Carbon monoxide competitively displaces oxygen on hemoglobin (~200–250× affinity) forming carboxyhemoglobin — live COHb elimination kinetics under room air, 100% O2, and hyperbaric oxygen, plus HBO referral criteria
Carbon monoxide is an odorless, colorless gas produced by incomplete combustion. It kills not by direct tissue corrosion but by competitively occupying the very binding site hemoglobin uses to carry oxygen — and by directly disabling mitochondrial respiration in every cell it reaches. Understanding this dual mechanism (hemoglobin-carrying-capacity loss plus direct cellular poisoning) explains why COHb percentage alone is an imperfect predictor of clinical severity and outcome.
Each hemoglobin tetramer carries four heme groups, each with a ferrous (Fe²⁺) iron center that reversibly binds a single small diatomic ligand — normally O2. Carbon monoxide fits the same pocket, but the CO–Hb complex dissociates far more slowly than O2–Hb, so at any given partial pressure CO progressively accumulates on hemoglobin at O2's expense. The net affinity advantage is roughly 200–250-fold, meaning even trace ambient CO (a few hundred ppm) can drive COHb saturation well above 50% given enough exposure time.
Two compounding effects worsen tissue oxygenation beyond the simple loss of carrying capacity:
• Reduced carrying capacity — every heme occupied by CO cannot carry O2, directly lowering arterial oxygen content. • Left-shifted oxyhemoglobin dissociation curve — CO binding at one or more subunits of a tetramer stabilizes the high-affinity "R" (relaxed) conformation of the remaining subunits, so any O2 still bound elsewhere on that same molecule is held more tightly and released less readily to tissues. This means a patient with 30% COHb suffers a functional oxygen deficit substantially worse than "30% of hemoglobin is unavailable" would suggest — the remaining 70% also unloads O2 less efficiently.
Common exposure sources:
• House fires and smoke inhalation — the leading cause of severe/fatal CO poisoning • Faulty or unvented furnaces, water heaters, and boilers • Indoor use of charcoal grills, propane heaters, or generators (classic post-storm power-outage presentation, often multiple family members simultaneously) • Industrial exposure — foundries, steel mills, exposure to methylene chloride (paint stripper), which is metabolized to CO by the liver with a delayed toxic course • Poorly maintained or blocked vehicle exhaust systems (idling in enclosed garages)
CO does not stop at hemoglobin. It also binds cardiac and skeletal myoglobin with high affinity, directly impairing oxygen storage and handling within muscle — a mechanism implicated in the myocardial injury (troponin elevation, wall motion abnormalities, arrhythmia) frequently seen in moderate-to-severe CO poisoning, independent of any coexisting coronary disease.
More importantly, CO diffuses into cells and binds mitochondrial complex IV (cytochrome c oxidase), the terminal enzyme of the electron transport chain. This blocks oxidative phosphorylation directly — forcing anaerobic metabolism, generating lactate, and producing reactive oxygen species upon subsequent reoxygenation/reperfusion. This oxidative and inflammatory cascade is now understood to be the substrate for the lipid peroxidation injury implicated in delayed neuropsychiatric sequelae (Stage 5), and it proceeds largely independent of the peak measured COHb% — which is one reason symptom severity and long-term outcome correlate so poorly with a single COHb blood level drawn at one point in time.
CO poisoning is notoriously difficult to recognize because its symptoms mimic viral illness, migraine, or food poisoning, and because the one bedside monitor clinicians reflexively trust — the pulse oximeter — reports a falsely reassuring number. Diagnosis requires a high index of suspicion (especially with multiple simultaneously ill household members, or symptoms improving on leaving the exposure environment) and specific co-oximetry testing.
Textbooks often present a tidy dose–response relationship, but real patients frequently deviate from it substantially — especially with chronic low-level or subacute exposure, in which severe toxicity can appear at levels that look mild on a single blood draw obtained after some room-air elimination has already occurred en route to care.
Approximate staging (for orientation, not for triage in isolation):
• ~10–20% COHb: headache, nausea, dizziness, mild dyspnea on exertion — easily mistaken for a viral syndrome • ~20–40% COHb: confusion, syncope, chest pain/angina (especially with underlying coronary disease), visual disturbance, ataxia • >40–50% COHb: seizures, coma, cardiovascular collapse, malignant arrhythmia • Very high levels: death, often before reaching care
Key caveats: elderly patients, those with underlying cardiopulmonary disease, fetuses, and infants tolerate any given COHb level far worse than healthy adults. Symptom severity and even outcome (including delayed neuropsychiatric sequelae) correlate only weakly with the single COHb value measured at presentation — treatment decisions should weigh the whole clinical picture, not a number alone.
Standard pulse oximeters estimate oxygen saturation using the differential light absorbance of oxyhemoglobin and deoxyhemoglobin at just two wavelengths (~660nm red and ~940nm infrared). Carboxyhemoglobin absorbs light at these two wavelengths in a pattern the device interprets as "oxygenated" — so a patient who is critically poisoned, with COHb of 40% or higher, can display a perfectly reassuring SpO2 of 97–100% on a standard monitor. This is a frequently missed diagnostic trap, particularly in smoke-inhalation patients where clinicians may be falsely reassured by a normal-looking monitor.
Correct diagnosis requires either:
• Co-oximetry — a laboratory blood gas analyzer using 4 or more wavelengths, able to separately resolve COHb, methemoglobin, oxyhemoglobin, and deoxyhemoglobin fractions from an arterial or venous sample • Noninvasive pulse CO-oximetry (e.g., Masimo Rad-57) — a specialized multi-wavelength fingertip device that estimates COHb noninvasively, useful for rapid field/ED screening though less precise than direct blood co-oximetry and validated mainly for screening rather than definitive diagnosis
The classic "cherry-red" skin discoloration taught in textbooks is in practice a rare and late finding, more often described post-mortem; most poisoned patients appear pale, mildly cyanotic, or simply unwell — do not rely on skin color to include or exclude the diagnosis.
Regardless of whether hyperbaric oxygen will ultimately be pursued, every patient with significant suspected or confirmed CO poisoning should receive high-flow 100% oxygen immediately — before COHb results return, before hyperbaric transfer is even considered. Normobaric 100% O2 dramatically accelerates CO elimination by mass-action displacement at the hemoglobin binding site.
CO does not detach from hemoglobin faster simply because ambient oxygen tension rises — its intrinsic dissociation rate from the heme pocket is largely unchanged. What changes is the competition for re-binding: as alveolar and arterial PO2 rises toward the levels achieved with 100% inspired oxygen, every CO molecule that transiently dissociates faces overwhelmingly more oxygen molecules competing to occupy that same site, and is far less likely to rebind. Freed CO is then eliminated by exhalation. This mass-action competition is why raising FiO2 from 21% to 100% cuts the COHb elimination half-life roughly four- to five-fold, from several hours down to about an hour.
Practical treatment points:
• Start immediately in any patient with suspected significant exposure — a tight-fitting non-rebreather mask at 15 L/min, or endotracheal intubation with FiO2 1.0 if the patient cannot protect the airway or has depressed consciousness • Continue until the patient is asymptomatic and COHb has normalized — commonly a minimum of about 6 hours, longer while a hyperbaric-transfer decision is being made • 100% O2 is never withheld while awaiting a decision about hyperbaric therapy — it is the universal baseline treatment, with HBO layered on top for patients meeting specific criteria (Stage 4) • Reassess mental status, cardiac rhythm, and repeat COHb serially; falling COHb with resolving symptoms supports continued normobaric management without hyperbaric transfer for lower-risk patients
Hyperbaric oxygen (HBO) at 2.4–3.0 atmospheres absolute (ATA) further accelerates COHb clearance to a half-life of roughly 20–30 minutes and dramatically increases physically dissolved plasma oxygen, delivering oxygen even to tissue with severely impaired hemoglobin-mediated delivery. A set of accepted clinical indications guides referral, but the randomized trial evidence for reducing long-term neurologic harm remains genuinely mixed — making HBO one of the more actively debated therapies in emergency and toxicology practice.
No single universal protocol exists, but the following criteria are widely cited across hyperbaric-capable centers and poison control guidance as reasonable indications for HBO referral, generally requiring transfer to a hyperbaric-capable facility if the treating hospital lacks a chamber:
• Loss of consciousness at any point during or after exposure — even if brief or witnessed only by history, this is one of the strongest and most consistently cited indications • COHb level >25% (some protocols use >20%) at any point, not necessarily at the time of the treatment decision • Pregnancy with COHb >15–20% — a substantially lower threshold than for non-pregnant adults, because fetal hemoglobin binds CO with even greater affinity than adult hemoglobin, and fetal COHb clearance lags maternal clearance significantly — the fetus may remain significantly poisoned well after the mother's COHb has normalized, so pregnant patients are treated more aggressively at lower measured levels • Severe metabolic acidosis, generally pH <7.1–7.2 • Evidence of cardiac ischemia or dysrhythmia — troponin elevation, ischemic ECG changes, new arrhythmia • Age above a threshold (commonly cited between 36 and 60 depending on institutional protocol) reflecting reduced physiologic reserve • Persistent abnormal neurologic exam or focal neurologic deficits after initial stabilization
The decision is ultimately case-by-case, typically made in consultation with regional poison control and a hyperbaric medicine specialist, explicitly weighing the risk and delay of transport against the plausible but unproven benefit — particularly when transfer would take many hours.
The physiological rationale for HBO is strong: it clears COHb roughly three times faster than normobaric 100% O2, dramatically raises physically dissolved oxygen independent of hemoglobin saturation, and is proposed to reduce neutrophil-mediated lipid peroxidation and adduct formation implicated in delayed neuropsychiatric sequelae. But the randomized controlled trial literature is genuinely mixed:
• Weaver et al., NEJM 2002 — a rigorously blinded trial (three hyperbaric sessions over 24 hours vs. sham) found a reduced incidence of delayed neuropsychiatric sequelae at 6 weeks (roughly 23% vs. 46% in the sham/normobaric group) — the most frequently cited positive trial supporting HBO • Other major trials (e.g., Scheinkestel et al. 1999 in Australia, Raphael et al. 1989 in France) found no benefit, and in one case a suggestion of possible harm associated with repeated hyperbaric exposures • Cross-trial comparison is hampered by real heterogeneity: differing COHb severity thresholds for enrollment, differing numbers of hyperbaric sessions and pressures used, differing time-from-exposure to treatment, and differing tools used to assess delayed neuropsychiatric outcome • Systematic reviews (including Cochrane) have generally concluded the evidence is inconclusive due to these methodological inconsistencies, rather than affirmatively refuting benefit
In practice, most centers continue to offer HBO to patients meeting the Stage 4 criteria above on the basis of physiological plausibility and the positive trial data that does exist, while acknowledging this remains an active area of clinical research and reasonable clinicians can differ on borderline cases — especially when transfer delay is substantial.
Perhaps the most clinically important reason to take CO poisoning seriously even after a patient looks well again is delayed neuropsychiatric sequelae (DNS): a syndrome of cognitive and neurological decline that can emerge weeks after a patient has seemingly fully recovered from the acute poisoning. Because DNS is common, delayed, and can be missed if no one is looking for it, structured follow-up is a core part of comprehensive CO poisoning care.
DNS classically follows a "lucid interval" — the patient recovers from the acute encephalopathy of poisoning, returns to an apparently normal neurological baseline, and then days-to-weeks later (most commonly within 3 to 6 weeks, though reported as early as 3 days and as late as 240 days after exposure) develops a new and often insidious neuropsychiatric decline.
Characteristic features include:
• Cognitive deficits — impaired memory and concentration, executive dysfunction • Personality and mood change — irritability, apathy, disinhibition, depression • Parkinsonism — bradykinesia, rigidity, gait disturbance, sometimes with basal ganglia imaging correlates • Urinary incontinence and gait/balance disturbance • In severe cases, a persistent vegetative or minimally conscious state
Imaging correlate: MRI may reveal bilateral globus pallidus necrosis (sometimes visible even on CT as bilateral hypodensities) and deep cerebral white matter changes consistent with delayed post-hypoxic leukoencephalopathy — a distinct radiographic and pathological pattern from acute hypoxic-ischemic injury.
Proposed mechanism: rather than simple oxygen deprivation, DNS is thought to arise from a delayed inflammatory and oxidative injury cascade — lipid peroxidation triggered by CO-induced mitochondrial dysfunction and reperfusion oxidative stress, leukocyte-mediated vascular injury, and adduct formation on myelin basic protein that may trigger an autoimmune-like demyelinating process. This distinct mechanism, separate from straightforward hypoxia, is the proposed (though not definitively proven) basis for HBO's benefit extending beyond simply correcting oxygen delivery.
Because DNS can emerge after a patient looks and feels well, and because early recognition allows for supportive management and realistic prognostic counseling, structured follow-up should be arranged for anyone with significant CO poisoning:
• Formal neurocognitive assessment — a brief bedside mental status exam is insufficient to exclude subtle DNS; a fuller neuropsychological battery is preferred, ideally with a baseline assessment near the time of the acute event and a repeat evaluation at approximately 4–6 weeks, spanning the highest-risk window for DNS onset • Patient and family education — explicit counseling on the warning signs of DNS (memory problems, personality change, new gait or urinary symptoms) so patients know to seek prompt re-evaluation rather than attributing new symptoms to stress or unrelated causes • Psychiatric screening — depression and anxiety are common after significant poisoning events, independent of DNS, and warrant proactive screening • Source remediation — identifying and correcting the exposure source (furnace inspection, generator placement counseling, chimney/flue evaluation) is essential to prevent re-exposure; installation of a functioning carbon monoxide detector remains the single most effective and inexpensive prevention measure • Occupational cases warrant workplace safety review and, where applicable, formal reporting