High flow nasal cannula oxygen versus noninvasive ventilation in adult acute respiratory failure: a systematic review of randomized-controlled trials

We reviewed the use of noninvasive ventilation (NIV) versus high flow nasal cannula (HFNC) oxygen in adult acute respiratory failure (ARF). We searched major databases and included randomized trials comparing at least NIV with HFNC or NIV+HFNC with NIV in ARF. Primary outcomes included intubation/re-intubation rates. Secondary outcomes were ICU mortality and morbidities. Five trials were included; three compared HFNC with NIV, one compared HFNC, NIV and oxygen whereas one compared HFNC+NIV with NIV. Patients had hypoxaemic ARF (PaO 2 /FiO 2 ≤300 mmHg). Heterogeneity prevented result pooling. Three and two studies had superiority and noninferiority design, respectively. Patients were postcardiothoracic surgery, mixed medical/surgical patients and those with pneumonia. Two trials were conducted after extubation, two before intubation and one during intubation. Three trials reported intubation/re-intubation rates as the primary outcomes. The other two trials reported the lowest peripheral capillary oxygen saturation readings during bronchoscopy or intubation. In the former three trials, the odds ratio for intubation/re-intubation rates between HFNC versus the NIV group ranged from 0.80 (95% confidence interval: 0.54–1.19) to 1.65 (95% confidence interval: 0.96–2.84). In the latter two trials, only one reported a difference in the lowest peripheral capillary oxygen saturation between NIV+HFNC versus the NIV group during intubation [100% (interquartile range: 95–100) vs. 96% (interquartile range: 92–99); P =0.029]. The secondary outcomes included differences in ICU mortality and patient tolerability, favouring HFNC. Results were conflicting, but highlighted future research directions. These include patients with hypercapneic ARF, more severe hypoxaemia (PaO 2 /FiO 2 ≤200 mmHg), a superiority design, an oxygen arm and patient-centred outcomes.