Treating children with autism can be remarkably frustrating for everyone involved. Its impact on children and their families is life-altering, its cause is unclear, and its therapies are minimally effective at best and costly, dangerous, or exploitative at worst. It is easy to understand how studies which may show a benefit for afflicted children will rapidly draw attention and proponents, regardless of its plausibility.

One such therapy is hyperbaric oxygen (HBO) treatment. There have been some reports that a minority of children with autism have abnormal blood vessels feeding various areas of their brain, potentially limiting the amount of oxygen available to that area of the brain and possibly causing dysfunction of the affected neurons. Notice the hedge terms I use; the evidence is quite weak at this point. Regardless, if this is true, one might expect better functioning of the oxygen deficient neurons with more oxygen delivered. Anecdotes exist describing such a response, but until recently no well designed studies had evaluated the claim.

Enter a recent study published in BioMed Central Pediatrics by Rossignol and colleagues "Hyperbaric treatment for children with autism: a multicenter, randomized, double-blind, controlled trial." (1) Rossignol randomized 62 children with autism into two groups, one of which would receive two 1-hour treatments of HBO per day, 5 days per week for 4 weeks, and a control group subjected to the same environment for the same spans of time with a very brief and minimal pressure change to blind the participants to their group.

Children's outcomes were followed using three behavior scores generated by questionnaires completed by the parents, one of which was also independently completed by the primary care physicians of the children. The study reports significant improvement in the autistic behaviors of children treated with HBO therapy including better eye contact, improved verbal skills, receptive language, reduced stereotypic behavior, reduced irritability, and globally improved function, among others. The authors reported 9/30 children in the treatment group "much or very much improved," vs 2/26 of the controls, and 24/30 in treatment group had any degree of improvement vs 10/26 controls.

On its face, this is a very interesting and potentially exciting finding. A double blinded, randomized, controlled trial with a small but sufficient number of patients studied to make valid statements about the outcome. This is, in fact, exactly what we proponents of evidence and science based medicine have been asking for, and what a relief it would be to have an effective tool to improve the functioning of autistic children. Why then am I not overly enthusiastic about the results?

The first serious concern is the design and execution of the study, which Dr Novella addressed in his blog (2). The numbers involved were not large, the blinding is questionable, there was no follow up beyond the study period, and there may be some conflict of interest for Dr Rossignol.

My second concern is that of biological plausibility. There are a lot of claims and hand waving in the discussion of the article explaining how HBO therapy may decrease inflammation, improve oxygen delivery, prevent brain ischemia, etc. However, most of the claims bottleneck down to one common pathway: more oxygen in the bloodstream delivered to the brain. Instead of getting mired down in detailed discussions of studies which are tangentially topical at best, let's examine HBO's ability to increase blood oxygen content.

HBO can without a doubt increase the amount of oxygen in your blood, and the amount of the increase is directly related to the concentration of oxygen being breathed (room air is 20.9%), and the amount of pressure to which you are subjected. The pressures and concentrations used in the study, 24% and 1.3 atm, are quite low, and we can easily calculate how much extra oxygen patients received during the trial (3).

If we make the reasonable assumption that the children had normal levels of hemoglobin in their blood and live near sea level, then the controls had ~20.7 ml of oxygen in each 100ml of blood, 98.6% of which is bound to hemoglobin, 1.4% dissolved within the blood.

The children given HBO on the other hand would have the underwhelming increase in total blood oxygen content to 21.4 ml of oxygen per 100 ml of blood (this assumes fully saturated hemoglobin). That's a 3% increase in blood oxygen content over baseline that will be lost within seconds of leaving the HBO chamber. Let me put this minimal increase into perspective.

Given the normal range of hemoglobin for humans (~12-16 gm/dl of blood), healthy people have a blood oxygen content ranging from 17 - 22.5 ml of oxygen per 100 ml of blood. It's therefore entirely normal for some healthy people to have 32% more oxygen in their blood than others, which is an order of magnitude greater variance than the 3% increase achieved by HBO.

You can look at this another way. In order for a person with a resting heart rate of 80 to increase their oxygen delivery by 3% (all other things being equal), they could increase their heart rate to 82.4 beats per minute. Contrast that to the fact most people can double and athletes can more than triple their oxygen delivery by this mechanism. A 3% increase is hardly a taxing demand even in critically ill people.

How about another perspective? Living in Denver, I can increase my blood content of oxygen (or that of an autistic child) by almost exactly 3% by descending from Denver's altitude of 5280 ft down to sea level.

Finally, I can increase an average person's blood oxygen content by 3% by providing them with 30% oxygen, which is easily delivered, portable, and can even be worn 24/7.

Given all of the compensatory mechanisms your body has in place to ensure that an adequate amount of oxygen gets to your various organs and tissues, each having wildly different and variable oxygen needs, a 3% increase in blood oxygen content is physiologically insignificant except in the most extreme of conditions. As the hinge upon which most of Rossignol's mechanisms of action swing, the scant increase in blood oxygen provided by HBO is implausible at best.

My other concern with this study is the author's poor handling of the article's discussion. Read without an eye to their citations the discussion flows well, however, most of the assertions made by the authors are supported with the weakest of evidence.

Rossignol asserts that children with autism have a high level of baseline inflammation and that hyperbaric oxygen therapy "possesses strong anti-inflammatory properties." That autistic children have inflammation as a matter of course is contentious, and I'll not cover it here. However, that hyperbaric oxygen therapy can reduce inflammation is a gross misinterpretation of the literature. Of the three studies cited to support their assertion, one concerned traumatic wounds and two evaluated it as a treatment for forms of colitis. All three of these scenarios have significant infectious and ischemic components in superficial areas which would indeed be treated by HBO therapy, and could plausibly reduce inflammation. In the absence of infection or ischemia accessible to HBO treatment, there is no reason to suspect HBO therapy would reduce inflammation, much less that it inherently has "strong anti-inflammatory properties."

Rossignol also ignores the copious body of literature in cellular, animal model, and human studies which establishes that hyperoxia (high levels of oxygen) is one of the more efficient methods of inducing inflammation and oxidative injury, not abating it. There is therefore scant evidence to support HBO therapy's ability to reduce inflammation, and ample evidence to be concerned about the induction of inflammation.

To support his hypothesis that HBO therapy improves cerebral blood flow, Rossignol cites just two articles. The first article is written by Rossignol himself, and is not in fact a research article at all but in fact an opinion piece wherein Rossignol describes the use of HBO therapy in autism (based upon what evidence? This was the first trial of any potential value!). The second citation was also not a peer-reviewed study but instead an author's reply to an actual study published in the Lancet that examined the effect of HBO on children with cerebral palsy and found no effect. Rossignol's best supportive evidence for this hypothesis appears to be nothing more than speculation.

These criticisms (and many others I have not expounded upon) and the lack of physiologic plausibility aside, what if Rossignol is right? Well, as the saying goes, "truth will out." Rossignol is free to build the case for his hypothesis, others will attempt to replicate his results. If HBO therapy actually helps autistic children, then the data from well conducted studies will validate it, and we will have gained a tool to help afflicted children and learned something new in the process. If it doesn't, then we should direct the limited resources at our disposal to more promising avenues of autism research. At this time, HBO therapy remains an expensive, difficult to administer, implausible, and unproven treatment for children with autism.


1) Rossignol DA et al. Hyperbaric treatment for children with autism: a multicenter, randomized, double-blind, controlled trial. BMC Pediatrics 2009, 9:21


2) Neurologica Blog: Hyperbaric Oxygen for Autism:


3) Calculating blood oxygen content:

PaO2 (Pressure of oxygen within arterial blood)) =

[(Atmospheric Pressure - H2O Vapor Pressure) x % Oxygen in atmosphere] - (Arterial CO2 / Respiratory Quotient)

Sea Level: [(760mmHg - 47mmHg) x 0.2093] - (40 / 0.8) = 100mmHg PaO2

We can then translate this pressure of dissolved oxygen to the volume of dissolved oxygen by knowing that there are 0.003 mls of O2 per mmHg of O2 dissolved in 100 ml of arterial blood, so:

0.003 x 100 mmHg = 0.3 ml O2 dissolved in every 100 ml of arterial blood at sea level

Hemoglobin binds 1.39 ml of oxygen per gram fully saturated, and has a non-linear saturation curve. For my calculations I used 95% saturation at 5280 ft, 98% saturation at sea level, and 15 gm hemoglobin per 100 ml of blood. Add the ml of oxygen bound to hemoglobin to the ml dissolved in blood for the total oxygen content.


Disclaimer: Dr Albietz has no ties to industry and no conflicts of interest to disclose.   The views expressed by Dr Albietz are his alone, and do not necessarily represent the views of his department or institution.  The information provided is for educational purposes only and should not replace a therapeutic relationship with a licensed and accredited medical professional.