) is more effective than a high-flow face mask (HFFM) in severe hypoxemia. DESIGN: Randomized, single-center, open-labeled, controlled trial. SETTING: University Hospital of Nantes, France. PARTICIPANTS: Cardiac surgery patients presenting oxygen saturation <96% with Venturi mask 50%. INTERVENTION: Oxygenation by HFNC (45 L/min, FIO2 100%) or Hudson RCI non-rebreather face mask with a reservoir bag (15 L/min). MEASUREMENTS AND MAIN RESULTS: The co-primary outcomes were the PaO2/FIO2 ratio at 1 and 24 hours. In the intent-to-treat analysis (90 patients), the mean (standard deviation) PaO2/FIO2 ratios were: after 1 hour, 113.4 (50.2) in HFFM versus 137.8 (57.0) in HFNC (mean difference 24.4, CI 97.5% [2.9-45.9], pâ¯=â¯0.03), and after 24 hours, 106.9 (62.6) in HFFM versus 129.9 (54.0) in HFNC (mean difference 23.0, CI 97.5% [1.5-44.6], pâ¯=â¯0.04). After adjustment on baseline PaO2/FIO2, this difference persisted at 24 hours (pâ¯=â¯0.04). For secondary outcomes, the PaO2/FIO2 ratio after 6 hours was 108.7 (47.9) in HFFM versus 136.0 (45.2) in HFNC (pâ¯=â¯0.01), without difference after 48 hours (pâ¯=â¯0.95). Refractory hypoxemia requiring noninvasive ventilation occurred in 13 (28%) patients in HFNC versus 24 (56%) patients in HFFM (pâ¯=â¯0.007). The HFNC improved satisfaction (pâ¯=â¯0.0002) and reduced mucus dryness (pâ¯=â¯0.003) compared with HFFM. CONCLUSION: In patients with severe hypoxemia after cardiac surgery, PaO2/FIO2 at 1 and 24 hours were higher and the use of noninvasive ventilation was reduced in HFNC compared with HFFM.