Wednesday, August 11, 2010

British scientists develop brain scan to detect autism - but don't get too excited just yet

The recent news story about the autistic spectrum, involving a study done by the Institute of Psychiatry at King's College in London is more news and a less bulls*** than most news stories about the spectrum, but it is still seriously over-hyped. Check the last paragraph of the text and if I am interpreting the story correctly, it means that 10% of autists tested will incorrectly come up with a result that indicates that they are not autistic, and 20% of neurotypical (normal) people undergoing this test will give a result that incorrectly indicates that they are autistic. I don't think there are too many non-autistic people who would be happy to take a test that gives them a 1 in 5 chance of being labelled as autistic. And when you consider the meaning of a positive result from the viewpoint of a subject who's neurological status is genuinely unknown, this test looks even more worthless. One must remember that the autism spectrum is an uncommon condition, it affects something like 6 in 1,000 people at a diagnosable level. So I think this means that a vast majority of people getting a POSITIVE result on this test would in fact NOT BE AUTISTIC by clinical standards, if this test was inflicted on the general population, simply because autistic people are vastly outnumbered by the non-autistic, and thus we must assume that by far most of the people taking the test would be non-autistic. When considering statistical matters one must always keep in mind base rates - what one would expect to find normally. Please fell free to correct me if you know more about statistics than me. Add to this reality-check the fact that this study was done on only 20 autistic study subjects, and all of them were male adults. So, like much of the research on autism, we are far from sure of what relevance it has for autistic females, of which I am convinced there are many more out there than the autism experts realise. Back to the lab, boffins! Please let us know when you have a firm grasp on what it actually means to be autistic.

This news story has appeared as a video news story on ABC's news channel, and will probably pop up on TV news broadcasts tonight.

"British scientists develop brain scan to detect autism"

There has been a lot of discussion of this study at the New Republic blog:


Anonymous said...

Thanks for the perspective! That's what I thought when I read it. 10% of people on the spectrum go undiagnosed. not good.

Rachel Cohen-Rottenberg said...

Let's see...They did brain scans on 20 men, to diagnose a condition first analyzed by looking at the behavior of boys...hmmmm...I wonder whether there's a brain scan to test for a propensity to forget that females exist. That seems like an approach with social consequences far more devastating than not being able to hold eye contact.

In any case, I'd probably decline to have my brain scanned. It just continues to perpetuate the myth that autism is a disorder and that something is wrong with our brains. I have the same problem with scientific research into brain markers for homosexuality. Reducing who people are to visible phenomena seems altogether wrong headed to me.

BTW, I love your blog, especially your tag line "Born to Offend." Ha! Moi aussi.

Lili Marlene said...

Yes, I would think some vulnerable people could fall through the cracks.

Lili Marlene said...

My comments don't come out in order. That's life.

My view on the brain scan research is that it can only be valid for diagnosis in combination with measures or behaviour or life experiences. I do like the idea of a physical marker for autism, but I'm sure this idea cannot be applied in a general way, as there is much heterogeneity among those diagnosed as autistic at present.