Table 1.

Most relevant studies of high flow nasal cannula oxygen (HFNC)

Condition	Author [ref]	Design	Patients	Main results
Acute respiratory failure (ARF)	Roca et al. [19]	Retrospective cohort	37 lung transplant recipients readmitted to ICU due to ARF (40 episodes)	The absolute risk reduction for MV with HFNC was 29.8%, and the NNT to prevent one intubation with HFNC was 3. Multivariate analysis showed that HFNC therapy was the only variable at ICU admission associated with a decreased risk of MV (OR 0.11 [95% CI 0.02–0.69]; p = 0.02)
	Frat et al. [20]	Randomized controlled trial	313 ARF patients randomly assigned to HFNC, conventional oxygen therapy or NIV	The hazard ratio for death at 90 days was 2.01 (95% CI 1.01 to 3.99) with conventional oxygen therapy vs HFNC (p = 0.046) and 2.50 (95% CI, 1.31 to 4.78) with NIV vs HFNC (p = 0.006). In the subgroup of patients with a PaO2/FiO2 ≤ 200 mmHg, the intubation rate was significantly lower in the HFNC group
	Rello et al. [31]	Retrospective cohort	35 patients with ARF due to H1N1 viral pneumonia	After 6 h of HFNC, non‐responders had a lower PaO2/FiO2. All 8 patients on vasopressors required intubation
	Lemiale et al. [29]	Randomized controlled trial	100 immunocompromised patients with ARF randomized to a 2 h of HFNC vs conventional oxygen	No differences in NIV or invasive MV during the 2 h period were observed. No differences in secondary outcomes (RR, HR, comfort, dyspnea and thirst) were observed
	Mokart et al. [42]	Retrospective propensity‐score analysis	178 cancer patients admitted to the ICU due to severe ARF	HFNC‐NIV was associated with more VFD and less septic shock occurrence. Mortality of patients treated with HFNC 35 vs 57% for patients never treated with HFNC, p = 0.008
	Kang et al. [30]	Retrospective cohort	175 patients who failed on HFNC and required intubation	In propensity‐adjusted and ‐matched analysis, early intubation (<48 h) was associated with better overall ICU mortality [adjusted OR = 0.317, p = 0.005; matched OR = 0.369, p = 0.046]
	Frat et al. [23]	Post‐hoc analysis of a randomized controlled trial	82 immunocompromised patients of the FLORALI study	NIV was associated with higher risk of intubation and mortality
	Roca et al. [32]	Prospective cohort	157 patients with severe pneumonia	ROX index, defined as ratio of SpO2/FiO2 to respiratory rate, ≥ 4.88 measured after 12 h of HFNC was significantly associated with lower risk for MV (HR 0.273 [95% CI 0.121–0.618])
	Coudroy et al. [25]	Retrospective cohort	115 immunocompromised patients	The rates of intubation and 28‐day mortality were significantly higher in patients treated with NIV than with HFNC
Cardiac surgery	Parke et al. [18]	Randomized	60 patients with non‐severe hypoxemic ARF were randomized to receive HFNC or oxygen therapy	HFNC patients tended to need NIV less frequently (10 vs 30%; p = 0.10) and had significantly fewer desaturations (p = 0.009)
	Parke et al. [38]	Randomized controlled trial	340 patients after cardiac surgery randomized to HFNC vs conventional oxygen therapy for 48 h	No differences in oxygenation on Day 3 after surgery were observed, but HFNC did reduce the requirement for escalation of respiratory support (OR 0.47, 95% CI 0.29–0.7, p = 0.001)
	Corley et al. [39]	Randomized controlled trial	155 extubated patients with BMI ≥ 30 kg/m2 received conventional oxygen therapy or HFNC	No difference was seen between groups in atelectasis. There was no difference in mean PaO2/FiO2 ratio or RR. Five patients failed allocated treatment in the control group compared with three in the treatment group (OR 0.53; 95% CI 0.11, 2.24, p = 0.40)
	Stephan et al. [28]	Randomized noninferiority trial	830 cardiothoracic surgical patients who developed ARF or were deemed at risk for respiratory failure after extubation. HFNC vs BiPAP	The treatment failed in 87 (21.0%) of 414 patients with HFNC and 91 (21.9%) of 416 patients with BiPAP (p = 0.003). No significant differences were found for ICU mortality (23 patients with BiPAP [5.5%] and 28 with HFNC [6.8%]; p = 0.66)
Pre‐intubation	Vourc’h et al. [43]	Randomized controlled trial	124 patients with PaO2/FIO2 ratio < 300 mmHg, RR ≥ 30 bpm and FIO2 ≥ 0.5. Randomized to HFNC or facial mask	No differences in the lowest saturation was observed (HFNC 91.5% vs high flow facial mask 89.5%; p = 0.44). There was no difference in difficult intubation, VFD, intubation‐related adverse events including desaturation < 80% or mortality
Post‐extubation	Maggiore et al. [37]	Randomized controlled trial	105 patients with PaO2/FiO2 ≤ 300 before extubation who were randomized to 48 h of conventional oxygen therapy or HFNC	HFNV improved the PaO2/FiO2 ratio, comfort, airway dryness, episodes of interface displacements, oxygen desaturations, reintubation rate, or any form of ventilator support
	Rittayamai et al. [44]	Randomized crossover study	17 patients were randomized after extubation to sequential HFNC and conventional oxygen therapy for 30 min periods	At the end of the study, patients with HFNC reported less dyspnea and lower RR and HR. Most of the subjects (88.2%) preferred HFNC to conventional oxygen therapy
	Tiruvoipati et al. [36]	Randomized crossover study	50 patients were randomized to sequential HFNC and facial mask after extubation	There was no significant difference in gas exchange, RR or hemodynamics. HFNC was better tolerated (p = 0.01) and tended to be more comfortable (p = 0.09)
	Hernandez et al. [1]	Randomized controlled trial	527 extubated patients without any high‐risk factor for reintubation were randomized to either HFNC or conventional oxygen therapy for 24 h	Reintubation within 72 h was less common in the high‐flow group, 4.9 vs 12.2% in the conventional group (p = 0.004). Postextubation respiratory failure was less common in the high‐flow group 8.3 vs 14.4% in the conventional group (p = 0.03)
	Hernandez et al. [5]	Randomized non‐inferiority trial	604 extubated patients with at least one high‐risk factor for reintubation were randomized to either HFNC or NIV for 24 h	Reintubation within 72 h was noninferior in the HFNC group compared to the NIV group (22.8 vs 19.1%, absolute difference, −3.7%; 95% CI −9.1% to ∞); postextubation respiratory failure was lower in the HFNC group (26.9 vs 39.8%, risk difference, 12.9%; 95% CI 6.6% to ∞)
Invasive procedures	Lucangelo et al. [45]	Randomized controlled trial	45 patients were randomly assigned to 3 groups: Venturi mask, nasal cannula, and HFNC during bronchoscopy	At the end of bronchoscopy, HFNC‐treated patients had higher PaO2/FiO2, and SpO2
	Simon et al. [46]	Randomized controlled trial	40 patients with hypoxemic ARF received NIV or HFNC during bronchoscopy	NIV group had better oxygenation. Two patients with HFNC were unable to proceed to bronchoscopy due to progressive hypoxemia
Heart failure	Roca et al. [9]	Prospective cohort	10 adult patients with NYHA class III and left ventriclar ejection fraction 45% or less	Median inspiratory IVC collapse significantly (p < 0.05) decreased from baseline (37%) to HFNC with 20 l/min (28%) and HFNC with 40 l/min (21%). Changes in IVC inspiratory collapse were reversible
Emergency department	Rittayamai et al. [15]	Randomized comparative study	40 hypoxemic patients were randomized to receive HFNC or conventional oxygen for 1 h	HFNC improved dyspnea and comfort. No serious adverse events related to HFNC were observed
	Jones et al. [47]	Randomized controlled trial	303 hypoxemic and tachypneic patients admitted to the lemergency department	5.5% of HFNC patients vs 11.6% of conventional oxygen therapy patients required MV within 24 h of admission (p = 0.053)

MV mechanical ventilation, RR respiratory rate, ICU intensive care unit, ARF acute respiratory failure, VFD ventilator‐free days, NNT number needed to treat, COT conventional oxygen therapy, NIV non‐invasive ventilation, HR heart rate, BMI body mass index, IVC inferior vena cava