Carbon Monoxide Poisoning

Carbon monoxide (CO) poisoning is a toxicologic emergency caused by inhaling a colourless, odourless, non-irritating gas produced whenever a carbon-based fuel burns incompletely — a gas that is, by design, impossible to detect without an alarm, which is exactly why recognition here depends entirely on clinical suspicion rather than on how a patient or their environment "seems."^[1] CO is absorbed through the lungs and binds haemoglobin with roughly 200-250 times the affinity of oxygen, forming carboxyhaemoglobin (COHb) and sharply reducing the blood's oxygen-carrying capacity; it also left-shifts the oxyhaemoglobin dissociation curve, impairing the release of whatever oxygen the blood is still carrying to the tissues.^[1] CO does not stop at haemoglobin — it also binds myoglobin and, directly and independently of any haemoglobin effect, the mitochondrial enzyme cytochrome c oxidase, poisoning cellular respiration itself. The heart and brain, the two most oxygen-hungry organs, bear the brunt of this combined carrying-capacity-plus-cellular-poisoning injury.^[2][1]

The oxyhaemoglobin dissociation curve. Carbon monoxide left-shifts this curve — impairing release of the oxygen still bound to haemoglobin — on top of its far larger effect of simply displacing oxygen from haemoglobin outright. (Ratznium/Aaronsharpe, public domain, Wikimedia Commons)

The single most dangerous pitfall in this chapter is not clinical, it is instrumental: a standard pulse oximeter cannot distinguish carboxyhaemoglobin from oxyhaemoglobin and will report a falsely normal or near-normal SpO₂ in a patient who is, in reality, profoundly hypoxic at the tissue level; a routine arterial blood gas is equally misleading, since its reported oxygen saturation is typically calculated rather than directly measured.^[1] The only way to actually detect CO poisoning is to specifically order co-oximetry for a directly measured carboxyhaemoglobin — a test that must be requested by name, because nothing about a normal-looking pulse oximeter or a normal-looking arterial blood gas will ever prompt it on its own.

In Canada, CO poisoning causes an estimated 300 deaths and 200 hospitalizations every year, with almost two-thirds of emergency department presentations occurring in the fall and winter and close to a quarter involving children under 4.^[3] Faulty furnaces, portable generators run indoors or in an attached garage, poorly ventilated space heaters, and house fires are the classic, recurring Canadian winter sources.^[3][2] Management is emergency-department- and, for more severe cases, ICU- or hyperbaric-unit-level care, never an outpatient decision — the pharmacist's core role is rapid recognition (especially the vague, flu-mimicking presentation and household clusters), getting 100% oxygen started immediately, and knowing exactly which patients need urgent hyperbaric oxygen referral.

The classic — and classically missed — presentation

Acute CO poisoning most often begins with a vague, nonspecific triad of headache, nausea (with or without vomiting), and dizziness or generalized malaise — a picture that is genuinely difficult to distinguish from a viral illness on history alone, and is exactly why CO poisoning is so often missed, especially during cold-and-flu season when both conditions are circulating.^[1] As exposure and COHb rise, the picture progresses to confusion and altered mental status, chest pain from myocardial ischemia, dyspnea, syncope, seizures, and coma.^[1] "Cherry red" skin is taught as the classic sign of CO poisoning, but in practice it is rare, appears late, and is unreliable at the bedside — it is described far more often as a postmortem finding than as something a clinician can count on to make the diagnosis in a living patient. Do not wait for it.^[1]

The clue that actually catches it: the household cluster

Chronic, low-level exposure

Chronic or intermittent low-level exposure — a poorly tuned furnace, a leaking chimney — produces a subtler picture: recurring headache, fatigue, and cognitive complaints that are easy to misattribute to depression, migraine, or a recurring viral syndrome across several visits before CO is ever considered. A history of symptoms that consistently track with time spent at home, or that improve on vacation, should prompt specific questions about home heating and appliance maintenance.

• Faulty, poorly maintained, or improperly vented furnaces, water heaters, and other fuel-burning household appliances.
• Portable generators run indoors, in a garage (even with the door open), or near a window or door — a classic, recurring source during winter storm-related power outages.^[3]
• Poorly ventilated or malfunctioning space heaters, wood stoves, and fireplaces.
• Motor vehicle exhaust in an attached or enclosed garage.
• House fires and smoke inhalation — the most common cause of fire-related death.^[2]
• Charcoal briquettes or a propane/gas barbecue used indoors or in an enclosed space such as a tent.^[3]
• Occupational exposure — indoor combustion-engine equipment (e.g. forklifts), welding, firefighting.
• Hookah/shisha use in a poorly ventilated space.
• No functioning, correctly certified carbon monoxide alarm in the home.^[4]
• Pre-existing cardiac or cerebrovascular disease, chronic anemia, pregnancy, and the extremes of age all increase the risk of serious harm at a given level of exposure, even though they do not increase the risk of exposure itself.

The test that actually makes the diagnosis

Carbon monoxide poisoning is diagnosed by directly measuring carboxyhaemoglobin with co-oximetry on a venous or arterial blood sample — a specific test that must be ordered by name. A standard pulse oximeter and a standard (non-co-oximetry) arterial blood gas are both unreliable here: pulse oximetry cannot distinguish COHb from oxyhaemoglobin, and a routine ABG's reported oxygen saturation is typically calculated rather than measured, so both can look reassuringly normal in a genuinely poisoned patient.^[1] Baseline COHb is roughly 1-3% in non-smokers and can run as high as 10-15% in heavy smokers — worth confirming smoking status for when interpreting a borderline result, though this should never be used to talk yourself out of treating a symptomatic patient with a convincing exposure history.

Supporting investigations

• ECG and high-sensitivity troponin in essentially every symptomatic patient — myocardial injury from CO poisoning is common and is often clinically silent until specifically looked for.^[2]
• Venous or arterial lactate — a markedly elevated lactate in a smoke-inhalation victim, especially one out of proportion to the COHb level, should raise concern for concomitant cyanide toxicity.
• Pregnancy test in every patient of reproductive age — it changes the oxygen-duration and hyperbaric-referral thresholds (see Special Populations).
• Head CT if there is a focal neurologic deficit, prolonged or unexplained altered mental status, or diagnostic uncertainty.
• Trend COHb serially once oxygen therapy has started rather than relying on a single value — COHb falls quickly on high-flow oxygen, so a normal repeat level reflects treatment working, not that the original exposure was mild.

Differential diagnosis

Because the classic presentation mimics a viral illness so closely, actively consider CO poisoning — rather than simply treating empirically for the more common alternative — in a patient with headache, nausea, and malaise, particularly with any of: symptom onset that clusters with time spent at home, more than one affected household member, a recent change in heating or generator use, or a picture that does not quite fit a documented local viral illness pattern. Migraine, acute coronary syndrome, viral gastroenteritis, and viral meningitis/encephalitis are the most common alternative diagnoses actually being excluded when the differential is not obvious.

• Recognize CO poisoning promptly, especially in its vague, flu-mimicking form and in a household-cluster presentation, rather than defaulting to a viral-illness diagnosis.
• Remove the patient — and any other affected occupants — from the source immediately, and do not allow re-entry until the source has been found and corrected.
• Start 100% oxygen immediately in any patient with a plausible exposure and consistent symptoms, before co-oximetry results are back, not after.
• Identify every patient who meets hyperbaric oxygen referral criteria and refer without delay — this is a genuinely time-sensitive decision.
• Screen for and manage cardiac injury, which is common and often clinically silent.
• Recognize and treat concomitant cyanide toxicity in smoke-inhalation victims with disproportionate shock, coma, or lactic acidosis.
• Apply the lower COHb and referral thresholds that specifically apply to pregnancy and to children, rather than the general adult criteria.
• Counsel every patient on the real possibility of delayed neurological/neuropsychiatric sequelae and arrange follow-up before discharge.
• Counsel on carbon monoxide alarm installation, certification, and maintenance as secondary prevention — for the patient and, where a household cluster is involved, for the whole affected household.

A patient who meets none of these criteria and is improving can continue on high-flow oxygen alone — see Non-Pharmacologic Treatment for oxygen-duration and discharge criteria, and Special Populations for the separate pregnancy and pediatric pathways.

Scene safety and source removal

Remove the patient — and every other occupant, including pets — from the source immediately, and involve the fire department or gas utility to identify and correct it before anyone is allowed to re-enter. This is not a detail to leave to the patient to sort out later: a patient discharged back into an uncorrected source will simply be re-poisoned.

Oxygen delivery

Deliver 100% oxygen by non-rebreather mask at minimum, escalating to high-flow nasal cannula, CPAP/BiPAP, or intubation for a patient who cannot protect their airway or maintain adequate ventilation. Continue until COHb has fallen to 3% or less and symptoms have resolved — in practice, this generally means a minimum of 6-8 hours of high-flow oxygen for a symptomatic adult before considering discharge.^[2] Oxygen sharply shortens the COHb half-life — from roughly 4-6 hours on room air to about 40-90 minutes on 100% oxygen — which is exactly why serial COHb, not a single value, is what actually tells you treatment is working.^[1]

Hyperbaric oxygen therapy

Hyperbaric oxygen further shortens the COHb half-life to roughly 15-30 minutes, and is believed to work through mechanisms that go well beyond simply clearing CO faster — increasing plasma-dissolved oxygen delivery, reducing CO binding to cytochrome c oxidase and myoglobin, and limiting the lipid-peroxidation/inflammatory cascade thought to drive delayed neurological injury.^[2][1] The best-known randomized trial protocol delivered 3 hyperbaric sessions at 2.8 atmospheres absolute — an initial 90-minute session followed by two 60-minute sessions on the following two days — though the exact number and timing of sessions is set by the treating hyperbaric unit's own protocol.^[5] Evidence for a mortality or long-term-outcome benefit from hyperbaric oxygen is genuinely mixed across the available randomized trials, but current North American practice still refers patients who meet the criteria above given the plausible benefit for delayed neurocognitive outcome and the low risk of a short, monitored hyperbaric course.^[5]

Supportive care in severe poisoning

• Continuous cardiac monitoring for any patient with cardiac involvement or significant exposure.
• Fever prevention/treatment — hyperthermia appears to worsen CO-related neurologic injury.
• Maintain adequate cardiac output and ventilation; manage seizures if they occur (see Pharmacologic Treatment Options).
• Evaluate every smoke-inhalation victim for concomitant cyanide toxicity, particularly with unexplained shock, coma, or a disproportionately elevated lactate — see Pharmacologic Treatment Options for empiric hydroxocobalamin.

Discharge readiness

A patient is ready for discharge once COHb is 3% or less, symptoms have resolved, any indicated cardiac workup is reassuring, and — critically — the source has actually been found and corrected and the home confirmed safe for re-entry. See Monitoring & Follow-Up for delayed-neurological-sequelae counseling and CO alarm counseling, both of which belong in this same discharge conversation, not as an afterthought.

Pregnancy

Pediatrics

Children use the same lower HBOT threshold as pregnancy — COHb over 15%, not 25% — reflecting their higher metabolic rate and oxygen consumption relative to body size.^[2] Presentation in young children is often nonspecific (irritability, poor feeding, lethargy) rather than a reported headache, which likely contributes to children under 4 accounting for close to a quarter of CO-related emergency department visits captured in Canadian surveillance data.^[3] Maintain a low threshold to test any young child with vague, non-localizing symptoms and any plausible household exposure history.

Older adults and pre-existing cardiac or cerebrovascular disease

The same COHb level causes more harm in a patient with reduced cardiac or cerebrovascular reserve — obtain an ECG and troponin routinely in this group even with modest symptoms, and hold a lower threshold for hospital-level observation.

Smokers

Baseline COHb runs higher in smokers (up to roughly 10-15%, versus roughly 1-3% in non-smokers), which can make a modestly elevated level look falsely reassuring — interpret the result in light of the clinical picture and exposure history, not the raw number in isolation, and never use smoking status alone to withhold treatment from a symptomatic patient.

Household and cluster presentations

Any suspected case should prompt an evaluation of every other occupant of the same building, including pets — pets can show signs before people do and are a genuinely useful early clue. A pharmacist who identifies this pattern in the community should treat it as an emergency for the whole household, not only for the person in front of them.

Continue oxygen and reassess until COHb is 3% or less and symptoms have resolved — trend COHb serially rather than relying on a single value, since it falls quickly once oxygen therapy starts.^[2] Repeat troponin and ECG before discharge for anyone with initial cardiac involvement.

Delayed neurological sequelae

Counsel every patient — this is not optional, and it is easy to forget once the acute presentation has resolved — that delayed neurological/neuropsychiatric sequelae occur in roughly a quarter to just under half of hospitalized or significantly symptomatic patients, typically emerging 2 to 40 days after apparent recovery, and can include memory and concentration problems, mood changes, movement disorders, and personality change.^[1][2] Arrange a follow-up contact and, where available, neuropsychological assessment at roughly 2-6 weeks, and give explicit return precautions for any new cognitive, mood, or motor symptom in that window rather than leaving the patient to decide on their own whether it is "just stress" from the original event.

Source correction and household safety

Confirm before discharge — not after — that the CO source has actually been identified and corrected and that the home has been cleared as safe for re-entry, typically by the fire department or gas utility.

Carbon monoxide alarm counseling

Recommend a correctly certified CO alarm — look for the CSA, cUL, ULC, or cETL mark — installed outside every sleeping area and on every level of the home, and confirm any existing alarm is actually certified and not simply a smoke alarm alone, since an uncertified device may fail to operate correctly.^[4] Alarms should be tested regularly and replaced on the manufacturer's schedule.^[4] Provide the regional poison control centre's number for any future concern — in Ontario, the Ontario Poison Centre is available 24 hours a day at 1-800-268-9017.^[7]

A certified residential carbon monoxide alarm — the only reliable way to detect CO before symptoms begin, since the gas itself cannot be seen, smelled, or tasted. (Tony Webster, CC BY 2.0, Wikimedia Commons)

1. Symptoms not improving after 4 hours of high-flow oxygen: this itself is a hyperbaric oxygen referral criterion — refer, rather than simply continuing oxygen and reassessing later.^[2]
2. COHb not declining as expected: confirm the oxygen delivery is actually effective (mask seal, flow rate, patient tolerance) before assuming ongoing endogenous production, and reconsider whether the source has genuinely been removed or corrected.
3. New cardiac or neurologic findings during observation: escalate immediately and re-evaluate for hyperbaric oxygen — do not wait for the next scheduled reassessment.
4. A smoke-inhalation patient with persistent shock, coma, or acidosis despite adequate oxygen and source control: treat empirically for concomitant cyanide toxicity with hydroxocobalamin rather than continuing to escalate CO-directed therapy alone.
5. A repeat presentation from the same household after being cleared to return home: the source was not actually corrected — re-involve the fire department/gas utility, and do not allow re-occupancy on the strength of the household's own reassurance alone.

• Trusting a normal pulse oximetry reading to exclude CO poisoning — standard pulse oximetry cannot distinguish carboxyhaemoglobin from oxyhaemoglobin.^[1]
• Relying on a routine arterial blood gas's calculated oxygen saturation instead of specifically ordering co-oximetry for a directly measured COHb.^[1]
• Waiting for "cherry red" skin as a diagnostic sign — it is rare, late, and unreliable, and is described far more often after death than in a patient you can still treat.
• Attributing a household cluster of vague, flu-like symptoms to a circulating viral illness without specifically considering CO, especially when symptoms improve away from home.
• Delaying 100% oxygen while waiting for co-oximetry confirmation, rather than starting it immediately on clinical suspicion.^[2]
• Applying the general adult COHb threshold (25%) to a pregnant patient or a child instead of the lower 15% threshold that actually applies to them.^[2]
• Discharging a pregnant patient after the standard adult oxygen course instead of the required minimum 24 hours with fetal assessment.^[2]
• Missing concomitant cyanide toxicity in a smoke-inhalation/fire victim presenting with shock or a disproportionately elevated lactate.
• Allowing a patient to return to a home where the CO source has not actually been confirmed corrected.
• Discharging without counseling on delayed neurological sequelae and without arranging follow-up.
• Treating CO alarm counseling as optional patient education rather than a core part of the discharge plan.

How to use this table: Doses are adult doses and require patient-specific confirmation, including pregnancy/pediatric and renal/hepatic adjustment where relevant. Select any drug name to open its complete Pharmasist Drug Database page. There is no pharmacologic antidote for carbon monoxide itself — oxygen, escalating to hyperbaric oxygen where indicated, is the actual treatment; the agents below are symptomatic/supportive care, or address the specific, real co-toxicity of cyanide in smoke-inhalation victims.

Oxygen Therapy

The immediate, first-line treatment for every patient with suspected or confirmed CO poisoning — start on clinical suspicion alone, never delayed for co-oximetry confirmation.

Cyanide Antidote (Suspected Concomitant Toxicity in Smoke Inhalation)

Reserve for a smoke-inhalation/fire victim with shock, coma, or lactic acidosis out of proportion to the measured COHb — give empirically, without waiting for a send-out cyanide level, alongside continued CO-directed care.

Symptomatic and Supportive Care

Treat pain, nausea, and (if it occurs) seizure activity alongside CO-directed therapy — none of these agents treat the CO poisoning itself.