HBOT Radar: 847 Alzheimer's Patients, and Cognition Improved More Than With Drugs Alone (March 2024)

This article is part of the HBOT Radar series, where we summarize the latest published hyperbaric oxygen therapy research.

Fresh Science, Clearly Explained — Brought to you by Brain Spa Hyperbaric

Disclaimer: This article is intended for educational and informational purposes only. It summarizes published medical research conducted in clinical settings and does not evaluate Brain Spa Hyperbaric products. The hyperbaric chambers offered on this website are non-medical wellness devices and are not intended to diagnose, treat, cure, or prevent any disease. Do not make medical decisions based on this article — consult a qualified healthcare professional.

📌 847 Alzheimer's Patients. Cognition Improved More Than With Drugs Alone.

🔍 Why this topic matters

Alzheimer's disease is the cruellest arithmetic in medicine. Roughly 6.2 million Americans over 65 are living with it today, and that number is projected to reach 13.8 million by 2060. China alone spent $248 billion on Alzheimer's care in 2020, with projections reaching $1.89 trillion by 2050. And the drugs we have — donepezil, memantine, rivastigmine — slow the decline modestly at best.

That gap between what patients need and what medicine offers is why researchers keep exploring unconventional approaches. One that keeps producing interesting results is time spent in a hyperbaric chamber, breathing oxygen under pressure.

The rationale isn't fringe. Alzheimer's brains show reduced cerebral blood flow, chronic neuroinflammation, oxidative damage, and impaired glucose metabolism — often years before symptoms appear. Hyperbaric oxygen affects all four. Animal work has shown it promotes clearance of amyloid-beta, reduces hippocampal inflammation, and blocks the mitochondrial apoptosis pathways that kill neurons. A 2020 PET imaging study found HBOT improved glucose metabolism in specific brain regions of AD patients — the left medial frontal gyrus, right visual cortex, right inferior parietal lobule.

But mechanisms aren't outcomes. So a team at Longhua Hospital, Shanghai University of Traditional Chinese Medicine, pooled every randomized controlled trial they could find on HBOT in people with a confirmed Alzheimer's diagnosis.

🔬 What the researchers reviewed

A PRISMA-compliant systematic review and meta-analysis, registered with PROSPERO (CRD42023483726), searching nine databases — six English-language, three Chinese — up to November 2023.

From 373 initial records, strict filtering left 11 randomized controlled trials with 847 participants (433 receiving HBOT, 414 controls). Studies mixing Alzheimer's with vascular dementia were excluded, as were studies that didn't state diagnostic criteria. Twelve otherwise-relevant studies were thrown out for exactly these reasons — a rigorous filter that explains why the final pool is small.

The trials were published between 2007 and 2021, sample sizes from 43 to 98, treatment durations from two weeks to six months. Ten compared HBOT plus a standard Alzheimer's drug against the same drug alone. One — Zhu 2017 — tested HBOT entirely on its own against a wait-list control, which turns out to matter.

Worth noting: one of the peer reviewers was Lindell K. Weaver of Intermountain Healthcare, who has run some of the most rigorous sham-controlled HBOT trials in existence. Not a reviewer who waves things through.

These studies were conducted in clinical hospital settings under medical supervision. Results from clinical research cannot be directly applied to other settings or devices — always consult a healthcare professional.

📊 What the evidence shows

Cognition: consistent improvement across nearly every trial

The pooled MMSE improvement was 3.08 points (95% CI: 2.56–3.61, p < 0.00001), with low heterogeneity (I² = 20%) — the trials largely agreed with each other.

The MMSE is the standard 30-point test doctors use to track dementia — questions about the date, remembering three words, simple arithmetic, following instructions. Patients in these trials scored around 17–20 at the start, placing them in the moderate range.

For context: donepezil, one of the most widely prescribed Alzheimer's drugs, typically produces around 1.4 points of MMSE improvement over placebo. And untreated Alzheimer's usually costs a patient 2–4 MMSE points per year.

So a 3.08-point gain on top of standard medication is roughly equivalent to giving someone back a year of decline. That's not a rounding error.

What's more convincing than the pooled number is the pattern underneath it. Looking at the individual trials in the forest plot, seven of eight showed a statistically significant improvement, with effects ranging from 2.11 to 4.90 points. Only the smallest and oldest (Huang 2007) failed to reach significance. That's not one outlier dragging an average — that's a consistent direction across independent samples.

HBOT on its own also worked

This is the finding that's easy to miss. Because ten of the eleven trials tested HBOT as an add-on to medication, the obvious objection is: maybe the drug is doing the work.

But Zhu 2017 compared HBOT alone against a wait-list control — no medication in either arm — and found a 2.50-point MMSE improvement (95% CI: 0.52–4.48) after just two weeks of daily sessions at 2.3 ATA. Smaller than the pooled effect, but independent of any drug.

One small trial isn't proof. It does mean the "it's just the medication" explanation doesn't fully hold.

The second memory test agreed — but almost all of that came from one trial

The researchers also pooled results from ADAS-Cog, a longer and more detailed cognitive assessment than the MMSE. It tests word recall, naming objects, following commands, and language comprehension, and it's the standard measure used in Alzheimer's drug trials. Lower scores are better.

The pooled result was an improvement of 4.53 points — larger than the 2–3 points typically reported in drug trials.

But here's the catch: only three of the eleven trials measured ADAS-Cog at all. And of those three, one single study accounts for 96% of the result. The other two barely register — one of them didn't even reach statistical significance.

So this isn't really a finding from three studies. It's one study's result, with two small ones attached to it. It's still a real result — but it carries the weight of one trial, not eleven.

Daily function improved — and the studies disagreed sharply

Scores for activities of daily living — dressing, eating, washing, managing personal care — improved by 10.12 points. For families, this is often the number that matters most, because it determines whether someone can stay in their own home.

But only three trials measured it, and they disagreed almost completely. One found an improvement of 4.45 points (too small to be statistically meaningful). Another found 8.95. The third found 15.12 — more than three times the first.

The tempting explanation is dose: longer treatment, bigger effect. The data says otherwise. The trial with the biggest improvement used the shortest course — just two weeks.

That's not a fluke of these three studies, and it points toward something interesting about how oxygen doses work in an ageing brain. We'll come back to it once we've looked at the protocols.

For now: with only three trials producing three very different answers, treat this number as a rough hint rather than a reliable figure.

Blood markers moved in the expected direction

Oxidative stress markers improved: MDA (oxidative damage) dropped, SOD (the body's antioxidant enzyme) rose. Inflammatory signalling shifted: IL-1β fell, TGF-β1 rose.

These are consistent with the proposed mechanism — hyperbaric oxygen easing the oxidative and inflammatory load on ageing neurons. It's useful confirmation that something biological is happening, not merely something behavioural.

The caveat is scale: each marker was measured in only two or three trials, with 79 to 122 patients. TGF-β1 — the one with perfect agreement between studies (I² = 0%) — rests on just two trials.

What pressure, how many sessions?

This is the question most readers want answered, and the supplementary data gives us a clear picture.

Chinese hyperbaric research reports pressure in megapascals rather than atmospheres. Converting: 0.1 MPa is roughly 1 ATA (normal sea-level pressure). Here's what the eleven trials actually did:

Most air breaks lasted 10 minutes; three trials didn't use them.

A few things stand out. Every trial used clinical-range pressure — 2.0 ATA or above in almost all cases. None tested lower pressures. Sessions were long, 80 to 120 minutes. Frequency was high — daily in seven trials, five days a week in three.

But the total number of sessions varied enormously: from roughly 14 (Zhu 2017 and Wang 2021, two weeks daily) to well over 100 (Wang 2020, six months at five days a week). That's a nearly tenfold difference in total exposure.

Does more treatment mean better results? The data says no — and there may be a reason

Here's where the protocol details upend an assumption.

The standard story in hyperbaric research is that effects accumulate: more sessions, better outcome. These trials point the other way.

The three trials measuring daily-living scores, ranked by how much total treatment patients received:

• Wang 2021: 14 sessions at 2.3 ATA, 80 minutes → improvement of 15.12 points
• Gao 2017: ~40 sessions at 2.0 ATA, 120 minutes → improvement of 4.45 points (not statistically significant)
• Zhao 2021: ~90 sessions at 2.0 ATA, 90 minutes → improvement of 8.95 points

The shortest course produced by far the biggest effect. The longest produced about half of it. The middle one performed worst of all.

There's a framework that makes sense of this, and it comes from Dr. Paul Harch — a hyperbaric physician who has spent three decades documenting both the benefits and the harms of the therapy.

Harch measures total hyperbaric exposure in "atmosphere-hours" — pressure multiplied by hours multiplied by number of sessions. In his own clinical records spanning 1989 to 2023, he documented patients developing signs of oxygen toxicity at cumulative doses averaging around 103 to 116 atmosphere-hours, at pressures well below the 2.0 ATA that textbooks list as the toxicity threshold.

Run these three trials through that calculation:

• Wang 2021: 43 atmosphere-hours — well below Harch's toxicity range
• Gao 2017: 160 atmosphere-hours — above it
• Zhao 2021: 270 atmosphere-hours — far above it

The only trial that stayed below the threshold produced the largest improvement. Both trials that exceeded it produced substantially less.

Why an ageing brain might tolerate less, not more

The mechanism behind this is well established, even if its application here is speculative.

Hyperbaric oxygen works partly by generating reactive oxygen species — the same molecules that cause oxidative damage. At the right dose, this creates a controlled stress that triggers the body's repair and adaptation machinery. Beyond that dose, ROS production outpaces the antioxidant systems meant to contain it, and the same process becomes destructive. Researchers call this a hormetic dose-response: an inverted-U curve, where the peak is not the highest dose.

Alzheimer's brains are, by definition, oxidatively compromised. Mitochondrial function is impaired, antioxidant enzyme activity is reduced, and the tissue is already under chronic inflammatory load — which is precisely why this meta-analysis found HBOT improving those markers. But a system with depleted reserves may also reach its ceiling faster. A short, well-placed course could provide a therapeutic push; a long, aggressive one could exhaust the very capacity it's meant to restore.

If that's what's happening, the pattern in these three trials makes sense: not "more didn't help," but "more began to work against itself."

Three caveats before anyone takes this too far. First, three trials cannot establish a dose-response curve — they differ in medication, patient severity, assessment methods, and research team, any of which could produce the same ordering by chance. Second, Harch's toxicity threshold was derived from patients with traumatic brain injury, stroke, and carbon monoxide poisoning, not dementia; whether it transfers is untested. Third, this pattern shows up in the daily-living scores but not in the main memory test — there, trials with very different protocols produced broadly similar results. If dose were the whole story, you'd expect to see it in both.

What is clearly true: none of these eleven trials tested the low-pressure, shorter-session protocol that Harch recommends for chronic neurological conditions — 1.5 ATA, 60 minutes, around 40 sessions. Every trial here used 2.0 ATA or above, with sessions of 80 to 120 minutes. They found benefits anyway. Whether a gentler protocol would find more, less, or the same is an open question that nobody has run the trial to answer.

That question — what is the right dose for a brain that's already struggling to handle oxidative stress? — may matter more than another study confirming that something happens.

No increase in side effects

Adverse events showed no significant difference between groups (OR = 1.17, 95% CI: 0.68–2.03, p = 0.58, I² = 0% — perfect agreement across six trials).

No seizures. No middle ear barotrauma. No pulmonary complications. Reported side effects were minor — headache, earache, insomnia, diarrhoea — at similar rates in both arms.

For a frail, poly-medicated patient population, a clean safety profile matters as much as efficacy.

⚖️ What to keep in mind

Geography. All 11 trials were conducted in China. The authors applied no geographic restriction — no trials from elsewhere met inclusion criteria. Independent replication in other populations would strengthen the case considerably.

Two 2007 trials look unusually alike. Huang 2007 and Tian 2007 report the same sample size (92), the same group split (50/42), the same year, the same intervention (HBOT plus rivastigmine), and identical treatment-group MMSE values (20.6 ± 6.1). Their control groups differ. This may be two reports drawing on overlapping data. Neither carried much weight in the pooled analysis, so it doesn't change the conclusions — but it's the kind of detail worth noticing in any meta-analysis.

Blinding. None of the 11 trials blinded participants or outcome assessors. This matters less for patient expectation — you don't improve delayed word recall by believing you will — and more for assessor scoring. ADAS-Cog in particular involves judgment calls in grading recall and comprehension. Methodological research on unblinded outcome assessment suggests effects are typically overstated by around a third in such designs. Applied here, a 3.08-point MMSE gain might realistically sit closer to 2 points — still above what standard drugs achieve alone, but worth factoring in.

Publication bias could not be assessed. Sample sizes were too small for Begg's or Egger's tests, leaving open the possibility that negative trials went unpublished.

The bottom line: a consistent multi-trial signal, in the right direction, across two independent cognitive measures, with supporting blood-marker data, a clean safety record, and at least one trial showing the effect without any drug attached. It is the strongest clinical evidence base HBOT has in Alzheimer's disease.

But the protocol data raises a sharper question than the pooled numbers do. If two weeks at a modest total dose outperforms three months at a much higher one, the field may be asking the wrong question. The useful trial isn't another one confirming that something happens — it's a blinded, sham-controlled study comparing doses directly, in a patient population whose brains are already carrying more oxidative load than they can comfortably handle.

📌 Takeaway for the community

• A meta-analysis of 11 randomized trials with 847 Alzheimer's patients found HBOT added to standard medication improved MMSE by 3.08 points — roughly double what standard drugs achieve alone, with seven of eight individual trials showing significant improvement
• One trial tested HBOT entirely without medication against a wait-list control and still found a 2.50-point MMSE gain, suggesting the effect isn't purely drug-driven
• Blood markers of oxidative stress and inflammation improved in the expected direction, and no increase in adverse events was found in any trial
• The second cognitive measure showed an even larger improvement, but 96% of that result comes from a single trial — and the daily-living result varied wildly between only three studies
• Every trial used clinical-range pressure (2.0–2.5 ATA), long sessions (80–120 minutes), and daily or near-daily attendance — none tested lower pressures
• More treatment did not mean better results. The trial with the shortest course produced by far the largest improvement in daily functioning, and it was also the only one whose total oxygen exposure stayed below the toxicity threshold Harch documented in his own patients — consistent with an inverted-U dose curve, though three trials cannot prove one
• Nobody has tested the lower-pressure, shorter-session protocol Harch recommends for chronic neurological conditions in Alzheimer's patients — which leaves the most interesting dosing question unanswered
• All trials were conducted in China and none used blinding, which likely inflates the effects somewhat without erasing them
• These studies were conducted in clinical hospital settings under medical supervision — results cannot be directly applied to other settings or devices

Source: https://www.frontiersin.org/journals/aging-neuroscience/articles/10.3389/fnagi.2024.1360148/full

Lin G, Zhao L, Lin J, Li X, Xu L. Clinical evidence of hyperbaric oxygen therapy for Alzheimer's disease: a systematic review and meta-analysis of randomized controlled trials. Front Aging Neurosci. 2024;16:1360148. doi: 10.3389/fnagi.2024.1360148.

Educational disclaimer

This content summarizes findings from published medical research for educational purposes only.

The hyperbaric chambers sold on this website are non-medical wellness devices and are not intended to diagnose, treat, cure, or prevent any disease.

The studies discussed here were conducted in clinical medical settings using medical-grade interventions. The inclusion of research summaries does not imply that similar outcomes can be achieved using non-medical wellness devices.