bleep.guide · Foundation Emergencies
Wednesday, 30 September 2026
A structured approach to falling oxygen saturations: oxygen delivery and targets, arterial blood gas interpretation, focused examination, initial investigations, and differentiation of the common causes.
Foundation Emergencies · Session 4 · 30 September 2026 · Oliver Devine
These notes come from a live recording, transcribed and written up using automated tools. Use them to recap, not as guidance. If something looks wrong, let us know.
All patient examples are hypothetical.
Hi guys, it's Ollie. Welcome back to another of our now classic Wednesday evening Foundation Emergencies talks, hosted by Bleep Guide. For those who don't know me, I'm a medical registrar in Cambridge, about to start a clinical pharmacology post in a couple of months.
Today we're talking about a new oxygen requirement. It's a pretty common bleep on ward cover, and a common presentation on the take. Sometimes the patient comes into hospital with it; sometimes they're already here for something else and their oxygen requirement starts going up.
We're going to work through two fictional patients. I'll tell you now: one has a postoperative pneumonia and the other has a COPD exacerbation. No great mystery about the diagnoses. What matters is how we assess them, what the gas actually tells us, and when we need help with respiratory support.
Mrs L is 72 and on day three after abdominal surgery. She had no respiratory symptoms on admission and was previously on air. Now the nurses call because her saturations are 88% on air, then 91% on 2 L/min, and she's been escalated to 6 L/min through a simple face mask.
Before you even get to the bedside, ask for the rest of the observations. Her respiratory rate is 30, heart rate 112, and temperature 38.2°C. The blood pressure sounds all right, but, as we discussed last week, compare it with her usual pressure and the trend.
That's a concerning patient. Don't let a saturation that improves with oxygen talk you out of that. How much oxygen is she needing, and how quickly has that changed? NEWS2 records supplemental oxygen as present or absent; it doesn't keep adding points every time you increase the flow. A patient can therefore deteriorate without much change in the score.1
You get a brief history. The oxygen requirement has risen over the last hour. She's developed a new cough and is bringing up green, rather gunky sputum. There's no reported aspiration event. She's come from home and isn't described as frail.
Pneumonia is high on the list. But coming from home, being relatively fit and having no witnessed aspiration don't prove that swallowing is safe or exclude aspiration. Postoperative patients can also have atelectasis, retained secretions, pulmonary oedema or a PE. Keep those possibilities in mind while you go and assess her.2
This sounds basic, but for God's sake, check the circuit is connected. Imagine attending a patient whose nasal oxygen keeps being increased, following the tubing all the way back to the wall, and finding that an extension has come apart. You can otherwise spend a lot of time worrying about an oxygen requirement that isn't being delivered.
Check the source is on, the tubing isn't kinked or disconnected, and the device is fitted properly. If they're on a reservoir mask, check that the reservoir bag is inflated and doesn't collapse completely as they breathe in. A simple face mask doesn't have a reservoir bag to inflate.
Then check the oximeter. Is there a reasonable waveform? Does its pulse match the patient's pulse? A cold hand, movement or a poorly positioned probe can produce a misleading reading. Try another site if necessary, but don't spend several minutes troubleshooting a probe while the patient is struggling to breathe. Oximetry can also overestimate saturation, including in people with darker skin; interpret a borderline result alongside the clinical picture.3
You repeat Mrs L's saturation: 95% on oxygen. Good, we've improved the number. She's still breathing at 30 and needing substantially more oxygen than an hour ago. That's still concerning.
Let's separate a few things. Ventilation is movement of gas into and out of the alveoli. Gas exchange is oxygen and carbon dioxide moving between alveolar air and the blood in the pulmonary capillaries. Oxygenation is getting oxygen into the blood.
!Conceptual alveolus and pulmonary capillary showing oxygen moving from alveolar air into blood and carbon dioxide moving from blood into alveolar air.
Original slide illustration: ventilation and gas exchange. A simplified teaching schematic, not a patient image. The arrows show directions of gas movement.
You can have a major oxygenation problem while still getting rid of CO₂ quite effectively. That's going to matter for Mrs L. Conversely, someone can have apparently satisfactory saturations on oxygen while becoming dangerously hypercapnic. That's going to matter in the COPD case.
For most acutely unwell adults, the usual target is 94–98%. For people at risk of hypercapnic respiratory failure, including many patients with COPD, use 88–92% initially while you obtain a gas, unless there's an established individual target.4
A raised bicarbonate can be a clue to chronic CO₂ retention. The kidneys retain bicarbonate in compensation for a persistently raised PaCO₂. But a bicarbonate above 30 doesn't prove that somebody is a chronic retainer: metabolic alkalosis, for example with diuretics or vomiting, can also raise it. Look at the pH, CO₂, previous gases and the history together.
And one distinction: the 88–92% oxygen target isn't the same decision as selecting NEWS2 Scale 2. Scale 2 is for confirmed hypercapnic respiratory failure, with a competent clinician's decision documented. Don't automatically put everyone with COPD onto Scale 2.5
If somebody is critically ill or peri-arrest, treat the immediate oxygenation and ventilation threat and get emergency help. A history of COPD isn't a reason to leave them severely hypoxaemic.
Her airway is patent and she's alert, but she's breathing quickly and speaking in short phrases. She shows you tissues with some fairly unpleasant green-brown sputum. You hear right basal crackles. There's air entry on both sides, no wheeze, and the JVP doesn't look elevated.
That supports a chest infection, although sputum colour alone doesn't tell you the organism. Crackles and fever don't exclude a second problem either. Continue through A–E, including circulation, consciousness and a look at her fluid balance and drug chart. Recent opioids, sedatives and IV fluids can change what you're dealing with.
In terms of investigations, I'd want:
Viral testing can affect isolation, antiviral treatment and the working diagnosis. It isn't just to satisfy ourselves academically that we've found something. Equally, a positive viral test doesn't automatically exclude bacterial infection.
If the history and examination suggest heart failure, BNP or NT-proBNP may help; we'll come back to that in the questions. If PE is a realistic possibility, follow the PE pathway rather than ordering a D-dimer just because the saturations are low.
Do those investigations in parallel with treatment and escalation. With these observations and suspected infection, Mrs L has a high NEWS2 and needs the high-risk sepsis pathway: urgent senior assessment and IV antibiotics within an hour of the initial NEWS2 assessment at deterioration. Don't wait for the film or for cultures to come back.7
Everyone asks what you can use a venous gas for. It's useful for a quick assessment of pH, bicarbonate and lactate, interpreted in context. Venous CO₂ can flag possible hypercapnia, but it isn't interchangeable with arterial CO₂ and doesn't reliably quantify how much CO₂ the patient is retaining. Venous oxygen isn't a substitute for PaO₂.8
So in Mrs L, with a rapidly increasing oxygen requirement, I'd do an ABG. In somebody admitted with a COPD exacerbation, NICE also recommends an arterial gas, recording the inspired oxygen concentration. That's particularly important if they're drowsy, acidotic or potentially need NIV.9
It isn't a rule that every stable patient on a little oxygen needs repeated arterial sampling. The question is whether you need an accurate assessment of oxygenation, ventilation or respiratory failure to make the next decision.
When you take the sample, write down the device, flow or set FiO₂, and the time. Record temperature if required locally. With a cannula or simple mask, the FiO₂ is an estimate; don't turn it into a precise measured value on the machine. And don't take the oxygen off an unwell patient simply to get an air gas.
Mrs L tolerates the procedure well. Here's her gas:
She's alkalaemic, her PaCO₂ is slightly below the usual adult arterial range, and her bicarbonate isn't raised. Taken together, that's compatible with a mild respiratory alkalosis: she's breathing quickly and blowing off CO₂. A CO₂ near the edge of a laboratory's reference range doesn't make the whole gas normal.
The oxygen is the worrying part. A PaO₂ of 9.5 while she's receiving oxygen isn't reassuring. Interpret it against her oxygen delivery and clinical state, rather than treating an air reference range as the answer. The normal lactate doesn't cancel out the respiratory deterioration.
Her ECG shows sinus tachycardia, the film is pending, and you've already called your registrar. You don't need to wait for all the investigations before making that call.
We've talked about escalating oxygen, so let's look at the devices. A nasal cannula, simple face mask and reservoir mask all deliver oxygen, but the flow number doesn't mean the same thing for each one.
!Conventional nasal oxygen cannula with nasal prongs and tubing on a white background.
Conventional nasal cannula. Original slide photograph, Intersurgical. Manufacturer source.
!Simple oxygen face mask with side vents and retaining strap on a white background.
Simple face mask. Original slide photograph, Intersurgical. Manufacturer source.
!Adult high concentration oxygen mask with a reservoir bag.
Reservoir mask. Original slide photograph, AirLife / Salter Labs. Manufacturer source.
A simple mask normally needs at least 5 L/min to reduce rebreathing. For a severely hypoxaemic patient needing a reservoir mask, use 15 L/min initially and check the bag. You don't have to work slowly through every device while somebody deteriorates.
The slide's nasal-cannula rule of thumb was 20 + 4 × flow in L/min, giving roughly 24% at one litre, 28% at two, 32% at three and 36% at four. That's a teaching approximation. Actual FiO₂ depends on the patient's inspiratory demand, respiratory pattern and the fit of the device; mouth breathing doesn't translate into a predictable correction.10
Here's the comparison to save somewhere useful:
These ranges describe device performance, not a prescription for an individual patient. Record the device and flow alongside SpO₂; record both flow and FiO₂ for HFNO.
What I particularly want you to remember is the Venturi valve. It entrains room air to provide a specified oxygen concentration. Look at the actual valve: it tells you the nominal FiO₂ and the minimum oxygen driving flow. Don't memorise a colour and assume every manufacturer's valve is identical.11
!Venturi oxygen mask kit with interchangeable coloured valves.
Venturi mask and valves. Original slide photograph, Intersurgical. Use the printed FiO₂ and minimum-flow instructions on the actual device. Manufacturer source.
If the patient is breathing very quickly, a trained clinician may increase the driving flow above that minimum to meet inspiratory demand. With the same valve, that increases total gas flow rather than selecting a different nominal oxygen concentration. Titrate to the patient's target and reassess.
The portable film shows right lower zone consolidation. We now have a working diagnosis of hospital-acquired pneumonia. That's pneumonia developing at least 48 hours after admission, provided it wasn't incubating on arrival. It fits the history we're working with. You've sent a respiratory sample and started antibiotics according to local policy.6
The exact choice depends on severity, local resistance patterns, recent antibiotics and the patient's circumstances. Check allergy history and renal function. Review microbiology and the response when results arrive; don't leave a broad initial regimen unquestioned.
If the JVP had been raised and the overall picture suggested congestion, we'd be thinking about pulmonary oedema. If she'd been wheezy, we'd consider an obstructive exacerbation, though wheeze can also occur in heart failure. Salbutamol, ipratropium and steroids may be appropriate for an obstructive presentation, with treatment directed by the diagnosis. Don't wait for a bronchodilator response before calling for help in a patient who's exhausting themselves.
Postoperatively, people can have a weak cough and retained secretions. Pain, sedation and immobility can all contribute. If somebody suddenly and severely desaturates, mucus plugging and atelectasis belong on the differential. Respiratory physiotherapy may help when retained secretions are contributing, and severe obstruction may need suction or specialist intervention. It isn't a diagnosis you make by excluding everything else, and a physiotherapy referral doesn't replace immediate assessment of a major deterioration.
Two other possibilities are PE and tension pneumothorax.
With PE, you may have breathlessness, pleuritic pain, hypoxaemia or tachycardia. You can also get fever. Syncope or collapse can be a presentation, particularly with haemodynamic compromise, so don't dismiss PE because someone has collapsed or because the chest sounds fairly unremarkable.
Look at VTE risk: recent surgery, immobility, previous VTE and whether prophylaxis has been prescribed and given. Prophylaxis reduces risk; it doesn't rule PE out.
For a stable patient, use the two-level Wells pathway. A score above 4 means PE is likely and you arrange CTPA directly. At 4 or below, D-dimer is usually the next test; a positive result leads to imaging. D-dimer is often raised after surgery or during infection, so its specificity is poor in Mrs L's situation.12
If indicated imaging is delayed, discuss interim therapeutic anticoagulation. Recent abdominal surgery makes bleeding risk particularly relevant, so this needs an individual senior decision. An unstable patient needs emergency assessment; don't let calculating Wells or waiting for a D-dimer delay that.
Tension pneumothorax is a clinical emergency. Imagine someone deteriorating rapidly, with markedly reduced air entry on one side and severe respiratory compromise or hypotension. A distended JVP or tracheal deviation may be present, but they're not findings you must wait for, and their absence doesn't exclude it.
Put out the emergency call and get immediate skilled help. If tension pneumothorax is causing severe compromise, don't wait for a chest X-ray before decompression. The technique depends on the setting, equipment and available expertise.13
This isn't a benign procedure, and “put a cannula somewhere in the safe triangle” isn't an adequate description. We'll come back to the FY1 question at the end. The useful preparation is supervised emergency training and knowing your local approach, kit and escalation route.
Back to Mrs L. Suppose she's now on a reservoir mask and still isn't doing well. At one point her saturation is 91% on 10 L/min, and she's breathing very quickly. Make sure the delivery is effective, increase the reservoir-mask flow to 15 L/min, and get urgent senior and critical care review while you reassess her.
Her CO₂ was 4.2, so this is predominantly an oxygenation problem at that point, rather than established hypercapnic failure. That doesn't guarantee she'll keep ventilating effectively if she becomes exhausted.
Think about the consolidated part of the lung as a wet sponge. It can still receive blood, but some of that blood passes through units that aren't being ventilated properly. It then mixes with blood from better-ventilated lung and lowers the oxygen content of the blood leaving the lungs.
That's ventilation–perfusion mismatch. If an area has essentially no ventilation but still has perfusion, it's a shunt, the extreme of low V/Q. It helps explain why oxygen alone may not fully correct the problem.
By this stage, your registrar should already be involved. A reasonable next discussion is whether she needs a trial of high-flow nasal oxygen, with close monitoring and a clear escalation plan.14
!High flow nasal oxygen system showing the nasal interface, circuit and humidification equipment.
HFNO equipment. Original slide photograph, Fisher & Paykel Healthcare. Manufacturer brochure. The image illustrates the interface, circuit and humidification system.
HFNO delivers a warmed, humidified air–oxygen mixture at a high flow. It can provide a controlled high oxygen concentration, wash out upper-airway dead space and generate some positive airway pressure, although that pressure is variable and isn't the same as set CPAP.14
It isn't reserved for a dry mouth or poor tolerance of a face mask, and it isn't the only non-invasive device capable of delivering a very high FiO₂. The question is whether it provides suitable support for this patient's respiratory failure. Persistent distress, worsening gases or exhaustion need reassessment; don't let a trial of HFNO delay intubation when that's indicated.
We'll leave Mrs L with the senior and critical care teams reviewing that decision, and move to the second case.
Our second fictional patient is Mrs K, who has known COPD and has come in with an exacerbation. Infection may or may not be the trigger; here, the immediate issue is the gas and how she's breathing.
The slide shows:
She's acidotic and hypercapnic. The raised bicarbonate suggests a chronic component, but you'd check previous gases before declaring that established. A saturation of 97% doesn't make this a reassuring patient. The oximeter doesn't tell you the CO₂ or pH.
Drowsiness with a low respiratory rate in this setting raises concern for hypercapnic encephalopathy, sometimes called CO₂ narcosis. But look for other contributors too: opioids, sedatives, hypoglycaemia and other acute illness. Don't assume that everything is explained by the oxygen prescription.
Get urgent senior and critical care help. Assess whether she's protecting her airway and ventilating adequately. If breathing is inadequate, she may need assisted ventilation immediately. Don't wait an hour to see what happens to a drowsy patient who's barely breathing.
Titrate oxygen towards 88–92%, with urgent gas reassessment. Don't simply turn it off and leave her hypoxaemic. The point is controlled oxygen alongside appropriate ventilatory support.
I wanted to spend a little time on this because “they lose their hypoxic drive” is an incomplete explanation. A reduction in ventilation can contribute, but two important mechanisms are worsening V/Q matching and the Haldane effect.15
Normally, a poorly ventilated alveolus has a low local oxygen tension. The small pulmonary arteries supplying that area constrict: hypoxic pulmonary vasoconstriction. That diverts some blood towards better-ventilated parts of the lung.
Give a susceptible patient excessive oxygen and you can reverse some of that vasoconstriction. More blood now goes through poorly ventilated lung. Matching worsens, and the patient may be unable to increase ventilation enough to clear the extra CO₂ burden. This isn't just oxygenated blood being diluted; the important consequence here is impaired CO₂ elimination.
!Two panel conceptual diagram showing a poorly ventilated alveolus with less blood flow during hypoxic vasoconstriction and more blood flow after high oxygen, while ventilation remains poor.
Original slide illustration: reversal of hypoxic pulmonary vasoconstriction. The same poorly ventilated unit is shown in both panels. Vessel widths and arrows are qualitative, not measured values.
Then there's the Haldane effect. Deoxygenated haemoglobin can carry more CO₂ and buffer more hydrogen ions. When haemoglobin becomes oxygenated, its CO₂ carriage and hydrogen-ion buffering capacity fall. That promotes release of CO₂ into the blood, including through the bicarbonate buffering system.
!Conceptual comparison of deoxygenated and oxygenated haemoglobin showing greater carbon dioxide carriage before oxygenation and lower carriage with release of carbon dioxide and hydrogen ions after oxygenation.
Original slide illustration: the Haldane effect. Symbols show qualitative differences in CO₂ carriage and H⁺ buffering; they aren't fixed chemical quantities or a depiction of separate oxygen and CO₂ reservoirs.
If the lungs can increase ventilation, that CO₂ can be exhaled. In someone with limited ventilatory reserve, PaCO₂ can rise. Most CO₂ is carried as bicarbonate, so saying “haemoglobin stores all the CO₂” would be misleading.16
It's a double contribution: oxygen can worsen regional matching and alter CO₂ carriage. The practical consequence is to prescribe oxygen to a target and check the gas when you need to know how the patient is ventilating.
For the COPD exacerbation, start appropriate treatment: bronchodilators such as salbutamol and ipratropium, systemic steroids, and antibiotics when indicated. Steroids aren't an immediate rescue for inadequate ventilation. In a hypercapnic patient, nebulisers are usually driven by air, with controlled oxygen supplied separately as needed.17
Reassess the patient and repeat the arterial gas. In a patient stable enough for an initial treatment trial, a gas at about an hour helps determine the response. The commonly used COPD threshold for NIV is pH below 7.35 with PaCO₂ above 6.5 kPa persisting despite optimal initial treatment.18
On the repeat slide, Mrs K's pH is 7.30 and PaCO₂ 8.1 kPa. There's some improvement, but she's still acidotic and hypercapnic. Suppose she's now cooperative and protecting her airway. That makes a closely monitored trial of bilevel NIV reasonable, with an experienced team and a plan if it fails.
!Non invasive ventilation full face mask photographed alone on a white background.
NIV interface. Original slide photograph, Intersurgical / Sleepnet. The mask must be compatible with the circuit and its exhalation arrangement. Manufacturer source.
Bilevel NIV gives a higher pressure during inspiration and a lower pressure during expiration. The difference supports ventilation and helps clear CO₂. Cooperation matters, but so do airway protection, secretion clearance, vomiting risk and haemodynamic stability.
Reduced consciousness is a reason for urgent expert assessment; it isn't something to cover with a tight mask and leave on an ordinary ward. Some patients with hypercapnic encephalopathy can receive NIV under specialist supervision, while others need invasive ventilation. The decision depends on the full situation.
Departments that provide acute NIV need trained staff and suitable monitoring. At FY1 or FY2 level, your job is to recognise the potential indication, start the safe initial measures and get the appropriate team involved. You aren't expected to choose unfamiliar settings unaided.
After NIV starts, assess comfort, respiratory effort and consciousness, and repeat the gas early, usually within an hour under the local protocol. If the pH, breathing or consciousness worsens, escalate promptly. Starting NIV isn't the end of the assessment.19
We've talked about HFNO and bilevel NIV. CPAP provides one continuous pressure. In acute cardiogenic pulmonary oedema it can recruit alveoli and reduce cardiac loading. It doesn't simply push fluid back into the pulmonary capillaries, and it has uses beyond pulmonary oedema.
For acute pulmonary oedema, non-invasive support is particularly relevant when there's severe breathlessness and acidaemia, or failure to respond adequately to treatment. You don't have to wait until every other option has failed to discuss it. Treat the underlying heart failure at the same time.20
Intubation and invasive ventilation may be needed when someone can't protect their airway, is exhausted, remains hypoxaemic or has worsening ventilatory failure. It's a senior and critical care decision, but recognise the possibility early. The choice is about the type and severity of failure, not simply stepping through devices until you've reached the one with the highest FiO₂.
We haven't got time for a full discussion of ceilings of care, but in long-term respiratory disease it's useful to have those conversations early. Discuss the patient's wishes, likely benefit and burdens, and the plan if NIV fails. Dementia or delirium may affect tolerance or decision-making capacity; neither automatically excludes treatment. Involve the patient wherever possible, and family or an appropriate representative when needed.
A DNACPR recommendation is about CPR. It doesn't itself mean “no NIV” or “no active treatment”. Record the actual escalation plan and review it when the situation changes.21
We'll skip the worked SBAR today and go straight to the questions.
It's a difficult question, and it's something to think about before you meet the emergency. Suppose you've put out a 2222 call and you're with a patient who has very poor air entry on one side, a pressure of 80/40 and rapidly worsening breathing. You think this is tension pneumothorax. What do you do while help is coming?
You call for immediate skilled help, make the suspected diagnosis explicit, and prepare the kit. In severe compromise, decompression mustn't wait for imaging. But being an FY1 doesn't, on its own, mean you're competent to perform it, and an online video doesn't establish competence. Get supervised training, know the local emergency technique and seek real-time senior support.
If you're trained and competent to perform emergency decompression, follow that training and local protocol. If you're not, say so clearly and urgently get someone who is. Needle decompression has risks and can fail; it isn't the complete definitive treatment. This is an emergency skill to learn properly.1322
Roy asks whether BNP is relevant in the acute scenario, because he's heard different things from different doctors.
Yes, it can be useful, particularly in new suspected acute heart failure. BNP and NT-proBNP are released in response to cardiac wall stress. They aren't a direct measurement of how much excess fluid the patient has.
NICE uses BNP below 100 ng/L or NT-proBNP below 300 ng/L to help rule out new acute heart failure. A raised result supports further assessment, usually including echocardiography, but doesn't prove the diagnosis. Kidney dysfunction, AF, sepsis and right-heart strain can raise the result; obesity can lower it. Interpret it alongside the history, examination, ECG and imaging.23
Previous results may provide context, but a changing number doesn't by itself tell you how much to diurese. And if the patient clearly has acute pulmonary oedema and needs urgent treatment, don't wait for BNP to come back.
Yes, that's the idea behind the Haldane effect. At a given CO₂ pressure, oxygenated blood carries less CO₂ than deoxygenated blood. Oxygenation promotes CO₂ release. If the patient can't increase ventilation enough to exhale it, the dissolved CO₂ and PaCO₂ rise.
So when we say “less carriage”, we're talking about the form in which CO₂ is carried, not CO₂ disappearing from the body. That's one reason excessive oxygen can worsen hypercapnia in a susceptible patient.
All right, folks, that's us for this week. Next week we're doing heart failure and acute pulmonary oedema, including where CPAP fits and the causes of sudden, severe pulmonary oedema. Thank you very much once again.
References support the clinical corrections and explanatory additions; they were not spoken citations in the webinar. Device photographs remain the copyright of their manufacturers and illustrate examples rather than brand recommendations. The three physiology illustrations are generated teaching schematics from the original slide deck.