Reading time: approximately 10 minutes. Last updated: 30 May 2026.
TL;DR. A new 2026 systematic review and meta-analysis pooled 17 studies (n=920 adults) on hyperbaric oxygen therapy in psychiatric disorders. HBOT produced a large pooled effect on depressive symptoms (Cohen’s d = -0.82) and on PTSD symptoms in sham-controlled trials (d = -1.96), and a moderate effect on anxiety (d = -0.59). The depression signal was dose-gated: large benefit at 2.0 ATA, no benefit at 1.2 ATA. Adverse events were mild and transient. The work was authored partly out of Canada and is the most rigorous evidence synthesis to date for HBOT in mental-health indications.
Hyperbaric oxygen therapy (HBOT) is a medical treatment that delivers 100% oxygen at supra-atmospheric pressure (typically 2.0 to 2.5 ATA) inside a sealed chamber. Its use for major depressive disorder, posttraumatic stress disorder (PTSD), and anxiety is currently investigational in Canada; none of these psychiatric indications are recognised by Health Canada in 2026. A new meta-analysis published in Psychiatry and Clinical Neurosciences in May 2026 is the largest and most methodologically rigorous synthesis of the evidence to date, and it changes how this literature should be framed for Canadian researchers and referring clinicians.
What did the 2026 Al-Shamali meta-analysis actually test?
The Al-Shamali et al. 2026 systematic review and meta-analysis searched three databases for studies evaluating HBOT in adults with major depressive disorder, PTSD, or anxiety. Seventeen studies (n = 920 patients) met inclusion criteria, of which nine were randomised controlled trials that contributed to the quantitative meta-analysis. The prespecified primary endpoint was psychiatric symptom severity at the end of the HBOT course, with random-effects models implemented in R (v4.4.0). PubMed record: PMID 42169234.
What were the pooled effect sizes by disorder?
| Disorder | Studies pooled | Pooled Cohen’s d | Heterogeneity (I-squared) | Effect magnitude |
|---|---|---|---|---|
| Depressive symptoms | 9 RCTs | -0.82 | 82% | Large |
| PTSD (sham-controlled) | 2 RCTs | -1.96 | 0% | Large |
| Anxiety | 3 studies | -0.59 | 42.8% | Moderate |
For depressive symptoms, the pooled effect was large (Cohen’s d = -0.82) but accompanied by high heterogeneity (I-squared = 82%), which the authors investigated through subgroup analyses (see dose-response and comparator sections below). For PTSD, only two sham-controlled RCTs were eligible for pooling, but they showed a very large effect (d = -1.96) with zero heterogeneity (I-squared = 0%). For anxiety, three studies produced a moderate effect (d = -0.59) with moderate heterogeneity.
Why is the dose-response signal at 2.0 ATA important?
One of the most consequential findings was the dose-response analysis for depression. Robust benefit was observed at 2.0 ATA (Cohen’s d = -0.93), but no benefit was observed at 1.2 ATA. This is directly relevant to the long-running sham-debate in HBOT research, where 1.2 ATA has often been used as a “sham” arm on the assumption that it is therapeutically inert. Al-Shamali’s pooled data are consistent with that assumption for depression: at sub-therapeutic pressure, the intervention behaves like a placebo. The corollary is that mild-HBOT protocols (typically 1.3 ATA and below) cannot be assumed to deliver the depression effect documented at 2.0 ATA.
How does HBOT compare against sham, usual care, and antidepressants?
The meta-analysis also stratified the depression effect by comparator arm, with significant heterogeneity between groups (P = 0.038):
- Versus sham: moderate effect (d = -0.68)
- Versus usual care or psychotherapy: large effect (d = -1.33)
- Versus antidepressants: no significant difference (d = -0.60)
The comparator analysis matters for clinical framing. The largest apparent effect (versus usual care or psychotherapy) likely incorporates expectation and attention effects that the sham-controlled estimate (d = -0.68) partially controls for. The antidepressant comparison did not show a significant difference, but the point estimate (d = -0.60) was moderate in size; this is best read as HBOT showing comparable, not superior, antidepressant performance in the available trials, and it warrants direct head-to-head trials at adequately powered sample sizes before any pharmacotherapy substitution is contemplated.
What does the PTSD signal look like in practice?
The d = -1.96 pooled effect for PTSD in sham-controlled trials is unusually large for a psychiatric meta-analysis, but it rests on only two RCTs and the authors flag this in their conclusions. A useful real-world reference point comes from Levitt et al. 2026 in Psychological Trauma, a prospective single-arm observational program evaluation of 87 U.S. veterans receiving HBOT at three Florida treatment centres. Fifty veterans (57%) completed all assessment points. Mean PTSD Checklist (PCL-5) scores fell from 51.00 to 20.62 (Cohen’s d = -1.91), which the authors describe as “a clinically significant reduction from diagnostic to subthreshold levels.” Improvements were sustained at 1-, 3-, and 6-month follow-up, with concurrent reductions in depression, anxiety, somatization, sleep disturbance, impulsivity, and stress. The Levitt design is open-label and uncontrolled, so the effect estimate is inflated by expectation and natural history; its value is as a signal of durability and breadth, not as confirmatory evidence.
Is there mechanistic evidence supporting a depression effect?
A 2026 mechanistic RCT adds biological plausibility for at least one mood-disorder phenotype. Tang et al. 2026 conducted a randomised double-blind sham-controlled trial in 61 post-stroke depression (PSD) patients (HBOT n=29, sham n=32). The primary outcome was change in the 17-item Hamilton Depression Rating Scale (HAMD-17) from baseline to week 4. HAMD scores decreased significantly in the HBOT group versus sham at weeks 2 (p=0.017) and 4 (p<0.01). Concurrently, serum Brain-Derived Neurotrophic Factor (BDNF) and beta-Nerve Growth Factor (beta-NGF) levels rose significantly in the HBOT group (both p<0.01), and the reductions in HAMD scores correlated with the BDNF rise (r = 0.66) and the beta-NGF rise (r = 0.47). PSD itself affects 29 to 52% of stroke survivors, which makes this a high-volume indication; the neurotrophic correlation is consistent with the inflammation-modulation and neuroplasticity mechanisms proposed in the broader HBOT literature, summarized for example in Bhargava et al. 2026.
Does HBOT improve sleep in PTSD populations?
Doenyas-Barak et al. 2026 is a retrospective longitudinal analysis of 395 patients (180 healthy aging, 92 long COVID, 123 PTSD) who received 60 HBOT sessions at 2.0 ATA, 100% oxygen, 90 minutes, 5 days per week at the Sagol Center. Baseline Pittsburgh Sleep Quality Index (PSQI) scores were highest in the PTSD group. Post-HBOT, total PSQI scores improved significantly in all three groups (p < 0.001), with Cohen’s d ranging from 0.37 to 0.91. Subjective sleep quality, sleep latency, and sleep disturbances improved across all groups. Notably, daytime dysfunction improved in the aging and long COVID groups but not in the PTSD group. Medication use was unchanged. The protocol matches the 2.0 ATA dose at which Al-Shamali found the largest depression effect, and the persistence of sleep architecture benefit alongside symptom reduction is consistent with the durability signal seen in Levitt.
What is the safety profile in the pooled data?
Al-Shamali report that adverse events across the included trials were mild and transient. This is consistent with the broader HBOT safety record at 2.0 to 2.5 ATA in monitored clinical settings: the principal risks (middle-ear barotrauma, transient myopia, rarely seizure from oxygen toxicity) are well-characterized, dose-related, and managed by trained staff at accredited centres. None of the pooled psychiatric trials reported a safety signal warranting halt criteria.
What are the methodological limitations Canadian researchers should weigh?
- Small RCT base. Nine RCTs across three psychiatric indications is a thin evidence base. The PTSD pool is two trials; a single new trial could shift the pooled estimate materially.
- High heterogeneity for depression. I-squared of 82% means most of the between-study variance is true difference, not sampling noise. Subgroup analysis attributed most of it to pressure (2.0 vs 1.2 ATA) and comparator type, which is the right direction but not a final answer.
- Sham design is unresolved. The dose-response data effectively retire 1.2 ATA as a credible sham for depression, because 1.2 ATA is inert. The field needs better blinded sham conditions that match the sensory experience of 2.0 ATA without delivering the active intervention.
- Antidepressant comparison is underpowered. The d = -0.60 versus antidepressants is non-significant but moderate in size; it does not support an inferiority conclusion and does not support a superiority claim.
- Demographic generalizability. The Levitt cohort is 90% White and 90% male U.S. veterans, which limits external validity to Canadian civilian psychiatric populations.
What is the Canadian research context?
The Al-Shamali meta-analysis was authored in part out of Canadian institutions including the University of Toronto and Defence Research and Development Canada, which puts Canadian methodology directly into the centre of this literature. The clinical implication for Canada is that mental-health indications for HBOT remain investigational under Health Canada’s regulatory framework, but the evidence base now warrants prospective Canadian-led RCTs at 2.0 ATA in MDD, PTSD, and treatment-resistant anxiety, with appropriately matched sham conditions and head-to-head antidepressant comparators. Researchers planning protocols should consult the research database for recent comparable trial designs and the directory of Canadian hospitals and regulated facilities for HBOT-capable trial sites.
Frequently asked questions
Is HBOT a Health Canada-recognised treatment for depression, PTSD, or anxiety?
No. As of May 2026, none of these psychiatric indications are on Health Canada’s list of approved HBOT indications. All use for mental-health conditions in Canada is investigational and should occur in the context of a clinical trial or with explicit informed consent acknowledging off-label use.
Why does the 2.0 ATA versus 1.2 ATA distinction matter?
Because 1.2 ATA showed no depression benefit in Al-Shamali’s pooled data, it can no longer be treated as therapeutically equivalent to 2.0 ATA. This affects two communities: trialists who use 1.2 ATA as a sham (now justified for blinding logic, but should not be characterized as “mild HBOT”) and clinics that market sub-therapeutic-pressure devices for psychiatric indications without disclosing the dose-response evidence.
How large is the PTSD effect, and how confident can we be in it?
The pooled sham-controlled PTSD effect was Cohen’s d = -1.96 with zero heterogeneity, which is statistically clean but rests on only two RCTs. The Levitt 2026 observational cohort returned a similar within-subject effect (d = -1.91), which adds convergent signal but not confirmatory evidence. A single new sham-controlled trial could move the pooled estimate substantially in either direction.
Did the meta-analysis find HBOT superior to antidepressants?
No. The pooled HBOT-versus-antidepressant comparison returned d = -0.60, which was not statistically significant. The correct reading is that HBOT performed comparably to antidepressants in the included trials, not that it outperformed them. Direct head-to-head trials at adequate sample size are still needed before any pharmacotherapy-substitution implication is warranted.
Were there safety concerns in the pooled trials?
The authors report that adverse events across the 17 included studies were mild and transient. No safety signal warranted halt criteria. This is consistent with the broader 2.0 to 2.5 ATA HBOT safety record at accredited Canadian Hyperbarics directory facilities.
Where should Canadian researchers go for the underlying study set?
Start with the Al-Shamali 2026 record, which synthesizes 17 studies and n=920 adult patients across major depressive disorder, PTSD, and anxiety, with nine of those studies (the RCTs) contributing to the pooled meta-analysis. Follow through to PubMed for protocol-level detail, and cross-reference the most rigorous primary trials from the included set against your local IRB requirements before designing a Canadian replication. Canada Hyperbarics indexes more than 14,000 HBOT studies in its research database and updates the index weekly.
Where to find HBOT-capable Canadian trial sites
Canada Hyperbarics maintains an independent directory of Canadian hospitals and regulated facilities offering hyperbaric oxygen therapy. For researchers planning prospective psychiatric trials, the directory is the starting point for identifying potential 2.0 ATA capable sites. Visit the facilities directory for the current list.
This content is for informational purposes only and does not constitute medical advice. HBOT for major depressive disorder, PTSD, and anxiety is investigational and is not currently a Health Canada-recognised indication. Patients should discuss any HBOT decision with their treating physician.