Hyperbaric oxygen therapy may be beneficial for autists

40 hours of treatment with oxygen-enriched air under elevated pressure showed promising improvement in some children diagnosed with autism, according to a new study.

Hyperbaric treatment with oxygen-rich air is a controversial treatment for children with autism, but this new study does have some appealing results. The BBC reports that 62 children diagnosed with autism and aged between two and seven were submitted at random to a treatment with air containing 24% of oxygen and an air pressure of 1.3 times the normal atmospheric pressure or to a dummy (placebo) treatment with normal air under a lower pressure of 1.03 times the normal atmospheric pressure. 30% of the children in the treatment group were very much or much improved, according to doctors whereas 8% of the children in the dummy group were very much or much improved. Overall, says the study, 80% of the children receiving the treatment improved, while 38% of the children receiving the dummy treatment improved. According to Richard Mills, research director at Research Autism, this is the first well-designed study of this therapy.

"We know this kind of therapy is useful in a number of neurological conditions and that's been well established."

"What we don't know is how useful it is in autism, what we could be seeing is an improvement in other neurological conditions that go alongside autism."

"We also don't know about long-term effects - it could be a transitory effect."

Professor Philip James, an expert in hyperbaric medicine at the University of Dundee, says that the pressure used was not higher than the pressure used in aircraft cabins on the ground. He also says that oxygen controls inflammation, and may have had other effects, such as gene expression or tissue regeneration. Even if the treatment is proven, it may not be for everybody.

"When you have any condition, there are people who have too much damage to get better."

"All the oxygen is doing is bringing things towards normal."

People who are not familiar with scientific testing must be advised caution. This study proves nothing. This is at best a pilot study that tries to see if there "is something to it," and if further research is warranted. Nothing more. 62 children were included. That sounds quite a few, but it is not. They are to be divided in two equal groups: 31 receiving the treatment, 31 who receive the dummy treatment. 80% of the children receiving the treatment improved. That is 24.8 children. that is not possible. Either 25 improved, or 24. Nitpicking? Subjective, negative, biased? No. This difference is important:

25 = 80.6%

24 = 77.4%

1 child makes a difference of 3.2%

38% of the dummy group improved. That's 11.78 children. That's not possible. It's either 12 or 11. The reason for these discrepancies may or may not be cause for alarm. It is quite possible, even likely, that the group was not split in groups of 31. They probably started out with more children, and ended up with 62 after everything was finished. Test subjects often quit, for numerous reasons. Nevertheless, this discrepancy must be explained, for it can invalidate the claims. More important is that the age differences were huge. Children change more rapidly when they are younger. A child that is two years old will change more rapidly and more significantly than one that is seven years old. That alone could invalidate the test. Depending on how the children were distributed in the two groups, the age discrepancies may already be enough to invalidate the test. There is also no information available on how "improvement" was evaluated in the study. People looking into this treatment must also be aware of the fact that the ICDRC in Florida where study leader Dan Rossignol works, does not have a spotless reputation. They have been associated with Andrew Wakefield, infamous for causing an unfounded worldwide vaccine-related autism scare. They are also known for peddling expensive, unproven treatments to desperate parents. As such, Dan Rossignol's objectivity is not guaranteed.