Non-invasive ventilation (NIV) and high-flow nasal oxygen (HFNO) are the two pillars of non-invasive respiratory support in adult critical care. Together they allow the clinician to rest fatigued respiratory muscles, correct gas-exchange abnormalities, and stabilise oxygenation without the sedation, airway instrumentation, ventilator-associated pneumonia, and diaphragmatic disuse that accompany endotracheal intubation. For acute hypercapnic respiratory failure in chronic obstructive pulmonary disease (COPD) and for acute cardiogenic pulmonary oedema, NIV is not an alternative to invasive ventilation but the treatment of first choice, with randomised evidence of reduced intubation and reduced mortality. HFNO, which delivers heated and fully humidified gas at flows that exceed the patient's peak inspiratory demand, has become the default oxygen interface for de-novo hypoxaemic failure because it is better tolerated and at least as effective as conventional oxygen therapy.

The two modalities are nevertheless neither interchangeable nor benign. NIV works only when the patient can tolerate a sealed interface, protect the airway, clear secretions, and show measurable physiological improvement within the first one to two hours. HFNO is more comfortable and virtually free of interface-related harm, but it applies far less pressure and no true inspiratory assistance, so it can only marginally augment alveolar ventilation. Used injudiciously, both can lull the team into watching a deteriorating patient instead of intubating one. The single most common lethal error in non-invasive respiratory support is not choosing the wrong device at the outset; it is persisting with a failing device after the signals of failure have become obvious.

This chapter examines the interfaces, ventilators, and modes of NIV with their typical initial settings; the conditions for which the evidence of benefit is strong; the contentious use of non-invasive support in de-novo hypoxaemic respiratory failure; the physiology and clinical evidence of HFNO; the monitoring tools that predict failure, above all the ROX index; and the complications and pitfalls that determine whether non-invasive support saves a patient from intubation or merely delays it.

Figure 1: High-flow nasal oxygen exerts its effects through several simultaneous mechanisms, including dead-space washout, a flow-dependent positive airway pressure, stable oxygen delivery, and heat and humidification, which together reduce work of breathing and improve oxygenation. (Source: Lodeserto FJ, Lettich TM, Rezaie SR. Cureus 2018. DOI: 10.7759/cureus.3639).

Interfaces, Ventilators, and Modes

Choosing the Interface

The interface determines tolerance, and tolerance determines success. The oronasal mask is the standard for acute hypercapnic respiratory failure: it is the most extensively studied, permits higher inspiratory pressures, and reduces oral leak. Its liabilities are claustrophobia, and pressure injury over the nasal bridge and cheeks, so a forehead spacer or a thin prophylactic dressing should be used from the outset. The nasal mask is more comfortable, leaves the mouth free for speech and expectoration, and is well suited to chronic nocturnal ventilation, but in the acutely distressed patient mouth breathing and leak undermine both synchrony and effective ventilation. The total-face mask encloses the eyes and mouth and spares the nasal bridge; it is useful when oronasal masks have already caused skin breakdown, though its larger dead space and lower efficiency at high pressures limit it to patients with modest ventilatory demands. The helmet, a transparent head enclosure with a soft collar seal, is the best-tolerated interface for prolonged continuous positive-pressure support. It permits very high pressure without facial skin injury and was used successfully in hypoxaemic failure in the helmet-NIV trial during the COVID-19 pandemic. Its drawbacks are noise, greater dead space, less effective triggering, and a risk of carbon dioxide rebreathing, which make it a specialised rather than a first-line interface.

Ventilators and Modes

Two device classes deliver NIV. Dedicated bilevel devices are flow generators with a single-limb circuit, an intentional leak, and active leak compensation; they cycle between an inspiratory positive airway pressure (IPAP) and an expiratory positive airway pressure (EPAP) and are simple, portable, and ideal for ward and emergency-department initiation. Intensive-care ventilators with a dedicated NIV mode also compensate for leak and use a single limb, but additionally measure tidal volume, leak, and pressure, allow fine control of inspiratory time and cycle-off criteria, and permit a higher fraction of inspired oxygen. For patients who are severely ill or at high risk of failing NIV, the ICU ventilator is preferred because it makes failure legible.

The mode is almost always pressure support with positive end-expiratory pressure (PSV-PEEP). Bilevel positive airway pressure (BPAP, or BiPAP) is the same mode expressed differently: IPAP is the sum of the pressure support and the EPAP, so a support of $10\ \text{cmH}_2\text{O}$ above an EPAP of $5\ \text{cmH}_2\text{O}$ is delivered as IPAP $15\ \text{cmH}_2\text{O}$. Adding a backup rate converts spontaneous pressure support into the spontaneous-timed (S/T) mode, which guarantees a minimum minute ventilation and is essential in patients with central hypoventilation or a failing respiratory drive. Pressure-controlled ventilation is reserved for unusual circumstances in which a guaranteed tidal volume is required at the cost of patient comfort.

Initial Settings

Parameter	Typical initial setting	Titration target
IPAP	10–14 $\text{cmH}_2\text{O}$	Increment by 2 $\text{cmH}_2\text{O}$ to comfort, tidal volume 6–8 $\text{mL/kg}$ predicted body weight, and a fall in respiratory rate; usual maximum 20–25 $\text{cmH}_2\text{O}$
EPAP	4–6 $\text{cmH}_2\text{O}$	Raise for hypoxaemia or cardiogenic pulmonary oedema up to 8–12 $\text{cmH}_2\text{O}$; use the lowest value that maintains oxygenation and counterbalances intrinsic PEEP
Backup rate	10–14 breaths/min	Set just below the patient's spontaneous rate to avoid stacking; ensure an adequate minimum minute ventilation
FiO₂	Titrate from 1.0	Target $\text{SpO}_2$ 92–96%, or 88–92% in chronic hypercapnic COPD
Interface and humidification	Oronasal mask; heated humidifier or heat–moisture exchanger	Ensure a leak of less than about 40 $\text{L/min}$ and a comfortable seal

Established Indications with Strong Evidence

Acute Hypercapnic Exacerbation of COPD

This is the archetypal indication. In a patient with an acute exacerbation and persistent hypercapnic acidosis (typically pH 7.25–7.35 with $P_{a\text{CO}_2} > 45\ \text{mmHg}$) despite optimal bronchodilators, corticosteroids, and controlled oxygen, bilevel NIV reduces intubation and mortality. The 2017 Cochrane review of 17 randomised trials and 1264 patients found a mortality risk ratio of 0.54 and an intubation risk ratio of 0.36, with a number needed to treat of 12 and 5 respectively, and a hospital stay shortened by about 3.4 days. Benefit is present across the range of acidosis, including pH below 7.30, and in both intensive-care and ward settings. The physiological rationale — resting the respiratory muscles, offsetting intrinsic PEEP, and improving alveolar ventilation — is set out in the chapter on dynamic hyperinflation.

Acute Cardiogenic Pulmonary Oedema

Positive-pressure support recruits flooded alveoli, reduces left-ventricular afterload and preload, and opposes the negative intrathoracic pressure swings of respiratory distress. Continuous positive airway pressure (CPAP) at 5–10 $\text{cmH}_2\text{O}$ is the simplest and best-validated therapy; the 3CPO trial showed rapid improvement in dyspnoea, heart rate, and acidosis, though no reduction in mortality. The 2019 Cochrane review confirms that either CPAP or bilevel NIV reduces intubation compared with standard oxygen, with no important difference between the two, so bilevel NIV is reserved for patients who remain hypercapnic or distressed on CPAP. Non-invasive support is an adjunct to, never a substitute for, nitrates, diuretics, and treatment of the precipitant.

Immunocompromised Patients with Hypoxaemic Respiratory Failure

The evidence here is genuinely mixed and depends on case selection. An early randomised trial in neutropenic patients with pulmonary infiltrates found that early NIV reduced intubation and intensive-care mortality, and the 2017 European Respiratory Society and American Thoracic Society (ERS/ATS) guideline therefore issued a conditional recommendation for NIV in immunocompromised patients. A larger subsequent randomised trial, however, found no mortality benefit from early NIV compared with oxygen alone in immunocompromised patients with acute respiratory failure, and reported more intensive-care-acquired complications, while HFNO is now widely preferred as the first step. The practical rule is to use non-invasive support only in patients who are haemodynamically stable, have no more than one failing organ, and improve within one hour; otherwise, proceed early to invasive ventilation.

Prophylaxis and Treatment of Post-Extubation Failure

Extubation failure carries a mortality of 25–50%, and the patients at highest risk — age above 65 years, cardiac or respiratory comorbidity, chronic hypercapnia, weak cough, or a failed spontaneous breathing trial before extubation — benefit from prophylactic support. A landmark trial in high-risk patients showed that NIV started immediately after extubation reduced respiratory failure and intensive-care mortality, and this remains the strongest use of NIV in the peri-extubation period. For patients at low risk of failure, HFNO is the preferred interface: a randomised trial found reintubation was halved with HFNO compared with conventional oxygen. In high-risk patients, and in particular those with hypercapnia, the combination of NIV with HFNO used during breaks from NIV reduced reintubation further than HFNO alone. The choice of interface after extubation should therefore be driven by the pre-extubation risk profile and the arterial carbon dioxide tension.

Chest-Wall and Neuromuscular Restrictive Disease

Patients with kyphoscoliosis, thoracoplasty, obesity hypoventilation, or neuromuscular disease (amyotrophic lateral sclerosis, Duchenne muscular dystrophy, post-polio syndrome) develop chronic hypercapnia from weak inspiratory muscles and a stiff chest wall. Nocturnal bilevel NIV rests the diaphragm, improves alveolar ventilation overnight, resets the carbon dioxide set-point, and improves symptoms and quality of life; in amyotrophic lateral sclerosis it prolongs survival. The evidence for benefit is strongest in chronic stable hypercapnia, and the acute use of NIV in these patients is a bridge to, or initiation of, domiciliary therapy.

Condition	Interface and mode	Strength of evidence	Principal evidence
Acute hypercapnic COPD exacerbation	Oronasal mask, bilevel/PSV-PEEP, IPAP 10–14 $\to$ 20, EPAP 4–6 $\text{cmH}_2\text{O}$	Strong — reduces intubation and mortality	Cochrane 2017; ERS/ATS 2017
Acute cardiogenic pulmonary oedema	CPAP 5–10 $\text{cmH}_2\text{O}$, or bilevel if hypercapnic	Strong — reduces intubation; mortality unchanged	3CPO 2008; Cochrane 2019
Immunocompromised hypoxaemic failure	Oronasal or helmet NIV; HFNO a common alternative	Conditional / mixed — benefit confined to selected patients	Hilbert 2001; Lemiale 2015; ERS/ATS 2017
Post-extubation in high-risk patients	NIV with HFNO during breaks; HFNO alone if low risk	Strong for prophylaxis in high-risk patients	Nava 2005; Hernández 2016; Thille 2019
Chest-wall and neuromuscular restrictive disease	Nocturnal bilevel NIV, IPAP titrated to normocapnia	Strong for chronic hypercapnia; variable in acute failure	ERS/ATS 2017
De-novo hypoxaemic failure (influenza, COVID-19)	CPAP or HFNO; avoid mask NIV as first line	Contentious — high failure rate; risk of delaying intubation	FLORALI 2015; RECOVERY-RS 2022

De-Novo Hypoxaemic Respiratory Failure: The Contentious Indication

When a patient without chronic lung disease develops acute hypoxaemic respiratory failure with a ratio of arterial oxygen tension to inspired oxygen fraction ($P_{a\text{O}_2}/\text{FiO}_2$) below 300 $\text{mmHg}$, the temptation to apply mask NIV is strong and the evidence is weak. In the FLORALI trial, patients randomised to NIV delivered through a face mask had the highest intubation rate of the three arms, at 50%, compared with 38% for HFNO and 47% for conventional oxygen. Failure rates in pandemic influenza and coronavirus disease 2019 (COVID-19) pneumonia have been similarly high, often exceeding 40–50% when mask NIV is used as the primary support.

The danger is not the failure itself but the delay. A large observational study of HFNO failure found that patients in whom intubation was delayed died more often than those intubated promptly, and the same pattern has been described repeatedly during COVID-19. Two mechanisms explain the harm: the sustained high tidal volumes and injurious transpulmonary pressures generated by an unsupported, vigorous inspiratory effort during pressure support (patient self-inflicted lung injury), and the failure to recognise that the patient is tiring before cardiovascular collapse or cardiac arrest occurs.

Failure is predictable. The features that should trigger early intubation rather than escalation of non-invasive support are a respiratory rate above 35 breaths/min after one hour of therapy, a $P_{a\text{O}_2}/\text{FiO}_2$ ratio below 150–200 $\text{mmHg}$, an FiO₂ requirement above 0.6 to maintain saturation, the need for vasopressors, an inability to cough or clear secretions, and above all failure to improve — or worsening — within one to two hours. A HACOR score (heart rate, acidosis, consciousness, oxygenation, respiratory rate) above 5 after one to two hours of NIV predicts failure with acceptable accuracy and should be calculated routinely.

High-Flow Nasal Oxygen: Mechanisms and Settings

Meeting Inspiratory Demand

An adult in respiratory distress generates peak inspiratory flows of 30–60 $\text{L/min}$ or more. A conventional nasal cannula delivers a fixed oxygen flow that the patient entrains with room air, so the actual inspired oxygen fraction falls as inspiratory flow rises and is impossible to know. HFNO delivers a blended, heated, humidified gas mixture at flows of up to 60–70 $\text{L/min}$ through wide-bore nasal prongs, matching the inspiratory demand and thereby preventing entrainment of room air. This produces a stable and predictable inspired oxygen fraction and reduces the work of breathing associated with inspiring chilled, dry gas.

Nasopharyngeal Dead-Space Washout

The nasopharynx and upper airway constitute a volume of about 50 $\text{mL}$ that is rebreathed with every breath. A high continuous flow of fresh gas flushes carbon dioxide from this anatomical dead space during expiration, so that the next inspiration draws on an oxygen-rich, carbon-dioxide-poor reservoir. The result is an increase in effective alveolar ventilation and a fall in arterial carbon dioxide tension, while anatomical dead-space washout also explains part of the improvement in oxygenation.

A Modest, Flow-Dependent PEEP

The gas flow delivered into the nose generates a small positive pressure in the nasopharynx whose magnitude is proportional to the flow. The relationship is approximately

$$P_{\text{nasopharyngeal}} \approx k \times \dot{V}$$

where $P_{\text{nasopharyngeal}}$ is the mean nasopharyngeal pressure ($\text{cmH}_2\text{O}$), $\dot{V}$ is the delivered gas flow ($\text{L/min}$), and $k$ is an empirical constant of roughly 0.05–0.10 $\text{cmH}_2\text{O}\cdot\text{min}\cdot\text{L}^{-1}$ — equivalently, about 0.5–1 $\text{cmH}_2\text{O}$ for every 10 $\text{L/min}$ when the mouth is closed. At 60 $\text{L/min}$ this can reach about 5–7 $\text{cmH}_2\text{O}$, but mouth opening collapses the effect to almost nothing. HFNO is therefore a source of modest, unstable CPAP, not a reliable method of applying a prescribed PEEP.

Heat and Humidification

The heater–humidifier delivers gas at 34–37 °C with close to 100% relative humidity, which preserves mucociliary function, prevents mucosal drying and epistaxis, thins secretions, and is a major reason HFNO is so well tolerated over hours to days. Humidification also reduces the metabolic cost of conditioning inspired gas.

Typical Settings

Flow is started at 30–40 $\text{L/min}$ for comfort and raised in 5–10 $\text{L/min}$ steps to 50–60 $\text{L/min}$ in de-novo hypoxaemic failure; a lower flow (about 30–40 $\text{L/min}$) is used when the aim is to target hypercapnia or when the patient finds high flow uncomfortable. The oxygen fraction is initially set to maintain saturation and then weaned, targeting $\text{SpO}_2$ 92–96%, or 88–92% in patients at risk of hypercapnic respiratory failure. Temperature is set to 34 °C for comfort and 37 °C when secretion retention is a problem. HFNO provides no inspiratory pressure support and a lower and less predictable carbon dioxide clearance than NIV, so it should not be used as the primary support for established hypercapnic acidosis.

The Clinical Evidence for High-Flow Nasal Oxygen

De-Novo Hypoxaemic Failure: FLORALI

The FLORALI trial randomised 310 immunocompetent patients with de-novo acute hypoxaemic respiratory failure to HFNO, conventional oxygen, or mask NIV. The primary endpoint, intubation within 28 days, did not differ significantly between the groups (38%, 47%, and 50% respectively). However, the 90-day mortality was lowest in the HFNO arm (12% versus 23% with conventional oxygen and 28% with NIV), and in the subgroup with a $P_{a\text{O}_2}/\text{FiO}_2$ ratio of 200 $\text{mmHg}$ or less, intubation was substantially less frequent with HFNO than with either comparator. FLORALI established HFNO as the preferred first-line non-invasive support in de-novo hypoxaemic failure and raised doubt about mask NIV in this population.

In COVID-19, the RECOVERY-RS trial of 1273 patients found that CPAP reduced the composite of intubation or mortality compared with conventional oxygen, whereas HFNO did not achieve statistical significance for that composite. The lesson is that in severe hypoxaemic failure with a stiff, poorly recruited lung, a sealed interface that can deliver reliable positive airway pressure (CPAP or helmet NIV) may be more effective than HFNO, and that the choice should be based on the severity of the oxygenation defect rather than on comfort alone.

Post-Extubation and Pre-Oxygenation

HFNO reduces reintubation in unselected patients after extubation, particularly those at low risk of failure, and improves comfort during breaks from NIV. It also has an established role in pre-oxygenation before emergency intubation. A large randomised trial found that pre-oxygenation with a non-invasive ventilation strategy reduced severe hypoxaemia during intubation compared with a bag-valve-mask or standard oxygen, and network meta-analysis supports non-invasive techniques over conventional oxygen for this purpose.

High-Flow Nasal Oxygen versus Conventional Oxygen and Non-Invasive Ventilation

Relative to conventional oxygen therapy, HFNO consistently reduces intubation, escalates support less often, and improves dyspnoea and comfort in hypoxaemic failure; the effect on mortality is small and is best demonstrated in the most hypoxaemic patients. Relative to mask NIV, HFNO is better tolerated and at least as effective in de-novo hypoxaemic failure, but inferior for hypercapnic acidosis, where NIV remains the treatment of choice. The two are best seen as complementary rather than competing: HFNO for hypoxaemia with preserved ventilatory drive, NIV for hypercapnia and for ventilatory pump failure.

Monitoring, Failure Prediction, and Escalation

The ROX Index

The respiratory rate–oxygenation (ROX) index combines oxygenation, oxygen requirement, and respiratory effort into a single dimensionless number that can be calculated at the bedside without a blood gas:

$$\text{ROX} = \frac{\text{SpO}_2 / \text{FiO}_2}{\text{RR}}$$

where $\text{SpO}_2$ is the peripheral oxygen saturation entered as a percentage (for example 94), $\text{FiO}_2$ is the inspired oxygen fraction as a decimal (for example 0.60), and $\text{RR}$ is the respiratory rate in breaths per minute. The oxygen saturation ratio is analogous to the $P_{a\text{O}_2}/\text{FiO}_2$ ratio, and division by the respiratory rate penalises the tachypnoea that signals impending failure, so the index carries units of $\text{min}\cdot\text{breaths}^{-1}$. A worked example makes the arithmetic transparent: a patient with $\text{SpO}_2$ 94%, $\text{FiO}_2$ 0.60, and a respiratory rate of 28 breaths/min has $\text{ROX} = (94 / 0.60) / 28 \approx 5.6$, which lies above the success threshold.

HACOR and Other Predictors

The HACOR score — heart rate, acidosis, consciousness, oxygenation, and respiratory rate — was derived to predict NIV failure in hypoxaemic patients and performs well when calculated after one to two hours of therapy; a score above 5 is associated with a failure rate of around 80%. Its components are a reminder that failure is a multi-system event: tachypnoea, worsening acidosis, falling conscious level, persistent hypoxaemia, and tachycardia commonly evolve together. Serial clinical examination, arterial blood gases, and the trajectory of the respiratory rate remain at least as valuable as any score.

Criteria for Invasive Ventilation

Proceed to invasive ventilation when non-invasive support fails to improve or worsens pH, carbon dioxide tension, respiratory rate, or conscious level within one to two hours; when the respiratory rate exceeds 35 breaths/min or the oxygen requirement climbs despite maximal flow; when haemodynamic instability requires vasopressors; when the patient cannot protect the airway, clear secretions, or tolerate the interface; or when respiratory arrest is imminent. An absolute contraindication to non-invasive support at the outset is itself an indication to intubate rather than to persist.

flowchart TD
    A["Adult with acute respiratory failure"] --> B{"Hypercapnic cause: pH < 7.35 with PaCO2 > 45 mmHg?"}
    B -->|"Yes"| C["Start bilevel NIV: IPAP 10-14, EPAP 4-6 cmH2O, backup rate"]
    B -->|"No"| D{"Hypoxaemic with P/F < 300 mmHg?"}
    D -->|"Yes"| E["Start high-flow nasal oxygen 50-60 L/min, FiO2 titrated"]
    D -->|"No"| F["Conventional oxygen and treat the precipitant"]
    C --> G{"Improved at 1-2 h: pH, PaCO2, RR, GCS?"}
    E --> H{"ROX >= 4.88 at 2-6 h?"}
    G -->|"Yes"| I["Continue, reassess every 2-4 h, plan weaning"]
    G -->|"No"| J["Optimise interface and settings; consider HFNO during breaks"]
    H -->|"Yes"| I
    H -->|"No"| K["Consider sealed-interface CPAP or NIV; anticipate failure"]
    J --> L{"Deterioration, exhaustion, or airway compromise?"}
    K --> L
    L -->|"Yes"| M["Invasive mechanical ventilation without delay"]
    L -->|"No"| I

Contraindications, Complications, and Practical Pitfalls

Contraindications

Absolute contraindications to NIV are cardiac or respiratory arrest; inability to protect the airway or clear secretions; undrained pneumothorax; and facial trauma, burns, or deformity that precludes a seal. NIV is also unsafe with vomiting or bowel obstruction and a high aspiration risk, with upper-airway obstruction, and with haemodynamic instability requiring vasopressors. Hypercapnic drowsiness is only a relative contraindication, because it often improves once ventilation begins, but agitation that requires sedation is a strong predictor of failure. HFNO has few contraindications: nasal obstruction, recent base-of-skull or mid-face surgery, and an inability to tolerate the prongs are the principal ones, and severe hypercapnic acidosis is a relative contraindication because HFNO cannot reliably augment alveolar ventilation.

Interface-Related Complications

Oronasal and total-face masks cause pressure injury over the nasal bridge and cheeks, the commonest NIV complication, which is largely preventable with correct sizing, a forehead spacer or pad, alternating interfaces, and a scheduled skin inspection every four hours. Gastric distension and aerophagia follow high inspiratory pressures or an improperly seated mask, can precipitate vomiting and aspiration, and are managed by reducing IPAP, repositioning the interface, and considering a nasogastric tube in high-risk patients. Conjunctival irritation from air leak under the mask edge, sinusitis and otitis from prolonged positive pressure, and claustrophobia are further interface effects. Helmet ventilation avoids facial injury but carries noise, reduced trigger sensitivity, and carbon dioxide rebreathing. HFNO causes epistaxis or nasal dryness if humidification fails, and rarely pneumothorax; its main hazard remains the delay of intubation in a patient who is failing.

The Cost of Persistence

The gravest pitfall is the failure to recognise and act on non-invasive failure. A trial of NIV or HFNO should begin with a defined reassessment time, a defined set of physiological endpoints, and an explicit plan for what will happen if they are not met. Persisting for hours with a rising respiratory rate, worsening acidosis, and a falling conscious level converts a manageable respiratory failure into a peri-arrest intubation and increases mortality. The literature on HFNO and NIV failure is unambiguous: delay, not the device, is what kills.

Key Takeaways

1. NIV is first-line therapy with strong evidence of benefit in acute hypercapnic COPD exacerbation and in acute cardiogenic pulmonary oedema; it reduces intubation, and in COPD also reduces mortality.
2. Bilevel NIV is pressure support with PEEP: IPAP is pressure support plus EPAP, and a backup rate should be added whenever the respiratory drive is unreliable.
3. Start IPAP at 10–14 $\text{cmH}_2\text{O}$ and EPAP at 4–6 $\text{cmH}_2\text{O}$, and titrate to comfort, a tidal volume of 6–8 $\text{mL/kg}$ predicted body weight, and a falling respiratory rate, with $\text{SpO}_2$ targeted at 92–96%.
4. In de-novo hypoxaemic respiratory failure — influenza and COVID-19 included — mask NIV has a high failure rate; HFNO or a sealed-interface CPAP strategy is preferred, and the danger of delaying intubation is greater than the risk of intubating unnecessarily.
5. The interface determines tolerance: oronasal for acute hypercapnia, nasal or total-face for comfort and skin sparing, and helmet for prolonged high-pressure support with the least facial injury.
6. HFNO works by matching inspiratory demand with stable FiO₂, washing out nasopharyngeal dead space, generating a modest and flow-dependent positive airway pressure of roughly 0.5–1 $\text{cmH}_2\text{O}$ per 10 $\text{L/min}$, and providing optimal heat and humidification.
7. FLORALI established HFNO as the default first-line support in de-novo hypoxaemic failure with a mortality benefit at 90 days, while CPAP was superior to HFNO for the composite of intubation or mortality in COVID-19 in RECOVERY-RS.
8. The ROX index, $(\text{SpO}_2/\text{FiO}_2)/\text{RR}$, measured at 2, 6, and 12 hours, predicts HFNO outcome: 4.88 or more predicts success, below 3.85 predicts failure, and the intervening zone demands reassessment.
9. HACOR above 5 at one to two hours predicts NIV failure; together with the ROX index it converts clinical vigilance into an actionable threshold.
10. Define an explicit reassessment deadline at the start of every trial of non-invasive support, and proceed to invasive ventilation without delay when the patient is not clearly improving.