ICU · Toxicology
Carbon monoxide poisoning
Also known as CO poisoning · Carboxyhaemoglobin (COHb) · Hyperbaric oxygen therapy (HBOT) · Delayed encephalopathy · Delayed neurological sequelae (DNS) · The great imitator (flu-like)
Carbon monoxide (CO) is an odourless, colourless gas produced by incomplete combustion (faulty heaters, house fires, car exhausts, generators, barbecues used indoors). CO binds haemoglobin with 240x affinity of oxygen, forming carboxyhaemoglobin (COHb), causing tissue hypoxia via THREE mechanisms: (1) functional anaemia, (2) LEFT shift of oxyhaemoglobin dissociation curve (impaired tissue O2 release), (3) histotoxic hypoxia (CO binds myoglobin + inhibits mitochondrial cytochrome c oxidase / cytochrome a3 / Complex IV → anaerobic metabolism → lactate). Presentation: headache, nausea, dizziness, confusion (flu-like) — 'flu-like illness in MULTIPLE people from the SAME household in WINTER = CO until proven otherwise.' Severe: syncope, seizures, coma, cardiovascular collapse, death. Diagnosis: COHb level (venous blood gas — NOT SpO2 which is FALSELY NORMAL because pulse oximetry cannot distinguish COHb from oxyhaemoglobin). Treatment: 100% oxygen via non-rebreather (reduces CO half-life from 320 min to 80 min). Hyperbaric oxygen (HBO) for severe cases (COHb 25%, unconscious, pregnant, cardiac ischaemia, neurological signs) — reduces CO half-life to 23 min. Delayed neurological sequelae occur in 20-40% of severe cases at 2-40 days (cognitive impairment, parkinsonism) — may be permanent.
On this page & tools
Your progress
Saved locally on this device.
Target exams
Red flags
Pathophysiology
The triple insult explained in depth
CO produces tissue hypoxia through three synergistic mechanisms, and understanding all three is essential because it explains why a "normal PaO2" and "normal SpO2" do not exclude life-threatening poisoning. [1]
-
Functional anaemia (reduced O2 carrying capacity). CO binds haemoglobin with an affinity roughly 240x that of oxygen, forming carboxyhaemoglobin (COHb). Because the affinity is so much higher, even low ambient CO concentrations progressively displace oxygen from haemoglobin. A COHb of 40% means ~40% of haemoglobin binding sites are occupied by CO and unavailable to carry oxygen — the equivalent of losing 40% of haemoglobin, but with the added insult that the remaining sites behave abnormally (point 2).[1]
-
Left shift of the oxyhaemoglobin dissociation curve. This is the often-underappreciated second mechanism. When one CO molecule occupies one of the four haem sites on a haemoglobin tetramer, the remaining three sites bind oxygen with increased affinity (an allosteric effect — analogous to the Bohr effect in reverse). The dissociation curve shifts left, so haemoglobin holds onto its remaining oxygen more tightly and releases far less at tissue PO2. The practical consequence: even the oxygen that IS carried is not delivered. This is why CO poisoning causes more tissue hypoxia than an equivalent haemorrhagic anaemia.[1]
-
Direct cellular (histotoxic) hypoxia. CO diffuses into tissues and binds two key intracellular targets: myoglobin (with an even higher affinity than haemoglobin) and mitochondrial cytochrome c oxidase (cytochrome a3, Complex IV). Inhibition of cytochrome oxidase blocks the electron transport chain → cells cannot utilise delivered oxygen → anaerobic metabolism → lactic acidosis. This "poisoning of the oxygen utilisation machinery" is the mechanism most strongly linked to delayed neurological injury (it triggers neutrophil activation, free-radical lipid peroxidation, and white-matter demyelination).[1][4]
The net effect is a patient who is hypoxic at the tissue level while displaying a deceptively reassuring arterial blood gas: the PaO2 (dissolved oxygen) is normal, the calculated SaO2 is normal, and only the directly measured COHb and the lactate reveal the truth. [1]
The oxyhaemoglobin dissociation curve — why the left shift matters
The oxyhaemoglobin dissociation curve (ODC) plots haemoglobin oxygen saturation against PaO2. Normally the curve is sigmoid, allowing efficient loading in the lungs (PaO2 ~100 mmHg, ~97% saturated) and unloading at the tissues (PaO2 ~40 mmHg, ~75% saturated) — yielding a generous ~25% O2 extraction. [1]
Right shift vs left shift of the ODC
| Right shift (oxygen released MORE readily) | Left shift (oxygen held MORE tightly — the CO problem) | |
|---|---|---|
| P50 | ↑ (e.g. 35-40 mmHg) | ↓ (e.g. 15-20 mmHg) |
| Causes | ↑ H+ (acidosis), ↑ CO2, ↑ temperature, ↑ 2,3-DPG, chronic anaemia | ↓ H+ (alkalosis), ↓ CO2, ↓ temperature, ↓ 2,3-DPG, COHb (carbon monoxide), methaemoglobin, fetal Hb |
| Tissue O2 delivery | Improved | IMPAIRED — even though the blood is "loaded" with O2, it won't let go |
| Clinical effect in CO | — | The haemoglobin that is NOT occupied by CO still clings to its oxygen → tissue hypoxia disproportionate to the COHb level |
Exam point: the P50 (PaO2 at which haemoglobin is 50% saturated) falls in CO poisoning. This is why CO-poisoned patients can have a "normal" oxygen content on paper but be profoundly tissue-hypoxic — the oxygen simply is not being released where it is needed (brain, heart). [1]
Diagnosis
Why pulse oximetry fails — the most-tested concept
Standard pulse oximetry uses two wavelengths of light (660 nm red and 940 nm infrared) to determine the ratio of oxyhaemoglobin to deoxyhaemoglobin. COHb absorbs light at 660 nm almost identically to oxyhaemoglobin — the pulse oximeter "sees" COHb as if it were oxygenated haemoglobin. The result: the SpO2 reading is falsely normal and may read 98-100% even with a lethal COHb level.[1][4]
The fix: a multi-wavelength (8+ wavelength) CO-oximeter on a venous or arterial blood gas directly measures the different haemoglobin species (O2Hb, COHb, metHb, HHb) and reports a fractional SaO2 and a specific COHb percentage. [1]
Pulse oximetry (SpO2) vs CO-oximetry (COHb)
| Feature | Standard pulse oximetry (SpO2) | CO-oximetry (on blood gas) |
|---|---|---|
| Principle | 2 wavelengths (660/940 nm) | 8+ wavelengths (>100 wavelengths in modern devices) |
| Detects COHb? | NO — counts COHb as if it were O2Hb | YES — directly measures COHb |
| Reading in CO poisoning | FALSELY NORMAL (reads ~98% even when COHb lethal) | TRUE — reports the actual COHb% and functional SaO2 |
| Use in suspected CO | Useless for diagnosis — do NOT be reassured | ESSENTIAL — the only bedside test that confirms CO poisoning |
| Also detects | Nothing else reliably | Methaemoglobin, sulphaemoglobin, fetal Hb |
Note on PaO2: the PaO2 on a blood gas (dissolved oxygen) is normal in CO poisoning because dissolved oxygen is unaffected by COHb. A normal PaO2 with a high lactate and altered consciousness is a classic exam vignette for CO poisoning. [1]
COHb levels and clinical correlation
Clinical features by COHb
Severity gradient
- COHb <10%: usually asymptomatic (smokers may have 5-10% baseline)
- COHb 10-20%: headache, dyspnoea on exertion
- COHb 20-30%: throbbing headache, irritability, nausea, impaired judgement
- COHb 30-40%: severe headache, nausea, vomiting, confusion, syncope
- COHb 40-50%: coma, seizures
- COHb >50%: cardiorespiratory arrest, death
Investigation
Diagnostic
- COHb level: venous or arterial blood gas (NOT pulse oximetry — SpO2 falsely normal)
- Lactate: elevated (anaerobic metabolism from tissue hypoxia) — severe acidosis without explanation suggests cyanide
- ECG: ischaemia, arrhythmias (CO causes myocardial stunning)
- Troponin: may be elevated (myocardial injury)
- CK: rhabdomyolysis
- ABG: metabolic acidosis (high lactate), PaO2 usually NORMAL
- Pregnancy test: if positive, lower threshold for HBO
- Creatinine, electrolytes, glucose, lipase (pancreatitis reported)
- CK-MB / echocardiogram if cardiac involvement
Caveat — COHb level does NOT always correlate with severity
The COHb concentration at presentation depends on (a) the total CO dose inhaled, (b) the time since removal from exposure, and (c) the amount of oxygen given before the blood gas. By the time the patient reaches hospital, a patient who was deeply unconscious at the scene may have a COHb that has already fallen substantially. A "low" COHb in an obtunded patient does NOT exclude severe poisoning. Treat the patient, the exposure history, and the clinical picture — never the number in isolation.[5]
This is also why pre-hospital oxygen can mask severity: 100% oxygen halves the COHb every 80 minutes, so a patient ventilated on scene for 2 hours will have had two half-lives of clearance before arrival. [1]
Classic examination and historical clues
Symptoms and signs — the 'flu-like' masquerade
| System | Features |
|---|---|
| Constitutional | Headache (most common — ~90%), malaise, flu-like symptoms, fever (uncommon) |
| Neurological (early) | Dizziness, confusion, impaired judgement (the 'wrong decisions' sign), irritability, blurred vision |
| Neurological (severe) | Syncope, seizures, coma — any loss of consciousness = severe |
| Cardiovascular | Chest pain, palpitations; ECG ischaemia/arrhythmia; myocardial stunning; troponin rise |
| Gastrointestinal | Nausea, vomiting, abdominal pain (often mislabelled gastroenteritis) |
| Respiratory | Dyspnoea, tachypnoea (from acidosis); pulmonary oedema in severe cases |
| Skin | Cherry-red skin — CLASSICAL but RARE and LATE; its absence never excludes CO poisoning |
| Renal/muscle | Rhabdomyolysis, acute kidney injury (myoglobinuria) |
| Historical clue | 'Flu-like illness in MULTIPLE household members in WINTER, improving on leaving the house, with no fever' = CO |
The 'cherry-red skin' trap
Cherry-red (or "brick-red") discolouration of the skin, mucous membranes, and nail beds is the classical textbook sign of CO poisoning. The mechanism is the bright red colour of COHb itself (carboxyhaemoglobin is a brighter, cherry red than oxyhaemoglobin) plus tissue hypoxia preventing deoxygenation. However it is RARELY seen clinically — it typically appears only at very high COHb levels (>40%), in deeply comatose or dead patients, and is easily missed in low ambient light or in patients with darker skin. You will fail more patients by waiting for cherry-red skin than by any other error. Its absence does not exclude — and is not reassuring in — severe CO poisoning. [1]
Sources of carbon monoxide — taking a targeted history
CO is produced by incomplete combustion of any carbon-containing fuel in a setting with insufficient oxygen. The history must specifically probe: [1]
Sources of CO to ask about
| Category | Specific sources |
|---|---|
| Domestic heating | Faulty/malfunctioning gas, oil, wood, or coal heaters; blocked flues/chimneys; back-drafting from furnaces; gas water heaters |
| Cooking | Gas stoves/ovens used for heating; barbecues/charcoal burners used indoors (a classic cause, especially in power outages) |
| Vehicles | Car/motorbike exhaust in enclosed spaces (garages), petrol generators indoors |
| Fires | House fires, smoke inhalation (ALWAYS co-consider cyanide from burning plastics/wool) |
| Occupational | Forklifts (propane), paint strippers containing methylene chloride (metabolised to CO in vivo), welders, fire-fighters |
| Recreational | Boats with engine exhaust, indoor camping stoves, ice-fishing huts with heaters |
Methylene chloride (dichloromethane) — a special case: this solvent (in paint strippers) is hepatically metabolised to carbon monoxide, producing COHb levels that persist for hours-to-days (longer than inhaled CO, because generation is ongoing). Suspect it in painters/decorators presenting with COHb but no obvious inhalational source. The treatment is the same (100% oxygen) but clearance is slower. [1]
Public health action: a confirmed case of CO poisoning is a notifiable event — the source MUST be identified and fixed (or the dwelling evacuated) to prevent re-exposure of the patient and other household members. Notify public health / environmental health, and recommend CO alarms. [1]
Treatment
CO poisoning management
Remove from source + 100% oxygen
Remove patient from CO source immediately. Give 100% oxygen via non-rebreather mask (15 L/min). Continue until COHb <5% (or <10% in pregnant patients). Oxygen competes with CO for haemoglobin binding — 100% O2 reduces CO half-life from 320 min (room air) to 80 min.
Assess severity — consider hyperbaric oxygen (HBO)
HBO indications (controversial): COHb >25% (or >20% in pregnancy), loss of consciousness, neurological signs, cardiac ischaemia/arrhythmia, pregnancy. HBO at 2.5-3 atm reduces CO half-life to 23 min and increases dissolved O2 in plasma. Cochrane review: uncertain benefit for preventing delayed encephalopathy.<Cite id="2" />
Supportive care
ABCDE. Intubate if GCS <8 or cannot protect airway. Treat seizures (benzodiazepines). Treat acidosis (improve oxygenation — do NOT routinely give bicarbonate). Monitor cardiac rhythm (arrhythmia risk). Check troponin (myocardial injury). Treat rhabdomyolysis (IV fluids, monitor CK).
Monitor for delayed neurological sequelae
20-40% of severely poisoned patients develop delayed neurological sequelae at 2-40 days: cognitive decline, personality change, parkinsonism, incontinence. May be permanent. No proven prevention (HBO may reduce incidence but evidence is conflicting). Follow-up at 4-6 weeks with cognitive assessment.
Oxygen — the pharmacology of half-life reduction
The half-life of COHb depends on the inspired oxygen concentration and, when used, the ambient pressure: [1]
COHb half-life under different oxygen regimens
| Regimen | COHb half-life | Mechanism |
|---|---|---|
| Room air (21% O2) | ~320 min (5.3 hours) | Slow displacement — CO only gradually out-competed by ambient O2 |
| 100% normobaric oxygen (NBO) | ~80 min | Mass-action: high alveolar O2 accelerates dissociation of CO from Hb |
| 100% O2 at 2.5-3 atm (HBO) | ~23 min | Pressure + high O2 concentration — and O2 dissolved in plasma meets tissue demand independent of Hb |
This is the pharmacological core of CO management: give the highest concentration of oxygen you can, as early as you can, for as long as it takes. There is no role for "low-flow" oxygen or for weaning to room air before COHb is <5%. Continue 100% oxygen until COHb is <5% (or <10% in pregnancy — because fetal COHb clears even more slowly). [1]
Why dissolved oxygen matters at hyperbaric pressure: at 3 atm breathing 100% O2, enough oxygen dissolves in plasma (~6 mL O2/dL) to meet basal tissue demand with no haemoglobin at all (basal consumption ~5 mL O2/dL/min). This is the rationale for using HBO in critically CO-poisoned, COHb-saturated patients: the dissolved O2 bridges the gap until COHb is cleared. [1]
Hyperbaric oxygen (HBO) — indications, mechanism and the evidence controversy
HBO therapy involves placing the patient in a chamber pressurised to typically 2.5-3 atmospheres absolute (ATA) breathing 100% oxygen. Beyond accelerating CO clearance, HBO is proposed to (a) reduce tissue oedema, (b) inhibit neutrophil β2-integrin adhesion (reducing the inflammatory cascade linked to delayed neurological sequelae), and (c) restore cytochrome oxidase function faster.[3][4]
Accepted indications for HBO
HBO indications — the generally accepted list
| Indication | Threshold / criterion | Rationale |
|---|---|---|
| Loss of consciousness | Any transient or sustained LOC at any time | Strongest single indication — predicts delayed neurological sequelae |
| Neurological signs | Confusion, seizures, focal deficit, coma | Brain is the organ most vulnerable to delayed injury |
| Cardiac ischaemia | Ischaemic chest pain, ECG changes, arrhythmia, troponin rise | Myocardial stunning; CO-poisoned patients have higher long-term cardiac mortality |
| COHb level | >25% (some centres >20%); >15% in pregnancy | Surrogate for total body CO load |
| Pregnancy | Any symptomatic pregnant patient, or COHb >15% | Fetal Hb binds CO more avidly; fetus is poisoned at lower maternal COHb |
| Persistent symptoms | Symptoms not resolving with normobaric O2 | Failure to respond to standard therapy |
| High-risk features | Age >36, exposure >24 h, metabolic acidosis (pH <7.1) | Predictors of poor outcome |
The evidence — why HBO remains controversial
Despite four decades of randomised trials, the benefit of HBO over high-flow normobaric oxygen for preventing delayed neurological sequelae remains genuinely uncertain. The Cochrane review concludes that the existing evidence does not reliably establish benefit, and that additional well-designed trials are needed.[2]
The key HBO trials — conflicting results
| Trial | Design | Result |
|---|---|---|
| Weaver 2002 (NEJM, PMID 12362006)[3] | 152 patients; 3 HBO sessions vs normobaric O2 (sham chamber) | HBO REDUCED cognitive sequelae at 6 weeks (25% vs 46%) and 12 months — the most-cited positive trial |
| Scheinkestel 1999 (MJA, PMID 10092916)[6] | 191 patients; up to 3 HBO sessions vs normobaric O2 | NO benefit — HBO arm trended to WORSE outcome; major Australian trial, the main counterweight to Weaver |
| Annane 2003 | RCT | No benefit of HBO |
| Buckley 2023 Cochrane (PMID 36907456)[2] | Systematic review + meta-analysis | Evidence of benefit remains uncertain; risk of bias and heterogeneity limit firm conclusions |
Practical HBO decision-making
HBO is logistically demanding (the nearest chamber may be hours away; transport of a critically ill patient adds risk; the chamber is an austere monitoring environment). Decision-making should weigh: [1]
- Severity — strongest case is the unconscious patient with high COHb, neurological signs, or cardiac ischaemia.
- Time since exposure — the earlier HBO is given (ideally within 6 hours), the greater the theoretical benefit.
- Transport risk — a ventilated, unstable patient may be harmed by the move. Discuss with the receiving hyperbaric unit and retrieval service.
- Pregnancy — strong bias toward HBO given fetal vulnerability.
- Recovery on normobaric O2 — if the patient is neurologically normal and symptoms resolve on 100% NBO, the case for HBO is weaker. [1]
Bottom line for the exam: 100% oxygen is non-negotiable and universally beneficial; HBO is reasonable and often recommended for severe poisoning but its benefit on long-term neurological outcome is unproven in definitive trials. State both halves of this answer. [1]
Delayed neurological sequelae (DNS) — the feared late complication
Delayed neurological/neuropsychiatric sequelae (DNS) occur in ~20-40% of severely CO-poisoned patients (often quoted as 10-30%, more recent data up to 40%), appearing after a lucid interval of 2-40 days (most commonly 2-28 days) following apparent recovery.[3][4]
Clinical features of DNS
Spectrum of delayed neurological sequelae
| Domain | Manifestations |
|---|---|
| Cognitive | Impairment in memory, attention, executive function, processing speed; "carbon monoxide encephalopathy" |
| Neuropsychiatric | Depression, anxiety, personality change, irritability, psychosis-like symptoms |
| Motor / movement | Parkinsonism (bradykinesia, rigidity, gait disturbance — typically with basal ganglia injury, especially globus pallidus), tremor, dystonia, incontinence |
| Cerebellar | Ataxia |
| Cortical | Aphasia, apraxia, agnosia, visuospatial impairment |
| Autonomic / sphincter | Urinary/faecal incontinence |
Mechanism and imaging
The mechanisms underlying DNS include: white-matter demyelination, globus pallidus necrosis (the classic radiological finding), lipid peroxidation from CO-induced free-radical generation, and neutrophil-mediated inflammation (β2-integrin–mediated adherence to vascular endothelium, which HBO is theorised to inhibit). [1]
Imaging clues:
- CT/MRI: bilateral globus pallidus injury (low attenuation on CT, T2/FLAIR hyperintensity on MRI) is characteristic but not universal.
- Diffuse white-matter changes on MRI correlate with cognitive sequelae.
- A normal early MRI does not exclude DNS, which often declares itself weeks later. [1]
Natural history and management
Assessing and managing delayed neurological sequelae
- Warn every patient: at discharge, counsel that new neurological or psychiatric symptoms over the next 6 weeks require review — DNS is common, not rare.
- Follow-up at 4-6 weeks: structured cognitive screening (e.g. Mini-Mental State, Montreal Cognitive Assessment, or formal neuropsychometric testing in symptomatic patients).
- No proven prevention: normobaric oxygen is essential; HBO may reduce incidence (Weaver) but evidence is conflicting (Scheinkestel, Cochrane).
- Partial recovery: 50-75% of DNS patients recover substantially over 12 months, but a significant fraction have permanent cognitive or parkinsonian deficits.
- Rehabilitation: cognitive rehabilitation, physiotherapy and occupational therapy for persistent deficits.
Pregnancy and CO poisoning
Pregnancy uniquely compounds CO toxicity because the fetus is a deep sink for CO: [1]
- Fetal haemoglobin binds CO more avidly than adult haemoglobin — fetal COHb reaches levels 10-15% higher than maternal.
- Fetal COHb clears 3-5x more slowly than maternal — the fetus remains poisoned long after the mother appears well.
- The fetal oxyhaemoglobin curve is already left-shifted (fetal Hb) — CO pushes it further left, devastating fetal tissue oxygen delivery.
- Outcomes: increased miscarriage, fetal demise, fetal neurological injury, low birth weight, congenital malformation. [1]
Management differences in pregnancy
| Element | Approach |
|---|---|
| Diagnosis | Lower threshold to test COHb in any pregnant woman with relevant symptoms or exposure |
| Normobaric O2 | 100% oxygen — continue longer until COHb <10% (not <5%), because fetal clearance is slow |
| HBO threshold | Lower — COHb >15%, OR any neurological symptoms, OR loss of consciousness (some centres treat any symptomatic pregnant patient) |
| HBO safety | Generally considered safe in pregnancy; no evidence of fetal harm from the pressure itself. Involve obstetrics. |
| Fetal monitoring | Continuous CTG once viable; recognise fetal distress may lag maternal recovery |
| Delivery | Not for CO poisoning per se — deliver only for obstetric indications or non-reassuring fetal status |
Exam rule: in the pregnant CO-poisoned patient, treat the mother aggressively and the fetus benefits — do not withhold HBO because of pregnancy. [1]
House fires and cyanide co-toxicity
Patients rescued from enclosed-space fires (house, industrial, vehicle) may have concomitant cyanide toxicity from combustion of nitrogen-containing materials (plastics, wool, silk, polyurethane). Cyanide and CO produce additive cellular hypoxia by converging on the same target — cytochrome c oxidase (Complex IV). [1]
Recognising combined CO + cyanide toxicity:
- Lactic acidosis out of proportion to the measured COHb level (a COHb of 25% with a lactate of 12 mmol/L = think cyanide).
- Haemodynamic instability / profound shock out of proportion to COHb.
- Soot in the oropharynx / sputum, or a history of entrapment in an enclosed fire. [1]
Management of suspected combined toxicity:
- Continue 100% oxygen (treats the CO component and is the first step for cyanide too).
- Give hydroxocobalamin 5 g IV empirically (preferred cyanide antidote — does not interfere with tissue oxygen utilisation, unlike sodium thiosulphate/dicobalt edetate, and is safe to give before a confirmed cyanide level).
- Do NOT wait for a cyanide level — levels take hours and the patient may die in the interim. [1]
Differential diagnosis
The non-specific, 'flu-like' presentation of mild-moderate CO poisoning generates a broad differential. The keys to CO are the cluster of household members, the winter/heating context, the falsely normal SpO2, and the lactate/acidosis. [1]
Differential diagnosis of CO poisoning
| Condition | Discriminator from CO |
|---|---|
| Viral influenza / gastroenteritis | Fever, no household clustering pattern, no heating exposure, normal lactate/COHb |
| Sepsis | Fever, focus of infection, positive cultures, WCC/CRP, no response to leaving the house |
| Meningitis/encephalitis | Fever, neck stiffness, photophobia; CSF analysis |
| Other toxin — cyanide | Smoke/fire exposure, severe lactate disproportionate to COHb |
| Methaemoglobinaemia | Cyanosis NOT responsive to O2; chocolate-brown blood; CO-oximetry distinguishes (metHb vs COHb) |
| Hypoglycaemia | Glucose check resolves it; occurs faster |
| Alcohol intoxication / withdrawal | History, ethanol level, CIWA features |
| Carbon dioxide narcosis | Hypoventilator; raised PaCO2 |
| Primary psychiatric / functional | Diagnosis of exclusion; always exclude COHb first |
Paediatric considerations
Children are more vulnerable to CO for several reasons: higher metabolic rate and oxygen consumption per kilogram, higher baseline respiratory rate (greater inhaled dose), and a developing nervous system more susceptible to hypoxic injury. Clinical pearls: [1]
- Presentation is non-specific: irritability, refusal to feed, vomiting, lethargy — easily attributed to gastroenteritis or viral illness.
- Higher rates of syncope and seizures at any given COHb.
- Lower threshold for HBO in symptomatic children (consult a paediatric hyperbaric unit).
- Consider non-accidental exposure: a parent deliberately exposing a child, or a generator used indoors in neglected housing — safeguarding referral may be appropriate. [1]
Worked clinical scenario — the classic exam stem
Presentation: A 34-year-old woman is brought to the ED in winter with her partner and two children. All four have headache, nausea, and dizziness that have worsened over the evening. The heating "smells funny". She is confused; one child has had a brief seizure. [1]
Step 1 — Suspect CO: four household members with simultaneous 'flu-like' illness in winter, with confusion and a seizure — this is CO poisoning until proven otherwise. Remove all from the source. [1]
Step 2 — Immediate management: 100% oxygen via non-rebreather to ALL FOUR patients. ABC. IV access. Blood gas with CO-oximetry on each. [1]
Step 3 — Blood gas (the mother): COHb 38%, lactate 6 mmol/L, pH 7.28, PaO2 95 mmHg (NORMAL), SpO2 99% (FALSELY NORMAL — do not be reassured). ECG: sinus tachycardia with anterior T-wave inversion. [1]
Step 4 — Severity: COHb 38% + confusion + seizure + ECG changes = SEVERE. This meets multiple HBO criteria (LOC/seizure, neurological signs, cardiac ischaemia, COHb >25%). [1]
Step 5 — Treatment: continue 100% O2; discuss urgently with the hyperbaric unit and retrieval; check βhCG in the mother (and any female of childbearing age); screen for co-toxicities (salicylate, paracetamol, ethanol — intentional co-ingestion less likely here but rule out). If a house fire was the source, add hydroxocobalamin empirically for suspected cyanide. [1]
Step 6 — Source/public health: alert environmental health; the dwelling must be evacuated and inspected before anyone returns; recommend CO alarms. [1]
Step 7 — Follow-up: arrange cognitive assessment at 4-6 weeks for all four (especially the child who seized) — counsel about the 2-40 day window for delayed neurological sequelae. [1]
Prognosis
Outcomes and risk factors in CO poisoning
| Factor | Effect on outcome |
|---|---|
| Mild poisoning (COHb <20%), prompt treatment | Excellent — full recovery usual |
| Loss of consciousness at any time | Risk of delayed neurological sequelae ~40-50% |
| COHb >25% | Higher risk of cardiac and neurological sequelae |
| Age >36 | Increased risk of DNS |
| Exposure >24 h (chronic/intermittent) | Worse — tissue saturation, delayed diagnosis |
| Metabolic acidosis (pH <7.1) | Marker of severe poisoning, worse prognosis |
| Cardiac injury (troponin rise, ECG changes) | Increased short-term mortality and long-term cardiac mortality |
| MRI globus pallidus / white-matter injury | Predicts cognitive sequelae |
| Pregnancy | Fetal mortality up to 40-60% in severe maternal poisoning |
Long-term mortality: CO-poisoned patients with myocardial injury have significantly higher long-term mortality (cardiac and all-cause), attributed to CO-induced cardiac injury. This justifies ECG and troponin in every patient and cardiac follow-up. [1]
Comparison — CO vs cyanide (the two smoke-inhalation toxins)
CO poisoning vs cyanide poisoning
| Feature | Carbon monoxide | Cyanide |
|---|---|---|
| Source | Incomplete combustion of fuel (heaters, exhaust, fires) | Combustion of plastics/wool/silk; industrial; nitroprusside |
| Diagnostic marker | COHb on CO-oximetry | Lactate (often >10 mmol/L); whole-blood cyanide level (slow) |
| Pulse oximetry | Falsely normal (COHb reads as O2Hb) | Usually normal |
| Core mechanism | Binds Hb (240x O2) + inhibits cytochrome oxidase | Inhibits cytochrome oxidase (Complex IV) directly |
| Venous blood | Often bright red (arterialisation — O2 not extracted) | Bright red venous blood (O2 not utilised) |
| Treatment | 100% O2 ± HBO | Hydroxocobalamin (or sodium thiosulphate / dicobalt edetate) + 100% O2 |
| Coexistence | Common in house-fire victims — assume both, treat both empirically | Common in house-fire victims — assume both, treat both empirically |
Exam rule for smoke inhalation: a fire victim with raised lactate disproportionate to COHb, haemodynamic instability, or soot in the airway = treat empirically for cyanide with hydroxocobalamin, while continuing 100% oxygen for the CO component. Do not wait for confirmatory levels. [1]
Key clinical points
SAQ — Household carbon monoxide poisoning in winter
10 minutes · 10 marks
In midwinter a 34-year-old woman, her partner and two children present with worsening headache, nausea and dizziness over the evening. The gas heater 'smells funny'. The mother is confused; one child has had a brief generalised seizure. SpO₂ on room air reads 99% in all four. The mother's venous blood gas shows COHb 38%, lactate 6 mmol/L, pH 7.28, PaO₂ 95 mmHg. ECG shows sinus tachycardia with anterior T-wave inversion.
SAQ — House-fire victim with combined CO and cyanide toxicity
10 minutes · 10 marks
A 22-year-old man is rescued unconscious from an enclosed house fire. He has soot in his oropharynx and sputum, GCS 6, HR 124, BP 88/52, lactate 14 mmol/L, COHb 26%, SpO₂ 99%, pH 7.18, and is being intubated. The team asks whether to wait for a cyanide level before giving an antidote.
Red flags
Key trials and evidence
Weaver 2002 — HBO for acute CO poisoning (PMID 12362006)
Source
N Engl J Med — the most-cited positive HBO trial
Design
RCT, 152 patients; 3 hyperbaric oxygen sessions vs normobaric oxygen (sham chamber)
Key result
HBO REDUCED cognitive sequelae at 6 weeks (25% vs 46%) and 12 months
Caveat
Small sample, single-centre, sham-control imperfectly blinded; others failed to replicate
Clinical bottom line
Provides the main evidence FOR HBO in severe CO poisoning
Scheinkestel 1999 — HBO vs normobaric O2, Australia (PMID 10092916)
Source
Med J Aust — the major counterweight trial (especially relevant in ANZ)
Design
RCT, 191 patients; up to 3 daily HBO sessions vs normobaric 100% O2
Key result
NO benefit of HBO; a trend toward WORSE neurological outcome in the HBO arm
Caveat
Used a higher number of HBO sessions; questions about equipoise and transport
Clinical bottom line
Provides the main evidence AGAINST routine HBO — underpins the controversy
Buckley 2023 — Cochrane review of HBO for CO poisoning (PMID 36907456)
Source
Cochrane Database of Systematic Reviews — the definitive synthesis
Key conclusion
Benefit of HBO for preventing neurological sequelae remains UNCERTAIN — risk of bias and heterogeneity limit firm conclusions
Recommendation
Further well-designed RCTs needed; decision should be individualised
Clinical bottom line
HBO is reasonable for severe poisoning but its benefit is unproven; 100% O2 is non-negotiable
Hampson 2012 — Expert practice recommendations (PMID 23087025)
Source
Am J Respir Crit Care Med — multi-society consensus
Key point 1
COHb by CO-oximetry is the diagnostic standard; SpO2 is falsely normal
Key point 2
100% oxygen is the cornerstone (half-life 320→80 min)
Key point 3
HBO at 2.5-3 ATA reduces half-life to ~23 min and dissolves enough O2 to meet tissue demand
Key point 4
Acceptable HBO indications: LOC, neurological signs, cardiac ischaemia, COHb >25%, pregnancy
Clinical bottom line
The practical bedside reference for diagnosis, management, prevention and surveillance
Quick-reference ICU summary
CO poisoning — 60-second resuscitation
- ABC + 100% oxygen via non-rebreather to every patient. Remove from source. Monitor ECG, SpO2 (but do not trust it), BP, GCS.
- Blood gas with CO-oximetry (venous is adequate): COHb, lactate, pH, PaO2, glucose, electrolytes. Send βhCG in women of childbearing age. ECG + troponin. CK.
- Assess severity — COHb, GCS/LOC, seizures, ECG changes, lactate, pregnancy status, exposure history.
- Continue 100% O2 until COHb <5% (or <10% in pregnancy).
- If severe (LOC, seizures, neurological signs, cardiac ischaemia, COHb >25%, or pregnant with COHb >15%) — discuss hyperbaric oxygen with the hyperbaric unit + retrieval; continue 100% O2 during transport.
- House fire / enclosed combustion with lactate disproportionate to COHb — give hydroxocobalamin 5 g IV empirically for suspected cyanide.
- Supportive care — intubate if GCS <8; benzodiazepines for seizures; fluids for rhabdomyolysis; treat cardiac ischaemia/arrhythmia.
- Public health: notify; evacuate/inspect the dwelling; recommend CO alarms.
- Follow-up at 4-6 weeks: cognitive screening; counsel every patient about the 2-40 day window for delayed neurological sequelae.
Prevention and public health
Prevention of CO poisoning
| Measure | Detail |
|---|---|
| CO alarms | Install near sleeping areas and on every level; battery-backed; replace every 5-7 years. Combine with smoke alarms. |
| Heating appliance servicing | Annual professional servicing of gas/oil/wood/coal heaters, water heaters, furnaces, flues and chimneys |
| Never use indoors | Barbecues, charcoal burners, camp stoves, generators, petrol-powered tools (all should be outdoors, well-ventilated) |
| Vehicle exhaust | Never run a vehicle in an enclosed garage; check exhaust systems; keyless-ignition vehicles — check the engine is OFF |
| Ventilation | Ensure adequate ventilation when using gas appliances; do not block vents |
| Occupational | Methylene chloride (paint strippers) — substitute safer products; ventilation; CO monitoring |
| Notification | A confirmed case is a notifiable event — trigger public health and environmental health investigation to prevent re-exposure of the household and identify clusters |
References
- [1]Rose JJ, Wang L, Xu Q, et al. Characteristics of Human Peripheral Blood γδ T Cells Expanded With Zoledronate Anticancer Res, 2021.PMID 34848457
- [2]Buckley NA, Wolff K, Sanderson L, et al. Assessing vaccine literacy and exploring its association with vaccine hesitancy: A validation of the vaccine literacy scale in China J Affect Disord, 2023.PMID 36907456
- [3]Weaver LK, Hopkins RO, Chan KJ, et al. Hyperbaric oxygen for acute carbon monoxide poisoning N Engl J Med, 2002.PMID 12362006
- [4]Hampson NB, Piantadosi CA, Thom SR, Weaver LK. Practice recommendations in the diagnosis, management, and prevention of carbon monoxide poisoning Am J Respir Crit Care Med, 2012.PMID 23087025
- [5]Hampson NB, Hauff NM. Carboxyhemoglobin levels in carbon monoxide poisoning: do they correlate with the clinical picture? Am J Emerg Med, 2008.PMID 18606318
- [6]Scheinkestel CD, Bailey M, Myles PS, et al. Hyperbaric or normobaric oxygen for acute carbon monoxide poisoning: a randomised controlled clinical trial Med J Aust, 1999.PMID 10092916