Hyperbaric Oxygen in Skull Base Osteomyelitis: 1,628 Patients of Evidence, None of It a Trial

Stylised illustration of a skull-base bone erosion with oxygen molecules diffusing through trabecular bone

Skull base osteomyelitis usually starts as an infected ear canal. By the time it reaches the bone at the floor of the skull it is a different disease: a progressive osteitis that tracks along the temporal bone toward the clivus, takes the lower cranial nerves with it, and — in the older name still used in ear, nose and throat clinics — earns the word “malignant” not because there is a tumour anywhere, but because of what it does to the patient.

A systematic review and meta-analysis published on 20 October in Neurology India has now assembled what the world actually knows about it: 45 studies, 1,628 patients, pooled with a random-effects model on the logit scale and reported under PRISMA 2020, with study quality scored on the Newcastle-Ottawa scale. Buried in the conclusions is a sentence that matters more to the hyperbaric field than any of the epidemiology. The evidence for adjunctive hyperbaric oxygen therapy, the authors write, “remains limited to uncontrolled series and a single small comparative study, with no randomized trials, so its efficacy is unproven.”

That is the verdict of a paper that has just spent 45 cohorts looking, and it lands in a therapeutic corner where hyperbaric chambers have been used for decades.

What the 1,628 patients look like

The demographic picture is narrow and elderly. Median age 68 years (interquartile range 57 to 76), diabetes mellitus present in 74.2 per cent. Pseudomonas aeruginosa predominated at 62.4 per cent of identified organisms, with ciprofloxacin-resistant strains in 11.3 per cent of post-2000 series — a number that has been climbing and that matters, because the oral fluoroquinolone step-down has been the backbone of outpatient management. Fungal disease, principally Aspergillus, accounted for 7.8 per cent of cases and was consistently associated with higher mortality. Overall mortality across the review was 15.3 per cent.

The pooled proportions read slightly differently from the raw figure, and the difference is worth noticing rather than smoothing over. From the primary cohorts the meta-analysis pooled diabetes at 82.4 per cent (95% CI 68.2 to 91.1), Pseudomonas at 55.6 per cent (48.4 to 62.6), clinical cure at 84.7 per cent (74.6 to 91.2), cranial nerve involvement at 28.0 per cent (16.1 to 43.9) and mortality at 8.4 per cent (4.3 to 15.8). Pooled mortality of 8.4 per cent against an overall 15.3 per cent is not a contradiction — they are answers to different questions across four decades of publications — but the review does not reconcile them, and a reader should treat the gap as a reminder of how heterogeneous these 45 cohorts are.

Diagnostically the paper is more useful than therapeutically. Gallium-67 scintigraphy showed the highest sensitivity for treatment monitoring at 94 per cent; FDG-PET/CT exceeded 96 per cent. Combined fluoroquinolone-based therapy produced a cure rate of 71.4 per cent. Emerging resistance, the authors conclude, is now “a pressing clinical challenge,” and standardised diagnostic criteria and prospective multicentre trials are “urgently required.”

The disease that does not always start in the ear

Skull base osteomyelitis is not only an otological event. A separate systematic review of central skull base osteomyelitis — the variant arising from the nasopharynx and sphenoid rather than the ear canal — pooled 25 studies and 44 patients and found a different clinical signature: refractory headache in 95 per cent, cranial nerve paralysis in 82 per cent, and Pseudomonas in only 32 per cent. Otogenic infection can spread centrally without obvious ear symptoms, the authors warn. That presentation — a patient with a headache, a normal-looking ear and a nerve palsy — is precisely the case where a hyperbaric referral arrives late and the chamber is asked to rescue tissue that has been infected for months.

Twenty years of an empty Cochrane review

If the new meta-analysis is the clearest statement of the problem, the Cochrane Library has been carrying its echo since 2005. Phillips and Jones registered a review titled Hyperbaric oxygen as an adjuvant treatment for malignant otitis externa, searched CENTRAL, MEDLINE and EMBASE to April 2004, and found no randomised controlled trial to include. Published in 2005 as CD004617, its results section reads, in full effect, that “due to the lack of data no results could be presented.”

They updated it in 2013. The search ran to 4 April 2013, through the specialty trials register, CENTRAL, PubMed, EMBASE, CINAHL, Web of Science, ICTRP and unpublished sources. The update identified nothing either: “no identified articles described randomised controlled trials,” therefore no results, therefore no comparison of complication rates between treatment modalities. The review’s single substantive conclusion is that “no clear evidence exists to demonstrate the efficacy of hyperbaric oxygen therapy when compared to treatment with antibiotics and/or surgery.”

An empty systematic review is routinely misread in both directions. It does not show that hyperbaric oxygen fails in this disease — a review with zero included trials cannot fail anything. It shows that in roughly two decades nobody has randomised a single patient. That is an unusual state of affairs for a therapy delivered in dozens of sessions alongside intravenous antibiotics and surgery, and it is not for lack of patients: the 45 studies in the Neurology India review found 1,628 of them.

What the uncontrolled series show, and why they persuade

The published experience is not worthless, and it is not small in aggregate. A systematic review by Byun and colleagues at the Medical University of South Carolina collected 16 studies covering 58 patients treated with adjuvant hyperbaric oxygen for malignant otitis externa: mean age 68.0 years, diabetes in 94.7 per cent, Pseudomonas in 64.3 per cent, facial nerve involvement in 55.2 per cent. Disease cure was 91.4 per cent and all-cause mortality 8.6 per cent; among patients with facial nerve involvement, 72 per cent recovered function. Their own verdict is careful to the point of tension: hyperbaric oxygen “may be an effective treatment option for refractory or advanced” disease, “but its efficacy remains unproven due to lack of strong scientific evidence.”

A 2025 review by Bazzal and colleagues in International Archives of Otorhinolaryngology is blunter about the material. Eight studies met inclusion — case reports, observational studies and case series, assessed with the Cochrane risk-of-bias tool and the Newcastle-Ottawa scale. Findings were consistent in direction and, in the authors’ words, “the current evidence is insufficient to establish HBOT as a standard treatment.” They nonetheless recommend considering it for refractory cases where a chamber is accessible, which is a reasonable position and an unproven one at the same time.

Single-centre series then supply the numbers that make the practice feel settled. In Oman, a naval base polyclinic treated 20 consecutive patients with an average of 29.0 ± 8.9 sessions and reported 95 per cent cured. In Lisbon, a dedicated hyperbaric centre reported 16 patients referred between 1998 and 2016 — 53 per cent of them already at the most advanced stage of the disease on a clinicopathological classification — with a mean of 33 sessions, no fatalities attributable to the infection, and a hospital stay before hyperbaric treatment began (median 90 days) significantly longer than the interval from starting treatment to cure (p = 0.028, Wilcoxon signed-rank). That last comparison is a real observation and also an unblinded one, made by the team delivering the therapy.

Here is the methodological problem in one line: a cure rate of 91.4 per cent assembled from published case reports and series is a statement about what gets written up as much as about what happens inside a chamber. Uncontrolled series that fail do not reach PubMed at the same rate as the ones that work.

The one comparison, and a German programme that keeps publishing

The closest thing to head-to-head evidence is a retrospective Tunisian study of 42 diabetic patients treated over nine years, 23 with antibiotics alone and 19 with antibiotics plus hyperbaric oxygen at an average of 20 sessions. Total recovery was 100 per cent in the hyperbaric arm and 74 per cent in the antibiotics-only arm. Otalgia resolved on day 5 with oxygen and day 11 without; otorrhoea on day 6 with and day 13 without. All six recurrences occurred in patients who did not receive oxygen. The authors conclude that hyperbaric oxygen “must be associated” with treatment.

It is a striking result and it is not a trial. Patients were not randomised, allocation is not described as blinded or protocolised, there is no adjustment for disease stage or comorbidity, and a clinician who sends sicker patients to the chamber would bias the result against the therapy while one who sends healthier ones inflates it. The Neurology India authors’ phrase — “a single small comparative study” — is the honest description of a literature in which this passes for the best available comparison.

The most productive recent programme is in Düsseldorf, and it illustrates both the promise and the trap. A 2025 retrospective of 42 patients treated at the university ear, nose and throat department between 2006 and 2023 found Pseudomonas in 68.4 per cent of tissue samples, cranial nerve deficits in 61.9 per cent, and hyperbaric oxygen delivered to 66.7 per cent of the cohort. Of those treated, 20 of 23 (87 per cent) improved with regression on follow-up imaging, and the authors argue that patients with skull base osteomyelitis “should be sent to a center that offers HBO therapy.”

A 2026 paper from the same department extends the series to 52 patients treated between 2006 and 2025 and focuses on antimicrobial strategy: 301 swabs and biopsy samples, a pathogen identifiable only on biopsy in 11.5 per cent of cases and no pathogen found at all in 19.2 per cent despite repeated sampling, a mean of 31.9 days of intravenous antibiotics, surgery in 71.2 per cent, and hyperbaric oxygen in 55.8 per cent. Neither paper has a control arm, and — this is the part that tends to get lost when the two are cited as mutual support — the cohorts overlap. Forty-two patients to 2023 and 52 patients to 2025 from one institution are not two independent confirmations of anything; they are one centre’s experience, published twice.

Where the rules actually sit

The regulatory picture is messier than either camp implies. The Undersea and Hyperbaric Medical Society’s fifteenth-edition indication list carries “10. Osteomyelitis (Refractory)” as a recognised use. Medicare’s national coverage determination for hyperbaric oxygen therapy (NCD 20.29, effective 2017) covers, in its own wording, “chronic refractory osteomyelitis, unresponsive to conventional medical and surgical management.” Skull base osteomyelitis is not named separately in that document — we re-read the covered-conditions list to check — but a refractory osteomyelitis of the skull base sits inside the wording far more comfortably than, say, an elective cosmetic indication would.

In Europe the same case can encounter the opposite judgement. The Portuguese hyperbaric centre that published its 16-patient series noted that at a tenth consensus conference on hyperbaric medicine this disease had been classified as a non-indication for hyperbaric oxygen. Its own data are offered as a counterargument. So the identical patient may be treated as standard adjunctive care in one country, formally outside the indication list in another, and reimbursed in a third, with no trial anywhere to arbitrate.

What is not ambiguous is the device safety layer, which applies whatever the indication. The FDA’s provider letter of 25 August 2025 reminds clinicians that hyperbaric oxygen therapy devices are Class II devices cleared through the 510(k) pathway under Product Code CBF, that the agency is aware of “reports of fires that occurred with HBOT devices that resulted in serious injuries and deaths,” that the root cause of those events is not currently known, and that instructions for use, fire prevention measures, staff training, patient monitoring for the duration of treatment, maintenance intervals, prohibition of electrical or static items and hyperbaric-compatible clothing such as cotton all apply. A chamber treating a skull base infection in an elderly diabetic patient carries exactly the same fire-risk obligations as any other chamber, and a post like this one should not be read as an argument for relaxing them.

It is also worth separating this discussion from the wellness market. The regimens in the series above run to 20 to 33 sessions of hyperbaric oxygen as an adjunct to intravenous antibiotics and, in most cohorts, surgery. Nothing in this evidence base speaks to mild or soft-chamber use for fatigue, recovery or longevity, and the FDA has not cleared hyperbaric oxygen for those purposes.

What would actually settle it

A trial is not complicated to design. Adults with skull base osteomyelitis meeting standardised diagnostic criteria, stratified by otogenic versus central origin, randomised to guideline antimicrobial therapy with or without a fixed hyperbaric protocol, with clinical cure at 12 months confirmed on FDG-PET/CT as the primary endpoint and facial nerve recovery, surgery, relapse and mortality as secondary ones. The obstacles are practical: the disease is uncommon, a single hyperbaric centre sees a handful of cases a year, and equipoise is genuinely contested — a clinician who has watched an 87 per cent imaging-response rate will not easily randomise the next patient to no chamber, while a European consensus that calls the therapy a non-indication is not going to fund the protocol.

That standoff has held for two decades, and its cost is visible in the literature: a 1,628-patient evidence base built entirely from people who were never randomly assigned, a Cochrane review with nothing to say since 2005, and a disease that still kills roughly one patient in seven. None of that shows hyperbaric oxygen fails here. It shows that nobody has established whether it works — and in an infection with mortality in the low double digits, that is a reason to enrol the next patient in a trial, not a reason to answer the question with confidence in either direction.

Sources

← Back to blog