Respiratory Failure and Non-Invasive Ventilation

The answer first

Acute respiratory failure is a failure of gas exchange, defined on arterial blood gas — and the entire clinical game is matching the type of failure to the right respiratory support, early. Four rules carry you through almost every DWE question and every after-hours ward call [1] [2]:

1. Type 1 is oxygenation failure; type 2 is ventilation failure. Type 1 is a PaO₂ below 8 kPa with a normal or low PaCO₂ — the problem is the lung (V/Q mismatch, shunt). Type 2 adds a PaCO₂ above 6 kPa — the problem is the pump (reduced ventilatory drive, muscle weakness, or a load the muscles cannot meet) [2] [20].
2. Oxygen is a drug with a target, not a reflex. Most acutely ill adults are targeted to 94–98%, but anyone at risk of CO₂ retention — COPD above all — gets controlled oxygen (24–28% Venturi) aiming for 88–92%, with a blood gas checked after any change [3] [4].
3. An acidotic COPD exacerbation is an NIV emergency. pH below 7.35 with hypercapnia despite maximal medical therapy means BiPAP now — it reduces intubation and death, and the evidence is among the strongest in acute medicine. Cardiogenic pulmonary oedema, by contrast, gets CPAP [2] [7] [9].
4. NIV is a trial with an exit, not a destination. Repeat the blood gas at 1–2 hours. Improving pH and PaCO₂ means continue; worsening acidosis, exhaustion, secretions or agitation means stop escalating and intubate — delayed intubation is the classic fatal error [2] [19].

Two failures, one physiology

The two types of respiratory failure are not arbitrary ABG labels — they are two different physiological catastrophes, and the therapy that fixes one can be irrelevant to the other [20].

Type 1 — hypoxaemic failure. Oxygen fails to cross from alveolus to blood. The dominant mechanism is ventilation–perfusion mismatch: alveoli that are perfused but poorly ventilated (pneumonia consolidation, pulmonary oedema, atelectasis) let blood pass without picking up oxygen; at the extreme this is true shunt. Diffusion limitation and low inspired oxygen (altitude) contribute in narrower settings. Crucially, CO₂ clearance is preserved — CO₂ is about twenty times more diffusible than oxygen and the surviving ventilated lung increases its minute ventilation, so the PaCO₂ is normal or low. The clinical signature is breathlessness, tachypnoea, and hypoxaemia that may partly correct with oxygen [20] [21].

Type 2 — ventilatory failure. The pump fails and alveolar ventilation falls; PaCO₂ rises in direct proportion to the fall in ventilation, and PaO₂ falls with it. The causes sort into three bins: central drive (opioids, sedatives, brainstem lesions), neuromuscular and chest wall (motor neurone disease, Guillain–Barré, myasthenia, kyphoscoliosis, obesity hypoventilation), and increased load with fatiguing muscles — above all COPD, where airways obstruction, hyperinflation and flattened diaphragms mean the muscles cannot sustain the work [2] [21].

Reading the ABG properly — oxygenation, acid–base, compensation

Examiners can tell within ten seconds whether you read a gas systematically or scan it for the number you fear. The method below is fast, defensible, and works for every gas you will ever be shown [21].

Acute versus chronic retention is read from the pH–HCO₃ pair. In acute CO₂ retention, the kidneys have had no time to compensate: bicarbonate is near normal (rises only about 1 mmol/L per 1.3 kPa rise in PaCO₂ from tissue buffering), so the pH falls sharply — pH 7.22 with PaCO₂ 9 kPa and HCO₃ 27. In chronic retention, renal bicarbonate retention (over 3–5 days) drags the pH back toward normal: pH 7.34 with the same PaCO₂ 9 kPa but HCO₃ 34. The acute-on-chronic gas — the decompensated COPD exacerbation — shows a high HCO₃ with a pH that has nonetheless fallen: the compensation was there, and the new insult has overwhelmed it [21] [2].

Oxygen therapy — a drug with a prescription

Oxygen is the most commonly prescribed drug in acute medicine and the least commonly prescribed properly. The modern framing — written into both the BTS guideline and the Thoracic Society of Australia and New Zealand guideline — is that oxygen is prescribed to a target saturation range, with the device and flow chosen to hit that range, and the response checked [3] [4].

Why controlled oxygen matters in the retainer — and why the old story is wrong. The teaching you inherited is that COPD patients "breathe off their hypoxic drive" and that oxygen switches off ventilation. That is a myth, and it is dangerous in both directions — it makes juniors terrified of oxygen in the hypoxaemic, and it makes them casual about it in the retainer. The measured mechanisms of oxygen-induced hypercapnia are, in order of importance: reversal of hypoxic pulmonary vasoconstriction (oxygen re-opens vessels in poorly ventilated lung, worsening V/Q mismatch and dead space), the Haldane effect (oxygenated haemoglobin carries less CO₂, raising PaCO₂ at a fixed ventilation), and only a small true fall in minute ventilation. The practical point stands despite the mechanism being mislabelled: uncontrolled oxygen in a susceptible retainer produces real, sometimes fatal, rises in PaCO₂ [16].

The evidence for controlled oxygen is not theoretical. In a randomised trial of prehospital care for suspected COPD exacerbations, titrated oxygen delivered by nasal cannulae to an SpO₂ of 88–92% reduced mortality by more than half compared with high-flow oxygen — with less acidosis and less hypercapnia on arrival [14]. And the IOTA meta-analysis of acutely ill adults found that liberal oxygen strategies increased mortality in a dose-dependent way compared with conservative targets — hyperoxia is a treatment exposure with a harm profile, not a safety blanket [15].

The acidotic COPD exacerbation — NIV is the treatment

This is the single highest-yield indication in the whole topic, and one of the strongest evidence bases in acute medicine. A COPD exacerbation that has progressed to respiratory acidosis — pH below 7.35 with a raised PaCO₂ — despite maximal medical therapy (controlled oxygen, nebulised bronchodilators, steroids, antibiotics where indicated) is an indication for NIV, and every hour of delay costs benefit [2] [7].

The trial sequence examiners expect, in order [5]:

What BiPAP actually does. The two pressures do two different jobs, and viva candidates who can say this cleanly stand out. IPAP (inspiratory positive airway pressure) is the ventilatory pressure — it augments tidal volume, rests fatigued inspiratory muscles, and clears CO₂; titrating IPAP up is how you treat the acidosis. EPAP (expiratory positive airway pressure) splints the airway against dynamic collapse and offsets intrinsic PEEP (auto-PEEP) so the patient does not have to overcome trapped-gas pressure before each breath can begin; it also recruits alveoli and helps oxygenation. The gradient between IPAP and EPAP is the pressure support — that is what does the ventilating [2] [1].

Cardiogenic pulmonary oedema — CPAP, and an honest evidence story

Acute cardiogenic pulmonary oedema is type 1 failure driven by flooded alveoli — but the positive-pressure story here is CPAP, not BiPAP, and the evidence needs stating precisely because the trials say slightly different things [8] [9].

A single continuous pressure recruits fluid-filled alveoli, improves compliance and oxygenation, and — by raising intrathoracic pressure — reduces venous return and left ventricular afterload, offloading the failing heart in both directions. The Cochrane synthesis of non-invasive positive pressure in cardiogenic pulmonary oedema found CPAP (and bilevel) reduced mortality and intubation compared with standard oxygen therapy [9]. The large pragmatic 3CPO trial then compared CPAP and bilevel NIV against standard oxygen in over a thousand emergency patients: neither modality improved mortality, but both produced faster relief of breathlessness, acidosis and hypercapnia — and there was no excess of myocardial infarction with bilevel, burying an old safety fear [8].

The honest summary for an examiner: CPAP is first-line respiratory support for cardiogenic pulmonary oedema with persisting hypoxaemia or distress despite oxygen and medical therapy; expect rapid symptomatic benefit, claim mortality benefit cautiously (meta-analyses yes, the largest single trial no), and prefer CPAP over bilevel unless there is concurrent hypercapnia with acidosis. And it is an adjunct — the definitive treatment is vasodilators (glyceryl trinitrate), diuretics where volume-overloaded, and treatment of the trigger [8] [9] [1].

Monitoring NIV — and recognising failure early

NIV is prescribed like a drug with a therapeutic window, and the repeat blood gas is the serum level. The BTS/ICS guideline is explicit: reassess clinically and repeat the ABG at 1–2 hours after starting, and after any change in settings or FiO₂ [2].

Success looks like: a rising pH (even a small rise matters), a falling PaCO₂, falling respiratory rate, improving synchrony and comfort, and a patient who can talk and eventually doze. Failure looks like: a pH that is static or falling at 1–2 hours, rising PaCO₂, persisting tachypnoea and exhaustion, copious secretions the patient cannot clear, agitation or deteriorating conscious level, haemodynamic instability, or a mask the patient simply cannot tolerate despite adjustment [2] [19].

For hypoxaemic indications the same discipline applies with harder stakes. In a multicentre study of NIV in acute hypoxaemic respiratory failure, NIV failure — and subsequent intubation — was common and was predicted by older age, higher severity scores, ARDS or pneumonia as the cause, and failure to improve oxygenation early; intubation after failed NIV carried high mortality, which is why the threshold to stop must be low and predefined [19].

Absolute and near-absolute contraindications — the list that wins or loses DWE marks [2]:

• Immediately life-threatening failure needing emergent intubation (respiratory arrest, peri-arrest, severe refractory hypoxaemia).
• Inability to protect the airway: vomiting or high aspiration risk, reduced conscious level (unless rapidly reversible, e.g., hypercapnic narcosis that wakes on NIV — a nuance examiners enjoy), uncontrolled upper GI bleeding, facial trauma or recent upper airway surgery.
• Haemodynamic instability or serious arrhythmia.
• Copious secretions the patient cannot clear.
• Untreated pneumothorax.
• Undrained facial burns / inability to fit a mask, and patient refusal [2].

Where NIV needs caution — pneumonia and the immunosuppressed

The NIV evidence base was built on COPD and cardiogenic pulmonary oedema. Extrapolating it to de novo hypoxaemic failure — pneumonia and ARDS — is where candidates get burned [1] [23].

The one RCT examiners cite is Confalonieri: in severe community-acquired pneumonia with acute respiratory failure, NIV reduced intubation and ICU mortality only in the subgroup with underlying COPD — patients without COPD derived no clear benefit. Read it as a COPD study wearing pneumonia clothes, not as a mandate to ventilate pneumonias non-invasively [23]. The ERS/ATS guideline, weighing this and the failure data, suggests NIV in de novo hypoxaemic failure only cautiously and selectively — with experienced teams, close monitoring, and a low, predefined intubation threshold — because failed NIV followed by late intubation is associated with harm [1] [19].

The immunosuppressed patient deserves its own line because the trial story flipped. Older teaching favoured early NIV in immunocompromised patients to avoid intubation-related infection; the Lemiale JAMA trial (2015) randomised immunocompromised patients with acute hypoxaemic respiratory failure to NIV versus standard oxygen and found no difference in mortality or intubation — stripping NIV of any special advantage in this group. Modern practice: HFNO is often chosen for comfort and oxygenation, but the survival lever is the underlying diagnosis and early intubation when indicated, not the interface [11] [22].

High-flow nasal oxygen — where it fits

HFNO delivers heated, humidified oxygen at up to 60 L/min through wide-bore nasal prongs, washing out nasopharyngeal dead space, providing a small flow-dependent positive pressure, and — because the FiO₂ is set rather than entrained — delivering a reliable inspired fraction with far better comfort than a mask [22].

The landmark trial is FLORALI: in patients with acute hypoxaemic (non-hypercapnic) respiratory failure, mostly pneumonia, HFNO was compared with standard oxygen and with NIV. The primary intubation endpoint did not differ overall, but in the more hypoxaemic stratum intubation was reduced, and 90-day mortality was lower with HFNO — plausibly because NIV delivered high tidal volumes that aggravated lung injury. FLORALI repositioned HFNO as the default high-end oxygen device in hypoxaemic failure [10]. The ERS HFNO guideline gives HFNO a conditional recommendation over conventional oxygen in hypoxaemic respiratory failure, and a conditional recommendation over NIV in that setting — while emphasising that HFNO must not delay intubation when it is failing (watch the ROX pattern: saturations, FiO₂ requirement, and respiratory rate trajectory) [22].

When to intubate — the triggers

Intubation is the answer when the disease is winning faster than the non-invasive strategy. The triggers below are deliberately written as bedside observations, not numbers alone, because that is how the decision is actually made [2] [19]:

• Failed NIV trial: pH or PaCO₂ worsening at 1–2 hours, or any persisting severe acidosis with exhaustion.
• Refractory hypoxaemia: oxygenation deteriorating despite HFNO or NIV — particularly with rising work of breathing.
• Exhaustion and deteriorating conscious level — including CO₂ narcosis that is not lifting on NIV.
• Airway threat: vomiting, aspiration, secretions the patient cannot clear.
• Haemodynamic collapse or malignant arrhythmia.
• Multi-organ failure where respiratory failure is one failing organ among several [19].

The corollary: intubation must also be the answer you are allowed to give. A patient with a documented decision against ICU admission is not "failed NIV" — they are a patient at their ceiling, and the job changes to symptom control [2].

Domiciliary NIV — who actually benefits at home

Long-term home NIV is not "acute NIV that never stopped" — it is a selected, titrated therapy for chronic ventilatory failure, and the evidence is diagnosis-specific [12] [13].

Weaning NIV and ceilings of care

Weaning is symptom- and gas-led, not clock-led. As the exacerbation resolves, move from near-continuous NIV to sessions (off for meals and nebulisers), then to nights only, then off — typically over 48–72 hours. The failure mode to name in the exam is the abrupt stop on day one because the first gas looked better, followed by relapse overnight; the acidosis took days to develop and the muscles have not recovered in an afternoon [2].

Ceilings of care are decided before the machine is switched on, not during failure. Every NIV order should carry two explicit, documented answers: is this patient for intubation and ICU if NIV fails?, and what was discussed with the patient and family? The BTS/ICS guideline treats NIV-as-ceiling as a legitimate, common plan in advanced COPD — NIV relieves dyspnoea and can return a drowsy hypercapnic patient to lucidity for a proper goals-of-care conversation — but it insists the status is recorded so the 3 a.m. team is not improvising someone's last hours [2].

The DCE angles

Long case — severe COPD with recurrent acidotic exacerbations. The examiner's interest moves quickly from the acute gas to the program. Your framework [2] [12]:

1. The exacerbation that brought them in: severity, pH, NIV course, ICU exposure, and what this second admission says about trajectory.
2. Optimising the disease: inhaler technique and escalation, pulmonary rehabilitation after every hospitalised exacerbation, vaccination, smoking cessation support, treatment of comorbid heart failure and OSA overlap.
3. The domiciliary decisions: a 2–4 week post-discharge gas to find persistent hypercapnia (HOT-HMV selection for home NIV); long-term oxygen assessment if persistently hypoxaemic on air — remembering LTOT and home NIV answer different problems [12].
4. Ceilings and advance care planning: prognosis stated honestly, NIV/intubation ceilings documented, an advance care plan made while well.
5. The social frame: function, carer strain, anxiety around breathlessness, and palliative symptom strategies for refractory dyspnoea.

Short case — the breathless patient. Work from the end of the bed: respiratory rate and pattern, accessory muscle use and tracheal tug, pursed-lip breathing, thoracoabdominal paradox (fatigue and impending arrest), ability to speak in sentences. Then the CO₂-retention cluster — coarse flap (asterixis), bounding pulse, warm peripheries, drowsiness — distinguished from the fine tremor of salbutamol, and central cyanosis at the tongue. Then the chest: hyperinflation, breath sounds, added sounds, and the cor pulmonale triad of raised JVP, loud P₂, parasternal heave and peripheral oedema. Close by saying what you would do next: sit them up, controlled oxygen, and a blood gas [2] [16].

Exam traps, collected