Carbon Monoxide Poisoning Interpreter
Carbon Monoxide Poisoning Interpreter
Read a carboxyhaemoglobin level against the clinical features, the exposure context and whether oxygen was given before the sample — starting from the fact that pulse oximetry reads falsely normal in carbon monoxide poisoning and cannot detect it at all.
CO poisoning assessment
COHb + features + contextCarboxyhaemoglobin 12%, high-flow oxygen already given before the sample, a witnessed syncope at the scene, pH 7.31, non-smoker, not pregnant
What the assessment rests on, and what it deliberately does not do
Carboxyhaemoglobin half-life: room air 320 min · 100% oxygen < 90 min · hyperbaric at 3 atm ≈ 23 min
Reference: non-smoker < 3% · smoker up to 10% · above 10% implies additional exposure even in a smoker
Significant poisoning: > 20% adults, > 15% children
- why the oximeter lies
- carboxyhaemoglobin absorbs light at almost the same wavelength as oxyhaemoglobin, and a two-wavelength pulse oximeter cannot tell them apart — “Standard peripheral pulse oximetry devices cannot distinguish COHb from oxyhemoglobin”, so SpO₂ reads normal or near-normal in a severely poisoned patient. The pO₂ on a blood gas is also normal, because dissolved oxygen is unaffected. Only co-oximetry measures the fractions separately
- no COHb symptom ladder here
- the familiar table pairing percentages with symptoms is not on this page. The primary source available states that “COHb levels correlate poorly with clinical symptoms” and gives no such table, so publishing one would mean asserting a correlation the source denies. What is banded instead is the reference range and the referral thresholds, which are published
- why oxygen ruins the level
- with the half-life falling from 320 minutes to under 90, a patient at 30% when the crew arrived can be at 12% when the blood is drawn. High-flow oxygen “can rapidly reduce the levels of carboxyhemoglobin… leading to an inaccurate interpretation”. A low level after oxygen is evidence that the treatment is working, not that the poisoning was mild
- more than haemoglobin
- carbon monoxide also shifts the oxyhaemoglobin dissociation curve leftward, so what oxygen remains is released to tissue less readily, and it binds cytochrome c oxidase and myoglobin, impairing mitochondrial respiration and myocardial function directly. That is why a metabolic acidosis, a troponin rise and neurological signs track severity better than a percentage does
- what to band a percentage with
- for the percentage across its full range, this record defers to the carboxyhaemoglobin interpreter, which is a single-input bander. This page is the multi-input version, and its job is the management decision rather than the number
Worked example
Carboxyhaemoglobin 12%, high-flow oxygen already given before the sample, a witnessed syncope at the scene, pH 7.31, non-smoker, not pregnant
Features: a witnessed syncope, which is the transient loss of consciousness category rather than the coma category — so the severe rule does not fire
But transient loss of consciousness is one of the commonly accepted hyperbaric oxygen criteria → criterion met, and the rule fires before the level is read at all
Now read the level in that light. 12% in a non-smoker is above the 3% reference range but well below the 20% that indicates significant poisoning and below the 25% referral threshold. A page that banded the percentage alone would call this mild
And 12% is not the exposure. High-flow oxygen was running before the sample. On room air the half-life is 320 minutes; on 100% oxygen it is under 90. Forty-five minutes of oxygen takes 25% down to roughly 17%, and ninety minutes takes it to about 12% — which is consistent with this patient having been at 25% or more at the scene
So two independent things point past the number: the syncope meets a referral criterion on its own, and the level cannot be interpreted because it was taken on oxygen
Continue 100% oxygen through a tight-fitting reservoir mask, get an ECG and a troponin, examine neurology and cognition and record it as a baseline, and discuss with a poisons centre
And the thing that changes nothing about the immediate management but matters for what happens next: the source. Find out what it was, whether it is still running, and who else — including children and pets — was exposed to it
Why the oximeter is the first thing on this page
| Measurement | In carbon monoxide poisoning | Why |
|---|---|---|
| Pulse oximeter SpO₂ | Falsely normal. Cannot detect it at all | Carboxyhaemoglobin absorbs light at nearly the same wavelength as oxyhaemoglobin, and a two-wavelength pulse oximeter cannot separate them. “Standard peripheral pulse oximetry devices cannot distinguish COHb from oxyhemoglobin”, so “bedside assessments of oxygen saturation cannot be relied upon” |
| pO₂ on a blood gas | Normal | Dissolved oxygen in plasma is unaffected. The problem is carriage and delivery, not dissolution — so a normal pO₂ beside a normal SpO₂ is exactly what a severely poisoned patient looks like |
| Calculated oxygen saturation | Falsely normal | Most analysers calculate it from the pO₂ and the pH rather than measuring it, so it inherits the same blind spot. Check whether your gas result is measured or calculated |
| Co-oximetry | Diagnostic | Measures the haemoglobin fractions at multiple wavelengths and reports carboxyhaemoglobin separately. “An arterial blood gas sample with co-oximetry analysis is necessary to determine COHb saturation.” Venous co-oximetry is adequate for the carboxyhaemoglobin fraction itself |
| Clinical appearance | Unreliable | The classic cherry-red skin is a late and largely post-mortem sign. Cyanosis is characteristically absent, which is part of why the diagnosis is missed: the patient does not look hypoxic |
Half-life, and what it does to a level taken later
| Treatment | Carboxyhaemoglobin half-life | What that means for the number in front of you |
|---|---|---|
| Room air | 320 minutes — over five hours | A long transport or a delayed presentation drops the level substantially on its own. A level taken four hours after leaving the source is roughly half the peak |
| 100% oxygen, tight-fitting reservoir mask | Under 90 minutes | This is the treatment, and it is also why the level is untrustworthy once it has started. Ninety minutes of oxygen halves it |
| Hyperbaric oxygen, 3 atmospheres | Approximately 23 minutes | The pharmacological rationale for hyperbaric treatment, though the clinical trials do not settle whether that faster clearance improves neurological outcome — see the row below |
Hyperbaric oxygen: the accepted criteria, and the evidence against them
| What it says | Source | |
|---|---|---|
| Commonly accepted referral criteria | Neurological deficits · transient loss of consciousness · cardiac ischaemia · altered mental status · persistent metabolic acidosis (pH below roughly 7.1–7.25) · end-organ ischaemia · hypotension · carboxyhaemoglobin above 25% in adults, or above 15–20% in pregnancy | StatPearls, Carbon Monoxide Toxicity |
| What the randomised evidence shows | Six trials, 1,361 participants. Pooled odds ratio for neurological deficit 0.78, 95% CI 0.54 to 1.12 — no statistically significant benefit | Cochrane systematic review, Buckley et al, 2011 |
| The Cochrane conclusion, in full | “Existing randomised trials do not establish whether the administration of HBO to patients with carbon monoxide poisoning reduces the incidence of adverse neurologic outcomes” | Cochrane, 2011 |
| Why the disagreement persists | The individual trials conflict rather than being uniformly negative, they used different pressures, session numbers and timings, their outcome measures for neurocognitive injury differ, and the follow-up rates were variable. That is a design problem rather than a clear negative — the review calls for better multicentre trials | Cochrane, 2011 |
| What is agreed | 100% oxygen at normal pressure, immediately and through a tight-fitting non-rebreathing reservoir mask, started before any level returns. Up to 40% of patients develop chronic neurocognitive impairment despite treatment, so a baseline neurological and cognitive assessment is worth recording in everybody | StatPearls; Cochrane |
The oximeter reads normal, and the level does not measure severity
Two facts have to come first, because both of them run against instinct. The first is that pulse oximetry cannot detect carbon monoxide poisoning. Carboxyhaemoglobin absorbs light at almost the same wavelength as oxyhaemoglobin, and a standard two-wavelength pulse oximeter cannot tell them apart, so the SpO₂ reads normal in a severely poisoned patient. The pO₂ on a blood gas is normal too, because dissolved oxygen is unaffected, and the saturation many analysers report is calculated from that pO₂ rather than measured, so it inherits the same blind spot. The patient is not cyanosed. Nothing at the bedside points at the diagnosis. It is made by taking the exposure history and sending co-oximetry — a blood gas analysed on a co-oximeter — and it is missed when nobody thinks to.
The second is that the carboxyhaemoglobin percentage, once you have it, is a poor measure of how unwell the patient is. StatPearls states it directly: carboxyhaemoglobin levels correlate poorly with clinical symptoms and should not be used in isolation to guide treatment. There are good mechanistic reasons. Carbon monoxide does not only occupy haemoglobin binding sites — it shifts the oxyhaemoglobin dissociation curve leftward, so the oxygen still carried is released to tissue less readily, and it binds cytochrome c oxidase and myoglobin, impairing mitochondrial respiration and myocardial function directly. None of that is measured by the percentage. It is why this record takes the clinical features, the acid-base picture and the exposure context as inputs alongside the level, and why four of its rules fire before the level is read at all.
The third fact follows from the first two and is the one that catches people out most often: a low level after oxygen has been given does not exclude significant poisoning. The half-life of carboxyhaemoglobin is 320 minutes on room air, under 90 minutes on 100% oxygen, and approximately 23 minutes on hyperbaric oxygen at three atmospheres. Those numbers are the treatment, but they are also an interpretation problem. A patient at 30% when the ambulance crew arrive, given high-flow oxygen for an hour and a half in transit and in triage, will be at about 15% when the sample is finally drawn — and 15% reads as mild. High-flow oxygen, as StatPearls puts it, can rapidly reduce carboxyhaemoglobin and lead to an inaccurate interpretation of the level. So the two questions that have to be asked of every result are how long oxygen had been running before the blood was taken, and whether anything was measured at the scene. A level taken on oxygen is a lower bound, not an answer.
What the published figures do support is narrower than the tables usually reproduced. A non-smoker runs below 3%, from endogenous haem breakdown; a smoker may be as high as 10%; and above 10% implies additional carbon monoxide exposure even in someone who smokes. Above 20% in an adult, or 15% in a child, indicates significant poisoning, and myocardial impairment has been observed at levels as low as 20%. Those are the bands this page uses. The familiar ladder pairing each ten per cent with a set of symptoms is deliberately absent, because the source that gives the reference range gives no such table and states that the correlation it depends on does not hold — and for the banding of the percentage itself across its whole range, the carboxyhaemoglobin interpreter is the page for that.
Hyperbaric oxygen is the genuinely contested part, and it deserves to be presented as contested rather than resolved. The commonly accepted referral criteria are neurological deficit, transient loss of consciousness, cardiac ischaemia, altered mental status, persistent metabolic acidosis, end-organ ischaemia, hypotension, and a carboxyhaemoglobin above 25% in adults or above 15 to 20% in pregnancy. The Cochrane systematic review of the randomised evidence — six trials, 1,361 participants — found a pooled odds ratio for neurological deficit of 0.78 with a confidence interval from 0.54 to 1.12, and concluded that existing randomised trials do not establish whether hyperbaric oxygen reduces adverse neurological outcomes. The trials conflict with each other, used different pressures and schedules, and measured neurocognitive injury differently, so this is an unresolved question rather than a clear negative. In practice the accepted criteria are what most poisons centres work to, the decision is time-critical, and the transfer logistics are part of it — so the answer is to discuss the patient rather than to apply either column mechanically. What is not in dispute is the normobaric half: 100% oxygen through a tight-fitting non-rebreathing reservoir mask, immediately, for everybody, before any number comes back. Nor is the follow-up: up to 40% of patients develop chronic neurocognitive impairment despite treatment, which makes a recorded baseline neurological and cognitive assessment worth the few minutes it takes. And the exposure itself has to be dealt with — the source found and made safe, and everyone else who shared it, including children and pets, assessed.
Frequently asked questions
Can pulse oximetry detect carbon monoxide poisoning?
No. This is the single most important fact about the diagnosis. Carboxyhaemoglobin absorbs light at almost the same wavelength as oxyhaemoglobin, and a standard two-wavelength pulse oximeter cannot distinguish them, so the SpO₂ reads normal or near-normal in a severely poisoned patient. The pO₂ on a blood gas is also normal, because dissolved oxygen is unaffected, and any saturation calculated from that pO₂ inherits the same blind spot. The patient is characteristically not cyanosed. A co-oximeter, or a blood gas analysed with co-oximetry, is required — it measures the haemoglobin fractions at multiple wavelengths and reports carboxyhaemoglobin separately. Venous blood is adequate for the carboxyhaemoglobin fraction.
Does a low carboxyhaemoglobin rule out carbon monoxide poisoning?
No, and this is the commonest interpretive error. Two things lower the level without lowering the risk. Carboxyhaemoglobin has a half-life of 320 minutes on room air and under 90 minutes on 100% oxygen, so a patient at 30% when the ambulance arrived may be at 15% by the time blood is drawn after ninety minutes of high-flow oxygen — high-flow oxygen can rapidly reduce the level and lead to an inaccurate interpretation of it. And separately, the level correlates poorly with clinical symptoms even when it is measured properly. So in a symptomatic patient with a plausible exposure, a low level is not a reason to stand down; ask how long oxygen has been running, ask what the scene reading was, and decide on the clinical picture.
What is a normal carboxyhaemoglobin, and what does smoking do to it?
A non-smoker runs below 3%, produced by the endogenous breakdown of haem. A smoker may be as high as 10%. Above 10% suggests additional carbon monoxide exposure even in someone who smokes, so a level above that confirms an exposure regardless of smoking history. Above 20% in an adult, or 15% in a child, indicates significant poisoning, and myocardial impairment has been observed at levels as low as 20%. The practical point about smokers is that they start every exposure from a higher baseline, so a given inhaled dose takes them to a higher absolute level — and a level inside the smoker range in a symptomatic smoker with an exposure history does not exclude poisoning.
When is hyperbaric oxygen indicated?
This is genuinely contested and worth knowing as a disagreement rather than a rule. The commonly accepted referral criteria are neurological deficit, transient loss of consciousness, cardiac ischaemia, altered mental status, persistent metabolic acidosis at a pH below roughly 7.1 to 7.25, end-organ ischaemia, hypotension, and a carboxyhaemoglobin above 25% in adults or above 15 to 20% in pregnancy. But the Cochrane systematic review of six randomised trials in 1,361 participants found a pooled odds ratio for neurological deficit of 0.78 with a confidence interval of 0.54 to 1.12, and concluded that existing randomised trials do not establish whether hyperbaric oxygen reduces adverse neurological outcomes. The trials conflict and differ in pressure, schedule and outcome measurement, so this is unresolved rather than settled either way. Discuss the individual patient with a poisons centre; what is not in dispute is high-flow 100% oxygen immediately for everybody.
Why is the referral threshold lower in pregnancy?
Because the fetus is at greater risk than the maternal level indicates. Fetal haemoglobin binds carbon monoxide more avidly than adult haemoglobin, the fetal carboxyhaemoglobin level lags behind the maternal one and then exceeds it, and it clears more slowly — so the fetus is still loaded when the mother has cleared. The commonly cited threshold for hyperbaric referral in pregnancy is above 15 to 20%, against 25% in a non-pregnant adult, and maternal oxygen is given for longer than the maternal level alone would justify. Fetal monitoring and joint discussion with obstetrics and a poisons centre are appropriate, and concern about the fetus is not a reason to withhold treatment: untreated maternal poisoning is worse for the fetus than the treatment is.
Why does this page not show a table of symptoms by carboxyhaemoglobin percentage?
Because the correlation that such a table depends on does not hold, and the source that gives the reference range says so. StatPearls states that carboxyhaemoglobin levels correlate poorly with clinical symptoms and should not be used in isolation to guide treatment, and it gives no percentage-to-symptom table of its own. The familiar ladder — headache at 15 to 20%, nausea at 20 to 30%, confusion at 30 to 40% — is widely reproduced but could not be traced here to a primary source, and publishing it would mean asserting a relationship the available evidence denies. What is published, and what this page uses, is the reference range, the significant-poisoning threshold and the hyperbaric referral thresholds.
Related calculators
References
- Carbon Monoxide Toxicity. StatPearls. NCBI Bookshelf. “Standard peripheral pulse oximetry devices cannot distinguish COHb from oxyhemoglobin” and “bedside assessments of oxygen saturation cannot be relied upon”; “An arterial blood gas sample with co-oximetry analysis is necessary to determine COHb saturation”; non-smokers below 3%, smokers up to 10%, above 10% suggesting additional exposure even in people who smoke; above 20% in adults and 15% in children suggesting significant poisoning, with myocardial impairment observed as low as 20%; “Since COHb levels correlate poorly with clinical symptoms, they should not be used in isolation to guide treatment decisions”; half-life 320 minutes on room air, under 90 minutes on 100% oxygen and approximately 23 minutes on hyperbaric oxygen at 3 atmospheres; hyperbaric candidates including neurologic deficits, transient loss of consciousness, cardiac ischaemia, altered mental status, persistent metabolic acidosis at pH below 7.1–7.25, end-organ ischaemia, hypotension, COHb above 25% in adults or above 15–20% in pregnancy; high-flow oxygen can rapidly reduce COHb “leading to an inaccurate interpretation of COHb levels”; up to 40% of patients develop chronic neurocognitive impairment despite treatment.
- Buckley NA, Juurlink DN, Isbister G, Bennett MH, Lavonas EJ. Hyperbaric oxygen for carbon monoxide poisoning. Cochrane Database Syst Rev. 2011;(4):CD002041. Six trials involving 1,361 participants; pooled odds ratio for neurological deficit 0.78 (95% CI 0.54 to 1.12); “Existing randomised trials do not establish whether the administration of HBO to patients with carbon monoxide poisoning reduces the incidence of adverse neurologic outcomes”; additional research needed to better define the role, if any, of hyperbaric oxygen.
- Weaver LK, Hopkins RO, Chan KJ, et al. Hyperbaric oxygen for acute carbon monoxide poisoning. N Engl J Med. 2002;347(14):1057–1067. The largest single trial reporting benefit, and one of the six included in the Cochrane review — cited here as one side of a disagreement rather than as a settled result.
- Annane D, Chadda K, Gajdos P, Jars-Guincestre MC, Chevret S, Raphael JC. Hyperbaric oxygen therapy for acute domestic carbon monoxide poisoning: two randomized controlled trials. Intensive Care Med. 2011;37(3):486–492. Two trials that did not find benefit, and part of the reason the pooled estimate is inconclusive.
- Centers for Disease Control and Prevention, National Center for Environmental Health. Toxidromes chart. Files carbon monoxide under “Cellular Hypoxia” rather than as a toxidrome, alongside cyanide, hydrogen sulfide and sodium azide — which is why carbon monoxide poisoning is not resolvable from pupils, skin and bowel sounds.
Medical Disclaimer: The tools and content provided here are for educational and reference purposes only. They are not intended to substitute for professional medical advice, diagnosis, or treatment. Clinical decisions should always be based on the comprehensive assessment of a qualified healthcare professional.