High-flow nasal cannula, non-invasive ventilation and CPAP have been used across acute respiratory failure with wide variation and no comprehensive guideline. The American Thoracic Society has now produced one, developed by a multidisciplinary panel using GRADE, addressing four questions and informed by systematic reviews and network meta-analyses.
For acute hypoxaemic respiratory failure, the panel makes a strong recommendation for high-flow nasal cannula and a conditional recommendation for non-invasive ventilation or CPAP, based primarily on effects on the need for intubation, with the qualifier that close monitoring for escalation is required. For acute hypercapnic respiratory failure the hierarchy reverses: a strong recommendation for non-invasive ventilation, on mortality and intubation, and a conditional recommendation for high-flow only in less severe hypercapnia with mild acidaemia — pH above 7.25 is the example given — and only where close monitoring and prompt escalation to non-invasive ventilation are available.
Two further recommendations are easy to overlook and immediately actionable. Preoxygenation before intubation should use high-flow or non-invasive ventilation rather than a bag or a facemask, and this is a strong recommendation. And after extubation from critical illness the choice should be risk-stratified — high-flow for low-risk patients, non-invasive ventilation for high-risk ones — to reduce reintubation.
The strength labels are doing real work here and should be read carefully. Strong recommendations mean most patients should receive the intervention; conditional ones mean the choice depends on circumstances and preference. The pattern across the document is that the modality with the best evidence differs by the physiology, and that using high-flow reflexively for everything — which is what has happened in many units as the equipment spread — is not supported in hypercapnia.
For Indian practice the constraint is oxygen. High-flow consumes very large volumes of oxygen, and in hospitals where supply is piped from a limited source or delivered in cylinders, that is a real institutional decision rather than a bedside one. The guideline itself acknowledges that recommendations should be informed by institutional capacity, and this is where that caveat bites hardest.
- Acute hypoxaemic failure: high-flow nasal cannula first, with close monitoring for escalation.
- Acute hypercapnic failure: non-invasive ventilation, strongly recommended, to reduce mortality and intubation.
- Use high-flow in hypercapnia only with milder acidaemia (pH above 7.25) and immediate escalation available.
- Preoxygenate before intubation with high-flow or non-invasive ventilation, not a bag-valve-mask alone.
- After extubation, stratify: high-flow for low-risk patients, non-invasive ventilation for high-risk.
The statistics, in plain English
GRADE separates how confident the panel is in the evidence from how strongly it recommends acting on it. A strong recommendation means the panel judged that most well-informed patients would choose the intervention, so it is reasonable to treat it as a quality standard. A conditional recommendation means the balance of benefits and harms is close, or the evidence is uncertain, and the right answer varies by patient — those should prompt a decision, not a protocol. The hypoxaemic recommendations rest primarily on intubation rates rather than mortality, which is a softer endpoint: avoiding intubation is worth having, but it is also influenced by clinician thresholds in unblinded trials.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for pulmonology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free