03 juin 2006

Publish or be damned

Ben Goldacre
The Guardian

MMR is back. "US scientists back autism link to MMR," squealed the Telegraph. "Scientists fear MMR link to autism," roared the Mail. "US study supports claims of MMR link to autism," croaked the Times, a day later.

Strap me to the rocket and print my home address in the paper, I'm going after them again. So what was this frightening new data? Well it's hard to tell, since it hasn't been properly published anywhere yet. This is now standard operating procedure for all scare stories, because journalists have learnt that informed and informative public debate on unpublished research is basically impossible. So it turns out that these three stories were all about a poster presentation at a conference that had yet to occur on research not yet completed by a man with a track record of announcing research that then does not appear in academic journals.

The story is that Arthur Krigsman may have found genetic material (RNA) from vaccine-strain measles virus in some gut samples from children with autism and bowel problems. Some believe that this could implicate the vaccine in causing health problems.

But let's not forget, the Mail was promoting Dr Krigsman's research back in 2002: at that time, he was putting endoscopes into the bowels of young children with autism, and said he had found evidence of inflammation. Four years later, looking on PubMed, the standard database for all medical papers, it seems this research still has not been published in a peer-reviewed academic journal. Forgive my bluntness, but it seems a shame to go poking around up there if you're not going to write up your findings properly.

Meanwhile the Telegraph says his latest unpublished claim replicates similar work from 1998 by Andrew Wakefield, and from 2002 by Professor John O'Leary. But there is no such work from 1998 by Dr Wakefield, at least not on PubMed. Meanwhile it is well documented that other labs have tried to reproduce the 2002 study and come back with different results, and that the protocol was likely to have problems with false positives because of the tests used: two perfect examples of the importance of research being fully written up and published, so it can be replicated and assessed.

I could go on, but instead, here is the news you didn't read: in the May issue of the Journal of Medical Virology there was a similar study, only this one has actually been published. It looked for measles RNA in children with regressive autism after MMR vaccination but found no evidence of the magic vaccine-strain measles RNA to implicate MMR, and perhaps because of that unfrightening result, the study was loudly ignored by the press. Like all science in the real world it has its flaws, but because it has been published in full, I can read it, and pick holes in it.

In the spirit of science, the least opponents of MMR could do is share their data, and most importantly publish their scientific work, in full, openly, before their peers, rather than the press.

Les adeptes de l'anti-vaccination sont toujours à leur travail néfaste. La bataille scientifique étant maintenant désespérée pour eux, ils continuent avec une bataille médiatique et utilisent leur arme favorite, la rumeur et son corrolaire la théorie du complot.