Respiratory failure

Resident Survival Guide

For patient information, click here

Respiratory failure Microchapters
Home
Patient Information
Overview
Historical Perspective
Classification
Pathophysiology
Causes
Differentiating Respiratory Failure from other Diseases
Epidemiology and Demographics
Risk Factors
Screening
Natural History, Complications and Prognosis
Diagnosis
Diagnostic Study of Choice
History and Symptoms
Physical Examination
Laboratory Findings
Chest X Ray
Electrocardiogram
CT
MRI
Echocardiography and ultrasound
Other Imaging Findings
Other Diagnostic Studies
Treatment
Medical therapy
Oxygen therapy
Mechanical ventilation
Primary Prevention
Secondary Prevention
Cost-Effectiveness of Therapy
Future or Investigational Therapies
Case Studies
Case #1
Respiratory failure On the Web
Most recent articles
Most cited articles
Review articles
CME Programs
Powerpoint slides
Images
American Roentgen Ray Society Images of Respiratory failure All Images X-rays Echo & Ultrasound CT Images MRI
Ongoing Trials at Clinical Trials.gov
US National Guidelines Clearinghouse
NICE Guidance
FDA on Respiratory failure
CDC on Respiratory failure
Respiratory failure in the news
Blogs on Respiratory failure
Directions to Hospitals Treating Type page name here
Risk calculators and risk factors for Respiratory failure

Editor-In-Chief: C. Michael Gibson, M.S., M.D. [1]; Associate Editor(s)-in-Chief: Omar Elshafei, MD[2]; M. Khurram Afzal, MD [3]; Hadeel Maksoud M.D.[4]

Synonyms and keywords:: Acute respiratory failure, type I respiratory failure, type II respiratory failure, type III respiratory failure, type IV respiratory failure, type 1 respiratory failure, type 2 respiratory failure, type 3 respiratory failure, type 4 respiratory failure, acute hypoxemic respiratory failure, acute hypercapnic respiratory failure, acute-on-chronic respiratory failure, ventilatory failure

Overview

Acute respiratory failure is the failure of the respiratory system to maintain adequate gas exchange, producing hypoxemia, hypercapnia, or both. Hypoxemic (type 1) respiratory failure is defined by a PaO2 less than 60 mm Hg or an SaO2 less than 88%, and hypercapnic (type 2) respiratory failure by a PaCO2 of 45 mm Hg or more with a pH less than 7.35. Hypoxemia results from ventilation-perfusion mismatch, shunt, diffusion limitation, hypoventilation, or low inspired oxygen tension, whereas hypercapnia reflects alveolar hypoventilation relative to carbon dioxide production ("pump failure"), increased dead space, or increased CO2 production. Traditional teaching also describes perioperative (type 3) and shock-related (type 4) respiratory failure. Common causes include pneumonia, acute respiratory distress syndrome (ARDS), cardiogenic pulmonary edema, pulmonary embolism, exacerbations of chronic obstructive pulmonary disease (COPD) and asthma, obesity hypoventilation syndrome, neuromuscular disease, and drug-induced hypoventilation.

Acute respiratory failure is one of the most common reasons for hospital and intensive care unit (ICU) admission. In the United States, hospitalizations with respiratory failure and mechanical ventilation rose from 249 to 455 per 100,000 adults between 2002 and 2017, while hospital mortality fell from 34% to 23%. ARDS accounts for 10.4% of ICU admissions worldwide, and hospital mortality reaches 46.1% in severe ARDS.

Diagnosis rests on arterial blood gas analysis, which remains the reference standard. Pulse oximetry is essential for monitoring but can overestimate oxygenation, particularly in patients with darker skin, and peripheral venous blood gas analysis is useful mainly to rule out hypercapnia. Chest radiography, point-of-care lung ultrasound, electrocardiography, echocardiography, and computed tomography identify the underlying cause.

Management combines treatment of the cause with oxygen titrated to a target saturation (94-98% for most patients, 88-92% for those at risk of hypercapnia) and the least invasive effective respiratory support. Current guidelines recommend high-flow nasal cannula (HFNC) oxygen as first-line support for acute hypoxemic respiratory failure, because it reduces intubation, and noninvasive ventilation (NIV) for acute hypercapnic respiratory failure with acidosis, because it reduces both intubation and mortality. Patients who fail noninvasive support should be intubated without delay, after preoxygenation with HFNC or NIV, and ventilated with lung-protective settings (tidal volume about 6 mL/kg predicted body weight and plateau pressure of 30 cm H2O or less). Prone positioning, neuromuscular blockade, corticosteroids, and venovenous extracorporeal membrane oxygenation (ECMO) are reserved for selected patients with moderate-to-severe ARDS. Survivors often have long-term physical, cognitive, and psychological impairment (post-intensive care syndrome) and benefit from structured follow-up.

Historical Perspective

The evidence base for acute respiratory failure has developed through a series of landmark studies:

• 1995: A multicenter randomized controlled trial showed that NIV for COPD exacerbations reduced intubation from 74% to 26% and in-hospital mortality from 29% to 9%.
• 2000: The ARDS Network ARMA trial established low tidal volume ventilation (6 mL/kg predicted body weight), reducing mortality from 39.8% to 31.0%. In the same year, ward-based NIV for COPD exacerbations was shown to halve in-hospital mortality (20% to 10%).
• 2000: A US national estimate reported 329,766 adults with acute respiratory failure in 1994 and a 31-day hospital mortality of 31.4%.
• 2008: The 3CPO trial clarified the role of CPAP and NIV in acute cardiogenic pulmonary edema.
• 2012: The Berlin definition replaced the American-European Consensus Conference definition of ARDS and graded severity by PaO2/FiO2.
• 2013: The PROSEVA trial showed that prolonged prone positioning reduced 28-day mortality in severe ARDS from 32.8% to 16.0%.
• 2015: The FLORALI trial positioned HFNC as a first-line option for acute hypoxemic respiratory failure.
• 2018: The EOLIA trial and its Bayesian reanalysis defined the role of venovenous ECMO in very severe ARDS.
• 2020-2022: The COVID-19 pandemic produced large trials of dexamethasone, CPAP, HFNC, helmet NIV, and awake prone positioning.
• 2024: A new global definition of ARDS added patients on HFNC, pulse-oximetry criteria, ultrasound imaging, and a version for resource-limited settings.
• 2025-2026: The RENOVATE and SOHO trials compared HFNC with NIV and standard oxygen, the American Thoracic Society (ATS) issued its first guideline covering noninvasive respiratory support across the whole course of acute respiratory failure, and the Surviving Sepsis Campaign updated its respiratory recommendations.

Classification

Classification by Gas Exchange Abnormality

Respiratory failure is classified by the predominant gas exchange abnormality. Physiologically, it can be viewed as "lung failure," which causes hypoxemia, or "pump failure" (failure of the ventilatory pump, including drive, nerves, muscles, and chest wall), which causes alveolar hypoventilation and hypercapnia. The arterial thresholds below are those used in current reviews and guidelines.

Type	Arterial blood gas definition	Main mechanisms	Typical causes
Type 1 (hypoxemic)	PaO2 < 60 mm Hg (8 kPa) or SaO2 < 88%; PaCO2 normal or low	V/Q mismatch, shunt, diffusion limitation, hypoventilation, low inspired oxygen tension	Pneumonia, ARDS, cardiogenic pulmonary edema, pulmonary embolism, atelectasis, pneumothorax
Type 2 (hypercapnic)	PaCO2 ≥ 45 mm Hg (6 kPa) and pH < 7.35	Alveolar hypoventilation, increased dead space fraction, increased CO2 production	COPD exacerbation, asthma, obesity hypoventilation syndrome, neuromuscular disease, chest wall disease, opioid or sedative toxicity
Type 3 (perioperative)	Traditional teaching category; not part of current guideline definitions	Atelectasis from reduced functional residual capacity, often with hypoventilation	Upper abdominal or thoracic surgery, obesity, postoperative pain and sedation
Type 4 (shock-related)	Traditional teaching category; not part of current guideline definitions	Respiratory muscle hypoperfusion and high ventilatory demand	Septic shock, cardiogenic shock, hypovolemic shock

Classification by Time Course

• Acute respiratory failure: develops over minutes to days; in acute hypercapnia the pH falls because renal bicarbonate compensation has not yet occurred.
• Chronic respiratory failure: develops over months to years; in chronic hypercapnia, renal bicarbonate retention keeps the pH near normal.
• Acute-on-chronic respiratory failure: an acute deterioration in a patient with chronic respiratory failure, typically a COPD exacerbation, with acidosis despite an elevated bicarbonate. British guidelines manage acute and acute-on-chronic hypercapnic respiratory failure under the same NIV criteria.

Classification of Acute Respiratory Distress Syndrome

ARDS is the most severe form of acute hypoxemic respiratory failure. The Berlin definition grades severity by PaO2/FiO2 on PEEP or CPAP of 5 cm H2O or more, and the 2024 global definition extends the diagnosis to non-intubated patients and resource-limited settings. In resource-limited settings, the earlier Kigali modification (SpO2/FiO2 ≤ 315, no PEEP requirement, bilateral opacities on ultrasound or radiograph) identified ARDS in 4.0% of hospital admissions, none of whom would have met Berlin criteria.

Feature	Berlin definition (2012)	Global definition (2024)
Respiratory support	Invasive ventilation, or noninvasive ventilation with PEEP or CPAP ≥ 5 cm H2O	Adds patients on HFNC with flow ≥ 30 L/min; in resource-limited settings, no PEEP, flow rate, or device requirement
Hypoxemia criterion	PaO2/FiO2 ≤ 300 mm Hg	PaO2/FiO2 ≤ 300 mm Hg, or SpO2/FiO2 ≤ 315 if SpO2 ≤ 97%
Imaging	Bilateral opacities on chest radiograph or CT	Bilateral opacities on radiograph or CT, or on lung ultrasound
Severity grades	Mild: 200 < PaO2/FiO2 ≤ 300 Moderate: 100 < PaO2/FiO2 ≤ 200 Severe: PaO2/FiO2 ≤ 100	Retains severity grading for intubated patients
Mortality by severity	27% (95% CI 24%-30%), 32% (29%-34%), and 45% (42%-48%) for mild, moderate, and severe ARDS	Not a mortality model

Pediatric ARDS is defined separately by the Pediatric Acute Lung Injury Consensus Conference (PALICC-2).

Pathophysiology

Mechanisms of Hypoxemia

Five mechanisms produce arterial hypoxemia: ventilation-perfusion mismatch (low V/Q units), shunt, diffusion limitation, alveolar hypoventilation, and low inspired oxygen tension. Shunt and low V/Q regions are the most frequent causes in clinical practice, whereas diffusion limitation, hypoventilation, and low inspired PO2 can cause hypoxemia even without V/Q mismatch. Shunt responds poorly to supplemental oxygen, which is why oxygenation in ARDS improves more with recruitment of collapsed lung (PEEP, prone positioning) than with increases in FiO2. Wasted ventilation (dead space) mainly increases the minute ventilation needed to maintain a normal PaCO2 rather than raising PaCO2 directly.

Mechanism	A-a gradient	Response to supplemental oxygen	Examples
V/Q mismatch (low V/Q)	Increased	Good	COPD, asthma, pneumonia, pulmonary embolism
Shunt (V/Q = 0)	Increased	Poor	ARDS, lobar pneumonia, atelectasis, pulmonary edema, intracardiac or pulmonary arteriovenous shunt
Diffusion limitation	Increased	Good	Interstitial lung disease, especially during exercise
Hypoventilation	Normal	Good (PaCO2 remains high)	Opioid or sedative toxicity, neuromuscular disease, chest wall disease
Low inspired PO2	Normal	Good	High altitude

The alveolar-arterial (A-a) oxygen gradient separates hypoventilation and low inspired PO2 (normal gradient) from intrinsic lung disease (widened gradient). The alveolar PO2 is estimated with the alveolar gas equation: PAO2 = FiO2 x (barometric pressure - water vapor pressure) - PaCO2/R, which at sea level on room air is approximately 0.21 x (760 - 47) - PaCO2/0.8.

Mechanisms of Hypercapnia

Hypercapnia occurs when alveolar ventilation is inadequate for the body's CO2 production. It may result from alveolar hypoventilation, an increased dead space fraction, or increased CO2 production. Ventilatory failure develops when the respiratory system can no longer meet ventilatory demand, and reflects reduced effective alveolar ventilation, increased physiologic dead space, airflow limitation, dynamic hyperinflation, altered respiratory drive, and ventilation-perfusion abnormalities, which together increase the work of breathing and promote respiratory muscle fatigue. The underlying failure can therefore lie in the central drive (drugs, brainstem lesions), the neuromuscular apparatus, the chest wall, or the airways and lungs.

In patients with COPD exacerbations, high-concentration oxygen can worsen hypercapnia. In a study of 22 patients breathing 100% oxygen, 12 retained CO2 (PaCO2 rose by more than 3 mm Hg; mean 8.3 ± 5.6 mm Hg), and their minute ventilation fell from 9.0 ± 1.5 to 7.2 ± 1.2 L/min, whereas ventilation in nonretainers was unchanged; both hypoventilation and V/Q redistribution contributed. This physiology underlies the controlled oxygen targets recommended for patients at risk of hypercapnia.

Acute Respiratory Distress Syndrome

ARDS is characterized by diffuse inflammatory injury to the alveolar-capillary barrier, protein-rich pulmonary edema, loss of aerated lung, and shunt. Patients are heterogeneous: latent class analysis of two ARDS Network trials (1022 patients) identified a hyperinflammatory subphenotype with higher mortality and fewer ventilator-free and organ failure-free days, and the effect of a higher PEEP strategy on mortality differed between subphenotypes (p = 0.049). An increased pulmonary dead space fraction is an independent marker of risk: in 179 intubated patients the mean dead space fraction was 0.58 ± 0.09, it was higher in nonsurvivors (0.63 ± 0.10 vs 0.54 ± 0.09), and each 0.05 increase raised the odds of death by 45% (odds ratio 1.45; 95% CI 1.15-1.83).

Ventilator-Induced and Patient Self-Inflicted Lung Injury

Mechanical ventilation can worsen lung injury through excessive stress and strain. In an analysis of 3562 patients from nine trials, each 1-SD increase in driving pressure (approximately 7 cm H2O) was associated with higher mortality (relative risk 1.41; 95% CI 1.31-1.51), even when plateau pressure and tidal volume were within protective limits. Spontaneously breathing patients with high respiratory drive, large tidal volumes, and large transpulmonary pressure swings may develop a similar injury, termed patient self-inflicted lung injury (P-SILI), which provides a rationale for not delaying intubation when noninvasive support fails. During NIV for de novo hypoxemic respiratory failure, an expired tidal volume above 9.5 mL/kg predicted body weight predicted NIV failure (sensitivity 82%, specificity 87%) in patients with PaO2/FiO2 of 200 mm Hg or less.

Causes

Acute respiratory failure is a syndrome with many causes. COPD exacerbations are the most common cause of hypercapnic respiratory failure, which also occurs with asthma, cardiogenic pulmonary edema, obesity-related hypoventilation, drug-induced hypoventilation, and neuromuscular disorders. In a cohort of 873 patients with hypercapnic respiratory failure, 83% had at least one prespecified cause; infection and neuromuscular disease were associated with higher in-hospital mortality, whereas obstructive lung disease and cardiac failure were associated with lower mortality. Among 78 ICU survivors of acute hypercapnic respiratory failure, 67% had COPD (only 19 previously diagnosed), patients without COPD were primarily obese, and severe obstructive sleep apnea was present in 51% of those with COPD and 81% of those without.

Site of failure	Predominant type	Causes
Alveoli and lung parenchyma	Hypoxemic	Pneumonia (bacterial, viral including COVID-19 and influenza, aspiration), ARDS (from sepsis, pneumonia, aspiration, pancreatitis, trauma, transfusion), cardiogenic pulmonary edema, diffuse alveolar hemorrhage, acute exacerbation of interstitial lung disease
Pulmonary vasculature	Hypoxemic	Pulmonary embolism, fat embolism, pulmonary hypertension with right ventricular failure
Pleura	Hypoxemic or mixed	Pneumothorax, large pleural effusion
Airways	Hypercapnic or mixed	COPD exacerbation, asthma exacerbation, upper airway obstruction (anaphylaxis, angioedema, foreign body), bronchiectasis
Central respiratory drive	Hypercapnic	Opioid, benzodiazepine or sedative toxicity, stroke, central hypoventilation, head injury
Spinal cord, nerves, neuromuscular junction, and muscles	Hypercapnic	Guillain-Barré syndrome, myasthenia gravis, amyotrophic lateral sclerosis, muscular dystrophies, cervical spinal cord injury, critical illness neuromyopathy
Chest wall	Hypercapnic	Obesity hypoventilation syndrome, kyphoscoliosis, flail chest
Increased CO2 production or ventilatory demand	Hypercapnic or mixed	Sepsis, fever, thyrotoxicosis, excessive carbohydrate load in patients with limited ventilatory reserve
Perioperative	Hypoxemic or mixed	Atelectasis, residual anesthetic or neuromuscular blockade, opioid analgesia, abdominal or thoracic surgery

Differentiating Respiratory Failure from other Diseases

Acute respiratory failure must be distinguished from conditions that cause dyspnea or tachypnea without failure of gas exchange, and from situations in which pulse oximetry is misleading:

• Metabolic acidosis (for example diabetic ketoacidosis, lactic acidosis, salicylate toxicity): compensatory hyperventilation with a low PaCO2 and normal oxygenation.
• Anxiety and hyperventilation syndrome: low PaCO2 with normal A-a gradient; a diagnosis of exclusion.
• Anemia, carbon monoxide poisoning, and methemoglobinemia: reduced oxygen content or delivery despite a normal PaO2; carbon monoxide poisoning requires CO-oximetry because standard pulse oximetry does not distinguish carboxyhemoglobin from oxyhemoglobin.
• Occult hypoxemia from pulse oximeter bias: among measurements with SpO2 of 92-96%, arterial saturation was below 88% in 11.7% of measurements in Black patients compared with 3.6% in White patients.

Once respiratory failure is confirmed, the priority is to identify its cause. Point-of-care lung ultrasound using the BLUE protocol correctly diagnosed the cause in 90.5% of 260 patients with acute respiratory failure. International recommendations support point-of-care lung ultrasound for this purpose. Evaluation of suspected acute heart failure and pulmonary embolism follows the 2026 European Society of Cardiology heart failure guideline and the 2026 AHA/ACC multisociety and 2019 ESC pulmonary embolism guidelines.

Condition	Key clinical features	Blood gas pattern	Imaging and lung ultrasound	Confirmatory or supportive tests
Pneumonia	Fever, cough, purulent sputum, pleuritic pain, focal crackles	Type 1; type 2 if exhausted or with underlying COPD	Focal consolidation; on ultrasound, anterior or posterior consolidation, asymmetric interstitial pattern, or effusion (sensitivity 89%, specificity 94%)	Cultures, respiratory pathogen testing, procalcitonin
ARDS	Known risk factor (sepsis, pneumonia, aspiration, pancreatitis, trauma) within 1 week; refractory hypoxemia	Type 1 with severe shunt; PaO2/FiO2 ≤ 300 mm Hg or SpO2/FiO2 ≤ 315	Bilateral opacities not fully explained by effusion, collapse, nodules, or cardiac failure	Echocardiography to exclude a primary cardiac cause
Cardiogenic pulmonary edema	Orthopnea, paroxysmal nocturnal dyspnea, raised jugular venous pressure, S3, peripheral edema, cardiac history	Type 1; hypercapnia in severe cases	Bilateral perihilar opacities, cardiomegaly, effusions; diffuse anterior B lines with lung sliding (sensitivity 97%, specificity 95%)	Natriuretic peptides, electrocardiogram, troponin, echocardiography
Pulmonary embolism	Sudden dyspnea, pleuritic pain, hemoptysis, syncope, signs of deep vein thrombosis	Type 1 with hypocapnia; widened A-a gradient	Often near-normal chest radiograph; normal anterior lung profile plus deep vein thrombosis on ultrasound (sensitivity 81%, specificity 99%)	Clinical probability, D-dimer, CT pulmonary angiography, echocardiography
COPD exacerbation	Known COPD, increased dyspnea, cough, and sputum, wheeze, hyperinflation	Type 2, often acute-on-chronic with raised bicarbonate	Hyperinflation; predominant A lines with lung sliding (asthma or COPD: sensitivity 89%, specificity 97%)	Blood gases; evaluate for mimics (pneumonia, pulmonary embolism, heart failure)
Asthma exacerbation	Known asthma, wheeze, reduced peak flow, silent chest in life-threatening attacks	Early hypocapnia; normal or rising PaCO2 signals impending failure	Hyperinflation; A-line profile	Peak expiratory flow, blood gases
Pneumothorax	Sudden pleuritic pain and dyspnea, reduced breath sounds, hyperresonance; hypotension in tension pneumothorax	Type 1	Visible pleural line on radiograph; absent lung sliding, A lines, and lung point on ultrasound (sensitivity 81%, specificity 100%)	Chest radiograph or CT
Neuromuscular disease (Guillain-Barré syndrome, myasthenia gravis, amyotrophic lateral sclerosis)	Limb or bulbar weakness, weak cough, paradoxical abdominal motion, orthopnea	Type 2 with normal A-a gradient	Often normal lungs; atelectasis or aspiration	Vital capacity and other pulmonary function tests; polysomnography or overnight oximetry
Obesity hypoventilation syndrome	Body mass index ≥ 30 kg/m2, daytime somnolence, snoring, right heart failure	Type 2, acute-on-chronic, raised bicarbonate	Low lung volumes	Serum bicarbonate < 27 mmol/L can be used to exclude OHS when clinical suspicion is below 20%; polysomnography
Drug-induced hypoventilation (opioids, benzodiazepines, alcohol)	Reduced consciousness, miosis with opioids, low respiratory rate	Type 2 with normal A-a gradient	Often normal; aspiration	Toxicology; response to naloxone in opioid toxicity

Epidemiology and Demographics

Incidence

• In 1994, an estimated 329,766 patients in the United States met criteria for acute respiratory failure, an incidence of 137.1 hospitalizations per 100,000 residents aged 5 years or older.
• US hospitalizations for acute respiratory failure increased from 1,007,549 in 2001 to 1,917,910 in 2009, and total costs rose from $30.1 billion to $54.3 billion.
• In the US National Inpatient Sample, 1,146,195 discharges in 2017 carried a respiratory failure diagnosis plus a mechanical ventilation code; incidence rose by 83%, from 249 to 455 cases per 100,000 adults, between 2002 and 2017. Using diagnosis codes alone, incidence rose from 429 to 1,275 per 100,000 adults.
• In six US states in 2005, 2.8% of hospitalizations (180,326 of 6,469,674) involved invasive mechanical ventilation, projecting to 790,257 hospitalizations nationally (2.7 per 1000 population) and $27 billion in costs, 12% of all hospital costs.
• In the Finnish FINNALI study, 39% of ICU admissions (958 of 2,473) received ventilatory support for more than 6 hours, and the incidence of acute respiratory failure, acute lung injury, and ARDS was 149.5, 10.6, and 5.0 per 100,000 population per year.
• In the LUNG SAFE study of 459 ICUs in 50 countries, ARDS occurred in 10.4% of ICU admissions and 23.4% of patients requiring mechanical ventilation; 30.0% of cases were mild, 46.6% moderate, and 23.4% severe. Clinicians recognized ARDS in only 51.3% of mild and 78.5% of severe cases.
• In a Rwandan referral hospital, 4.0% of admissions met the Kigali modification of ARDS criteria and only 30.9% of these patients were admitted to an ICU.

Mortality Trends

• Hospital mortality from acute respiratory failure in the US fell from 27.6% in 2001 to 20.6% in 2009, and among patients receiving mechanical ventilation from 34% in 2002 to 23% in 2017.
• Population-level, age-adjusted mortality related to acute respiratory failure nevertheless rose from 74.9 (95% CI 74.6-75.2) per 100,000 in 2014 to 85.6 (95% CI 85.3-85.9) per 100,000 in 2018.

Age

• The incidence of acute respiratory failure rises nearly exponentially with each decade of life until age 85.

Use of Noninvasive Ventilation

• Use of NIV among US hospitalizations for acute respiratory failure rose from 4% in 2001 to 10% in 2009. In FINNALI, ventilatory support was started noninvasively in 19% of patients.

Special Populations

• Children: pediatric ARDS occurred in 3.2% of pediatric ICU admissions (744 of 23,280) in an international study.
• Pregnancy: respiratory failure complicates up to 1 in 500 pregnancies, and in the US, ARDS requiring mechanical ventilation increased from 36.5 (95% CI 33.1-39.8) to 59.6 (95% CI 57.7-61.4) per 100,000 live births between 2006 and 2012.

Risk Factors

Risk factors for acute respiratory failure include:

• Advanced age: incidence rises nearly exponentially with each decade until age 85.
• Chronic lung disease: COPD is the most common cause of hypercapnic respiratory failure, and in ICU survivors it is frequently undiagnosed before admission.
• Obesity, obesity hypoventilation syndrome, and obstructive sleep apnea
• Neuromuscular and chest wall disease
• Sedative and opioid exposure
• Conditions predisposing to ARDS: sepsis, shock, pneumonia, aspiration, trauma, and high-risk surgery. In a multicenter cohort of 5,584 at-risk patients, 6.8% developed acute lung injury, and the Lung Injury Prediction Score (LIPS) discriminated those who did (area under the curve 0.80; 95% CI 0.78-0.82).
• Immunosuppression: hematologic malignancy, neutropenia, transplantation, and immunosuppressive therapy predispose to severe hypoxemic respiratory failure.
• Comorbid heart disease: undiagnosed cardiac dysfunction with preserved ejection fraction was present in 44% of ICU survivors of hypercapnic respiratory failure.

Screening

There is no population screening program for acute respiratory failure. Screening focuses on early recognition of deterioration in hospitalized and high-risk patients:

• Early warning scores: the National Early Warning Score 2 (NEWS2) incorporates SpO2 and uses a dedicated SpO2 scoring scale (Scale 2) for patients with confirmed hypercapnic respiratory failure who have a prescribed target range of 88-92%. For acutely ill hospitalized adults, the 2026 Surviving Sepsis Campaign recommends NEWS, NEWS2, MEWS, or SIRS over qSOFA as a single screening tool (strong recommendation, moderate certainty).
• Pulse oximetry and blood gases: pulse oximetry should be interpreted with awareness of its bias in patients with darker skin. The US Food and Drug Administration issued draft guidance in January 2025 recommending that clinical performance testing of pulse oximeters include at least 150 participants, with at least 25% in each Monk Skin Tone group; as of November 2026 it remains a draft.
• Risk prediction for ARDS: the Lung Injury Prediction Score identifies at-risk patients early (area under the curve 0.80); acute lung injury was associated with higher in-hospital mortality (odds ratio 4.1; 95% CI 2.9-5.7).
• Monitoring of patients on noninvasive support (screening for treatment failure):

Tool	Components	Thresholds and performance
ROX index (HFNC)	(SpO2/FiO2)/respiratory rate	ROX ≥ 4.88 at 2, 6, or 12 hours associated with lower risk of intubation (12 hours: hazard ratio 0.291; 95% CI 0.161-0.524); ROX < 2.85, < 3.47, and < 3.85 at 2, 6, and 12 hours predicted HFNC failure
HACOR score (NIV)	Heart rate, acidosis, consciousness, oxygenation, respiratory rate (maximum 25 points)	At 1 hour of NIV, area under the curve 0.88; a score > 5 had sensitivity 72.6% and specificity 90.2% for NIV failure
Expired tidal volume (NIV)	Expired tidal volume per kg predicted body weight	With PaO2/FiO2 ≤ 200 mm Hg, tidal volume > 9.5 mL/kg predicted body weight predicted NIV failure (sensitivity 82%, specificity 87%)

Natural History, Complications and Prognosis

Natural History

Untreated acute respiratory failure progresses to respiratory muscle fatigue, worsening acidosis, altered consciousness, cardiac arrhythmias, and cardiac arrest. With treatment, the course depends mainly on the reversibility of the cause. Patients who fail noninvasive support and are intubated late have worse outcomes:

• Among 175 patients failing HFNC, ICU mortality was 39.2% when intubation occurred within 48 hours and 66.7% when it occurred later (propensity-adjusted odds ratio 0.317).
• Among patients failing NIV with a HACOR score above 5 at 1 hour, hospital mortality was 66% with intubation within 12 hours and 79% with later intubation.
• In LUNG SAFE, NIV was used in 15.5% of patients with ARDS and failed in 22.2% of mild, 42.3% of moderate, and 47.1% of severe cases. Hospital mortality was 16.1% after NIV success and 45.4% after NIV failure; NIV was independently associated with higher ICU mortality (hazard ratio 1.446; 95% CI 1.159-1.805), and propensity-matched ICU mortality was higher than with invasive ventilation when PaO2/FiO2 was below 150 mm Hg.
• In a survey of 70 French ICUs, NIV failure was associated with ICU mortality in de novo acute respiratory failure (odds ratio 3.24; 95% CI 1.61-6.53) but not in cardiogenic pulmonary edema or acute-on-chronic respiratory failure.

Complications

Complication	Evidence
Peri-intubation adverse events	In 2964 critically ill patients from 29 countries, 45.2% had at least one major peri-intubation event: cardiovascular instability 42.6%, severe hypoxemia 9.3%, and cardiac arrest 3.1%; ICU mortality was 32.8%.
Ventilator-induced lung injury and barotrauma	Higher driving pressure is associated with mortality (relative risk 1.41 per 1-SD increase); prolonged recruitment maneuvers increase mortality and barotrauma.
Harm from excess oxygen	In acutely ill adults, liberal oxygen increased in-hospital mortality compared with conservative oxygen (relative risk 1.21; 95% CI 1.03-1.43).
Postextubation respiratory failure	Using NIV to treat established postextubation respiratory failure did not prevent reintubation (48% in both groups) and increased ICU mortality (25% vs 14%; relative risk 1.78), with longer time to reintubation (12 hours vs 2 hours 30 minutes).
Delirium and long-term cognitive impairment	In 821 critically ill adults, delirium occurred in 74%; global cognition was 1.5 SD below population means in 40% at 3 months and 34% at 12 months, and longer delirium was associated with worse cognition.
Physical impairment	ARDS survivors had a median 6-minute walk distance of 281 m at 3 months and 422 m at 12 months, below predicted, and had lost 18% of baseline body weight. At 5 years, the median 6-minute walk distance was 436 m (76% of predicted) and the SF-36 physical component score was 41 (population norm 50), despite near-normal lung function.
Post-intensive care syndrome	New or worsening physical, cognitive, or mental health impairment arising after critical illness and persisting beyond hospitalization, which can also affect family members.
Readmission and death after hypercapnic failure	Among ICU survivors of acute hypercapnic respiratory failure, 46% were readmitted or died over an average of 3.5 months after discharge.

Prognosis

Population	Mortality
US hospitalizations for acute respiratory failure	27.6% (2001) to 20.6% (2009) hospital mortality
US respiratory failure with mechanical ventilation	34% (2002) to 23% (2017) hospital mortality
Invasive mechanical ventilation, US, 2005	34.5% in-hospital mortality
Acute respiratory failure in Finnish ICUs	31% at 90 days
ARDS by Berlin severity (validation cohorts)	27% mild, 32% moderate, 45% severe
ARDS in LUNG SAFE (hospital mortality)	34.9% mild, 40.3% moderate, 46.1% severe
Hypercapnic respiratory failure (hospital cohort)	12.8% in-hospital mortality
Patients with do-not-intubate orders treated with NIV	Survival 56% (95% CI 49%-64%) to hospital discharge and 32% (95% CI 21%-45%) at 1 year; hospital survival 68% in COPD, 68% in pulmonary edema, 41% in pneumonia, and 37% in malignancy
Pediatric ARDS	17% overall; 33% in severe pediatric ARDS
ARDS in a low-resource setting (Kigali criteria)	50.0% hospital mortality

Adverse prognostic factors include greater severity of hypoxemia, high dead space fraction, the hyperinflammatory ARDS subphenotype, infection or neuromuscular disease as the cause of hypercapnia, multiorgan failure, and delayed intubation after failure of noninvasive support.

Diagnosis

Diagnostic Study of Choice |History and Symptoms | Physical Examination | Laboratory Findings | Electrocardiogram | Chest X Ray | CT | MRI | Echocardiography or Ultrasound | Other Imaging Findings | Other Diagnostic Studies

Diagnostic Study of Choice

Arterial blood gas analysis is the reference standard for diagnosing respiratory failure, classifying it as hypoxemic or hypercapnic, and distinguishing acute from acute-on-chronic hypercapnia.

Diagnosis	Criteria
Acute hypoxemic (type 1) respiratory failure	PaO2 < 60 mm Hg or SaO2 < 88%
Acute hypercapnic (type 2) respiratory failure	PaCO2 ≥ 45 mm Hg and pH < 7.35
Respiratory acidosis warranting NIV	pH ≤ 7.35 (ERS/ATS); pH < 7.35 and PaCO2 > 6.5 kPa persisting despite optimal medical therapy (BTS/ICS)
ARDS	PaO2/FiO2 ≤ 300 mm Hg (or SpO2/FiO2 ≤ 315 if SpO2 ≤ 97%) with bilateral opacities on radiograph, CT, or ultrasound

Diagnostic Approach

Suspected acute respiratory failure (dyspnea, tachypnea, accessory muscle use, altered consciousness, low SpO2)

Pulse oximetry and arterial blood gas Intubate immediately for apnea, cardiac arrest, coma, or inability to protect the airway

Hypoxemic (type 1) PaO2 < 60 mm Hg or SaO2 < 88%

Hypercapnic (type 2) PaCO2 ≥ 45 mm Hg and pH < 7.35

Calculate the A-a gradient Chest radiograph and lung ultrasound

Assess central drive, neuromuscular function, chest wall, and airways Review drugs, consciousness, and bicarbonate

Clear lungs or A-line profile: pulmonary embolism, asthma or COPD, intracardiac shunt Absent lung sliding: pneumothorax

Focal opacity or consolidation: pneumonia, atelectasis, aspiration

Bilateral opacities or diffuse B lines: cardiogenic pulmonary edema vs ARDS (echocardiography, natriuretic peptides)

Normal A-a gradient: hypoventilation (drugs, neuromuscular or chest wall disease) Raised A-a gradient: COPD, asthma, or mixed failure Raised bicarbonate: acute-on-chronic failure

History and Symptoms

• Symptoms of hypoxemia: dyspnea, agitation, confusion, and chest pain or palpitations.
• Symptoms of hypercapnia: morning headache, drowsiness, confusion, and reduced level of consciousness.
• History pointing to the cause: fever and cough (pneumonia), orthopnea and cardiac history (heart failure), sudden pleuritic pain or immobility (pulmonary embolism, pneumothorax), known COPD or asthma, progressive weakness or bulbar symptoms (neuromuscular disease), snoring and daytime somnolence (obesity hypoventilation), and drug or alcohol ingestion.
• Baseline function and goals of care: prior home oxygen or NIV, baseline blood gases, and the patient's wishes regarding intubation, which inform the ceiling of treatment.

Physical Examination

• Respiratory distress: tachypnea, use of accessory muscles, intercostal retraction, inability to speak in full sentences, and paradoxical (thoracoabdominal) breathing, which signals diaphragmatic fatigue.
• Hypoxemia: central cyanosis, tachycardia, and hypertension or hypotension.
• Hypercapnia: asterixis, warm peripheries, bounding pulse, and somnolence progressing to coma.
• Clues to the cause: focal crackles or bronchial breathing (pneumonia), bilateral crackles with raised jugular venous pressure (pulmonary edema), wheeze (asthma, COPD), absent breath sounds and tracheal deviation (pneumothorax, effusion), signs of deep vein thrombosis, and neuromuscular weakness.

Laboratory Findings

• Arterial blood gas: PaO2, PaCO2, pH, bicarbonate, lactate, and the A-a gradient.
• Peripheral venous blood gas: arterial pH is typically 0.03 higher than venous pH (95% CI 0.029-0.038), the 95% prediction interval for the venous-arterial PCO2 difference is -10.7 to +2.4 mm Hg, and arterial PO2 is typically 36.9 mm Hg higher (95% CI 27.2-46.6), so venous PO2 cannot replace arterial measurement. In a Cochrane review (6 studies, 805 participants), venous blood gas analysis had a sensitivity of 97.6% (95% credible interval 94.1-99.4) and specificity of 36.9% (17.1-60.1) for respiratory failure, and a sensitivity of 97.1% and specificity of 53.9% for isolated hypercarbia, supporting its use mainly as a rule-out test (low to very low certainty).
• Pulse oximetry and SpO2/FiO2 ratio: SpO2/FiO2 = 64 + 0.84 x (PaO2/FiO2); an SpO2/FiO2 of 235 corresponds to a PaO2/FiO2 of 200 (sensitivity 85%, specificity 85%) and 315 to 300 (sensitivity 91%, specificity 56%), valid only when SpO2 is 97% or less. The 2026 Surviving Sepsis Campaign notes that arterial blood gases remain the gold standard and that SpO2/FiO2 is less accurate in shock, in patients with darker skin tones, and when SpO2 is below 90% or above 97%. Pulse oximetry misses occult hypoxemia more often in Black patients (11.7% vs 3.6% of measurements with SpO2 of 92-96%).
• CO-oximetry: required when carbon monoxide poisoning or methemoglobinemia is suspected.
• Other tests: complete blood count, electrolytes, renal and liver function, natriuretic peptides and troponin (cardiac causes), D-dimer when pulmonary embolism is suspected, blood and respiratory cultures, respiratory pathogen testing, and toxicology screens.

Electrocardiogram

The electrocardiogram identifies arrhythmias, myocardial ischemia or infarction precipitating pulmonary edema, and signs of right ventricular strain in pulmonary embolism.

Chest X Ray

Chest radiography identifies consolidation, pulmonary edema, effusion, pneumothorax, and bilateral opacities consistent with ARDS, and confirms device position after intubation.

CT

Computed tomography characterizes parenchymal disease, CT pulmonary angiography diagnoses pulmonary embolism, and CT can establish the bilateral opacities required for ARDS.

MRI

Magnetic resonance imaging has no routine role in acute respiratory failure; it may be used to evaluate spinal cord or brainstem causes of hypoventilation.

Echocardiography or Ultrasound

• Lung ultrasound (BLUE protocol): predominant A lines with lung sliding indicated asthma or COPD (sensitivity 89%, specificity 97%); diffuse anterior B lines with lung sliding indicated pulmonary edema (97%, 95%); a normal anterior profile plus deep vein thrombosis indicated pulmonary embolism (81%, 99%); absent lung sliding with A lines and a lung point indicated pneumothorax (81%, 100%); and consolidation or asymmetric interstitial patterns indicated pneumonia (89%, 94%). Lung ultrasound is accepted as an imaging modality for ARDS in the 2024 global definition, especially in resource-limited settings.
• Echocardiography: assesses left and right ventricular function, valvular disease, and pericardial effusion, and helps distinguish cardiogenic pulmonary edema from ARDS.

Other Diagnostic Studies

• Neuromuscular disease: vital capacity and other pulmonary function tests, polysomnography, or overnight oximetry help determine when to start NIV.
• Suspected obesity hypoventilation: serum bicarbonate and polysomnography.
• After acute hypercapnic failure: lung function, echocardiography, and polysomnography uncover undiagnosed COPD, cardiac dysfunction, and obstructive sleep apnea in most survivors.

Treatment

Medical Therapy | Surgery | Primary Prevention | Secondary Prevention | Cost-Effectiveness of Therapy | Future or Investigational Therapies

Initial Management

Immediate Assessment

• Assess airway, breathing, and circulation, start continuous pulse oximetry, and obtain an arterial blood gas.
• Intubate immediately in respiratory or cardiac arrest, coma or inability to protect the airway, or refractory shock; noninvasive support should not delay necessary emergent intubation.
• Treat the underlying cause in parallel (for example antimicrobials for pneumonia or sepsis, bronchodilators for airway disease, diuretics and vasodilators for pulmonary edema, anticoagulation or reperfusion for pulmonary embolism, naloxone for opioid toxicity).
• Establish goals of care and the ceiling of treatment early.

Oxygen Therapy and Targets

Oxygen should be prescribed to a target saturation range and titrated, because both hypoxemia and hyperoxemia are harmful. In a meta-analysis of 25 randomized trials (16,037 acutely ill adults), liberal oxygen increased mortality in hospital (relative risk 1.21; 95% CI 1.03-1.43), at 30 days (relative risk 1.14; 95% CI 1.01-1.29), and at longest follow-up (relative risk 1.10; 95% CI 1.00-1.20), suggesting that supplemental oxygen may become unfavorable above an SpO2 of 94-96%. In patients with suspected COPD exacerbation, prehospital titrated oxygen reduced mortality compared with high-flow oxygen (4% vs 9%; relative risk 0.42; 95% CI 0.20-0.89), and to 2% vs 9% in confirmed COPD (relative risk 0.22; 95% CI 0.05-0.91).

Source	Most acutely ill adults	At risk of hypercapnic respiratory failure	Other recommendations
British Thoracic Society (2017)	SpO2 94-98%	SpO2 88-92%	Prescribe oxygen to a target range and monitor
Thoracic Society of Australia and New Zealand (2022)	SpO2 92-96%	SpO2 88-92%, including during NIV for acute respiratory acidosis (Grade A)	HFNC in selected severe hypoxemic respiratory failure (PaO2/FiO2 < 300; Grade B)
BMJ Rapid Recommendation (2018)	Target 90-94% is reasonable; avoid oxygen above SpO2 96% (strong)	SpO2 88-92%	Do not start oxygen in acute myocardial infarction or stroke if SpO2 ≥ 90% (strong for ≥ 93%, weak for 90-92%)
Surviving Sepsis Campaign (2026), sepsis with acute hypoxemic respiratory failure	Titrate FiO2 to a liberal or conservative target according to patient factors and resources (conditional, low certainty); trials used a lower target of about 90-93% and a higher target of 96% or more	Not addressed	Measure oxygenation by SpO2 or arterial blood gas
NEWS2 (Royal College of Physicians)	Standard SpO2 scoring scale	SpO2 Scale 2 with a prescribed target of 88-92% in confirmed hypercapnic respiratory failure	Track-and-trigger escalation

Noninvasive Respiratory Support for Acute Hypoxemic Respiratory Failure

The 2026 ATS guideline recommends HFNC over standard oxygen for adults with de novo acute hypoxemic respiratory failure (strong recommendation, moderate certainty), and suggests NIV or CPAP via facemask or helmet over standard oxygen (conditional recommendations, low certainty), with no recommendation between HFNC, CPAP, and NIV. In its network meta-analysis of 39 trials (7640 patients), compared with standard oxygen, intubation was reduced by HFNC (relative risk 0.76; 95% CI 0.64-0.89), helmet CPAP (0.53; 0.35-0.79), helmet NIV (0.22; 0.11-0.43), and facemask NIV (0.79; 0.67-0.93), while HFNC may reduce mortality (relative risk 0.81; 95% CI 0.67-0.97; low certainty). The ERS (2022) and international (2020) HFNC guidelines and the ESICM (2023) ARDS guideline also favor HFNC over conventional oxygen, and the 2026 Surviving Sepsis Campaign suggests HFNC over conventional oxygen when PaO2/FiO2 is below 200 or SpO2/FiO2 below 235, HFNC rather than NIV as initial therapy, and a trial of awake prone positioning in non-intubated patients without using sedation to promote tolerance (all conditional).

Trial or analysis	Population	Comparison	Main results
FLORALI (2015)	310 patients, non-hypercapnic AHRF	HFNC vs standard oxygen vs facemask NIV	Day-28 intubation 38% vs 47% vs 50% (P = 0.18); 90-day mortality hazard ratio 2.01 (95% CI 1.01-3.99) for standard oxygen and 2.50 (1.31-4.78) for NIV compared with HFNC
SOHO (2026)	1110 ICU patients; PaO2/FiO2 ≤ 200, respiratory rate > 25/min, pulmonary infiltrates	HFNC vs standard oxygen	Day-28 mortality 14.6% vs 14.6% (difference -0.05 percentage points; 95% CI -4.21 to 4.10); intubation 42.4% vs 48.4% (difference -5.93 percentage points; 95% CI -11.78 to -0.08); cardiac arrest or pneumothorax 2.3% vs 1.1%
RENOVATE (2025)	1766 patients in five groups, 33 Brazilian hospitals	HFNC vs facemask NIV	Intubation or death at 7 days 39% vs 38%; HFNC noninferior in non-immunocompromised hypoxemia (32.5% vs 33.1%), COPD with acidosis (28.6% vs 26.2%), cardiogenic pulmonary edema (10.3% vs 21.3%), and COVID-19 (51.3% vs 47.0%); immunocompromised group stopped for futility (57.1% vs 36.4%)
Helmet vs facemask NIV (2016)	83 patients with ARDS, single center, stopped early	Helmet NIV vs facemask NIV	Intubation 18.2% vs 61.5% (difference -43.3%; 95% CI -62.4% to -24.3%); 90-day mortality 34.1% vs 56.4%
Network meta-analysis (2020)	25 trials, 3804 patients	Noninvasive strategies vs standard oxygen	Mortality: helmet NIV relative risk 0.40 (95% credible interval 0.24-0.63), facemask NIV 0.83 (0.68-0.99); intubation: helmet 0.26 (0.14-0.46), facemask 0.76 (0.62-0.90), HFNC 0.76 (0.55-0.99)
Meta-analysis (2019)	9 trials, 2093 patients	HFNC vs conventional oxygen	Mortality relative risk 0.94 (95% CI 0.67-1.31); intubation 0.85 (0.74-0.99); escalation of oxygen therapy 0.71 (0.51-0.98)

Algorithm for Acute Hypoxemic Respiratory Failure

De novo acute hypoxemic respiratory failure

Oxygen titrated to the target SpO2 and treatment of the cause Intubate immediately for arrest, coma, or refractory shock

Persistent hypoxemia or high work of breathing (e.g., PaO2/FiO2 < 200 or SpO2/FiO2 < 235)

First line: high-flow nasal cannula (strong recommendation) Alternatives: NIV or CPAP by facemask or helmet (conditional) Consider a trial of awake prone positioning

Monitor closely where rapid intubation is available ROX index, HACOR score, respiratory rate, expired tidal volume

Improving: continue, then wean FiO2 and flow

Failing: ROX < 2.85 (2 h), < 3.47 (6 h), or < 3.85 (12 h) HACOR > 5 at 1 h Expired tidal volume > 9.5 mL/kg on NIV

Intubate without delay Preoxygenate with HFNC or NIV Lung-protective ventilation

Noninvasive Ventilation for Acute Hypercapnic Respiratory Failure

NIV via facemask is recommended for acute hypercapnic respiratory failure (ATS 2026: strong recommendation, moderate certainty). In 42 trials (4763 patients; median pH 7.31, mean PaCO2 65 ± 12 mm Hg), NIV reduced mortality (relative risk 0.47; 95% CI 0.34-0.65) and intubation (relative risk 0.43; 95% CI 0.32-0.56) compared with standard oxygen. HFNC is suggested as an alternative only in patients with mild acidemia (for example pH > 7.25) when monitoring and prompt escalation to NIV are available (conditional, low certainty); one-third of patients started on HFNC in the trials required escalation to NIV. The ERS/ATS guideline recommends bilevel NIV for COPD exacerbations with acute or acute-on-chronic respiratory acidosis (pH ≤ 7.35; strong recommendation, high certainty), the BTS/ICS guideline recommends starting NIV when pH < 7.35 and pCO2 > 6.5 kPa persist or develop despite optimal medical therapy (Grade A), and GOLD 2026 states that high-flow oxygen systems and NIV are indicated in COPD with acute respiratory failure.

Trial or analysis	Population	Main results
Brochard (1995)	85 patients with COPD exacerbation, 5 ICUs	Intubation 26% vs 74%; in-hospital mortality 9% vs 29%; complications 16% vs 48%
Plant (2000)	236 ward patients with mild-to-moderate acidosis	Treatment failure 15% vs 27%; in-hospital mortality 10% vs 20%
Cochrane review (2017)	17 trials, 1264 participants	Mortality relative risk 0.54 (95% CI 0.38-0.76; number needed to treat 12); intubation 0.36 (0.28-0.46; number needed to treat 5); hospital stay 3.39 days shorter; benefit similar for pH 7.30-7.35 and below 7.30
HFNC vs NIV in COPD (2020)	79 patients, pH 7.25-7.35 and PaCO2 ≥ 55 mm Hg	HFNC noninferior for PaCO2 reduction at 2 hours (-6.8 vs -9.5 mm Hg); 32% of HFNC patients switched to NIV by 6 hours

Algorithm for Acute Hypercapnic Respiratory Failure

Acute or acute-on-chronic hypercapnic respiratory failure (e.g., COPD exacerbation)

Controlled oxygen to SpO2 88-92% Optimal medical therapy Arterial blood gas

pH ≥ 7.35: continue medical therapy and controlled oxygen; repeat blood gas if worsening

Respiratory acidosis: pH ≤ 7.35 (ERS/ATS), or pH < 7.35 with PaCO2 > 6.5 kPa despite optimal therapy (BTS/ICS)

Start NIV via facemask (strong recommendation) HFNC only if acidemia is mild (pH > 7.25) and escalation to NIV is immediately available

Reassess clinical status and blood gases

Improving pH and PaCO2: continue NIV, then wean

Worsening acidosis, intolerance, or falling consciousness: intubation, unless NIV is the agreed ceiling of treatment

Acute Cardiogenic Pulmonary Edema

CPAP or bilevel NIV is recommended for cardiogenic pulmonary edema (ERS/ATS: strong recommendation, moderate certainty; TSANZ: CPAP in severe cases, Grade A). In the 3CPO trial (1069 patients), noninvasive ventilation improved dyspnea, acidosis, and hypercapnia at 1 hour but did not change 7-day mortality (9.5% vs 9.8% with standard oxygen), and CPAP and NIPPV were similar (death or intubation 11.7% vs 11.1%). In a Cochrane review, NIV reduced hospital mortality (relative risk 0.65; 95% CI 0.51-0.82; number needed to treat 17; low quality) and intubation (relative risk 0.49; 95% CI 0.38-0.62; number needed to treat 13; moderate quality). In RENOVATE, intubation or death occurred in 10.3% with HFNC and 21.3% with NIV in this group. The 2026 ATS guideline did not address cardiogenic pulmonary edema as a separate question and noted the distinct role of CPAP in this population.

Monitoring and Escalation

• Manage patients on HFNC or NIV in a setting capable of close monitoring and rapid intubation, with predefined escalation criteria.
• Use the ROX index for HFNC, the HACOR score and expired tidal volume for NIV, and clinical trajectory to detect failure early (see Screening).
• Do not delay intubation when noninvasive support is failing: late intubation after HFNC or NIV failure is associated with higher mortality. Noninvasive strategies appear safe in mild-to-moderate hypoxemia (PaO2/FiO2 > 150 mm Hg) but may delay intubation and increase mortality in moderate-to-severe cases.

Medical Therapy

Treatment of the Underlying Cause

• Sepsis and pneumonia: for probable or definite sepsis, or possible to definite septic shock, the 2026 Surviving Sepsis Campaign recommends antimicrobial therapy immediately, ideally within 1 hour of recognition.
• COPD exacerbation: bronchodilators, systemic corticosteroids, and antibiotics when indicated, with controlled oxygen, according to GOLD 2026.
• Cardiogenic pulmonary edema: management according to the 2026 ESC heart failure guideline.
• Pulmonary embolism: anticoagulation and risk-adapted reperfusion according to the 2026 AHA/ACC multisociety and 2019 ESC guidelines.

Corticosteroids

Indication	Guideline recommendations	Key trial evidence
Severe community-acquired pneumonia	ATS 2026: suggest systemic corticosteroids for inpatients with severe CAP (conditional, low-quality evidence) and recommend against them in non-severe CAP (strong, low-quality evidence). SCCM 2024: recommend corticosteroids for hospitalized adults with severe CAP (strong). ERS/ESICM/ESCMID/ALAT 2023: suggest corticosteroids if shock is present (conditional, low quality).	Hydrocortisone reduced 28-day mortality from 11.9% to 6.2% (difference -5.6 percentage points; 95% CI -9.6 to -1.7) and intubation (18.0% vs 29.5%; hazard ratio 0.59) in CAPE COD. Methylprednisolone did not significantly reduce 60-day mortality (16% vs 18%; adjusted odds ratio 0.90; 95% CI 0.57-1.40). In REMAP-CAP, a fixed 7-day hydrocortisone course was stopped for futility (day-90 mortality 15% vs 9.8%).
ARDS	ATS 2024: suggest corticosteroids (conditional, moderate certainty). SCCM 2024: conditional recommendation; no recommendation on agent, dose, or duration.	Dexamethasone increased ventilator-free days by 4.8 days (95% CI 2.57-7.03) and reduced 60-day mortality (21% vs 36%) in moderate-to-severe ARDS.
COVID-19 requiring oxygen or ventilation	Standard of care	Dexamethasone reduced 28-day mortality (22.9% vs 25.7%; rate ratio 0.83; 95% CI 0.75-0.93), with the largest effect in invasively ventilated patients (29.3% vs 41.4%; rate ratio 0.64; 95% CI 0.51-0.81).
Septic shock	Surviving Sepsis Campaign 2026: suggest IV corticosteroids (conditional, low certainty). SCCM 2024: conditional; recommend against high-dose, short-duration regimens.	See septic shock

Neuromuscular Blockade in ARDS

• In ACURASYS (340 patients with early severe ARDS), 48 hours of cisatracurium reduced the adjusted hazard of 90-day death (hazard ratio 0.68; 95% CI 0.48-0.98); crude 90-day mortality was 31.6% vs 40.7%.
• In ROSE (1006 patients), early continuous cisatracurium with deep sedation did not reduce 90-day in-hospital mortality compared with lighter sedation (42.5% vs 42.8%) and caused more cardiovascular adverse events.
• Guidelines: ATS 2024 suggests neuromuscular blockade in early severe ARDS (conditional, low certainty); ESICM 2023 recommends against routine continuous infusion (strong, moderate); SCCM 2026 suggests neuromuscular blocking agents for ARDS with PaO2/FiO2 below 150 (conditional); and the Surviving Sepsis Campaign 2026 suggests intermittent boluses over continuous infusion in moderate-to-severe sepsis-associated ARDS (conditional, moderate).

Fluid Management

• In FACTT (1000 patients with acute lung injury), a conservative fluid strategy did not change 60-day mortality (25.5% vs 28.4%) but increased ventilator-free days (14.6 ± 0.5 vs 12.1 ± 0.5), with a 7-day fluid balance of -136 ± 491 mL vs 6992 ± 502 mL.
• In septic shock after the acute resuscitation phase, the Surviving Sepsis Campaign 2026 suggests active fluid removal (conditional, very low certainty).

Sedation

• The PADIS guidelines and their 2025 focused update guide analgesia and sedation; the update conditionally suggests dexmedetomidine over propofol for sedation and enhanced mobilization. In the subsequent A2B trial (1404 patients), neither dexmedetomidine (subdistribution hazard ratio 1.09; 95% CI 0.96-1.25) nor clonidine (1.05; 0.95-1.17) shortened time to successful extubation compared with propofol.
• Inhaled sedation with sevoflurane in ARDS reduced ventilator-free days (median difference -2.1 days; 95% CI -3.6 to -0.7) and lowered 90-day survival (47.1% vs 55.7%; hazard ratio 1.31; 95% CI 1.05-1.62) compared with propofol, and is not recommended.
• Daily interruption of sedation reduced median duration of mechanical ventilation (4.9 vs 7.3 days) and ICU stay (6.4 vs 9.9 days), and pairing it with spontaneous breathing trials increased days breathing without assistance (14.7 vs 11.6) and reduced 1-year mortality (hazard ratio 0.68; 95% CI 0.50-0.92).

Therapies Not Recommended

Therapy	Evidence
Inhaled nitric oxide in ARDS	No mortality benefit (38.2% vs 37.5%; relative risk 1.04; 95% CI 0.9-1.19) despite a transient improvement in PaO2/FiO2 at 24 hours, with increased renal failure (relative risk 1.59; 95% CI 1.17-2.16).
Statins in ARDS	Simvastatin did not increase ventilator-free days (12.6 vs 11.5) or reduce 28-day mortality (22.0% vs 26.8%); rosuvastatin was stopped for futility (60-day mortality 28.5% vs 24.9%).
Intravenous salbutamol in ARDS	Stopped for harm: 28-day mortality 34% vs 23% (relative risk 1.47; 95% CI 1.03-2.08).
Intravenous vitamin C	No improvement in organ dysfunction in sepsis-associated ARDS; the Surviving Sepsis Campaign 2026 suggests against IV vitamin C in sepsis.
Inhaled sevoflurane sedation in ARDS	Lower 90-day survival (47.1% vs 55.7%).
Routine early continuous neuromuscular blockade with deep sedation	No mortality benefit and more cardiovascular adverse events.

Procedural / Surgical Therapy

Tracheal Intubation

Emergency intubation of critically ill patients is high risk: 45.2% of patients in an international cohort had at least one major peri-intubation adverse event. Guidance on airway management in critically ill adults is provided by the Difficult Airway Society, Intensive Care Society, Faculty of Intensive Care Medicine, and Royal College of Anaesthetists, the Difficult Airway Society 2025 guideline for unanticipated difficult intubation, the Society of Critical Care Medicine rapid sequence intubation guideline, whose only strong recommendation is to use a neuromuscular blocking agent, and the American Society of Anesthesiologists difficult airway guideline.

Step	Recommendation or evidence
Preoxygenation	ATS 2026: use HFNC or NIV via facemask rather than standard oxygen (strong recommendation, moderate certainty); in its network meta-analysis, peri-intubation hypoxemia was reduced by NIV (relative risk 0.51; 95% CI 0.39-0.65) and HFNC (0.69; 0.54-0.88), and NIV probably reduced hypoxemia compared with HFNC (0.73; 0.55-0.98). In PREOXI (1301 patients), NIV reduced hypoxemia (SpO2 < 85%) from 18.5% to 9.1% (difference -9.4 percentage points; 95% CI -13.2 to -5.6) compared with an oxygen mask. In hypoxemic patients, NIV and high-flow preoxygenation produced similar rates of severe hypoxemia (23% vs 27%), with less hypoxemia on NIV in the PaO2/FiO2 ≤ 200 subgroup (adjusted odds ratio 0.56; 95% CI 0.32-0.99).
Ventilation between induction and laryngoscopy	Bag-mask ventilation reduced severe hypoxemia (10.9% vs 22.8%; relative risk 0.48; 95% CI 0.30-0.77).
Laryngoscope	Video laryngoscopy increased first-attempt success (85.1% vs 70.8%; difference 14.3 percentage points; 95% CI 9.9-18.7) without increasing severe complications (21.4% vs 20.9%).
Bougie	First-attempt success with a bougie was not higher than with a tube and stylet (80.4% vs 83.0%; difference -2.6 percentage points; 95% CI -7.3 to 2.2).
Induction agent	In the RSI trial (2365 patients), day-28 in-hospital mortality did not differ between ketamine and etomidate (28.1% vs 29.1%; difference -0.8 percentage points; 95% CI -4.5 to 2.9), but cardiovascular collapse was more frequent with ketamine (22.1% vs 17.0%; difference 5.1 percentage points; 95% CI 1.9-8.3).
Fluid bolus	A 500-mL fluid bolus did not prevent cardiovascular collapse (21.0% vs 18.2%; difference 2.8%; 95% CI -2.2% to 7.7%).

Invasive Mechanical Ventilation

Intervention	Guideline recommendations	Key evidence
Low tidal volume (6 mL/kg predicted body weight) with plateau pressure ≤ 30 cm H2O in ARDS	Recommended by ATS/ESICM/SCCM 2017 (4-8 mL/kg; strong), ESICM 2023 (4-8 mL/kg; strong), and the Surviving Sepsis Campaign 2026 (6 mL/kg and plateau pressure goal of 30 cm H2O; strong, high certainty)	ARMA (861 patients): mortality 31.0% vs 39.8% with 12 mL/kg (P = 0.007).
Tidal volume without ARDS	Surviving Sepsis Campaign 2026: 6-8 mL/kg ideal body weight rather than 4 to < 6 mL/kg (conditional, low certainty), with regular screening for ARDS	PReVENT (961 patients without ARDS): ventilator-free days 21 in both low and intermediate tidal volume groups; 28-day mortality 34.9% vs 32.1%.
Higher PEEP in moderate-to-severe ARDS	ATS 2024: higher PEEP without lung recruitment maneuvers (conditional, low to moderate certainty); Surviving Sepsis Campaign 2026: higher PEEP (conditional, moderate) and against an incremental PEEP titration strategy (strong, moderate); ESICM 2023: no recommendation	Hospital or 28-day mortality with higher vs lower PEEP did not differ in ALVEOLI (27.5% vs 24.9%), LOVS (36.4% vs 40.4%; relative risk 0.90; 95% CI 0.77-1.05), or ExPress (27.8% vs 31.2%); in an individual patient data meta-analysis, higher PEEP reduced hospital mortality in ARDS (34.1% vs 39.1%; adjusted relative risk 0.90; 95% CI 0.81-1.00). Esophageal pressure-guided PEEP did not improve outcomes.
Lung recruitment maneuvers	ATS 2024 and ESICM 2023: against prolonged or high-pressure recruitment maneuvers (strong, moderate)	ART (1010 patients): 28-day mortality 55.3% with recruitment plus titrated PEEP vs 49.3% (hazard ratio 1.20; 95% CI 1.01-1.42). PHARLAP found no benefit and more new arrhythmias (29% vs 13%).
Driving pressure	Not a guideline target; used as a physiologic signal	Each 1-SD increase (about 7 cm H2O) was associated with higher mortality (relative risk 1.41; 95% CI 1.31-1.51).
Prone positioning	ATS/ESICM/SCCM 2017: more than 12 hours/day in severe ARDS (strong, moderate); ESICM 2023: moderate-to-severe ARDS (strong, high); Surviving Sepsis Campaign 2026: more than 12 hours daily (conditional, moderate)	PROSEVA (466 patients, PaO2/FiO2 < 150 mm Hg): 28-day mortality 16.0% vs 32.8% (hazard ratio 0.39; 95% CI 0.25-0.63); 90-day 23.6% vs 41.0%.
High-frequency oscillatory ventilation	ATS/ESICM/SCCM 2017: recommend against routine use (strong, high)	OSCAR: 30-day mortality 41.7% vs 41.1%; OSCILLATE: stopped early, in-hospital mortality 47% vs 35% (relative risk 1.33; 95% CI 1.09-1.64).
Oxygen targets during invasive ventilation	Surviving Sepsis Campaign 2026: liberal or conservative targets by patient factors (conditional)	ICU-ROX: no difference in ventilator-free days or 180-day mortality (35.7% vs 34.5%); HOT-ICU (PaO2 60 vs 90 mm Hg): 90-day mortality 42.9% vs 42.4%; PILOT (SpO2 90%, 94%, 98%): similar outcomes; UK-ROX (16,500 patients): 90-day mortality 35.4% vs 34.9% (adjusted difference 0.7 percentage points; 95% CI -0.7 to 2.0); in ARDS, LOCO2 was stopped early for safety with 90-day mortality 44.4% vs 30.4% in the conservative group. A meta-analysis of 9 trials (20,447 patients) found no mortality difference (relative risk 1.01; 95% CI 0.94-1.09).

Extracorporeal Support

• Venovenous ECMO: in CESAR, referral to an ECMO center increased survival without severe disability at 6 months (63% vs 47%; relative risk 0.69). In EOLIA (249 patients with very severe ARDS), 60-day mortality was 35% with ECMO and 46% with conventional ventilation (relative risk 0.76; 95% CI 0.55-1.04; P = 0.09), with 28% crossover to ECMO; a Bayesian reanalysis estimated a 96% posterior probability of mortality benefit, and an individual patient data meta-analysis of CESAR and EOLIA found lower 90-day mortality (36% vs 48%; relative risk 0.75; 95% CI 0.6-0.94). ECMO is recommended for EOLIA-defined severe ARDS by ESICM 2023 (strong, moderate), suggested in selected severe ARDS by ATS 2024 (conditional, low), and suggested in experienced centers when conventional ventilation fails by the Surviving Sepsis Campaign 2026 (conditional, low). Patient selection, cannulation, management, and weaning follow the Extracorporeal Life Support Organization guideline. Prone positioning during ECMO did not increase successful weaning by day 60 (44% vs 44%).
• Extracorporeal CO2 removal: in REST (412 patients with acute hypoxemic respiratory failure), ECCO2R to facilitate lower tidal volumes did not reduce 90-day mortality (41.5% vs 39.5%) and reduced ventilator-free days (7.1 vs 9.2).

Tracheostomy

• In TracMan (909 patients), early tracheostomy (within 4 days) did not reduce 30-day mortality compared with late tracheostomy (after 10 days) (30.8% vs 31.5%).

Long-Term Management

Liberation from Mechanical Ventilation

• The ATS/CHEST guidelines conditionally recommend, for patients ventilated more than 24 hours, protocolized rehabilitation directed toward early mobilization, a ventilator liberation protocol, a cuff leak test in patients at high risk of postextubation stridor, and systemic corticosteroids at least 4 hours before extubation if the cuff leak test fails; they also suggest inspiratory pressure augmentation for the initial spontaneous breathing trial (SBT) and sedation-minimizing protocols, and recommend preventive NIV immediately after extubation in high-risk patients (strong).
• SBT technique: a 30-minute pressure support SBT led to more successful extubations than a 2-hour T-piece trial (82.3% vs 74.0%; difference 8.2%; 95% CI 3.4%-13.0%) without more reintubation (11.1% vs 11.9%). In high-risk patients, pressure support and T-piece trials gave similar ventilator-free days and reintubation rates (14.9% vs 13.6%). In a factorial trial of 797 patients, neither screening frequency (hazard ratio 0.88; 95% CI 0.76-1.03) nor SBT technique (1.06; 0.91-1.23) changed time to successful extubation.
• Ventilator modes: proportional assist ventilation plus did not shorten time to successful liberation compared with pressure support (7.3 vs 6.8 days).
• Postextubation respiratory support (ATS 2026): HFNC or NIV via facemask is suggested over standard oxygen after extubation in the ICU (conditional, moderate certainty), and NIV via facemask over HFNC or standard oxygen in patients at high risk of extubation failure, such as those older than 65 years or with obesity, hypercapnia, prolonged mechanical ventilation, or severe illness (conditional, moderate).

Trial	Population	Comparison	Main results
Hernández (2016)	527 low-risk patients	HFNC vs conventional oxygen	Reintubation within 72 hours 4.9% vs 12.2% (difference 7.2%; 95% CI 2.5%-12.2%)
Hernández (2016)	604 high-risk patients	HFNC vs NIV	Reintubation 22.8% vs 19.1% (HFNC noninferior); postextubation respiratory failure 26.9% vs 39.8%
HIGH-WEAN (2019)	641 high-risk patients	HFNC plus NIV vs HFNC alone	Day-7 reintubation 11.8% vs 18.2% (difference -6.4%; 95% CI -12.0% to -0.9%)
Ferrer (2009)	106 hypercapnic patients with chronic respiratory disorders	NIV for 24 hours vs conventional oxygen	Postextubation respiratory failure 15% vs 48%; lower 90-day mortality with NIV (p = 0.0146)
Esteban (2004)	221 patients with established postextubation respiratory failure	NIV vs standard therapy	No reduction in reintubation (48% in both); higher ICU mortality with NIV (25% vs 14%; relative risk 1.78; 95% CI 1.03-3.20)

Algorithm for Postextubation Support

Planned extubation after a successful spontaneous breathing trial

Assess the risk of extubation failure

Low risk HFNC rather than conventional oxygen

High risk (age > 65 years, obesity, hypercapnia, prolonged ventilation, severe illness) Prophylactic NIV, which may alternate with HFNC

Hypercapnic chronic respiratory disease NIV for 24 hours after extubation

If respiratory failure develops after extubation despite these measures, reintubate promptly rather than using NIV to delay reintubation.

Rehabilitation

• Early physical and occupational therapy during mechanical ventilation increased functional independence at hospital discharge (59% vs 35%; odds ratio 2.7; 95% CI 1.2-6.1) and shortened delirium (2.0 vs 4.0 days).
• In TEAM (750 patients), an increased dose of early active mobilization did not increase days alive and out of hospital at 180 days (143 vs 145) and increased potentially mobilization-related adverse events (9.2% vs 4.1%).
• The PADIS 2025 update suggests enhanced over usual mobilization and rehabilitation (conditional).
• After hospital discharge, the Surviving Sepsis Campaign 2026 suggests physical rehabilitation for survivors who received invasive ventilation for more than 48 hours, follow-up services after critical illness, and mental health support (conditional, low certainty). A remote multicomponent rehabilitation program after critical illness produced only a borderline difference in 8-week quality of life (EQ-5D-5L 0.69 vs 0.67; adjusted difference 0.04; 95% CI -0.001 to 0.09).

Follow-Up of Survivors

• Screen for and manage post-intensive care syndrome: physical weakness, cognitive impairment, anxiety, depression, and post-traumatic stress, in patients and families.
• After acute hypercapnic respiratory failure, evaluate lung, heart, and sleep function: undiagnosed COPD, cardiac dysfunction with preserved ejection fraction, and severe obstructive sleep apnea are common, and 46% of survivors are readmitted or die over an average of 3.5 months after discharge.

Home Ventilation and Oxygen after Hypercapnic Respiratory Failure

Situation	Recommendation or evidence
Persistent hypercapnia after a COPD exacerbation	ATS 2020: do not start long-term NIV during the acute-on-chronic admission; reassess for persistent hypercapnia at 2-4 weeks (conditional, low certainty); nocturnal NIV in chronic stable hypercapnic COPD (conditional, moderate), with screening for obstructive sleep apnea first. ERS 2019: long-term home NIV after an exacerbation requiring acute NIV, with settings targeting CO2 reduction (conditional). In HOT-HMV (116 patients with PaCO2 > 53 mm Hg), adding home NIV to oxygen prolonged time to readmission or death (4.3 vs 1.4 months; adjusted hazard ratio 0.49; 95% CI 0.31-0.77).
Stable severe hypercapnic COPD (PaCO2 ≥ 7 kPa)	Long-term NIV targeting a PaCO2 reduction of at least 20% reduced 1-year mortality (12% vs 33%; hazard ratio 0.24; 95% CI 0.11-0.49).
Obesity hypoventilation syndrome	Discharge hospitalized patients with suspected OHS on NIV until titration, ideally within 2-3 months (conditional, very low); CPAP and NIV had similar long-term hospitalization rates in OHS with severe obstructive sleep apnea (1.63 vs 1.44 days per patient-year).
Neuromuscular disease	Pulmonary function tests every 6 months, individualized NIV settings, and polysomnography or overnight oximetry to guide NIV initiation.
Chronic resting hypoxemia	Long-term oxygen for severe chronic resting hypoxemia in COPD (strong, moderate) or interstitial lung disease (strong, low); not for moderate resting hypoxemia in COPD (conditional).

Special Populations

Immunocompromised Patients

Early NIV improved outcomes in a small trial from 2001 (52 patients; intubation 12 vs 20 and ICU death 10 vs 18), but larger modern trials have not confirmed a benefit of any noninvasive strategy on mortality:

• Early NIV vs oxygen alone: day-28 mortality 24.1% vs 27.3% (difference -3.2; 95% CI -12.1 to 5.6).
• HFNC vs standard oxygen (HIGH, 776 patients): day-28 mortality 35.6% vs 36.1% (hazard ratio 0.98; 95% CI 0.77-1.24); intubation 38.7% vs 43.8%.
• HFNC alone vs NIV alternating with HFNC (FLORALI-IM, 299 patients): day-28 mortality 36% vs 35% (difference 1.2%; 95% CI -9.6 to 11.9).
• In RENOVATE, the immunocompromised group was stopped for futility (intubation or death 57.1% with HFNC vs 36.4% with NIV).

The ATS 2026 guideline found no credible subgroup effect of immune status in acute hypoxemic respiratory failure, so the same approach (HFNC as first-line, with close monitoring and timely intubation) applies, while the earlier ERS/ATS suggestion of early NIV in immunocompromised patients rests on older evidence.

Postoperative Patients

• After cardiothoracic surgery, HFNC was noninferior to bilevel NIV for treatment failure (21.0% vs 21.9%) and caused less skin breakdown (3% vs 10%). HFNC is suggested for high-risk and obese patients after cardiac or thoracic surgery (conditional, moderate certainty).
• After abdominal surgery, NIV reduced reintubation within 7 days compared with standard oxygen (33.1% vs 45.5%; difference -12.4%; 95% CI -23.5% to -1.3%) and health care-associated infections (31.4% vs 49.2%).

Obesity and Obesity Hypoventilation Syndrome

• Obesity is a criterion for high risk of extubation failure, for which postextubation NIV is suggested. In 144 obese patients, humidified NIV and HFNC after extubation gave reintubation rates of 23.6% and 33.3% (difference 9.7; 95% CI -4.9 to 24.4).
• In acute hypercapnic respiratory failure due to obesity hypoventilation, NIV is started using the same criteria as in COPD (Grade B).

Neuromuscular and Chest Wall Disease

• In hypercapnic patients with neuromuscular or chest wall disease, NIV should almost always be trialled (Grade D). The CHEST guideline addresses monitoring with pulmonary function tests and individualized NIV settings. Because hypercapnia in these conditions is not rapidly reversible, HFNC is less likely to be effective.

Patients with Do-Not-Intubate Orders

• In patients with do-not-intubate orders treated with NIV, survival was 56% to hospital discharge and 32% at 1 year, and was highest in COPD (68%) and pulmonary edema (68%). NIV can therefore be offered as a ceiling of treatment when consistent with the patient's goals.

COVID-19

Trial	Comparison	Main results
RECOVERY-RS	CPAP or HFNC vs conventional oxygen	Intubation or death within 30 days: CPAP 36.3% vs 44.4% (difference -8%; 95% CI -15% to -1%); HFNC 44.3% vs 45.1%
HENIVOT	Helmet NIV vs HFNC	No difference in respiratory support-free days (20 vs 18); intubation 30% vs 51% (difference -21%; 95% CI -38% to -3%)
HiFlo-COVID	HFNC vs conventional oxygen	Intubation 34.3% vs 51.0% (hazard ratio 0.62; 95% CI 0.39-0.96); faster clinical recovery
HELMET-COVID	Helmet NIV vs usual respiratory support	28-day mortality 27.0% vs 26.1% (relative risk 1.04; 95% CI 0.72-1.49)
Awake prone positioning meta-trial	Awake prone positioning vs standard care in patients on HFNC	Treatment failure 40% vs 46% (relative risk 0.86; 95% CI 0.75-0.98); intubation hazard ratio 0.75 (0.62-0.91)
Awake prone positioning meta-analysis	10 trials, 1985 patients	Reduced intubation (relative risk 0.84; 95% CI 0.72-0.97), mainly in patients on advanced respiratory support
RECOVERY	Dexamethasone vs usual care	28-day mortality 22.9% vs 25.7%; 29.3% vs 41.4% in invasively ventilated patients

Pregnancy

• Respiratory failure complicates up to 1 in 500 pregnancies, and precise maternal oxygenation targets are unknown. The American College of Obstetricians and Gynecologists does not specify a numeric SpO2 or PaO2 target for ARDS; its early warning trigger is a room-air SpO2 below 95%. Most professional societies, including the Society for Maternal-Fetal Medicine, recommend maintaining SpO2 of 95% or more, although some experts propose 92-96% in admitted patients with acute respiratory failure.

Children

Topic	Evidence or recommendation
Pediatric ARDS	Occurred in 3.2% of pediatric ICU admissions with 17% mortality, 33% in severe pediatric ARDS; only 32% met Berlin criteria. PALICC-2 suggests tidal volumes of 6-8 mL/kg and plateau pressure of 28 cm H2O or less (conditional, very low certainty).
Bronchiolitis in infants	High-flow therapy reduced escalation of care (12% vs 23%; difference -11 percentage points; 95% CI -15 to -7).
First-line support in the pediatric ICU	HFNC was noninferior to CPAP for time to liberation (52.9 vs 47.9 hours).
Postextubation support in the pediatric ICU	HFNC was not noninferior to CPAP (50.5 vs 42.9 hours), and 180-day mortality was 5.6% vs 2.4% (adjusted odds ratio 3.07; 95% CI 1.1-8.8).
Ventilator liberation	International pediatric ventilator liberation guideline (PALISI).
Resource-limited settings	In African children with severe pneumonia and SpO2 < 80%, 48-hour mortality was 9.3% with high-flow therapy vs 13.4% with low-flow oxygen; the trial stopped early at 1852 of 4200 planned patients.

Neonates

• Management of neonatal respiratory distress syndrome follows the European Consensus Guidelines, 2025 update.

Resource-Limited Settings

• The 2024 global ARDS definition does not require PEEP, a specific oxygen flow, or a specific support device, and accepts SpO2/FiO2 and lung ultrasound. The ATS 2026 guideline notes that most evidence comes from resource-rich settings and that HFNC requires adequate oxygen supply, uninterrupted power, and disposable circuits.

Case Studies

Case #1

de:Ateminsuffizienz no:Respirasjonssvikt fr:Insuffisance respiratoire