18. HFNC vs. NIV in Acute Hypoxemic Respiratory Failure (FLORALI Evidence)

Quick Recap

Cross-cutting protocol — companion to Acute Respiratory Failure and ARDS protocols (Respiratory System), and to Awake Prone Positioning protocol. Addresses the choice of non-invasive respiratory support modality (high-flow nasal cannula vs. non-invasive positive-pressure ventilation vs. standard oxygen) in acute hypoxemic respiratory failure — grounded in a foundational trial with a genuinely important interpretive nuance: its striking mortality finding was a secondary, not primary, outcome, and a population-specific (immunocompromised) trial later failed to replicate any mortality benefit at all.

1. Definition

Acute hypoxemic respiratory failure (non-hypercapnic): respiratory failure characterized by impaired oxygenation without significant hypercapnia — the specific population studied in the foundational trial addressed here, distinct from hypercapnic respiratory failure (e.g., COPD exacerbation), where non-invasive positive-pressure ventilation (NIV) has long-established, separately validated benefit (cross-reference relevant Respiratory System protocols).

High-flow nasal cannula (HFNC): delivery of heated, humidified oxygen at high flow rates (up to 60 L/min) via nasal cannula, providing a stable, precisely titratable FiO2, modest positive airway pressure, reduced anatomical dead space washout, and improved patient comfort/tolerability compared to face-mask-based oxygen delivery or NIV.

Non-invasive positive-pressure ventilation (NIV): delivery of positive pressure ventilatory support via a tightly-fitted face or nasal mask — with well-established, separately validated benefit in hypercapnic COPD exacerbation and cardiogenic pulmonary edema, but with genuinely less certain benefit and safety in de novo hypoxemic acute respiratory failure/ARDS specifically, the population this protocol addresses.

2. Pathophysiology

In de novo hypoxemic acute respiratory failure (frequently evolving ARDS from pneumonia or other direct lung injury), the physiological goal of non-invasive support is to improve oxygenation and reduce work of breathing while avoiding both premature and delayed intubation — premature intubation exposes the patient to the risks of invasive ventilation and sedation unnecessarily, while delayed intubation after a failed trial of non-invasive support risks a more precipitous, higher-risk emergent intubation and potentially worse outcomes from ongoing patient self-inflicted lung injury (P-SILI) during the failed non-invasive trial.

A specific, mechanistically important concern with NIV in this population: NIV delivered via face mask in hypoxemic (rather than hypercapnic) respiratory failure has been associated with larger generated tidal volumes than the lung-protective ventilation targets established for invasive ventilation in ARDS (cross-reference ARDS protocol) — in the foundational trial addressed below, the NIV arm's tidal volumes (7–10 mL/kg) exceeded the 6 mL/kg lung-protective target associated with lower mortality in ARDS, providing a plausible mechanistic explanation for why NIV performed less favorably than HFNC in that trial, distinct from any inherent inferiority of positive-pressure support as a concept.

3. Immediate Stabilization (ABCDE) — Modality Selection as Part of Breathing Management

Not a standalone acute stabilization scenario; this protocol addresses the specific non-invasive modality selection within the Breathing component of acute respiratory failure management (cross-reference Acute Respiratory Failure and ARDS protocols):

Checklist:

4. Focused History

• Underlying cause of hypoxemic respiratory failure (pneumonia, evolving ARDS, other direct lung injury)
• Immunocompromised status (relevant to the population-specific evidence discordance, Section 11)
• Hypercapnia status (excludes this protocol's population if present, directing instead to established hypercapnic-failure NIV pathways)
• Comfort/tolerability considerations — genuinely relevant given HFNC's established comfort advantage over face-mask-based NIV

5. Comprehensive System-wise Examination

• Respiratory: work of breathing, PaO2/FiO2 ratio, respiratory rate
• Neurological: level of consciousness/cooperation (relevant to NIV tolerability and airway protection concerns)

POCUS integration: lung ultrasound may assist in characterizing the underlying pulmonary process, though not a primary component of the modality-selection decision itself; cross-reference Awake Prone Positioning protocol for lung ultrasound's specific, validated role in tracking treatment response during non-invasive support.

6. Syndrome Identification — Reframed as Modality-Selection Classification

• Non-hypercapnic hypoxemic respiratory failure, non-immunocompromised, PaO2/FiO2 ≤200: the population where HFNC showed the clearest intubation-rate benefit in subgroup analysis of the foundational trial (Section 11)
• Non-hypercapnic hypoxemic respiratory failure, immunocompromised: a population-specific evidence discordance exists (Section 11) — HFNC's mortality benefit was not replicated in this specific population in a dedicated, adequately powered trial
• Hypercapnic respiratory failure (e.g., COPD exacerbation): outside this protocol's scope — NIV retains well-established, separately validated benefit in this population, and should not be deprioritized based on this protocol's hypoxemic-failure-specific evidence
• COVID-19-specific hypoxemic respiratory failure: cross-reference Awake Prone Positioning protocol, which addresses the substantial, distinct COVID-19-specific evidence base for HFNC-supported patients specifically

7. Differential Diagnosis — Not a Traditional Differential

Cross-reference the relevant Respiratory System protocols (Acute Respiratory Failure, ARDS, Severe Pneumonia) for the underlying diagnostic differential of the hypoxemic respiratory failure itself; this protocol addresses the non-invasive support modality decision, not the underlying etiology.

8. Severity/Risk Assessment

PaO2/FiO2 ratio: the foundational trial's subgroup analysis found the intubation-rate reduction with HFNC was concentrated specifically in patients with PaO2/FiO2 ≤200 mmHg — the more severely hypoxemic subgroup — which may have been a specific driver of the trial's overall mortality finding, though this remains a subgroup (not primary, pre-specified) analysis and should be interpreted with corresponding caution.

Immunocompromised status: a specifically important risk-stratification factor given the population-specific evidence discordance detailed in Section 11 — this population's optimal non-invasive support strategy should not be assumed to follow the general (non-immunocompromised) population's evidence base.

9. Investigations

Immediate bedside: PaO2/FiO2 ratio, respiratory rate, work of breathing assessment

Repeat frequency: ongoing reassessment throughout the non-invasive support trial, with explicit, protocolized intubation criteria applied consistently rather than allowed to drift with clinical uncertainty

10. Point-of-Care Ultrasound — Not a Primary Component

Cross-reference Awake Prone Positioning protocol for lung ultrasound's validated role in tracking treatment response specifically in HFNC-supported patients.

11. Evidence-Based Management

FLORALI (2015) — The Foundational Trial and Its Genuinely Important Interpretive Nuance

• FLORALI trial (Frat et al., NEJM 2015, multicenter, n=310, non-hypercapnic acute hypoxemic respiratory failure with PaO2/FiO2 ≤300): randomized to HFNC, standard oxygen via face mask, or NIV — patients were not selected for immunocompromised status specifically (a general ICU hypoxemic respiratory failure population)
• Primary outcome (intubation rate at day 28): 38% (HFNC) vs. 47% (standard oxygen) vs. 50% (NIV) — not statistically significantly different across the three groups (P=0.18 for all comparisons) — this is the crucial, frequently under-emphasized point: the trial's primary, pre-specified outcome was negative
• Secondary outcome — 90-day mortality: significantly lower with HFNC — hazard ratio for death at 90 days was 2.01 (95% CI 1.01–3.99) with standard oxygen vs. HFNC (p=0.046) and 2.50 (95% CI 1.31–4.78) with NIV vs. HFNC (p=0.006) — a striking, clinically important secondary finding
• Ventilator-free days at day 28: significantly higher with HFNC (24±8 days) vs. standard oxygen (22±10) vs. NIV (19±12), p=0.02
• The specific, honest interpretive point this protocol emphasizes: FLORALI's celebrated mortality benefit was a secondary outcome finding in a trial whose primary, pre-specified outcome (intubation rate) was statistically null — this does not invalidate the mortality finding, which is clinically important and biologically plausible, but it does mean the finding should be held with somewhat more epistemic caution than if it had been the trial's primary, adequately-powered endpoint; secondary outcome findings, however striking, carry a structurally higher risk of chance findings or multiplicity-related false positives than a trial's designated primary outcome
• A specific, plausible mechanistic confound identified by expert commentary: the NIV arm's tidal volumes (7–10 mL/kg) exceeded the lung-protective 6 mL/kg target associated with lower ARDS mortality — suggesting that at least part of NIV's worse performance in this trial may reflect how NIV was delivered (larger tidal volumes than a lung-protective strategy would target) rather than an inherent property of non-invasive positive-pressure support as a concept
• A related, honestly-noted safety observation: two deaths occurred during the intubation process itself in the trial, and expert commentary has specifically raised the concern that the trial's protocolized intubation criteria, while methodologically reasonable for a research study, may have contributed to delayed intubation in some patients — the decision to intubate requires clinical judgment that is difficult to fully capture in a rigid protocol, a genuine tension this trial's design surfaces rather than resolves

Population-Specific Discordance — Immunocompromised Patients

• HIGH trial (Azoulay et al., JAMA 2018, n=776, immunocompromised patients specifically with acute respiratory failure): HFNC vs. standard oxygen — did not find a mortality benefit for HFNC in this specific population, in direct contrast to FLORALI's general-population finding
• FLORALI-IM trial (Coudroy et al., Lancet Respir Med, immunocompromised patients specifically): examined HFNC alone vs. HFNC alternating with NIV — a further, population-specific refinement of the non-invasive support question in this distinct patient group
• Practical significance: this represents a genuine, population-specific evidence discordance — the mortality benefit that made FLORALI so influential in general critical care practice did not replicate in a dedicated, purpose-built trial in immunocompromised patients specifically — clinicians should not assume FLORALI's findings automatically generalize to this distinct population, which has its own separate, and less uniformly favorable, evidence base

Current Guideline Status — A Notable Gap

• As of recent review (2024), no major guideline has been published that fully reflects FLORALI's specific findings into a formal, graded recommendation — despite FLORALI being one of the most frequently cited and practice-influential trials in this area, formal guideline incorporation has lagged, reflecting ongoing genuine uncertainty (including the primary-vs-secondary-outcome nuance above) about how strongly to weight this specific trial's findings
• A recent (2020, updated through 2025) meta-analysis incorporating FLORALI alongside subsequent trials (including COVID-19-era trials: HiFLo-COVID, HENIVOT, RECOVERY-RS, SOHO-COVID, Covid-High) found HFNC reduces intubation compared to standard oxygen with a non-significant possible mortality reduction — a more tempered, pooled-evidence conclusion than FLORALI's own striking single-trial secondary finding, and arguably a more appropriately cautious summary of the totality of current evidence

Practical Synthesis

HFNC is a reasonable, well-tolerated first-line non-invasive support strategy for non-hypercapnic, non-immunocompromised patients with acute hypoxemic respiratory failure, particularly those with PaO2/FiO2 ≤200, where the clearest intubation-rate benefit has been observed. The striking FLORALI mortality finding should be understood as a secondary outcome in a primary-outcome-negative trial — clinically important and worth taking seriously, but not with the same evidentiary weight as an adequately-powered primary endpoint, and the broader pooled evidence across subsequent trials supports a more modest, intubation-rate-focused benefit rather than a confidently established mortality benefit. In immunocompromised patients specifically, the mortality benefit has not been replicated, and management in this population should be guided by its own dedicated evidence base (HIGH, FLORALI-IM) rather than direct extrapolation from FLORALI. NIV retains its well-established, separately validated role in hypercapnic respiratory failure (COPD exacerbation, cardiogenic pulmonary edema) — this protocol's findings do not diminish that indication, which rests on an entirely separate evidence base.

12. Organ Support

Directly interacts with Acute Respiratory Failure and ARDS protocols (Respiratory System) for the broader ventilatory support decision framework, and with Awake Prone Positioning protocol as a complementary intervention for HFNC-supported patients specifically.

13. Disease-Specific Therapy — Not Applicable

This is a respiratory support modality selection protocol rather than a pharmacotherapy one.

14. Consultation Matrix

15. Monitoring Framework

• Clinical: PaO2/FiO2 trend, respiratory rate, work of breathing during non-invasive support trial
• Escalation triggers: explicit, protocolized intubation criteria applied consistently, with clinical judgment retained to avoid the delayed-intubation risk surfaced by FLORALI's own design limitations
• De-escalation criteria: improving oxygenation/work of breathing → gradual weaning of non-invasive support per standard respiratory support weaning principles

16. ICU Bundle Checklist

17. Complications

Early:

• Delayed intubation risk if non-invasive support trial is prolonged beyond a reasonable window without adequate reassessment — specifically surfaced by FLORALI's own protocolized-intubation-criteria limitation and the two intubation-related deaths in that trial
• NIV-specific: potential for excessive tidal volumes via face-mask delivery, a plausible contributor to NIV's less favorable performance in FLORALI specifically

Late:

• Not specifically characterized as a distinct late-complication profile beyond standard respiratory failure considerations

Prevention: explicit, consistently-applied intubation criteria balanced with clinical judgment; awareness of the tidal-volume concern with face-mask NIV in this specific population

Rescue: prompt intubation when independently indicated, regardless of which non-invasive modality was in use

18. Escalation & De-escalation

Escalation: failure to improve or worsening trajectory on non-invasive support → intubation per standard Acute Respiratory Failure/ARDS protocol criteria, applying clinical judgment rather than rigid adherence to a research-study-style protocol alone.

De-escalation: improving trajectory → gradual weaning of non-invasive support per standard principles.

19. ICU Discharge Criteria — Not Directly Applicable

Cross-reference relevant Respiratory System protocols and ICU Discharge Criteria & Step-Down protocol.

20. Documentation & Medicolegal Checklist

• Hypercapnic status assessment documented
• Non-invasive support modality selected and rationale documented, including immunocompromised status consideration where relevant
• PaO2/FiO2 trend documented throughout the support trial
• Intubation criteria and decision-making documented, particularly for any deviation from protocolized criteria based on clinical judgment

21. Key Guidelines

• No major guideline has fully incorporated FLORALI's specific findings into a graded formal recommendation as of recent review (2024) — clinicians should recognize this as a genuine gap between an influential, practice-shaping trial and formal guideline codification, rather than assuming graded guideline endorsement exists where it does not

22. Landmark Trials

23. Controversies

• The primary-vs-secondary-outcome distinction in FLORALI is the central, most important interpretive point for this entire protocol: FLORALI is frequently cited in practice and teaching as demonstrating a clear HFNC mortality benefit, but the trial's own primary, pre-specified outcome (intubation rate) was statistically null — this is not a minor technicality but a substantive reason for epistemic caution, since secondary outcomes (even striking, clinically plausible ones) are structurally more vulnerable to chance findings than a trial's designated primary endpoint; this protocol treats this distinction as essential context rather than an asterisk.
• The tidal-volume confound in the NIV arm offers a plausible, mechanistically grounded explanation for at least part of NIV's comparatively worse performance, suggesting the trial's finding may reflect how NIV was delivered in this specific trial rather than an inherent property of positive-pressure non-invasive support — this nuance matters for how strongly to generalize the finding to NIV delivered with more careful attention to tidal volume limitation.
• Guideline lag despite substantial practice influence: FLORALI has meaningfully shaped clinical practice patterns (HFNC increasingly favored as first-line non-invasive support in hypoxemic respiratory failure) without formal guideline codification reflecting its specific findings — an interesting case where practice has moved ahead of graded guideline endorsement, in the opposite direction from several other practice-lag patterns noted elsewhere in this library (e.g., fludrocortisone, stress ulcer prophylaxis), where guideline-supported evidence lagged behind in adoption rather than practice outrunning formal guidance.
• Immunocompromised-population discordance remains genuinely unresolved regarding mechanism: whether HIGH's null mortality finding reflects a true absence of benefit in this population, insufficient power, or population-specific factors (different underlying disease processes, different baseline risk) has not been definitively disentangled — clinicians should treat this population's optimal strategy as a distinct, less certain question rather than assuming either FLORALI's or HIGH's finding straightforwardly generalizes to all subtypes of immunocompromise.

24. References

1. Frat JP, Thille AW, Mercat A, et al; FLORALI Study Group and REVA Network. High-flow oxygen through nasal cannula in acute hypoxemic respiratory failure. N Engl J Med. 2015;372(23):2185-2196.
2. Azoulay E, Lemiale V, Mokart D, et al. Effect of high-flow nasal oxygen vs standard oxygen on 28-day mortality in immunocompromised patients with acute respiratory failure: the HIGH randomized clinical trial. JAMA. 2018;320(20):2099-2107.
3. Coudroy R, Frat JP, Ehrmann S, et al. High-flow nasal oxygen therapy alone or with non-invasive ventilation in immunocompromised patients admitted to ICU for acute hypoxemic respiratory failure: the FLORALI-IM randomised multicentre controlled trial protocol. BMJ Open. 2019;9(6):e029798; primary results, Lancet Respir Med.
4. Matthay MA. Saving lives with high-flow nasal oxygen [editorial]. N Engl J Med. 2015;372(23):2225-2226.
5. Grieco DL, Dell'Anna AM, Taccone FS, Antonelli M. High-flow oxygen therapy in hypoxemic respiratory failure: insights from the FLORALI trial. Minerva Anestesiol. 2017;83(2):125-127.
6. Spiegel RJ, Weingart SD. Blowing smoke: examining the benefits of high-flow nasal cannula in hypoxic respiratory failure [commentary]. Ann Emerg Med. 2015;66(5):556-557.
7. High-flow nasal cannula in acute hypoxemic respiratory failure: updated systematic review and meta-analysis incorporating COVID-19 era trials. 2025.
8. The Washington Manual of Critical Care, 4th ed. 2025 — respiratory support chapters.
9. ICU Protocols: A Step-wise Approach, 2nd ed. — acute respiratory failure chapter.