High-flow nasal cannula is comfortable, well tolerated, and easy to start — which is exactly why it drifts into use where non-invasive ventilation is what the patient needs. In hypercapnic respiratory failure that drift is dangerous, because high-flow provides washout of dead space and a little positive pressure but does not deliver the inspiratory support that offloads fatiguing respiratory muscles and clears carbon dioxide.
The safe version of using it looks like this. Take a blood gas before deciding, not after starting. If the pH is 7.25 or below, that is non-invasive ventilation, not high-flow. If the acidaemia is milder and you choose high-flow, set a time limit before you start — one to two hours is the usual practice — and repeat the gas at that point rather than when someone notices the patient looking worse. Decide in advance what constitutes failure: a falling pH, a rising carbon dioxide, a rising respiratory rate or a deteriorating conscious level. And confirm that escalation is actually available, meaning a machine, a mask, and a person competent to set it up, at the hour you might need it rather than in principle.
The failure mode this prevents is a specific and common one: a comfortable-looking patient on high-flow whose carbon dioxide is climbing unwatched, presenting hours later as a semi-conscious emergency intubation that a timely mask would have avoided.
- Get an arterial or venous gas before choosing the modality, not after.
- pH 7.25 or below means non-invasive ventilation; high-flow is not an adequate substitute.
- If you use high-flow in milder hypercapnia, set a 1 to 2 hour reassessment and repeat the gas then.
- Write down the failure criteria before starting: falling pH, rising carbon dioxide, rising rate, falling consciousness.
- Confirm escalation is available now — machine, mask and a competent operator — not in principle.
The statistics, in plain English
The pH threshold around 7.25 appears in guidance as an example rather than a hard cut-off, and it reflects where the trial evidence for non-invasive ventilation is strongest rather than a biological cliff. Below it, the mortality and intubation benefit of non-invasive ventilation is well established from randomised trials. Above it, the comparison between modalities is genuinely uncertain, which is why recommendations there are conditional — and why the monitoring requirement is doing most of the safety work, not the number itself.
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