Long COVID has driven a surge of HBOT-related interest unmatched since the autism wave of the early 2010s. The reason is one specific paper: a 2022 randomized controlled trial conducted at the Shamir Medical Center in Israel by Zilberman-Itskovich, Hadanny, Efrati, and colleagues, published in Scientific Reports.
This single study has become the citation backbone for thousands of clinic marketing pages, dozens of news articles, and an entire wave of cash-pay HBOT-for-long-COVID protocols at $200-$400 per session for 40-session courses.
The study is real. Its results are interesting. But what it does and doesn't show is different from what most clinic marketing claims it shows. This piece reads the paper carefully, separates the documented effects from the marketing extensions, and offers a calibrated framework for patients considering off-label HBOT for long COVID.
The Shamir RCT: what the study actually was
The full citation: Zilberman-Itskovich S, Catalogna M, Sasson E, et al. "Hyperbaric oxygen therapy improves neurocognitive functions and symptoms of post-COVID condition: randomized controlled trial." Scientific Reports (2022). Published in Scientific Reports, it remains the largest sham-controlled RCT of HBOT for long COVID to date.
The design was a single-center, double-blind, randomized, sham-controlled trial. Seventy-three adults with persistent post-COVID symptoms at least three months after diagnosis were randomized to either:
- Active HBOT: 40 sessions, 90 minutes each, at 2.0 atmospheres absolute (ATA) with 100% oxygen
- Sham control: 40 sessions in the same chamber design, but at 1.03 ATA (room air at very slightly elevated pressure) with mixed air
Outcomes were measured at baseline and after the 40-session protocol completion. Primary outcomes included neurocognitive function, sleep, pain, psychiatric symptoms, and brain MRI markers.
What the trial found
The published results showed:
Neurocognitive function — The active-HBOT group showed statistically significant improvements vs sham on measured cognitive scores, particularly in attention, executive function, and information processing speed. Effect sizes were modest but consistent across the cognitive battery used.
Psychiatric symptoms — Improvements in measures of energy, sleep quality, and certain mood-related symptoms. Some of these crossed clinical-significance thresholds.
Brain MRI — Changes in functional MRI patterns and certain microstructural measures, suggestive of altered brain connectivity. The clinical relevance of these imaging findings is still being debated.
Self-reported quality of life — Improvements in the active group on patient-reported outcome measures.
The paper concluded that HBOT "induces neuroplasticity and significantly improves neurocognitive functions, psychiatric symptoms, fatigue, sleep, and pain in patients suffering from post-COVID condition at the chronic neurocognitive symptomatic stage."
What the trial doesn't show
This is where careful reading matters. The Shamir RCT is a strong starting point. It is not the established-evidence base it gets cited as.
Sample size limitations. Seventy-three patients across two groups is small. For comparison, the major RCTs that established efficacy for FDA-approved HBOT indications typically enrolled hundreds of patients across multiple centers. A single 73-person study, even a well-designed one, is a starting point that needs replication — not a conclusion.
Single-center design. The trial was conducted entirely at one Israeli medical center, with one research team, on one specific patient population (mostly Israeli adults, mean age around 50). Results from a single-center trial may not generalize to other populations or healthcare contexts.
Short follow-up. The 3-month post-protocol assessment is the longest published follow-up. We don't yet know whether the improvements persist 6 months out, 12 months out, or longer. Long COVID itself can wax and wane spontaneously; without longer follow-up, durability is uncertain.
Heterogeneity of long COVID. Long COVID is not one disease. Different patient phenotypes (predominantly fatigue, predominantly cognitive, predominantly autonomic dysfunction, predominantly pain) may respond very differently. The Shamir RCT did not stratify patients by phenotype in ways that let us know who's most likely to benefit.
Effect sizes are modest. The improvements, while statistically significant, were modest in absolute terms. Some measures showed effect sizes around 0.4-0.6 (small to moderate). A statistically significant effect with a moderate effect size is meaningful research but should not be marketed as "transformative" or "life-changing."
No replication yet. As of mid-2026, no independent research group has published a comparably-designed replication study with similar findings. There are several active trials (see ClinicalTrials.gov) but published peer-reviewed replications are pending.
Conflicts of interest. Several authors of the original study are affiliated with the Sagol Center, which provides commercial HBOT services. The financial relationship between researchers and the treatment they're studying is a known concern in biomedical research and deserves disclosure even when properly handled methodologically.
Selection bias in patient cohort. Patients who self-select for an HBOT RCT may be different from the general long-COVID population — more motivated, more financially secure, more open to interventions. The results may generalize less well to patients without those characteristics.
How clinic marketing reframes the Shamir RCT
In the year since publication, the Shamir RCT has been cited in thousands of clinic marketing pages. The transformation pattern is consistent:
Original paper: "HBOT improved neurocognitive function in this specific cohort of 73 patients at 3-month follow-up; replication needed."
Marketing translation: "HBOT is the first proven treatment for long COVID."
Original paper: "Effect sizes were modest but consistent."
Marketing translation: "Patients report life-changing improvements."
Original paper: "Mechanistic findings suggest neuroplasticity; clinical relevance unclear."
Marketing translation: "HBOT heals the brain damage caused by COVID."
The marketing version is not what the paper says. It's what the paper might support after replication, larger trials, longer follow-up, and FDA review. None of those things have happened yet.
What's in flight
Multiple active or recently-completed clinical trials may update the evidence base:
- ClinicalTrials.gov NCT04953468 — A US trial of HBOT for long COVID
- NCT05084664 and related — Various sham-controlled designs in different patient populations
- A prospective registry at Shamir Medical Center continues to track long-term outcomes in their patient population
- Yale Medicine and other US academic centers have ongoing observational studies
Results from these may shift the picture in either direction. Some may confirm the Shamir findings; some may show smaller effects; some may show no benefit. The honest stance is: we should wait for results before treating the Shamir RCT as the established conclusion.
What major institutions actually say
Yale Medicine has published a balanced overview of HBOT for long COVID — describing the Shamir results, acknowledging the limited evidence base, and noting the need for replication. They do not endorse HBOT as standard treatment.
The CDC has not recommended HBOT for long COVID in any of its published guidance for clinicians or patients managing post-COVID conditions.
The NIH lists HBOT among the experimental interventions being studied for long COVID but has not endorsed routine use.
Major insurers have not added HBOT to their covered services for long COVID. This is the most concrete market signal: insurers underwrite treatments where evidence supports payment, and they haven't moved.
The FDA has not approved HBOT for long COVID. The 14 FDA-approved indications do not include any post-viral condition.
This is the bridge framing: the Shamir RCT is real evidence, it deserves serious attention, and it doesn't yet meet the threshold institutional medicine uses for endorsement of routine treatment.
What a long-COVID patient considering HBOT should know
-
The single RCT is real. Don't dismiss HBOT for long COVID. The Shamir study is methodologically credible, sham-controlled, and showed statistically significant improvements in cognitive function. This isn't snake oil.
-
One trial isn't replication. Pause before treating a single-center study as established consensus. Replication is the basic standard of medicine for a reason.
-
Effect sizes were modest. If you go in expecting "life-changing" results, your expectations exceed what the published evidence supports. Set expectations to "may provide modest improvement on some cognitive measures with effect sizes around 0.4-0.6 standard deviations."
-
The cost is significant. Forty-session protocols at $200-400 per session run $8,000-$16,000+ out of pocket. Insurance doesn't cover it because the evidence base hasn't yet met underwriting standards.
-
Long COVID can improve on its own. Some patients improve over months even without intervention. Distinguishing HBOT effects from natural recovery requires the kind of controlled comparison the Shamir RCT provided — but for any individual patient, you won't have your own control.
-
Other interventions have evidence too. Graded exercise (carefully — this is controversial in some long-COVID phenotypes), pacing strategies, cognitive rehabilitation, sleep optimization, and specific symptom-targeted medications all have evidence bases. Some are cheaper and more accessible.
-
Look for FDA-cleared facilities. If you do pursue HBOT, verify the clinic uses FDA-cleared chambers and ideally UHMS-accredited facilities. The chamber-safety baseline matters even for off-label use.
-
Consider participating in research. If a clinical trial in your area is enrolling, participation is a way to access HBOT (often free or low-cost) while contributing to the evidence base. ClinicalTrials.gov has the active trial list.
-
Talk to a physician without financial stake. A clinic selling the protocol is not the best source for weighing tradeoffs. Your primary care physician, a long-COVID specialist if accessible, or a neurologist if cognitive symptoms are primary, can offer perspective independent of the clinic's revenue.
The honest synthesis
HBOT for long COVID is not snake oil. It's also not a proven treatment.
The Shamir RCT is the best evidence we have, and it shows real but modest improvements in a specific cohort under specific conditions, without yet being replicated or followed up long-term. The marketing extends well beyond what the single study supports.
A patient considering HBOT for long COVID is making a real bet — there's meaningful evidence the treatment may help, and meaningful uncertainty about whether the help will be substantial, durable, or worth the cost for their specific situation.
The institutional position (CDC, NIH, FDA, major insurers, most academic centers) is: not yet. That's not a verdict that HBOT doesn't work. It's a verdict that the evidence is preliminary. For some patients, "preliminary" plus their specific suffering may make the bet worth taking. For others, the cost and uncertainty may not justify the choice.
This page exists to help patients make that decision with full information. Anyone marketing HBOT for long COVID as "proven" is overstating the case. Anyone dismissing it as worthless is understating the evidence. Reality sits in the carefully-mapped middle.
Sources and further reading
- Zilberman-Itskovich et al., 2022 — Hyperbaric oxygen therapy improves neurocognitive functions and symptoms of post-COVID condition — The Shamir RCT, original paper
- Yale Medicine — HBOT and Long COVID — Balanced clinical perspective
- CDC — Post-COVID Conditions — Public health guidance (HBOT not included)
- ClinicalTrials.gov — HBOT for long COVID trials in progress
- UHMS — Hyperbaric Oxygen Therapy Indications — The 14 FDA-approved uses (long COVID not on list)
- FDA — Hyperbaric Oxygen Therapy: Get the Facts
Frequently asked questions
Has the Shamir RCT been replicated? As of mid-2026, no independent research group has published a comparable replication study. Several trials are in progress at other centers (ClinicalTrials.gov has the active list). The Shamir team has continued to publish follow-up work, but independent replication by a different research group is what the field needs.
Why don't insurers cover HBOT for long COVID if the trial showed benefit? Insurers typically require multiple replicated trials and FDA approval before adding a treatment to covered services. A single-center RCT, even a well-designed one, doesn't meet that threshold. This is the standard process, not a conspiracy.
Could HBOT make long COVID worse? The Shamir RCT didn't report serious adverse events at meaningfully different rates from sham. HBOT does carry baseline risks (barotrauma, oxygen toxicity, ear injury) at any indication. The risk profile for long-COVID patients specifically is not yet well-characterized in published literature.
Is the 40-session protocol the only one studied? The Shamir RCT used 40 sessions at 2.0 ATA for 90 minutes each. Other protocols (shorter, lower pressure, more sessions) have not been evaluated against this in head-to-head trials. Clinics may offer different protocols based on cost or convenience — without published data showing equivalence.
What if I can't afford $10,000+ for HBOT? Several alternatives have evidence bases: pacing strategies (free), cognitive rehabilitation programs (often insurance-covered), targeted medications for specific symptoms, sleep optimization, and graded activity programs where appropriate. Consider whether your symptoms might respond to any of these before paying out-of-pocket for HBOT.
Medical disclaimer: This article is for informational purposes only and does not constitute medical advice. Long COVID is a serious condition with real impact on quality of life. Decisions about any treatment, including off-label HBOT, should be made in consultation with a qualified physician with expertise in post-COVID conditions. The information here reflects published evidence as of 2026; new trial results may change the picture.
— The HBOT Finder Team