Not all evidence is equal. A double-blind sham-controlled trial is worth more than a review. Below, the nine papers are grouped by the strength of their design — strongest first — with effect sizes quoted exactly as the authors reported them. Nothing here is stronger than the paper it came from.
| Study | Journal | Design | n | Key finding & effect size |
|---|---|---|---|---|
| Weaver 2025 | Scientific Reports | Double-blind, sham-controlled RCT | — | Real HBO₂ beat sham on brain-injury symptoms: MD 7.0 (95% CI 1.7–12.3, p = 0.01); gains also in olfaction, anxiety, sleep, vestibular function. |
| Harch 2020 | Medical Gas Research | Randomised controlled crossover · FDA IND #113823 | 63 | 40 sessions improved post-concussion symptoms, memory, depression, anxiety, PTSD, sleep, quality of life; gains held at 2-month follow-up. |
| Boussi-Gross 2013 | PLoS ONE | Randomised crossover | 56 | Post-concussion syndrome 1–5 yrs after mild TBI: significant cognition & quality-of-life gains; SPECT confirmed increased brain activity. |
| Shahid 2025 | Systematic review & meta-analysis | Pooled analysis (4 studies) | 250 | Memory MD 10.13 (p < 0.00001); attention MD 7.99 (p < 0.00001); plus executive function, processing speed, motor skill. |
| Hadanny 2018 | BMJ Open | Retrospective cohort | 154 | Largest chronic-TBI cohort of its kind: measurable improvement on objective computerised cognitive tests and metabolic brain imaging. |
| Harch 2017 | Medical Gas Research | Case-control with SPECT imaging | 30 | Veterans with PCS + PTSD: SPECT normalised in 75% of abnormal regions; improved cognition, IQ, memory, QoL; reduced suicidal ideation. |
| Gottfried 2021 | Biomolecules | Mechanism review | — | HBOT raises blood & tissue oxygen and drives cellular repair — better mitochondrial function, less oxidative stress and inflammation — underpinning cognitive gains. |
| Bin-Alamer 2024 | Frontiers in Neurology | Mechanism review | — | Details recovery via mitochondrial biogenesis, neurogenesis, synaptogenesis and angiogenesis — the neuroplasticity account. |
| Ding 2014 | Interventional Neurology | Review (adjacent: ischaemic stroke) | — | Surveys HBOT's neuroprotective mechanisms in stroke; notes the 2013 FDA arterial-occlusion indication; flags unsettled dose & timing. |
Effect sizes reproduced as reported by the authors. MD = mean difference; CI = confidence interval. Full PDFs on the Storehouse.
Randomisation is what separates a treatment effect from wishful thinking. Three trials here randomised patients — and one went further, hiding real oxygen behind a sham chamber so neither patients nor doctors knew who got what.
Scientific Reports
The hardest design in the set — the one built specifically to cancel the placebo effect. Adults with persistent symptoms after non-stroke brain injury received 40 real HBO₂ sessions or 40 sham sessions.
Primary outcome (Neurobehavioral Symptom Inventory): mean difference 7.0, 95% CI 1.7–12.3, p = 0.01 in favour of real oxygen — with further gains in olfaction, anxiety, sleep and vestibular function.
Medical Gas Research
63 civilian and military subjects with persistent post-concussion syndrome; 40 HBOTs at 150 kPa for 60 min, run under FDA IND #113823 with IRB and US Army human-research approval.
Statistically significant improvement across post-concussion symptoms, memory, cognition, depression, anxiety, PTSD, sleep and quality of life. The crossed-over control group reproduced near-identical gains; benefits persisted at 2-month follow-up.
PLoS ONE
56 patients with prolonged post-concussion syndrome 1–5 years after mild TBI. Protocol: 40 sessions, 60 min, 100% O₂ at 1.5 ATA.
Significant improvement in cognition and quality of life following HBOT but not during the control period; SPECT imaging showed elevated brain activity matching the cognitive gains. The authors conclude HBOT "can induce neuroplasticity leading to repair of chronically impaired brain functions."
A meta-analysis stacks the trials together to see whether the signal survives pooling. It does.
Four studies, 250 patients. HBOT produced significant, large-magnitude improvements across cognitive domains. The included RCTs carried low risk of bias. Proposed mechanisms: enhanced oxygenation, reduced inflammation and matrix-metalloproteinase expression, decreased cerebral oedema.
Pooled mean-difference improvement, by domain — larger bars are larger gains:
Memory and attention shown as reported mean differences (MD 10.13 and 7.99; both p < 0.00001). Other domains reported as significant without a single pooled MD; bars are indicative, not to the same scale. Source: Shahid et al., 2025.
Observational studies can't randomise — but they can be large, and they can anchor to objective brain-imaging endpoints rather than questionnaires alone.
BMJ Open
The largest cohort of its kind — 154 chronic-TBI patients of all severities — assessed with objective computerised cognitive testing and metabolic brain imaging before and after treatment. Improvement demonstrated at the late chronic stage. The retrospective design is offset by objective, imaging-anchored endpoints.
Medical Gas Research
30 military veterans with post-concussion syndrome and PTSD. SPECT imaging normalised in 75% of abnormal regions; significant improvement in neurological examination, IQ, memory, attention, quality of life and PTSD — including reduced suicidal ideation and reduced psychoactive-medication use.
Reviews don't test the therapy — they explain the biology, and they place it against the wider injury class. They matter because a treatment with a coherent mechanism is not the same as a folk remedy.
Biomolecules
HBOT raises oxygen in blood and tissue and drives cellular repair — improved mitochondrial function, reduced oxidative stress and inflammation — the account of why cognition improves after brain trauma and stroke.
Frontiers in Neurology
A recent synthesis of how HBOT promotes recovery and neuroplasticity: mitochondrial biogenesis, neurogenesis, synaptogenesis and new blood-vessel growth — the molecular substrate of repair.
Interventional Neurology
Reviews HBOT in acute ischaemic stroke — oxygen deprivation by a cerebrovascular route — summarising neuroprotective mechanisms and the FDA's 2013 recognition of arterial occlusion, while noting unsettled dose and timing.
Sample sizes are modest; several trials are crossover designs; one central cohort is retrospective (Hadanny 2018).
Protocols vary — 1.5 ATA vs 150 kPa vs 2–3 ATA, different session counts — so an optimal dose is not yet standardised (Ding 2014; Shahid 2025).
Blinding HBOT is genuinely hard, which is exactly why the Weaver double-blind sham trial matters — and it too found a significant effect.
The authors themselves call for larger, standardised, multi-centre trials. That is the ordinary language of a maturing field — not evidence of absence.
These are the limitations of an evidence base in transition — from accepted-indication medicine toward a broader neurological application. Six of the nine papers, including two RCTs and the imaging studies, were already on the public record by 2021.