Carbon monoxide (CO) poisoning is the most contested of the FDA-recognized HBOT indications. The reason is not that trials are missing. The reason is that the trials disagree.
Two large blinded trials reached opposite conclusions. The 2002 Weaver trial showed that HBOT lowered the rate of late nerve and mood issues. The 1999 Scheinkestel trial showed no benefit at all.
The UHMS, the CDC, and most US poison control centers still endorse HBOT for severe acute cases. The 2011 Cochrane review takes a more cautious stance. This page lays out both sides.
Quick Facts
| Field | Value |
|---|---|
| FDA approval status | Recognized indication (one of 14) |
| UHMS classification | Tier 1 — approved, primary therapy in selected cases |
| Typical protocol | 2.5-3.0 ATA, single session (sometimes 1-3 sessions over 24 hours) |
| Medicare coverage | Yes — covered under NCD 20.29, 2017 update |
| Insurance prior auth | Not required for acute emergent cases |
| Evidence grade | GRADE B (mixed RCTs, strong mechanistic case) |
The evidence
The CO poisoning literature is uniquely split. Eight randomized trials have been run since 1989. They reach different conclusions.
The Weaver NEJM 2002 trial, PMID 12362006 is the most-cited positive trial. It randomized 152 patients with acute CO poisoning to three HBOT sessions or sham. At six weeks, the HBOT group had 25 percent cognitive issues versus 46 percent in the sham group.
The Scheinkestel 1999 Medical Journal of Australia trial reached a different result. It randomized 191 patients with any grade of CO poisoning. At one month, HBOT did not lower the rate of late nerve problems.
The Buckley Cochrane review, 2011 PMID 21491385 pooled seven trials of varying quality. Two found a benefit at one month; four did not. The review judged the overall evidence too weak to support routine use.
The methodological gap matters. Weaver used strict patient selection and a double-blind sham. Scheinkestel used broader inclusion and a different sham design.
A 2024 PMC review by Mintegui et al. argues for HBOT in severe cases based on the totality of evidence. The LITFL Critical Care Compendium summary, 2024 is a balanced clinical overview.
The Hampson et al. 2012 practice recommendations (American Journal of Respiratory and Critical Care Medicine) lay out the criteria most US poison centers now use.
Mechanism of action
CO poisoning kills in two ways. First, CO binds hemoglobin 200 to 250 times more tightly than oxygen, which blocks oxygen delivery to tissue. Second, CO binds mitochondrial cytochromes and triggers a delayed inflammatory cascade.
Hyperbaric oxygen helps on both fronts. At 2.5 to 3.0 ATA, the plasma oxygen content rises high enough to supply tissue even when hemoglobin is fully blocked.
High oxygen pressure also speeds CO release from hemoglobin. The half-life of CO drops from about four hours on room air to roughly 23 minutes at 3.0 ATA.
The delayed neuro-inflammatory pathway is where the science gets richer. HBOT lowers leukocyte adhesion to brain endothelium and blunts lipid peroxidation. These effects may explain why HBOT helps with late issues even when CO is already cleared.
The mechanism is not in doubt. Whether the mechanism translates to better long-term outcomes for every patient is the question the trials disagree on.
Typical protocol
The most common protocol is a single session at 2.5 to 3.0 ATA lasting 90 to 120 minutes. The Weaver trial used 3.0 ATA for 60 minutes followed by 2.0 ATA for 60 minutes, repeated up to three times over 24 hours.
Some centers do a single longer session. Others do two or three sessions in the first 24 hours after admission. The first session should happen as soon as possible, ideally within six hours.
CO half-life on 100 percent oxygen at sea level is about 60 to 90 minutes. So delayed HBOT (more than 12 hours after exposure) treats the inflammatory cascade more than the CO load itself.
Air breaks are used during long sessions to limit oxygen toxicity. Pulmonary and CNS oxygen toxicity become real risks above 2.5 ATA.
Insurance and cost
Medicare covers acute CO poisoning under NCD 20.29, last reviewed 2017. Acute CO is on the covered list. No prior authorization is needed for emergent care.
Commercial insurance generally follows. Cost per emergent session ranges from $1,500 to $4,000. Most cases need only one or two sessions, so the total bill stays modest.
The harder question is access. Many community hospitals do not have a hyperbaric chamber. Patients may be transferred to a regional center, which adds time.
Where to get it
Acute CO cases are routed by the local poison control center, which can be reached at 1-800-222-1222 in the US. The American Association of Poison Control Centers, 2024 coordinates chamber referrals.
The UHMS accredited facility directory, current 2026 lists chambers with verified safety standards. For non-accredited sites, our UHMS accreditation guide covers what to check.
CO treatment requires a chamber capable of 2.5 to 3.0 ATA, which means a hospital-grade hard-shell unit. Soft-shell wellness chambers do not reach this pressure. See our FDA-cleared chambers list for clearance specifics.
Limitations and contraindications
Untreated pneumothorax is an absolute contraindication. Pressurizing a patient with an unresolved pneumothorax can produce a tension pneumothorax.
Relative contraindications include severe COPD, recent thoracic or ear surgery, claustrophobia, and active seizures. Each requires case-by-case judgment.
For carbon monoxide specifically, the patient must be stable enough to spend 90 to 120 minutes in a chamber. Intubated patients require a chamber set up for ventilator support.
Pregnancy is not a contraindication. Pregnant patients with significant CO poisoning are at higher risk for fetal injury because fetal hemoglobin binds CO more avidly. HBOT is recommended at lower carboxyhemoglobin thresholds in pregnancy.
The patient selection question matters most. Severity criteria typically include loss of consciousness, neurologic deficit, cardiac involvement, pregnancy with carboxyhemoglobin above 15 percent, or carboxyhemoglobin above 25 percent in non-pregnant adults.
Active research
ClinicalTrials registry studies on CO and HBOT, current 2026 are looking at multiple-session protocols, biomarkers for who benefits, and long-term cognitive outcomes.
The 2022 network meta-analysis in PMC attempted to compare HBOT against normobaric oxygen across all available trials. The finding was that HBOT may reduce late nerve issues but the effect estimate has wide uncertainty.
Active research questions include the role of cardiac biomarkers (troponin) for selecting candidates, the optimal number of sessions, and whether HBOT helps in delayed presentation (more than 12 hours).
How this compares to off-label HBOT uses
CO poisoning sits in an unusual place. It is FDA-recognized and UHMS Tier 1, yet the strongest single trial (Scheinkestel) was negative. This shows that recognized indications are not always settled science.
The contrast with wellness-clinic off-label HBOT is still sharp. Off-label uses like long COVID, brain injury, and anti-aging do not have a Weaver-equivalent positive trial. They do not have UHMS endorsement or Medicare coverage.
For context on how off-label HBOT markets diverge from clinical evidence, read our analysis of institutional silence on HBOT and our breakdown of marketing-driven HBOT claims.
The honest summary: CO is recognized, mechanistically clear, and supported by one strong positive trial. It is also the indication with the most legitimate scientific debate. Both can be true.
Frequently asked questions
Does HBOT help every CO poisoning case?
No. The evidence is strongest for severe cases (loss of consciousness, neurologic deficit, cardiac involvement). Mild cases without those features may do equally well on normobaric oxygen alone.
How fast does HBOT need to start?
Within six hours is the operational target for highest benefit. After 12 hours, the treatment shifts from clearing CO to blunting late inflammation, with smaller effect size.
Is one HBOT session enough?
Many centers do a single session. The Weaver protocol used three sessions in 24 hours. Both approaches are in use. No trial has cleanly compared single versus multiple sessions head-to-head.
What about delayed neuropsychiatric sequelae?
These are the late cognitive and mood symptoms that can appear two to 40 days after recovery. The Weaver trial found HBOT lowered this rate. The Scheinkestel trial did not. This is the core area of disagreement.
Can a soft-shell wellness chamber treat CO poisoning?
No. CO treatment requires 2.5 to 3.0 ATA. Soft-shell chambers do not reach that pressure and are not cleared for this use.
Sources
- Weaver et al. NEJM 2002, PMID 12362006
- Scheinkestel et al. Med J Aust 1999
- Buckley et al. Cochrane Review 2011, PMID 21491385
- Hampson et al. Am J Respir Crit Care Med 2012
- LITFL Critical Care Compendium summary, 2024
- Network meta-analysis, PMC 2022
- Mintegui review, PMC 2024
- CMS NCD 20.29 Hyperbaric Oxygen Therapy, 2017
- UHMS accredited facility directory, current 2026
Medical disclaimer
This page is medical journalism, not medical advice. Suspected carbon monoxide poisoning is an emergency. Call 911 or your local poison control center at 1-800-222-1222 immediately. HBOT treats CO poisoning in selected cases; it does not cure all CO-related injury and outcome depends on severity, timing, and patient selection. Talk to an emergency physician or hyperbaric specialist about any HBOT decisions.