One Trial, Eighty-Three Patients, No Control Arm: Hyperbaric Oxygen and CRPS

Abstract medical diagram of a forearm and wrist mapped with a thermal gradient, cool teal shading into amber and red across the hand

Ask the staff at a hyperbaric centre whether oxygen under pressure helps complex regional pain syndrome and the answer tends to arrive with a number attached. In the largest published patient series, 86 per cent reported relief. Ask a methodologist the same question and the answer arrives with a much smaller number: of eleven randomised trials of hyperbaric oxygen in chronic pain identified in a systematic review published this spring, exactly one enrolled patients with CRPS, and not a single trial in the set was judged at low risk of bias.

Both numbers are in the literature. The gap between them is the story.

A syndrome defined by arguments about its own definition

CRPS is what happens when an ordinary injury, a fracture or a sprain or a surgery, refuses to settle. The limb becomes burning and hypersensitive, the skin changes temperature and colour, joints stiffen, and the whole thing can spread. The National Institute of Neurological Disorders and Stroke, in the consumer summary carried by MedlinePlus, is blunt about the state of the art: the cause is unknown, there is no specific diagnostic test, there is no cure, and treatment “focuses on relieving the pain” through medicines, physical therapy and nerve blocks. A 2024 review in The Lancet Neurology calls it “a rare pain disorder that usually occurs in a limb after trauma” and concedes that “effective treatment of CRPS remains a challenge”.

That review also notes something that matters for every trial in this field: the diagnosis is still moving. The Budapest criteria, which first gave researchers a validated clinical definition, were published in Pain in 2010. In September 2025 a group of Italian rheumatologists argued in Diagnostics that they should be reopened, adding bone marrow oedema on MRI and hyperalgesia as necessary features, and listed hyperbaric oxygen among the treatments worth adding. The proposal has not been adopted. But a trial that enrolled “CRPS” patients in 2003 and one that enrols them in 2026 with the CRPS Severity Score may not be studying the same population.

Why oxygen looked plausible in the first place

The rationale is not foolish. The Journal of Pain Research review sets it out carefully: microvascular dysfunction and prolonged inflammation leave tissue relatively hypoxic, hypoxia stabilises HIF-1α, HIF-1α drives TNF-α and IL-1β, and those cytokines lower the firing threshold of nociceptors. Local acidosis activates acid-sensing channels and TRPV1, and sustained input then remodels the dorsal horn and wakes up microglia, which is how a peripheral injury becomes a central pain state. Hyperbaric oxygen at 1.5 to 3.0 atmospheres raises dissolved plasma oxygen, suppresses HIF-1α signalling, and in animals produces an antinociceptive effect through nitric-oxide-dependent release of endogenous opioids.

The authors add the caveat themselves: this is “mostly based on preclinical evidence”. A mechanism that fits a disease is a reason to run a trial. It is not a result.

The 2004 trial that still carries the argument

The one randomised study in CRPS comes from a Turkish military medical academy in Istanbul, published in 2004 in the Journal of International Medical Research. Kiralp and colleagues enrolled 71 patients with post-traumatic CRPS and gave 37 of them hyperbaric oxygen at 2.4 atmospheres for 90 minutes, fifteen sessions over three weeks; the other 34 got the same chamber schedule breathing ordinary air. Everyone received paracetamol. Pain, wrist swelling and wrist range of motion were measured before treatment, at the end, and on day 45.

The results were strong. Pain on a visual analogue scale fell by 3.1 points in the oxygen group, with a significant between-group difference that held at 45 days. Wrist flexion improved by 19.3 degrees and wrist circumference fell by 1.9 centimetres, both highly significant; only wrist extension failed to separate the arms. As hyperbaric shams go, this one was better than most: both groups were pressurised, and the only difference was the gas.

Then the asterisk. When the Antwerp reviewers put the trial through the Cochrane risk-of-bias tool, it came out at high risk, and the reason given is specific: “unclear allocation procedures (alternate allocation)”. Alternating patients into treatment and control is not randomisation in any meaningful sense. Whoever decides who enters the study knows who gets the chamber next, and in an unblinded-to-allocation trial with subjective outcomes, that is the door through which bias walks.

The cohort was young, predominantly male, post-traumatic and treated early, which is not the population that fills CRPS clinics today. And the paper appeared six years before the Budapest criteria were validated, so its diagnosis rests on clinical judgement rather than a standard instrument.

Eighty-three patients and no control arm

The largest human dataset in CRPS hyperbaric medicine is a retrospective case series from the Centre of Hyperbaric Medicine at Ostrava City Hospital, published in 2024. Eighty-three patients treated in 2018 and 2019 at 2.0 to 2.4 atmospheres once daily, averaging 22 sessions, produced 86 per cent reporting relief. Pain at rest fell from 3.2 to 1.6, pain on activity from 6.1 to 3.7, and a functional score for the affected limb from 7.0 to 4.3, all with p values below 0.001. Of the 79 patients assessed at the end, 23 were classed as having a large clinically significant response and 48 a partial one. There were no serious complications.

None of that is nothing. Eighty-three patients is real clinical experience, and the direction of change is consistent. The authors also found that the earlier after diagnosis treatment began, the larger the effect, significant at the three and six month cut-offs.

But a case series cannot answer the question being asked of it. CRPS fluctuates, classically moving from a warm, swollen, acutely painful phase into a colder, stiffer chronic one, and people seek treatment at the worst moment of the curve, when some improvement is likely regardless. A single-arm series cannot separate the treatment from that trajectory, from regression to the mean, or from the attention that comes with twenty-plus sessions in a specialist centre. The 86 per cent figure describes what happened to those patients; it is not an estimate of what the chamber did.

Eleven trials, one of them CRPS, and a field arguing with itself

The March 2026 systematic review in The Journal of Pain Research is the most useful document in this field precisely because it is unexciting. Leys and colleagues at Antwerp followed PRISMA, searched MEDLINE, EMBASE and the Cochrane Central Register of Controlled Trials to 22 August 2025, and restricted themselves to randomised trials in adults with chronic pain. Sixty-nine records yielded eleven eligible trials and 480 patients.

Eight of those eleven trials were in fibromyalgia. One was in idiopathic trigeminal neuralgia, one in radiation-induced brachial plexopathy, and one in post-traumatic CRPS. Sample sizes ran from 17 to 71, protocols from 1.45 to 2.4 atmospheres, 15 to 60 sessions and 70 to 100 minutes. Heterogeneity was judged too great to pool, so the synthesis is narrative.

No trial was at low risk of bias. Seven were rated “some concerns”, four high risk, and the high-risk group included the two studies the review describes as having unclear, possibly alternate allocation: the CRPS trial and a fibromyalgia trial from the same Istanbul group. The review’s conclusion is quoted in full because the wording matters: “HBOT may hold potential as a safe and useful addition to current multimodal management strategies for selected chronic pain syndromes; however, the available data remain insufficient to support firm conclusions.” Robust conclusions, it adds, require large, well-designed randomised trials “before HBOT can be integrated in routine chronic pain management”.

The review has its own soft spots. Study selection and data extraction were done by a single reviewer, a known source of error, and the authors say so. They also screened ten previous reviews and meta-analyses for trials they might have missed and found none. So the count, thin as it is, appears to be the count: 71 of 480 randomised chronic-pain patients, fifteen per cent of the evidence base, carry the entire CRPS claim.

In April 2025 the Clinical Journal of Pain had published a review by Cha, Bosco, Moon and Melloni, including Duke’s Richard Moon, a leading hyperbaric authority. It cast a wider net than the systematic review, thirteen studies and 280 participants including case series and retrospective cohorts, and split the patients by pain mechanism: 42.5 per cent sympathetically maintained, 48.2 per cent sympathetically independent, 9.3 per cent indeterminate. Treatment ran from 3 to 63 sessions, typically 2.4 atmospheres for 90 minutes. The authors concluded that hyperbaric oxygen “may offer significant symptom relief, regardless of CRPS subtype or disease duration” and called it “a valuable noninvasive option” whose anti-inflammatory and neuroplasticity-promoting properties might make it worth combining with other treatments.

That is a stronger claim than the evidence hierarchy usually supports, and the correspondence pages have been busy since: two letters in the same journal, the first in September 2025 carrying the review authors’ own names, the second in January 2026 from Hájek and Klugar in Ostrava. Neither is indexed with an abstract, so the argument itself sits behind a paywall, but the exchange is evidence that the field knows it has not settled. The Ostrava group has also published a two-page article in Medical Gas Research whose title calls hyperbaric oxygen “a novel treatment option in complex regional pain syndrome and other chronic pain conditions” — a position statement from an experienced centre, not a study.

Nothing on the indication list, nothing on the payer list

Here the documentary record is unambiguous, and it is the part patients rarely hear.

The Undersea and Hyperbaric Medical Society’s fifteenth-edition indications list, the specialty’s own statement of what hyperbaric oxygen is for, carries fifteen indications across seventeen numbered entries. CRPS is not among them. Neither the word “sympathetic” nor “dystrophy” nor the phrase “chronic pain” appears anywhere in that document.

Medicare is equally silent. National Coverage Determination 20.29 lists fifteen covered conditions and twenty-two that are explicitly not covered, and CRPS appears on neither list, because the rule is that anything unlisted is excluded: “No program payment may be made for any conditions other than those listed.” Arthritic diseases sit on the non-covered side.

The Food and Drug Administration regulates the devices rather than the indications, and its most recent word on them is a safety letter dated 25 August 2025. Devices are Class II, cleared through the 510(k) pathway under product code CBF. The letter reports fires “that resulted in serious injuries and deaths”, says the root cause is undetermined, and asks providers to follow instructions for use, ground equipment, train staff, monitor throughout, exclude electrical and static-producing devices, and dress patients in cotton. It says nothing about CRPS.

The society’s own warning is worth repeating here, because chronic pain patients are exactly the customers the wellness market is looking for. Low-pressure “mild hyperbaric chambers” in spas and wellness centres are, in the UHMS’s words, generally not physician-prescribed or supervised, are used for conditions “for which there is no scientific proof for any type of hyperbaric oxygen treatments”, and put “those exposed at risk for serious injury and even death”.

What would actually settle it

Something is finally running. A randomised controlled trial in CRPS is recruiting at two sites in Ostrava, registered in February 2026 as NCT07446712. It plans to enrol 60 adults with CRPS type I, randomised one-to-one to immediate hyperbaric oxygen or to usual care followed by the same treatment four to six weeks later, twenty to thirty sessions in all. The primary outcomes are active wrist range of motion and the WHODAS 2.0 disability score; pain intensity, a 30 per cent responder analysis, the CRPS Severity Score and SF-12 quality of life are secondary. Estimated primary completion is April 2028.

The design is the right shape in the ways that count: a modern diagnosis limited to type I, a validated severity instrument and a functional primary outcome rather than pain alone, registered before enrolment, which protects against the outcome-switching that has muddied so much hyperbaric research. It is also, by the standards of a definitive trial, small and unblinded. A delayed-start design controls for time and expectation but not for the experience of being in a chamber with staff attention five days a week, and sixty patients will not yield a precise estimate. The trial is run by the same centre that published the 83-patient series and the “novel treatment option” article, which is normal in a field this small and a reason to read the result closely.

Until then, what a patient can honestly be told is this. Hyperbaric oxygen at 2.0 to 2.4 atmospheres appears safe in this population; the commonest problem in the pain trials was mild middle-ear barotrauma, reported in 12 to 40 per cent of participants, and the review found no serious or permanent adverse events. It is expensive, it takes weeks, and an unlisted indication usually means paying out of pocket. Whether it works for CRPS is a question that has had one underpowered, quasi-randomised answer for twenty-two years, and will not have a better one before 2028.

The field has been unusually candid about that. It is the marketing that has not been.

Sources

← Back to blog