Last week, the American Medical Association urged the US government to establish a national registry of clinical trials. This is to counter the growing problem of drug companies failing to publish the results of trials that don't show their drugs in a good light, either through demonstrating that they don't actually work very well or through uncovering side effects. Journals would refuse to publish the results of trials that were not registered, and it would be very obvious if results were not being reported.
The New Scientist reports this week that although the big companies are making noises of agreement, in practice what they are prepared to do falls far short of this. Merck, for example, supports the proposal in theory, but according to the New Scientist is only prepared to register smaller trials, but only "large, pivotal trials". And while GlaxoSmithKline (currently contesting legal action with regard to its failure to disclose information on the effectiveness of an antidepressant) planning to set up a website showing the results of their clinical trials, this would only apply to drugs that have already been released. Information on drugs which have yet to be approved would still be hidden. Which rather misses the point of it.
In the same issue, and on the same page, the New Scientist also reports that drug companies want to ban people who respond to placebos from clinical trials of antidepressants. This would make their results look better, for sure. But surely the whole point of a trial is to see how well the drug compares to the placebo? If we rule out the placebo-responders from the start, then how can we possibly tell how the drug will perform in a real-life situation? The drug companies have apparently forgotten what the purpose of a medicine is.