"Publication bias" is not a phrase widely familiar to people outside the world of academic research. Yet it can explain how a drug launched as a safe and effective treatment can later turn out to be useless, or even deadly.
Pharmaceutical companies invest millions of pounds in drug research and have a powerful commercial interest in publishing positive findings for the medicines they have spent years developing. But they are equally keen to keep quiet about those trials which show no effect.
Medical journals comply in this process of self-censorship because, like the lay media, they are competing for readers and positive results – the more dramatic the better – attract more attention. The result is that over months and years, an impression is created that a drug is more effective, and has fewer side-effects, than is really the case.
This is known as "publication bias", the selection of only positive studies for publication. If all the studies conducted, positive and negative, were reviewed the overall impression might be very different. Publication bias has been blamed for the debacle over the powerful painkiller Vioxx, dramatically withdrawn from the market in 2004, after it was suspected of causing heart attacks. The fatal side-effect had not been picked up despite years of research in thousands of patients. Now it is being blamed for the revelation that two decades after their launch, the new-generation anti-depressants, including Prozac and Seroxat, may be no better than placebos.
Data obtained from the Food and Drug Administration in the United States under freedom of information legislation showed that when all the trials, published and unpublished, submitted at the time the drugs were licensed were analysed, it showed no clinically significant effect. The finding makes the review of the present Nice guidelines on the treatment of depression "all the more urgent", according to Tim Kendall, the head of group responsible for drawing them up.
The present guidelines, issued in 2004, recommend psychological treatments be offered as an alternative to drugs, especially in mild depression, a change from the original guidelines which recommended drugs as the first line of treatment. Revised guidelines are due at the end of the year.
Dr Kendall, a consultant psychiatrist in Sheffield, said: "The doubt the study raises is how much confidence we can have in our current data set, which is much bigger [than in the study] but may not be complete. The drug industry says they are being much more open but I am not convinced we are seeing the data we should see, and we are certainly not seeing what the licensing authorities are seeing."