HBOT for Traumatic Brain Injury: What a New Meta-Analysis Found

A systematic review and meta-analysis published in Complementary Therapies in Medicine in October 2026 pooled eight randomised controlled trials testing hyperbaric oxygen therapy against sham treatment, standard care, or no treatment in adults with traumatic brain injury. Across a combined 570 participants, the researchers, led by Wang and colleagues at the First Affiliated Hospital of Zhengzhou University in China, searched PubMed, Embase, the Cochrane Library, and Web of Science from inception to March 2026, and assessed the quality of included trials using the Cochrane Risk of Bias tool.

The result is not a simple yes or no. HBOT showed a statistically significant benefit for computerised cognitive performance, for executive function and processing speed, for memory function, and for sleep quality. For four other outcomes, overall functional recovery measured on the Glasgow Outcome Scale, PTSD symptoms, broader neurobehavioural symptoms, and emotional distress, the pooled data showed no significant benefit.

This kind of domain-specific result is exactly what section 2b of our editorial approach asks us to report honestly rather than smooth over: HBOT is not a uniform success or failure for TBI in this analysis. It appears to help with certain measurable, test-based outcomes relating to thinking speed, memory, and sleep, while showing no detectable effect on the broader measures of overall recovery, psychological symptoms, and day-to-day function that matter most to patients and families.

The Glasgow Outcome Scale is one of the oldest and most widely used measures of overall recovery after a brain injury, ranking outcomes from death through to good recovery with return to normal life. It is, in a real sense, the outcome that matters most to a patient and their family: not whether a specific cognitive test score improved, but whether the person has substantially recovered their life.

The meta-analysis found no significant benefit on this measure (risk ratio 1.57, with a wide confidence interval from 0.55 to 4.44, and very high heterogeneity between trials at 87%). A risk ratio above 1 nominally favours HBOT, but the confidence interval crossing 1, combined with the small number of pooled trials and the high heterogeneity, means this finding cannot be read as evidence that HBOT improves overall functional recovery. It is, honestly, an inconclusive result on the single outcome most people would care about most.

This is the central tension in the 2026 findings: real, statistically significant improvements on cognitive test batteries, sitting alongside a non-significant, highly uncertain result on the broader measure of whether someone has actually recovered. Both things can be true at once, and neither should be used to argue the other away.

The improvements that did reach statistical significance were measured using standardised test batteries rather than self-report. Computerised cognitive performance improved with a standardised mean difference of 0.23 (a small-to-moderate effect size), with essentially no heterogeneity between trials (I-squared of 0%), meaning the included studies agreed closely on this result. Executive function and processing speed showed a larger effect (standardised mean difference of -0.59, where the negative sign reflects faster processing times on the underlying test), with modest heterogeneity of 30%. Memory function improved with a standardised mean difference of 0.33, again with no heterogeneity.

Sleep quality improved by a mean difference of 1.98 points on the measurement scale used, but with considerably higher heterogeneity of 65%, meaning the individual trials varied more in how much benefit they found. The consistency of the cognitive and memory findings, in particular, is notable precisely because it is unusual: when multiple independent trials agree this closely (0% heterogeneity), it is a meaningfully stronger signal than a single trial on its own, even a well-designed one.

The authors were explicit, however, that these domain-specific benefits were seen predominantly in patients with mild traumatic brain injury, which in practice means concussion and similar injuries rather than the more severe end of the TBI spectrum. They stated directly that the applicability of the evidence to moderate-to-severe TBI populations is restricted, a caveat that matters enormously for anyone trying to apply this finding to a specific real-world situation.

The two outcomes with the weakest, most contradictory pooled data were neurobehavioural symptoms (mean difference of -9.06, with a confidence interval spanning -32.13 to 14.00, and heterogeneity of 97%) and emotional distress (standardised mean difference of 0.25, with a confidence interval from -0.32 to 0.81, and heterogeneity of 85%). A heterogeneity statistic above 75% is generally considered very high in meta-analysis, indicating that the individual trials being combined were measuring something different enough, in different populations or with different protocols, that pooling them into a single number risks obscuring more than it reveals.

PTSD symptoms showed a similar pattern, with a mean difference of -3.05 and a confidence interval from -7.05 to 0.95, again not reaching statistical significance, with heterogeneity of 67%. In each of these three cases, the honest conclusion is not that HBOT has no effect on neurobehavioural symptoms, PTSD, or emotional distress, but that the current trial evidence is too inconsistent to draw a reliable conclusion either way.

This matters because neurobehavioural and emotional symptoms, including irritability, mood changes, and post-traumatic stress, are often among the most disruptive, long-lasting consequences of a brain injury for patients and their families, arguably more disruptive day to day than a few points on a cognitive test. The fact that the strongest, most consistent findings in this meta-analysis are on cognitive test performance, while the outcomes with the most real-world emotional weight remain genuinely uncertain, is an important distinction to hold onto.

The review reported that most adverse events across the eight included trials were mild and transient, which is broadly consistent with the known safety profile of HBOT at the pressures typically used in clinical and rehabilitation settings. The researchers used the Cochrane Risk of Bias tool, a standard, widely respected method for assessing whether a randomised trial's design, blinding, and reporting are likely to have introduced bias, with two independent reviewers screening and extracting data from each trial.

The review does not report in its published abstract how many of the eight trials used a genuine sham or placebo HBOT control, as opposed to standard care or no treatment, which is a meaningful methodological detail for anyone trying to judge how much of the cognitive improvement might reflect a placebo response rather than a true physiological effect of the oxygen itself. Sham-controlled trials, where patients cannot tell whether they are receiving real or placebo pressure, are widely regarded as the most rigorous way to isolate a treatment's real effect from the psychological impact of receiving any kind of intensive, supervised care.

Without that detail being specified in the abstract, it would be inaccurate to claim this meta-analysis definitively rules out a placebo contribution to the cognitive and sleep findings. That is a genuine open question the published summary does not settle, and it is one worth asking about if more detailed results become available.

Standard concussion and brain injury care in Ireland does not currently include hyperbaric oxygen therapy. The Department of Health's 2024 national guidance for recognising and managing concussion in grassroots sport, developed with input from Dublin City University's School of Health and Human Performance, centres on the principle of "if in doubt, sit them out": immediate removal from play for a suspected concussion, structured rest, and a graded return to activity overseen by a GP or sports medicine clinician. The published guidance makes no reference to hyperbaric oxygen as part of recommended care.

This does not mean HBOT is prohibited or discouraged in Ireland; it means it sits outside the officially recommended pathway for concussion and TBI recovery, in the same space as a range of adjunct wellness therapies that some private clinics offer alongside, not instead of, standard medical management. Anyone considering HBOT after a head injury should do so as a supplement to, not a replacement for, the structured rest and graded return-to-activity approach their GP, physiotherapist, or sports medicine doctor recommends.

For people with a more serious, moderate-to-severe brain injury, the 2026 meta-analysis's own caveat is the single most important takeaway: the evidence base for benefit is concentrated in mild TBI, and its relevance to more severe injury is explicitly limited by the authors themselves.

At private Irish wellness clinics, a single HBOT session typically costs between €100 and €150, with €135 a commonly quoted single-session rate, dropping to roughly €100 to €120 per session for a package of 10 to 20 sessions. These wellness-clinic chambers generally run at lower pressures (often described as "mild" HBOT) than the higher-pressure medical chambers used in hospital settings, such as the HSE's hyperbaric unit at University Hospital Galway, which treats recognised medical indications like decompression sickness and non-healing wounds.

Published research protocols for HBOT in brain injury, including many of the trials likely pooled in this kind of meta-analysis, have historically used considerably longer courses than the 10 to 20 session packages marketed for general wellness use, sometimes running to several dozen sessions over multiple weeks. Anyone considering HBOT specifically for cognitive symptoms after a head injury should ask a prospective clinic what protocol, in terms of pressure, session length, and total number of sessions, they are proposing, and how that compares to what was actually studied in the trials showing benefit.

Most wellness uses of HBOT, including for TBI-related symptoms, are not covered by the HSE or by private health insurance in Ireland, so this is an out-of-pocket cost, and a longer, research-aligned protocol could represent a meaningful financial commitment. Prices last checked October 2026.

This is a genuinely mixed result, and it deserves to be reported as one. Hyperbaric oxygen therapy showed consistent, statistically significant benefits for cognitive performance, executive function, memory, and sleep quality in the pooled trial data, concentrated in people with mild traumatic brain injury. It showed no significant benefit for overall functional recovery, PTSD symptoms, neurobehavioural symptoms, or emotional distress, several of which carried very high heterogeneity between trials, meaning the underlying evidence is inconsistent rather than simply negative.

Neither the encouraging cognitive findings nor the inconclusive functional and psychological findings should be flattened into a single headline. For someone with a mild TBI or concussion weighing up whether to try HBOT for lingering cognitive fog or sleep disruption, this meta-analysis offers a reasonable, if not definitive, basis for cautious optimism on those specific symptoms. For someone hoping HBOT will meaningfully change their overall recovery trajectory, their PTSD symptoms, or their emotional wellbeing after a brain injury, the current evidence does not yet support that expectation.

As with the long COVID HBOT trials covered elsewhere on this site, the right response to a mixed, honestly reported result is not to dismiss the treatment outright or to oversell it, but to match expectations to what the specific evidence, for the specific outcome, actually shows.