Publication

• Title: High-Flow Oxygen through Nasal Cannula in Acute Hypoxemic Respiratory Failure
• Acronym: FLORALI (High Flow Oxygen therapy for the Resuscitation of Acute Lung Injury)
• Year: 2015
• Journal published in: New England Journal of Medicine
• Citation: Frat JP, Thille AW, Mercat A, et al. High-Flow Oxygen through Nasal Cannula in Acute Hypoxemic Respiratory Failure. N Engl J Med. 2015;372:2185-96.

Context & Rationale

• Background

  Initial support for de novo acute hypoxaemic respiratory failure (AHRF) had centred on conventional oxygen, early endotracheal intubation, and (in selected patients) noninvasive ventilation (NIV).
  NIV has clear benefit in hypercapnic exacerbations of COPD and cardiogenic pulmonary oedema, but benefit and safety in de novo hypoxaemic AHRF/ARDS had remained uncertain, particularly in relation to high tidal volumes and delayed intubation.
  High-flow nasal oxygen (HFNO/HFNC) was an emerging technique delivering heated, humidified high flows with stable inspired oxygen fraction, modest positive airway pressure, and improved comfort; robust randomised evidence for clinically important outcomes in ICU AHRF was limited.
• Research Question/Hypothesis

  In adults with non-hypercapnic de novo AHRF treated in ICU, does HFNC (or NIV) reduce the need for endotracheal intubation compared with standard oxygen therapy, and does choice of initial oxygenation strategy influence ventilator-free days and survival?
• Why This Matters

  Avoiding intubation may prevent complications of sedation and invasive ventilation, but delayed intubation after failed noninvasive support may worsen outcomes.
  FLORALI directly compared three commonly used early strategies (standard oxygen, NIV, and HFNC) using clinically meaningful endpoints and prespecified intubation criteria, informing modern “front-end” respiratory support pathways for ICU AHRF.

Design & Methods

• Research Question: Among ICU adults with de novo, non-hypercapnic acute hypoxaemic respiratory failure, does high-flow nasal oxygen (or NIV) reduce endotracheal intubation by day 28 compared with standard oxygen therapy?
• Study Type: Multicentre, randomised, controlled, open-label, three-group parallel trial conducted in 23 ICUs (France and Belgium).
• Population:
  • Setting: ICU; randomisation occurred within 3 hours after validation of inclusion criteria.
  • Key inclusion criteria: Adults (≥18 years) with de novo AHRF; respiratory rate >25 breaths/min; PaO2/FiO2 ≤300 mm Hg while breathing oxygen at ≥10 L/min for ≥15 minutes; PaCO2 ≤45 mm Hg; absence of chronic respiratory failure.
  • Key exclusions (examples): PaCO2 >45 mm Hg; asthma or chronic respiratory failure; cardiogenic pulmonary oedema; severe neutropenia; haemodynamic instability requiring vasopressors; Glasgow Coma Scale ≤12; contraindication to NIV; immediate need for intubation; do-not-intubate order.
• Intervention:
  • High-flow nasal oxygen (HFNC/HFNO): Heated, humidified high-flow oxygen via large-bore binasal prongs (Optiflow); flow 50 L/min; FiO2 1.0 at initiation then adjusted to maintain SpO2 ≥92%; applied continuously for at least 2 calendar days (then could be stopped and switched to standard oxygen).
• Comparison:
  • Standard oxygen: Nonrebreather face mask; flow ≥10 L/min; FiO2 adjusted to maintain SpO2 ≥92% until recovery or intubation.
  • Noninvasive ventilation (NIV): Face mask connected to ICU ventilator; pressure support to target expired tidal volume 7–10 mL/kg predicted body weight; PEEP 2–10 cm H2O; FiO2 and PEEP adjusted to maintain SpO2 ≥92%; minimum 8 hours/day for at least 2 calendar days; between NIV sessions, patients received high-flow nasal oxygen.
• Blinding: Open-label (no blinding of clinicians or patients); intubation was guided by prespecified criteria to mitigate discretionary bias.
• Statistics: Sample size assumed a 60% intubation rate in the standard oxygen group; 300 patients provided 80% power (two-sided alpha 0.05) to detect a 20 percentage point absolute difference in intubation by day 28 between standard oxygen and either HFNC or NIV; primary analysis was intention-to-treat.
• Follow-Up Period: Primary outcome assessed to day 28; mortality followed to day 90.

Key Results

This trial was not stopped early. Recruitment reached the planned sample size (310 analysed vs 300 planned).

• The primary endpoint (intubation by day 28) did not differ significantly across the three strategies (38% vs 47% vs 50%; P=0.18).
• In patients with more severe hypoxaemia (PaO2/FiO2 ≤200 mm Hg), intubation differed across groups (35% vs 53% vs 58%; P=0.009), favouring HFNC.
• Despite the non-significant primary outcome, HFNC was associated with lower 90-day mortality than either standard oxygen (HR 2.01; 95% CI 1.01–3.99) or NIV (HR 2.50; 95% CI 1.31–4.78) when HFNC served as the reference group.

Internal Validity

• Randomisation and allocation: Centralised, web-based randomisation; permuted blocks of six; stratified by centre and history of cardiac insufficiency; randomisation within 3 hours after validation of inclusion criteria.
• Post-randomisation exclusions: 313 randomised; 3 withdrew consent and were excluded; 310 analysed (HFNC n=106; standard oxygen n=94; NIV n=110).
• Blinding and bias risk: Open-label design increases susceptibility to performance bias, particularly for the clinician-driven primary outcome (intubation).
• Mitigation of discretionary intubation: Prespecified intubation criteria were used; the investigators reported no deviations and no intubation in patients who failed to meet criteria.
• Baseline comparability (selected): Age 61±16 (HFNC) vs 59±17 (standard) vs 61±17 (NIV).
• Baseline severity (selected): SAPS II 25±9 (HFNC) vs 24±9 (standard) vs 27±9 (NIV).
• Baseline oxygenation (selected): PaO2/FiO2 157±89 (HFNC) vs 161±73 (standard) vs 149±72 (NIV).
• Baseline respiratory distress (selected): Respiratory rate 33±6 (HFNC) vs 32±6 (standard) vs 33±7 (NIV).
• Heterogeneity: Aetiologies included pneumonia and ARDS; treatment effects differed by severity subgroup (PaO2/FiO2 ≤200 vs >200), consistent with clinically meaningful effect modification by baseline hypoxaemia severity.
• Timing: Early randomisation (≤3 hours after inclusion validation) supports testing “front-end” strategies before prolonged exposure to potentially failing noninvasive support.
• Dose (HFNC): Protocolised initiation at flow 50 L/min and FiO2 1.0 with titration to SpO2 ≥92% supports a consistent early HFNC “dose”.
• Dose (NIV): Median NIV exposure was 8 hours/day (IQR 4–12) during the first 2 days; median expired tidal volume 9.2 mL/kg predicted body weight (IQR 7.7–10.9); median PEEP 5 cm H2O (IQR 5–5).
• Separation of the variable of interest: NIV arm received HFNC between NIV sessions by design, reducing “purity” of the comparison and potentially diluting between-group differences.
• Crossover / rescue therapy: NIV was used as rescue in 40 patients: 26 (28%) in the standard oxygen group and 14 (13%) in the HFNC group; subsequent intubation occurred in 19/26 and 9/14, respectively.
• Adjunctive therapy imbalance: The protocolised availability of NIV rescue (and HFNC between NIV sessions) implies co-intervention patterns that may have influenced downstream outcomes.
• Outcome assessment: Mortality outcomes (ICU and day 90) are objective; ventilator-free days are less discretionary but can be influenced by extubation and reintubation practices; intubation decisions are the most vulnerable to bias despite prespecified criteria.
• Statistical rigour: The trial was powered to detect a large absolute reduction (20 percentage points) in intubation; the observed absolute differences in intubation were smaller, raising the possibility of limited power for more modest but clinically relevant effects.

Conclusion on Internal Validity: Overall internal validity is moderate-to-strong: randomisation and baseline balance were strong, and prespecified intubation criteria likely reduced discretionary bias, but the open-label design and protocolised co-interventions (HFNC in the NIV arm; NIV rescue in other arms) complicate causal attribution for the primary endpoint and may dilute between-group separation.

External Validity

• Population representativeness: Applies most directly to ICU adults with de novo, non-hypercapnic AHRF with high oxygen requirements (≥10 L/min) and PaO2/FiO2 ≤300 mm Hg.
• Important exclusions: Results should not be extrapolated to hypercapnic respiratory failure (e.g., COPD exacerbation), cardiogenic pulmonary oedema, profound neurological impairment, or severe haemodynamic instability requiring vasopressors at baseline.
• Applicability across systems: Conducted in 23 ICUs (France and Belgium); applicability is strongest for similar high-resource ICUs with ready access to HFNC devices, experienced staff, and protocols for escalation.
• Intervention fidelity in practice: The NIV strategy used (face mask NIV with relatively low PEEP and higher expired tidal volumes) may differ from current practice in some units (e.g., helmet NIV, higher PEEP strategies), which could alter comparative effectiveness.

Conclusion on External Validity: External validity is moderate: findings are broadly applicable to ICU de novo non-hypercapnic AHRF populations, but generalisability is limited in hypercapnic disease and cardiogenic oedema, and comparative inferences about “optimal NIV” depend on local NIV interfaces, dosing, and clinician expertise.

Strengths & Limitations

• Strengths: Multicentre ICU trial; randomised allocation with centralised concealment; clinically meaningful primary endpoint; objective prespecified intubation criteria; follow-up to day 90; clear protocolisation of HFNC settings and initial oxygenation targets.
• Limitations: Open-label design with a clinician-driven primary outcome; primary endpoint not statistically significant while mortality signals emerged in secondary analyses; three-arm design with protocolised co-interventions (HFNC between NIV sessions; NIV rescue) reduces between-group separation; NIV “dose” and delivered ventilatory pattern (expired tidal volume 9.2 mL/kg predicted; PEEP 5 cm H2O) may not represent all contemporary NIV approaches.

Interpretation & Why It Matters

• Clinical implication

  HFNC is a reasonable first-line strategy for ICU adults with de novo non-hypercapnic AHRF, offering comparable intubation rates overall and favourable signals in ventilator-free days and 90-day survival compared with standard oxygen and face-mask NIV in this trial.
• Severity matters

  The strongest signal for reduced intubation was observed in more severe hypoxaemia (PaO2/FiO2 ≤200 mm Hg: 35% vs 53% vs 58%), supporting severity-stratified escalation pathways rather than a one-size-fits-all approach.
• Mechanistic plausibility

  HFNC may provide a blend of improved oxygen delivery, reduced dead space, and better tolerance than face mask NIV, potentially lowering injurious inspiratory effort while maintaining feasibility for prolonged use.

Controversies & Subsequent Evidence

• Primary endpoint neutrality vs secondary mortality signal: Intubation by day 28 was not significantly different (P=0.18), yet 90-day mortality differed with HFNC as the apparent “best” strategy; interpreting secondary mortality effects as causal is complicated by multiplicity and the fact that the trial was powered for a large intubation difference rather than mortality.
• Open-label intubation decisions: Even with prespecified criteria (and reported absence of deviations), clinicians could have influenced timing and threshold for intubation through subjective interpretation of work of breathing and “failure to improve”.
• NIV strategy and biological risk: Delivered NIV was associated with expired tidal volumes of 9.2 mL/kg predicted body weight (IQR 7.7–10.9) and PEEP 5 cm H2O (IQR 5–5), raising concern that NIV may have permitted injurious lung stress/strain in some patients with de novo AHRF.
• Contamination by design: HFNC between NIV sessions reduces contrast between NIV and HFNC arms, and NIV rescue in the standard oxygen group (26/94) and HFNC group (14/106) blurs strict separation of strategies.
• Subgroup interpretation: The apparent intubation benefit in PaO2/FiO2 ≤200 mm Hg (P=0.009) must be interpreted in the context of subgroup analysis and potential interaction effects; nonetheless, it aligns with physiological reasoning that more severe hypoxaemia may be more sensitive to early support strategy and effort modulation.
• Subsequent evidence: Later trials, meta-analyses, and guidelines have generally supported HFNC as an important first-line option in acute hypoxaemic respiratory failure, while emphasising careful monitoring and timely escalation to avoid harm from delayed intubation; comparative effectiveness versus optimised NIV (including different interfaces and PEEP strategies) remains context-dependent.

Summary

• Multicentre ICU trial (23 ICUs in France and Belgium) randomised adults with de novo non-hypercapnic AHRF to HFNC, standard oxygen, or NIV (with HFNC between NIV sessions).
• Primary outcome was intubation by day 28: 38% (HFNC) vs 47% (standard) vs 50% (NIV); P=0.18.
• In PaO2/FiO2 ≤200 mm Hg subgroup, intubation differed (35% vs 53% vs 58%; P=0.009), favouring HFNC.
• Ventilator-free days (days 1–28) were higher with HFNC: 24 (18–26) vs 22 (10–25) vs 19 (0–25); P=0.02.
• 90-day mortality was lower in HFNC group (12% vs 23% vs 28%), though this was not the primary endpoint and must be interpreted cautiously.

Further Reading

Other Trials

• 2018Azoulay E, et al. Effect of High-Flow Nasal Oxygen vs Standard Oxygen on outcomes in immunocompromised adults with acute respiratory failure (HIGH trial). JAMA. 2018.
• 2015Lemiale V, et al. Noninvasive ventilation vs oxygen therapy in immunocompromised patients with acute respiratory failure. JAMA. 2015.
• 2016Hernández G, et al. Postextubation high-flow nasal cannula vs noninvasive ventilation in high-risk patients: randomised clinical trial. JAMA. 2016.
• 2014Maggiore SM, et al. Nasal high-flow versus Venturi mask oxygen therapy after extubation: effects on oxygenation, comfort, and clinical outcome. Am J Respir Crit Care Med. 2014.
• 2022Perkins GD, et al. Continuous positive airway pressure vs high-flow nasal oxygen vs conventional oxygen for acute respiratory failure: RECOVERY-RS randomised clinical trial. JAMA. 2022.

Systematic Review & Meta Analysis

• 2020Ferreyro BL, et al. Noninvasive oxygenation strategies and mortality in adults with acute hypoxaemic respiratory failure: systematic review and meta-analysis. JAMA. 2020.
• 2019Rochwerg B, et al. High-flow nasal cannula compared with conventional oxygen therapy for acute hypoxaemic respiratory failure: systematic review and meta-analysis. Intensive Care Med. 2019.
• 2017Leeies M, et al. High-flow nasal cannula in acute respiratory failure: systematic review and meta-analysis. BMC Syst Rev. 2017.
• 2020Marjanovic NS, et al. High-flow nasal cannula versus conventional oxygen therapy in acute respiratory failure: systematic review and meta-analysis. Am J Emerg Med. 2020.

Observational Studies

• 2016Roca O, et al. ROX index (oxygenation and respiratory rate) to predict outcome of high-flow nasal oxygen therapy. J Crit Care. 2016.
• 2015Kang BJ, et al. Failure of high-flow nasal cannula may delay intubation and increase mortality. Intensive Care Med. 2015.
• 2012Sztrymf B, et al. High-flow nasal oxygen therapy in acute respiratory failure: prospective observational study. J Crit Care. 2012.
• 2016Coudroy R, et al. High-flow nasal oxygen therapy (alone or with NIV) in immunocompromised patients with acute respiratory failure: observational cohort. Ann Intensive Care. 2016.
• 2016Carteaux G, et al. Tidal volume during NIV and outcomes in acute hypoxaemic respiratory failure: observational cohort. Crit Care Med. 2016.

Guidelines

• 2022Oczkowski S, et al. European Respiratory Society clinical practice guidelines on high-flow nasal cannula in acute respiratory failure. Eur Respir J. 2022.
• 2017Rochwerg B, et al. ERS/ATS clinical practice guidelines: noninvasive ventilation for acute respiratory failure. Eur Respir J. 2017.
• 2024Helms J, et al. Oxygen therapy in acute hypoxemic respiratory failure: guidelines from the SRLF-SFMU consensus conference. Ann Intensive Care. 2024.
• 2021Evans L, et al. Surviving Sepsis Campaign: international guidelines for management of sepsis and septic shock 2021. Intensive Care Med. 2021.
• 2020Alhazzani W, et al. Surviving Sepsis Campaign: guidelines on the management of critically ill adults with coronavirus disease 2019 (COVID-19). Intensive Care Med. 2020.

Notes

• HFNC and NIV are not interchangeable “oxygen devices”: both influence inspiratory effort and lung mechanics, so close monitoring for failure and timely escalation remain central to safe use in AHRF.
• Interpretation of the mortality signal in FLORALI should be integrated with subsequent trials and meta-analyses, and with local expertise in NIV interfaces, dosing, and monitoring.

Overall Takeaway

FLORALI is a landmark ICU trial because it directly compared three front-line oxygenation strategies in de novo, non-hypercapnic acute hypoxaemic respiratory failure using clinically meaningful endpoints and prespecified intubation criteria. Although the primary intubation outcome was neutral, the trial helped shift practice towards HFNC as a default initial strategy in many ICUs—particularly for more severe hypoxaemia—while sharpening attention to NIV “dose”, patient effort, and the risk of delayed intubation.

Bibliography