This is the tech companion to our autism parents guide. It walks through the published research on HBOT for autism in detail. We cover trial designs, results, critiques, and what the data does and does not support.
If you are a parent looking for a decision guide, start with the parents-guide article. If you are a clinician, researcher, or technical reader looking for the methods, this article is for you.
We will cover the five published RCTs, the meta-analyses, the methods critiques, and the subgroup claims for mitochondrial dysfunction, inflammation, and oxidative stress.
The published RCTs
Five randomized controlled trials of HBOT for autism have been published in peer-reviewed journals.
The Rossignol trial (Rossignol et al., BMC Pediatrics 2009) is the most cited and most contested. It randomized 62 children to 40 sessions at 1.3 ATA with oxygen vs 1.03 ATA with room air (sham). The HBOT arm showed gains on three of five outcome measures.
The Granpeesheh trial (Granpeesheh et al., Res Autism Spectr Disord 2010) randomized 46 children with autism to 80 sessions at 1.3 ATA vs sham (1.03 ATA). It found no significant differences on core autism measures.
The Thai 2012 trial (Diving Hyperb Med 2012) tested 1.5 ATA HBOT vs sham in 60 children. No significant advantage of HBOT over sham on the primary outcomes.
The Jepson trial (Jepson et al., J Autism Dev Disord 2011) randomized 11 children to 80 sessions at 1.3 ATA vs no treatment. No significant changes on standardized tests.
The Lerman trial (Lerman et al., J Autism Dev Disord 2008) tested 1.3 ATA HBOT in 18 children. Open-label, no control. Parent-rated improvements but no objective measure changes.
The 2009 trial in detail
A closer look at the most-cited trial. The 2009 trial enrolled 62 children and randomized them to 40 sessions at 1.3 ATA with 24% oxygen vs 1.03 ATA, room air. The trial was double-blinded.
The HBOT arm showed gains on overall function, language, social interaction, and eye contact (parent-rated CGI). The trial reported no significant changes on standardized autism scales (ABC, ATEC).
Several methodology issues complicate interpretation. The sham used 1.03 ATA — essentially room air — but some argue any pressure gap could give subtle cues to participants.
The trial was also funded by an HBOT-promoting nonprofit, raising conflict-of-interest concerns. The 1.3 ATA dose is below medical HBOT (2.0 to 2.5 ATA), which makes the biological rationale unclear.
The trial has not been replicated. Later larger trials used similar 1.3 ATA pressures and did not find the same effect.
The Cochrane and meta-analyses
The 2016 Cochrane review covered the available trials. Its conclusion: there is no evidence that HBOT improves core or related symptoms of autism.
The review is seen as a gold standard for data synthesis. It used standard methods, included the relevant trials, and reached a clear negative conclusion.
A 2022 systematic review (Gattino et al., Front Psychiatry 2022) covered later data and reached the same conclusion. The authors noted that positive findings were limited to parent-rated outcomes, not objective measures.
The pattern across reviews is consistent. Parent-rated outcomes show small positive effects. Objective measures show no effect.
The mitochondrial subgroup claim
Some HBOT proponents argue autism may have a mitochondrial subgroup that responds to oxygen. The hypothesis is interesting but not yet backed by trial data.
Frye and others have published papers proposing mitochondrial dysfunction as a treatable autism subgroup (Frye et al., Mol Psychiatry 2013). The proposed idea is that HBOT may improve cell-level oxygen flow in children with abnormal mitochondria.
No randomized trial has tested HBOT in this subgroup. The published trials enrolled children with autism broadly, not children with documented mitochondrial dysfunction. Subgroup analyses in existing trials have been exploratory, not pre-set.
Until a trial enrolls children with documented mitochondrial dysfunction and tests HBOT vs sham, claims that HBOT works for this subgroup are speculation.
The inflammation and oxidative stress claims
A second proposed subgroup. Children with raised inflammation markers or oxidative stress markers may respond to HBOT. The theory is that hyperoxia lowers inflammation in some tissues.
Cell and animal data link HBOT to inflammation changes in some contexts. The link to autism outcomes is not established.
No RCT has stratified autism patients by inflammation or oxidative stress markers and tested HBOT vs sham. The claim that HBOT works for inflammation-driven autism is a hypothesis, not a finding.
What about brain imaging studies
A small number of studies have used MRI, SPECT, or PET imaging before and after HBOT in autism. Results are mixed.
Some studies report changes in brain blood flow patterns after HBOT. The clinical meaning of these imaging changes is unclear. Imaging changes do not reliably track with behavior or thinking changes.
Imaging studies are not randomized trials. They are useful for mechanism research but not for proving whether HBOT helps autism patients in measurable ways.
How the wellness clinic claims compare
A reasonable comparison. Wellness clinics often claim HBOT helps with autism symptoms — speech, eye contact, and behavior. These claims rest mostly on the 2009 trial and on parent stories.
The trials after that have not replicated those findings. The systematic reviews explicitly say there is no evidence HBOT improves core autism symptoms.
For parent-facing red flags in autism HBOT marketing, see our autism parents guide. For broader warning signs, see our clinic red flags article.
What the trials do and do not show
A clear summary. The trials show: parent-rated outcomes sometimes improve with both HBOT and sham, with no consistent advantage of HBOT over sham. The placebo and expectation effects are large.
The trials do not show: clear, replicated improvement in objective autism measures. No specific subgroup has been identified as a responder. The proposed mitochondrial and inflammation subgroups remain hypotheses.
The trials cannot rule out: small effects in specific subgroups that current trials are underpowered to detect. The literature is thin enough that future research could change the picture.
What rigorous future research would look like
The HBOT-for-autism literature has structural gaps. A trial that would meaningfully advance the field would have several features.
Larger sample size. The existing trials range from 11 to 62 participants. A definitive trial would need 200+ participants per arm to detect modest effects.
Pre-specified subgroup analysis. If mitochondrial dysfunction or inflammation is the right subgroup, the trial should enroll patients with documented biomarkers and test HBOT vs sham within that subgroup.
Independent replication. The Sagol Center research on adult HBOT outcomes has not been replicated outside the same research group. A meaningful autism trial would be conducted by independent investigators.
Objective primary outcomes. The trials should use standardized, observer-rated, validated measures as the primary outcome, not parent-rated CGI.
Bottom line
The published research on HBOT for autism is limited and mixed. The single most-cited positive trial (2009) has not been replicated. Systematic reviews found no evidence of benefit for core symptoms.
Subgroup claims (mitochondrial dysfunction, inflammation) are hypotheses, not findings. Until pre-specified subgroup trials are conducted, these claims should be treated as speculative.
For clinicians, the research does not support recommending HBOT for autism. For parents, the decision framework article covers the practical implications.
Related Reading
- HBOT for autism spectrum disorder: what parents should know
- HBOT for ADHD: what the evidence shows
- What major institutions are not saying about HBOT
- HBOT for long COVID: Shamir RCT analysis
- FDA-cleared HBOT chambers: complete list
Frequently asked questions
How many randomized trials of HBOT for autism have been published?
Five RCTs have been published. The 2009 trial is the most-cited positive trial. The four later trials found no significant advantage of HBOT over sham.
What did the Cochrane review conclude?
The 2016 Cochrane review concluded there is no evidence that HBOT improves core or associated symptoms of autism. A 2022 systematic review reached a similar conclusion.
Is there evidence for HBOT in a mitochondrial autism subgroup?
The mitochondrial subgroup hypothesis is interesting but has not been tested in a pre-specified RCT. No trial has enrolled children with documented mitochondrial dysfunction and tested HBOT vs sham.
Why are parent-rated outcomes positive but objective measures negative?
Parent-rated outcomes are subject to placebo, expectation, and observer effects, which are particularly strong in pediatric trials. Objective measures (standardized tests, observer ratings) reduce these effects.
What would a definitive trial look like?
A definitive trial would have 200+ participants per arm, pre-specified subgroup analysis, independent replication, and validated objective primary outcomes. No such trial has yet been conducted.
Medical disclaimer: This article is informational and does not constitute medical advice. HBOT carries real risks including ear injury, oxygen-related harm, and chamber fire. The FDA has cleared HBOT for 13 specific uses; autism is not on that list. Discuss any HBOT plan with your child's pediatrician or autism specialist before starting.
-- The HBOT Finder Team