HBOT for Carbon Monoxide Poisoning: 2024-2026 Evidence Update

Reading time: about 11 minutes. Audience: Canadian referring physicians and emergency clinicians.

TL;DR: Hyperbaric oxygen therapy (HBOT) is a Health Canada and UHMS recognised treatment for acute carbon monoxide poisoning, but the 2024-2026 literature is more nuanced than headline summaries suggest. A 2026 meta-analysis of randomised controlled trials found no significant mortality or neurological benefit at any pressure, while observational dosing data show higher chamber pressure (2.8 vs 2.0 ATA) reduces delayed neuropsychiatric sequelae more than longer regimens at lower pressure. A small exploratory study supports an early (≤3 hour) initiation window. Glasgow Coma Scale <9 and rhabdomyolysis predict delayed neurological sequelae; initial lactate >2.8 mmol/L flags need for hyperbaric referral; altered mental status, low albumin, and low magnesium predict middle ear barotrauma. Canada Hyperbarics summarises the practical referral implications for Canadian emergency clinicians and hospitalists.

Carbon monoxide poisoning is one of the few hyperbaric indications where the referral window is measured in hours rather than days. For Canadian referring physicians, the question is no longer whether HBOT helps after acute CO poisoning, but which patients to refer, how quickly, and how to triage when the nearest chamber is hundreds of kilometres away. The 2024-2026 evidence base has refined each of those answers.

This evidence update from Canada Hyperbarics synthesises 12 recent peer-reviewed studies into practical guidance for emergency physicians, hospitalists, family physicians, and intensivists who refer to a Canadian hyperbaric programme. Every citation links to the underlying study summary on our research bank.

What is hyperbaric oxygen therapy for carbon monoxide poisoning?

Hyperbaric oxygen therapy is the delivery of 100 percent oxygen at a pressure greater than one atmosphere absolute (ATA), typically 2.4 to 3.0 ATA, inside a sealed monoplace or multiplace chamber. For acute carbon monoxide poisoning, HBOT accelerates the dissociation of carboxyhaemoglobin (COHb), reduces leucocyte adhesion in cerebral microvasculature, and limits the lipid peroxidation cascade that underlies delayed neuropsychiatric sequelae (DNS).

Carbon monoxide poisoning is a Health Canada and Undersea and Hyperbaric Medical Society (UHMS) approved indication for hyperbaric oxygen therapy. The Canadian Undersea and Hyperbaric Medical Association (CUHMA) endorses the UHMS indication list, and acute CO poisoning is listed on the Canada Hyperbarics conditions index as an established indication.

What does the 2024-2026 evidence say about HBOT efficacy in acute CO poisoning?

A 2026 systematic review and meta-analysis by Fujita and colleagues in Acute Medicine and Surgery synthesised six randomised controlled trials of hyperbaric oxygen therapy for acute CO poisoning, including a subgroup analysis of treatments at 2.5 ATA or higher. The pooled analysis found no significant benefit for mortality or neurological outcomes overall, and no advantage for the higher-pressure subgroup. The authors rated the quality of evidence as low to very low with moderate to significant heterogeneity, concluding that the efficacy of HBOT for acute CO poisoning, even at 2.5 ATA or above, remains unproven and warrants further investigation (Fujita et al., 2026).

The negative pooled signal does not remove HBOT from clinical practice. UHMS continues to list acute CO poisoning as a Category 1 (peer-reviewed) indication, and Health Canada recognises hyperbaric medicine programmes for this indication across provinces. What the Fujita meta-analysis sharpens is the question of which subgroups derive the most benefit, which is the focus of the dosing, timing, and risk-stratification literature reviewed in the sections that follow.

A 2025 narrative review by Afzal and colleagues in Diagnostics surveys the diagnostic landscape, prognostic factors, and treatment strategies for CO poisoning. The authors highlight the limitations of current diagnostic tools (blood carboxyhaemoglobin levels and pulse CO-oximetry both miss important cases), and note that prognosis is shaped chiefly by exposure severity and delayed treatment, both of which raise the risk of long-term neurological damage. The review positions HBOT as the primary treatment while acknowledging access constraints, and highlights emerging portable CO-oximeters and biomarker panels as a route to earlier diagnosis (Afzal et al., 2025).

The 2026 multicentre prospective ITA-OTI study by Ippolito and colleagues in the Journal of Anesthesia, Analgesia and Critical Care provides current real-world practice data from 327 patients across ten Italian hyperbaric centres. Acute CO poisoning was the second most frequent indication at 19.9% of treatments (after sudden hearing loss at 35.8%), with treatments delivered at a median of 2.5 ATA and no serious adverse events reported across the cohort. The study confirms that contemporary European hyperbaric practice for CO poisoning is consistent with national and international guidelines (Ippolito et al., 2026).

How does the timing of HBOT initiation affect neurological outcomes?

A 2026 study by Xu and colleagues in Open Medicine examined the optimal initiation window for HBOT after acute CO poisoning in a single-centre retrospective cohort of 12 moderate-to-severe ACOP patients treated within six hours of admission, split into early (≤3 hours, n=8) and delayed (>3 hours, n=4) groups. Early HBOT was associated with faster organ-function recovery (greater SOFA score reduction, p<0.05), and lactate clearance above 50% within 24 hours correlated with better neurological outcomes (p=0.002). The authors explicitly caveat that the small sample size and absence of a non-HBOT control group make these findings exploratory rather than definitive, and they call for confirmation in larger studies (Xu et al., 2026).

For Canadian referring physicians, the three-hour finding has direct workflow implications. In urban centres with a hospital-based hyperbaric programme such as Toronto General, Vancouver General, or the Foothills Medical Centre, the three-hour window is operationally achievable. In regional and rural settings where the nearest chamber is two or three hours away by ground or air, the window narrows the moment first responders identify suspected CO exposure. Pre-hospital high-flow oxygen and early advance notification of the receiving hyperbaric programme are the levers within the referring physician’s control.

Which biomarkers predict which CO-poisoned patients need HBOT?

Peak carboxyhaemoglobin alone is a weak predictor of delayed neuropsychiatric sequelae. The 2024-2026 literature has moved toward a multi-marker approach combining initial lactate, inflammatory and cardiac biomarkers, acute MRI findings, comorbid disease burden, and clinical features such as the Glasgow Coma Scale.

A 2026 study by Satilmis and colleagues in Clinical Laboratory compared initial lactate against lactate clearance as predictors of which CO-poisoned patients ultimately required hyperbaric referral, in a cohort of 169 emergency department admissions. Initial lactate above 2.8 mmol/L predicted HBOT need with an area under the curve of 0.705 (sensitivity 64%, specificity 73%), while lactate clearance showed no significant association with treatment modality (p=0.596). The practical takeaway is that a single early venous lactate above 2.8 mmol/L is a useful triage signal in itself: referring physicians do not need to wait for serial clearance kinetics before initiating the hyperbaric referral conversation (Satilmis et al., 2026).

A 2025 prospective observational study by Gulen and colleagues in Internal and Emergency Medicine evaluated plasma biomarker panels for early myocardial injury in 82 CO-poisoned patients presenting to a tertiary emergency department. Neutrophil-to-lymphocyte ratio (NLR) was an independent predictor of myocardial injury (OR 1.15), and on ROC analysis D-dimer (AUC 0.79) and NLR (AUC 0.79) were the most accurate biomarkers for predicting injury, while D-dimer (AUC 0.74) and NT-proBNP (AUC 0.70) best predicted the need for hyperbaric referral. Where these markers are elevated, cardiology co-management alongside hyperbaric referral is appropriate (Gulen et al., 2025).

A 2026 study by Gong and colleagues in BMC Neurology constructed a predictive model for delayed encephalopathy in 102 patients hospitalised for acute CO poisoning, of whom 15 developed delayed encephalopathy. The four-variable logistic-regression model combined length of stay, comorbidities (hypertension and diabetes), carboxyhaemoglobin, and post-discharge HBOT, achieving high discrimination (AUC 0.933, 95% CI 0.877-0.991) with good calibration. Length of stay and comorbid disease burden, both routinely captured at admission, give referring physicians an early stratification signal beyond COHb alone (Gong et al., 2026).

A separate 2025 study by Ho and colleagues in the Journal of the Chinese Medical Association retrospectively analysed 502 acute CO poisoning patients from the Taiwan National Poison Control Center to identify risk factors for delayed neurological sequelae and myocardial injury. Glasgow Coma Scale below 9 (OR 2.55) and rhabdomyolysis (OR 2.68) independently predicted delayed neurological sequelae. Risk of myocardial injury was driven by the same two factors plus acute renal impairment (OR 2.43) and leukocytosis (OR 9.55). HBOT showed a directionally protective effect for delayed neurological sequelae (OR 0.64) but not for myocardial injury, consistent with the dominant role of HBOT in mitigating CNS rather than cardiac injury (Ho et al., 2025).

What does acute MRI add to the referral decision?

A 2026 retrospective cohort study by Jia and colleagues in European Journal of Medical Research evaluated the predictive value of acute MRI-detected brain lesions (ABLs) in patients with acute CO poisoning. Acute brain lesions were associated with lower clinical cure (adjusted OR 0.35) and higher incidence of delayed encephalopathy (adjusted OR 2.54). Time from onset to first HBOT greater than six hours independently worsened both outcomes (cure aOR 0.38; delayed encephalopathy aOR 3.50), and Glasgow Coma Scale below 9 was a strong independent predictor of delayed encephalopathy (aOR 7.50). White matter lesions in particular were associated with lower clinical cure (aOR 0.23) and higher delayed-encephalopathy risk (aOR 3.17). Acute MRI is therefore most useful as a positive predictor: white matter lesions on early imaging argue strongly for hyperbaric referral and structured neurology follow-up (Jia et al., 2026).

A 2026 prospective observational study by Lee and colleagues in Scientific Reports compared non-invasive pulse co-oximetry (SpCO) against arterial carboxyhaemoglobin in 81 acute CO poisoning patients. SpCO showed only limited agreement with arterial COHb at baseline, with substantial positive bias and wide limits of agreement at higher concentrations. After HBOT, agreement deteriorated markedly, with unstable regression slopes and persistent heteroscedasticity across all severity strata. Arterial blood gas with co-oximetry therefore remains the reference standard for triage and referral, particularly in the post-treatment monitoring window (Lee et al., 2026).

How should HBOT be dosed for acute carbon monoxide poisoning?

A 2026 study by Gur and colleagues in Clinical Toxicology retrospectively compared four hyperbaric dosing regimens in 312 patients treated for acute CO poisoning at a large regional hyperbaric referral facility over 30 years. The dominant variable was chamber pressure, not session count: delayed neuropsychiatric sequelae occurred in 36% of patients treated at 2.0 ATA once, 37% at 2.0 ATA thrice, 20% at 2.8 ATA once, and 19% at 2.8 ATA thrice (overall p=0.011). Pooled, the lower-pressure groups had a 36.2% sequelae rate versus 19.3% for the higher-pressure groups (p=0.0013). Adding a second or third session at the same pressure did not further reduce delayed sequelae, suggesting that achieving the higher chamber pressure matters more than extending protocol length (Gur et al., 2026).

The table below summarises the predictive features most consistently associated with DNS across the 2024-2026 cohort and case-control studies and the corresponding referral implication for emergency physicians.

Feature on presentation	Evidence source	Referral implication
Glasgow Coma Scale below 9	Ho 2025 (OR 2.55 for DNS); Jia 2026 (aOR 7.50 for delayed encephalopathy)	Refer urgently for HBOT
Rhabdomyolysis	Ho 2025 (OR 2.68 for DNS)	Refer for HBOT
Acute white matter MRI lesions	Jia 2026 (aOR 0.23 for clinical cure; aOR 3.17 for delayed encephalopathy)	Refer for HBOT + neurology follow-up
Initial venous lactate above 2.8 mmol/L	Satilmis 2026 (AUC 0.705 for HBOT need)	Refer for HBOT
Elevated D-dimer or NLR	Gulen 2025 (AUC 0.79 each for myocardial injury)	Refer + cardiology co-management
Long anticipated length of stay or comorbidity (hypertension, diabetes)	Gong 2026 (four-variable model AUC 0.933)	Lower threshold to refer
Altered mental status at presentation	Lee 2025 (MEB OR 3.16); also Ho 2025 risk profile	Refer; flag for ear monitoring during HBOT
Time from onset above 6 hours	Jia 2026 (cure aOR 0.38; delayed encephalopathy aOR 3.50)	Refer immediately even if delayed
Pregnancy	UHMS guidance (foetal HbCO kinetics)	Refer regardless of maternal COHb
Persistent neurological signs after normobaric oxygen	Clinical convention	Refer urgently

What are the risks and rare presentations to anticipate during HBOT referral?

The single most common adverse event during HBOT for CO poisoning is middle ear barotrauma. A 2025 retrospective cohort study by Lee and colleagues in the Journal of Clinical Medicine identified independent predictors of middle ear barotrauma among CO-poisoned patients undergoing monoplace HBOT. Altered mental status at presentation increased risk (OR 3.16), while higher serum albumin (above 4.3 g/dL, OR 0.26) and higher magnesium levels (OR 0.21) were independently protective on multivariate analysis. The clinical pattern is intuitive: obtunded patients cannot equalise their ears, and lower albumin or magnesium tracks with more severe overall illness. Pre-treatment otoscopic assessment remains routine; closer monitoring is warranted in patients presenting with altered mentation or low nutritional reserves (Lee et al., 2025).

Rare but clinically relevant presentations reported in the recent literature include paroxysmal sympathetic hyperactivity (PSH) syndrome after severe CO poisoning. A 2026 retrospective analysis by Yang and colleagues in Medical Gas Research identified PSH in 3 of 53 moderate-to-severe ACOP admissions, with coma duration above 72 hours and irregular HBO treatment after onset as the principal risk factors. Antiepileptic drugs had poor effect on PSH attacks, but the attacks were effectively controlled after HBOT, supporting hyperbaric referral even when autonomic instability complicates the clinical picture. (Yang et al., 2026).

Delayed-onset parkinsonism after CO poisoning is uncommon but well-described. A 2026 case report by Ingiardi and colleagues in Neurocase documented a patient who developed delayed-onset parkinsonism and cognitive deficits after a symptom-free interval of approximately one month following CO poisoning. Combined pharmacological treatment, cycles of HBOT, and tailored neuropsychological rehabilitation produced progressive improvements in sustained attention, memory, and executive functioning, although residual difficulties persisted. The case reinforces the value of structured neurology and neuropsychology follow-up after discharge regardless of the apparent acute response (Ingiardi et al., 2026).

A separate consideration is venous thromboembolism risk in immobilised CO-poisoned patients. A 2025 retrospective study by Zhao and colleagues in American Journal of Translational Research reported a 12.8% incidence of lower-extremity deep vein thrombosis among 180 acute CO poisoning admissions. Coma duration, carboxyhaemoglobin concentration, C-reactive protein, procalcitonin, D-dimer, and a range of biochemical markers (creatinine, blood urea nitrogen, lactate dehydrogenase, myoglobin, creatine kinase) all correlated positively with DVT risk, while earlier HBOT initiation and higher albumin levels were protective. Standard VTE prophylaxis is appropriate for any CO-poisoned patient with prolonged coma or markedly elevated inflammatory markers, alongside hyperbaric referral (Zhao et al., 2025).

How do you refer a CO-poisoned patient for HBOT in Canada?

The Canadian referral pathway for acute CO poisoning is time-critical and varies by region. The following sequence reflects the operational pattern at major Canadian hyperbaric programmes.

1. Confirm the diagnosis. Obtain arterial blood gas with co-oximetry to measure COHb. Pulse co-oximetry (SpCO) is a useful screen but should not replace arterial measurement near treatment thresholds.
2. Start high-flow normobaric oxygen immediately. Use a non-rebreather mask at 15 L per minute while arranging definitive care. This step does not wait for the referral conversation.
3. Identify the nearest hyperbaric programme. Use the Canada Hyperbarics directory of hospitals and regulated facilities with hyperbaric capability across all provinces.
4. Phone the hyperbaric programme directly. Provide the COHb value, time of exposure, time of removal from source, clinical features, troponin if measured, and a contact number. Do not rely on faxed referrals for acute presentations.
5. Stabilise for transport. Continue high-flow oxygen, manage seizures and haemodynamic instability per local protocols, and document baseline neurological examination clearly. Pre-treatment otoscopic findings help the hyperbaric team plan ear-equalisation strategy.
6. Document the indication. Acute CO poisoning is a Health Canada and UHMS approved indication. Clear documentation supports the referral decision and downstream insurer or institutional review.
7. Plan follow-up. Arrange neurology or neuropsychology review at four to six weeks regardless of the apparent acute response, as DNS can emerge weeks after the index exposure.

Province-by-province logistics differ. Physicians in any province can review the regional pathway pages or the coverage overview for province-specific guidance on insured access and referral routing.

Frequently asked questions about HBOT for carbon monoxide poisoning

When should I refer a CO-poisoned patient for hyperbaric oxygen therapy?

Refer for HBOT when any of the following apply: loss of consciousness at any point, persistent neurological signs after 100 percent normobaric oxygen, COHb above approximately 25 percent (lower threshold in children and pregnancy), evidence of cardiac ischaemia, pregnancy regardless of maternal COHb, or an abnormal acute MRI. Recent evidence supports a three-hour initiation window when geography allows.

Does peak COHb predict who develops delayed neuropsychiatric sequelae?

Peak COHb is an imperfect predictor on its own. The strongest predictors in the 2024-2026 literature are Glasgow Coma Scale below 9, rhabdomyolysis (Ho 2025), white matter lesions on acute MRI (Jia 2026), initial venous lactate above 2.8 mmol/L (Satilmis 2026), and the combination of length of stay, comorbidities, and post-discharge HBOT in the four-variable Gong 2026 model (AUC 0.933). Multi-factor risk stratification consistently outperforms COHb-only triage.

What HBOT protocol is standard for acute CO poisoning?

Most Canadian programmes use a 2.4 to 3.0 ATA protocol with 100 percent oxygen, with the precise pressure profile and number of sessions varying by centre and patient severity. The 2026 Gur dosing study found that higher chamber pressure (2.8 ATA) reduced delayed neuropsychiatric sequelae more than longer regimens at lower pressure, suggesting protocols that prioritise reaching the higher pressure are likely the most protective. The referring physician’s role is to facilitate timely arrival; the hyperbaric physician selects the specific protocol.

Should pregnant patients with CO exposure be referred?

Yes. Foetal haemoglobin binds CO with even greater affinity than adult haemoglobin and clears more slowly, so foetal COHb is typically higher than maternal COHb at presentation. Refer pregnant patients regardless of maternal COHb. UHMS and Canadian hyperbaric programmes recognise this elevated foetal risk in standard referral guidance.

What complications should the referring physician anticipate?

Middle ear barotrauma is the most common complication. The 2025 Lee retrospective cohort study found altered mental status at presentation independently increased risk (OR 3.16), while higher serum albumin and magnesium levels were protective. Less common but reported events include sinus barotrauma, transient myopia, and rare oxygen-induced seizures. Pre-treatment otoscopy and clear handover of neurological baseline and nutritional indices help the hyperbaric team mitigate these risks.

Is HBOT for CO poisoning covered by Canadian provincial health plans?

Acute carbon monoxide poisoning treated in a Canadian hospital-based hyperbaric programme is covered by the provincial health plan in every province with a hospital chamber, including OHIP in Ontario, MSP in British Columbia, AHCIP in Alberta, RAMQ in Quebec, and MSI in Nova Scotia. Inter-facility transport for emergency hyperbaric care is generally insured under standard emergency transfer arrangements.

What follow-up should be arranged after acute HBOT?

Arrange neurology or neuropsychology review at four to six weeks even when the acute presentation resolves, because delayed neuropsychiatric sequelae can emerge weeks after the index exposure. Patients with persistent symptoms warrant repeat MRI and consideration of additional hyperbaric sessions in consultation with the hyperbaric programme.

Key takeaways for Canadian referring physicians

• RCT pooled evidence is mixed. The 2026 Fujita meta-analysis of six RCTs found no significant mortality or neurological benefit at any pressure, with low to very low overall quality of evidence. HBOT remains a UHMS Category 1 and Health Canada recognised indication for acute CO poisoning, but the clinical question is increasingly which subgroups benefit rather than whether to refer broadly.
• Chamber pressure matters more than session count. The 2026 Gur dosing study (n=312) showed treatment at 2.8 ATA reduced delayed neuropsychiatric sequelae to 19-20%, versus 36-37% at 2.0 ATA, regardless of session number.
• A three-hour initiation window is supported by exploratory data. The 2026 Xu single-centre study (n=12) reported faster organ-function recovery with HBOT initiated within three hours, though confirmation in larger studies is required.
• Glasgow Coma Scale below 9 and rhabdomyolysis are the strongest clinical predictors of delayed sequelae (Ho 2025, Jia 2026). White matter lesions on acute MRI are the strongest imaging predictor (Jia 2026).
• Initial venous lactate above 2.8 mmol/L is a useful single-measurement triage signal (Satilmis 2026, AUC 0.705); lactate clearance does not add predictive value over the initial reading.
• D-dimer, NLR, and NT-proBNP predict myocardial injury and HBOT need (Gulen 2025); co-manage with cardiology when elevated.
• Middle ear barotrauma is the dominant adverse event. Altered mental status, low albumin, and low magnesium predict higher risk (Lee 2025); pre-treatment otoscopy informs mitigation.
• Pregnant patients warrant referral regardless of maternal COHb due to foetal haemoglobin kinetics.
• Neurology or neuropsychology follow-up at four to six weeks is essential to capture delayed sequelae, which can emerge weeks after the index exposure.

For the directory of Canadian hyperbaric programmes accepting acute CO referrals, see the Canada Hyperbarics list of hospitals and regulated facilities. For the broader evidence base across hyperbaric indications, see the research bank, which is filterable by condition.

Medical disclaimer

This content is for informational purposes only and does not constitute medical advice. Hyperbaric oxygen therapy decisions must be made by qualified clinicians on a case-by-case basis. Always refer to current Health Canada and UHMS guidance, your institutional protocols, and your hyperbaric medicine consultant for individual patient decisions. Canada Hyperbarics is an independent informational resource and does not endorse any specific facility.