Publication

• Title: Effect of Postextubation High-Flow Nasal Cannula vs Noninvasive Ventilation on Reintubation and Postextubation Respiratory Failure in High-Risk Patients: A Randomized Clinical Trial
• Acronym: Not reported
• Year: 2016
• Journal published in: JAMA
• Citation: Hernández G, Vaquero C, Colinas L, Cuena R, González P, Canabal A, Sanchez S, Rodriguez ML, Villasclaras A, Fernández R. Effect of postextubation high-flow nasal cannula vs noninvasive ventilation on reintubation and postextubation respiratory failure in high-risk patients: a randomized clinical trial. JAMA. 2016 Oct 18;316(15):1565-74.

Context & Rationale

• Background

  • Postextubation respiratory failure and reintubation are frequent after planned extubation and are associated with prolonged ICU stay and worse outcomes.
  • Prophylactic NIV after extubation had evidence of benefit in selected high-risk populations, but requires an interface, skilled titration, and is limited by intolerance and adverse effects.
  • HFNC delivers heated, humidified oxygen at high inspiratory flows; physiologic rationale includes more stable FiO₂, reduced work of breathing, low-level PEEP effects, and improved secretion clearance.
  • Whether HFNC could replace NIV as postextubation prophylaxis in high-risk ICU patients remained uncertain.
• Research Question/Hypothesis

  • In high-risk ICU patients after planned extubation, is HFNC noninferior to prophylactic NIV for preventing (1) reintubation and (2) postextubation respiratory failure within 72 hours?
• Why This Matters

  • If HFNC is comparably effective, it offers a simpler and more tolerable postextubation strategy than NIV, with potential reductions in treatment discontinuation and ICU resource use.
  • Defines whether routine NIV prophylaxis is necessary in broad “high-risk” populations or whether HFNC can be a default support strategy in many patients.

Design & Methods

• Research Question: Among adults at high risk of reintubation after planned extubation, is postextubation HFNC noninferior to prophylactic NIV for preventing reintubation and postextubation respiratory failure within 72 hours?
• Study Type: Randomised, multicentre, open-label, noninferiority trial; 3 ICUs in Spain; recruitment September 2012 to October 2014; registered (NCT01191489).
• Population:
  • Setting: Adult ICU; extubation after successful spontaneous breathing trial.
  • Inclusion criteria: Adults mechanically ventilated >12 hours; planned extubation after successful spontaneous breathing trial; ≥1 predefined high-risk feature for reintubation (age >65 years; APACHE II >12 at extubation; BMI >30; difficult/prolonged weaning; moderate-to-severe COPD; heart failure as indication for ventilation; >2 comorbidities; airway patency problems; inability to clear secretions; mechanical ventilation >7 days).
  • Exclusion criteria: Low-risk for reintubation; do-not-reintubate order; unscheduled extubation; no informed consent; hypercapnia during spontaneous breathing trial.
  • Enrolment: 1211 assessed as weanable; 604 randomised (314 NIV; 290 HFNC).
• Intervention:
  • HFNC started immediately after extubation; initial flow 10 L/min, increased by 5 L/min increments until discomfort.
  • Heated humidification at 37°C; FiO₂ titrated to maintain SpO₂ >92%.
  • Applied for 24 hours, then switched to conventional oxygen (nasal prongs or Venturi) as required.
• Comparison:
  • NIV started immediately after extubation via full-face mask and BiPAP Vision ventilator; planned continuous delivery for 24 hours.
  • PEEP and inspiratory pressure support adjusted to target respiratory rate 25/min and adequate gas exchange (SaO₂ 92%, pH 7.35); FiO₂ titrated for SpO₂ >92%.
  • Sedatives to improve tolerance were not allowed; after 24 hours, NIV was withdrawn and oxygen given by Venturi mask.
• Blinding: Unblinded (clinicians and patients aware of treatment); outcomes included objective endpoints (reintubation, mortality) but also a composite “postextubation respiratory failure” construct with clinician-dependent elements.
• Statistics: A total of 600 patients (300 per group) were required to demonstrate noninferiority with an absolute margin of 10% (unilateral 95% CI; one-sided 5% alpha), assuming a reintubation rate of 20%–25%, with 80% power and 15% allowance for attrition; co-primary outcomes analysed by intention-to-treat and per-protocol; secondary outcomes analysed by intention-to-treat.
• Follow-Up Period: Co-primary outcomes to 72 hours postextubation; secondary outcomes included time to reintubation, ICU/hospital length of stay, and ICU/hospital mortality (to discharge).

Key Results

This trial was not stopped early.

Outcome	High-flow nasal cannula (HFNC)	Noninvasive ventilation (NIV)	Effect	p value / 95% CI	Notes
All-cause reintubation within 72 h (co-primary)	66/290 (22.8%)	60/314 (19.1%)	Risk difference (NIV − HFNC): −3.7%	One-sided 95% CI −9.1 to ∞	Noninferiority margin 10% (met; ITT and per-protocol analyses directionally consistent)
Postextubation respiratory failure within 72 h (co-primary)	78/290 (26.9%)	125/314 (39.8%)	Risk difference (NIV − HFNC): 12.9%	One-sided 95% CI 6.6 to ∞	Lower incidence with HFNC
Postextubation respiratory failure due to hypoxaemia	13/290 (4.5%)	31/314 (9.9%)	Risk difference (NIV − HFNC): 5.4%	95% CI 1.6 to 9.1; P=0.003	Component contributing to lower composite failure with HFNC
Postextubation respiratory failure due to inability to clear secretions	30/290 (10.3%)	52/314 (16.6%)	Risk difference (NIV − HFNC): 6.2%	95% CI 1.5 to 10.8; P=0.007	Interpretation complicated by dependence on clinician assessment of secretions clearance
Adverse events requiring treatment discontinuation	0/290 (0%)	135/314 (42.9%)	Not reported	P<0.001	Only in NIV group (facial skin erythema 6.4%, abdominal distension 4.1%, discomfort 14.0%)
ICU length of stay after randomisation (survivors)	3 (2–6) days	4 (2–6) days	Absolute difference: 1 day	95% CI −0.1 to 2.1; P=0.048	Median difference small; total ICU LOS was not significantly different
ICU mortality	9/290 (3.1%)	10/314 (3.2%)	Risk difference (NIV − HFNC): 0.1%	95% CI −2.7 to 3.0; P=0.93	Not powered for mortality differences
Hospital mortality	23/290 (7.9%)	25/314 (8.0%)	Risk difference (NIV − HFNC): 0.1%	95% CI −4.2 to 4.4; P=0.94	No signal of harm with either strategy

• HFNC met noninferiority for 72-hour reintubation (risk difference NIV − HFNC: −3.7%; one-sided 95% CI −9.1 to ∞; margin 10%).
• HFNC was associated with lower postextubation respiratory failure (26.9% vs 39.8%; risk difference 12.9%; one-sided 95% CI 6.6 to ∞), with lower rates of hypoxaemia and fewer secretion/fatigue criteria met.
• NIV tolerance and delivery were materially different: 42.9% required treatment discontinuation vs 0% in HFNC, while mortality was similar.

Internal Validity

• Randomisation and allocation:
  • Allocation was concealed via an independent telephone call centre.
  • Simple randomisation (no blocking/stratification) led to unequal group sizes (314 NIV vs 290 HFNC), increasing susceptibility to chance imbalances.
• Dropout/exclusions (post-randomisation):
  • Intention-to-treat population: 604/604.
  • Per-protocol population: 312/314 NIV and 288/290 HFNC (2 excluded per group; minimal attrition).
• Performance and detection bias:
  • Open-label design; clinicians were aware of assigned treatment.
  • Reintubation is clinically meaningful but threshold-driven; postextubation respiratory failure criteria include clinician-dependent elements (e.g., secretion clearance, respiratory muscle fatigue), creating potential for differential classification.
• Protocol adherence and “dose” delivered:
  • Both strategies were intended for 24 hours postextubation.
  • NIV delivery was attenuated: median duration 14 hours (IQR 8–23); adverse events requiring discontinuation occurred in 135/314 (42.9%).
  • HFNC delivered high flows: mean gas flow at 12 hours 50 (SD 5) L/min; FiO₂ at 12 hours was lower with HFNC (median 35% [IQR 30–40]) vs NIV (40% [35–50]).
  • Prohibition of sedatives for NIV tolerance may have increased discontinuation and reduced effective separation of ventilatory support exposure.
• Baseline characteristics and risk enrichment:
  • High-risk profile achieved: median number of risk factors 3 (IQR 2–4) in both groups.
  • Severity at extubation was similar: APACHE II median 10 (8–12) NIV vs 11 (8–12) HFNC; PaO₂/FiO₂ during SBT 194 (148–261) vs 191 (149–251).
  • Imbalances were present: heart failure as indication for ventilation 9.9% NIV vs 5.5% HFNC; surgical diagnosis at admission 33.4% NIV vs 43.8% HFNC.
• Timing and co-interventions:
  • Both interventions started immediately postextubation, aligning with the peak risk window captured by the 72-hour co-primary endpoints.
  • Rescue NIV after onset of postextubation respiratory failure was not allowed, limiting contamination but potentially influencing escalation thresholds.
• Statistical rigor:
  • Noninferiority assessed using one-sided 95% CI and a prespecified absolute margin (10%).
  • Primary outcomes reported in both ITT and per-protocol populations (appropriate for noninferiority designs).
  • Power was adequate for the co-primary noninferiority endpoints, not for mortality or small subgroup effects.

Conclusion on Internal Validity: Moderate to strong. Central randomisation, minimal attrition, and concordant ITT/per-protocol analyses support credibility; open-label delivery and substantial NIV discontinuation introduce meaningful risks of performance/detection bias and dose dilution for the active comparator.

External Validity

• Population representativeness:
  • Adult ICU cohort from 3 Spanish centres; all patients were deemed “weanable” and extubated after successful spontaneous breathing trial.
  • Enriched for high-risk extubations; low-risk patients (466) were excluded at screening, limiting inference to the broader ICU extubation population.
  • Hypercapnia during spontaneous breathing trial was an exclusion (59 patients), limiting applicability to chronic hypercapnic respiratory disorders.
• Applicability across settings:
  • Requires HFNC infrastructure and staff familiarity; the protocol used high flows (commonly ~50 L/min) with heated humidification.
  • NIV protocol mandated continuous application for 24 hours without sedative support; real-world NIV schedules and tolerance strategies vary and may change the balance of effectiveness and harms.
  • Findings are most applicable to ICUs with similar extubation-readiness and reintubation thresholds; translation to step-down/ward environments is uncertain.

Conclusion on External Validity: Good generalisability to non-hypercapnic, high-risk ICU extubations in centres with HFNC capability; limited generalisability to hypercapnic COPD populations or systems that deliver NIV with different dosing/tolerance practices.

Strengths & Limitations

• Strengths:
  • Pragmatic multicentre randomised design with a clinically relevant high-risk population.
  • Co-primary outcomes focused on the early postextubation period (72 hours), when preventable failure is most likely.
  • Central allocation concealment; prespecified noninferiority margin; both ITT and per-protocol analyses for primary outcomes.
  • Very low loss to follow-up and clear reporting of adverse events and treatment discontinuation.
• Limitations:
  • Open-label intervention and outcomes that include clinician-dependent components (especially the postextubation respiratory failure construct).
  • Simple randomisation caused unequal group sizes (314 vs 290) and some baseline imbalances.
  • Marked NIV intolerance and dose dilution (median 14 hours; 42.9% discontinuation), with sedatives prohibited.
  • Intervention duration fixed at 24 hours; optimal duration may differ by risk strata.
  • Exclusion of hypercapnia during spontaneous breathing trial limits inference for patient groups where NIV benefit may be greatest.

Interpretation & Why It Matters

• Clinical implications

  • HFNC provided comparable (noninferior) protection against early reintubation to prophylactic NIV, while reducing postextubation respiratory failure and demonstrating markedly better tolerability.
  • Hard outcomes (ICU and hospital mortality) were similar; the principal value proposition is reducing early respiratory deterioration and improving comfort, rather than mortality benefit.
  • Supports a risk-stratified approach: HFNC as a default prophylactic support in many high-risk but non-hypercapnic patients, with NIV reserved for patients needing explicit ventilatory unloading (e.g., hypercapnic physiology or very high-risk phenotypes).

Controversies & Other Evidence

• Assay sensitivity in a noninferiority comparison:
  • The correspondence emphasised that prophylactic NIV benefit is strongest in chronic hypercapnic respiratory disorders, while patients with hypercapnia during spontaneous breathing trial were excluded in this trial; this complicates noninferiority interpretation if the active comparator effect is attenuated in the enrolled population.¹
• Outcome construct and clinician-dependent thresholds:
  • Concerns were raised that several elements used to classify “postextubation respiratory failure” (e.g., secretion clearance, reduced level of consciousness) have limited specificity for a respiratory cause and may be sensitive to clinician behaviour in an unblinded trial, potentially influencing both composite failure classification and reintubation decisions.¹
• Post-publication correction and sensitivity analysis:
  • The investigators clarified that “inability to clear respiratory secretions” was intended as a criterion but was omitted from the printed definition of postextubation respiratory failure and later corrected; reanalysis excluding patients defined by secretion-related criteria reported reintubation in 73/314 (23.2%) NIV vs 48/290 (16.5%) HFNC (risk difference 6.7%; one-sided 95% CI 1.3 to ∞).²
• Subsequent synthesis and guideline integration:
  • ERS/ATS clinical practice guidelines recommend prophylactic NIV after extubation in high-risk patients and recommend against routine NIV in low-risk patients; they also recommend against NIV as rescue therapy once postextubation respiratory failure is established.³
  • A large systematic review and network meta-analysis (36 RCTs; 6806 patients; searches through October 2021) found that, compared with conventional oxygen, both NIV (OR 0.65; 95% CI 0.52–0.82) and HFNC (OR 0.63; 95% CI 0.45–0.87) reduced reintubation, with no difference between NIV and HFNC (OR 1.04; 95% CI 0.78–1.38), and no short-term mortality benefit for either strategy.⁴
  • The HIGH-WEAN trial evaluated alternating NIV with HFNC vs HFNC alone after extubation in high-risk patients and reported lower reintubation with the combined strategy, supporting the concept that NIV may add benefit in higher-risk strata beyond HFNC monotherapy.⁵

Summary

• In 604 high-risk ICU patients, HFNC for 24 hours met noninferiority vs prophylactic NIV for 72-hour reintubation (22.8% vs 19.1%; risk difference NIV − HFNC −3.7%; one-sided 95% CI −9.1 to ∞; margin 10%).
• HFNC was associated with lower postextubation respiratory failure (26.9% vs 39.8%; risk difference 12.9%; one-sided 95% CI 6.6 to ∞), with lower rates of hypoxaemia and fewer secretion/fatigue criteria met.
• NIV was frequently not tolerated (42.9% required discontinuation), whereas no HFNC discontinuations were reported; key NIV adverse events included discomfort (14.0%), facial skin erythema (6.4%), and abdominal distension (4.1%).
• ICU and hospital mortality were similar (ICU 3.1% vs 3.2%; hospital 7.9% vs 8.0%).
• HFNC reduced post-randomisation ICU length of stay by a small margin (median 3 vs 4 days; P=0.048), with no clear signal for broader length-of-stay reductions.

Overall Takeaway

This noninferiority trial supports HFNC as a practical, better-tolerated alternative to prophylactic NIV after extubation in high-risk (but non-hypercapnic) ICU patients, achieving similar reintubation rates while reducing the composite of postextubation respiratory failure. Its landmark contribution is reframing postextubation prophylaxis towards HFNC as a default strategy in many high-risk patients, while subsequent evidence supports escalation to combined NIV/HFNC approaches for those at the very highest risk.

Bibliography

• Tomazini BM, Besen BAMP. High-flow oxygen vs noninvasive ventilation for postextubation respiratory failure. JAMA. 2017 Feb 28;317(8):855-6.
• Hernández G, Colinas L, Canabal A. In reply. JAMA. 2017 Feb 28;317(8):855-6.
• Rochwerg B, Brochard L, Elliott MW, Hess D, Hill NS, Nava S, et al. Official ERS/ATS clinical practice guidelines: noninvasive ventilation for acute respiratory failure. Eur Respir J. 2017;50(2):1602426.
• Fernando SM, Tran A, Sadeghirad B, Burns KEA, Fan E, Brodie D, et al. Noninvasive respiratory support following extubation in critically ill adults: a systematic review and network meta-analysis. Intensive Care Med. 2022;48:137-47.
• Thille AW, Muller G, Gacouin A, Coudroy R, Sonneville R, Beloncle F, et al. Effect of postextubation high-flow nasal oxygen with noninvasive ventilation vs high-flow nasal oxygen alone on reintubation among patients at high risk of extubation failure: a randomized clinical trial. JAMA. 2019 Oct 15;322(15):1465-75.