Recognising Respiratory Failure Before the Numbers Say So
There is a moment in respiratory decline where the patient looks better and is doing worse. The noise settles. The rate comes down. The patient stops struggling and starts dozing. Every one of those changes reads as improvement, and none of them is.
Effort is not the same as effectiveness
Early distress is loud because the patient still has reserve to spend. Fast breathing, wheeze, accessory muscle use, sitting forward, refusing to lie flat and refusing to stop talking about it.
Failure is quiet because the reserve has run out. Muscles fatigue like any other muscle, and a diaphragm that has been working at maximum for an hour cannot keep it up indefinitely.
So the direction of travel is the finding. A respiratory rate falling toward normal in a patient who has not been treated is not a return to normal, and this is the trap the item is built around.
Ask what changed and why. Improvement after an intervention is credible. Improvement after nothing is a warning.
The findings that come before any number moves
Speech is one of the most sensitive bedside measures available and it costs nothing. Full sentences, then short phrases, then single words, then nothing is a decline you can hear across the room.
Posture tells you the same story. A patient who was upright and forward, then slumps or agrees to lie flat, has usually not become comfortable.
Then look at the chest wall itself. Accessory muscle use, intercostal retraction, and see-saw movement between chest and abdomen all say the ordinary mechanics are no longer sufficient.
Colour, sweating and a rising heart rate belong here too. Diaphoresis in a breathless patient is a sympathetic response, not a warm room.
Count the rate yourself, for a full minute, while the patient is unaware you are counting. Respiratory rate is easy to estimate and easy to record without ever counting, and it is the observation that carries the most warning.
Count the depth as well as the number. Shallow fast breathing moves less air than the figure suggests, and two patients with the same recorded rate can be in completely different trouble.
Who is at risk before any of this starts
Anticipating the patient is easier than rescuing them, and stems often hand you the risk in the first sentence. Recognising it early changes how closely you plan to observe.
- Anyone who has received opioids or sedatives, especially in combination
- Post-operative patients, particularly after abdominal or chest surgery
- Patients with existing lung disease and little reserve to spend
- Neuromuscular conditions where the muscles of breathing are already weak
- Chest trauma, rib fractures and anything that makes breathing hurt
- Reduced consciousness from any cause, including a first seizure
Pain deserves a note of its own. A patient who cannot take a deep breath because it hurts will breathe shallowly for hours and then develop a problem that better analgesia would have prevented.
What the monitor tells you, and when it tells you late
Saturation monitoring is useful and it is not an early warning system. StatPearls gives a normal arterial oxygen saturation of 95 to 100% on a blood gas, and a patient can be working extremely hard while the bedside number still sits inside that band.
So the number falling is confirmation, not detection. Poor perfusion, cold fingers, movement and nail polish all interfere with the reading as well, which is another reason to trust the patient over the display.
Level of consciousness is a respiratory observation
Students file consciousness under neurology and then miss it in respiratory stems. Both gases affect the brain, and each one does it in a recognisable way.
Falling oxygen tends to make people restless, anxious and hard to settle. Rising carbon dioxide tends to make them drowsy, headachy and slow, and eventually unrousable.
That gives you a usable pairing. Agitation early, somnolence late.
So a patient who has become peaceful after being agitated has either been treated effectively or has moved further along, and the stem will tell you which by telling you what was done. If nothing was done, treat the peace as deterioration.
The gas values behind that reasoning, and why oxygenation is assessed separately from acid-base status, are handled in step four, oxygenation is a separate question.
A sleepy patient is not always a lung problem, either. Rising pressure inside the skull produces its own sequence of changes, described in increased intracranial pressure and the changes that matter.
Sudden neurological change deserves the same speed of response. Facial asymmetry, one-sided weakness or new difficulty speaking sends you to stroke assessment and the time-sensitive actions, where the clock starts immediately.
What you do first, and what you do not do
The first move is almost never leaving. Stay with the patient, sit them upright, apply oxygen as ordered or per protocol, and get help to come to you rather than going to find it.
Then work the sequence, in this order.
- Position upright and support the patient forward if that is what they are already doing
- Apply or adjust oxygen within the ordered parameters
- Call for help early, using the rapid response pathway your facility has
- Stay and keep observing, because the next change is the information the team needs
- Anticipate suction, bag-valve-mask and airway support being needed at the bedside
Notice what is missing from that list. No option that involves leaving to document, fetching equipment alone, or waiting for the next scheduled observation round is correct here.
Sedation is worth flagging. A breathless, agitated patient is often uncomfortable, and an option offering sedation to settle them is offering to remove the drive that is keeping them ventilated.
If opioids are part of the story, the reversal agent and its own timing become relevant, and that timing is less forgiving than people assume. Escalating rather than improvising is the safe answer in a written item.
Handover language matters here too. Saying the patient is tiring, with the observations that support it and the trend across the shift, gets a faster response than reporting one abnormal number without context.
Chronic disease changes the baseline, not the principles
A patient with long-standing lung disease may live at values that would alarm you in someone else, and their usual observations are the comparison that matters. The specifics of that, including how oxygen is titrated, sit in COPD and asthma oxygen decisions.
The principle survives the baseline. Deterioration is a change from this patient's own normal, and a stem that gives you their baseline is telling you to use it.
The one-line version
Loud patients are frightening. Quiet patients are dangerous. Work out which one you have before you decide anything else, and keep the wider system order from the med-surg reasoning order underneath it.