Thursday, January 01, 2009

Can we trust pharmaceutical trials?

Ora sends me this interesting email:

You might be interested in this article from the New York Review of Books, titled “Drug Companies & Doctors: A Story of Corruption” (

A main point of this article is that positive results are far more likely to be published than negative results. Although this article concerns the distorting effect of drug companies’ sponsorship of drug trials on the published results of those trials, the same effect is likely in other fields (such as stuttering research) because of the tendency to publish positive results and ignore negative results. And of course, even without the corrupting influence of the Big Pharma money, stuttering research is pervaded by conflicts of interest simply because people have a personal interest in positive results.

Here are a couple of highlights (italics added):
  • In view of this control [the control that drug companies typically have over drug trials] and the conflicts of interest that permeate the enterprise, it is not surprising that industry-sponsored trials published in medical journals consistently favor sponsors' drugs—largely because negative results are not published, positive results are repeatedly published in slightly different forms, and a positive spin is put on even negative results. A review of seventy-four clinical trials of antidepressants, for example, found that thirty-seven of thirty-eight positive studies were published.[8] But of the thirty-six negative studies, thirty-three were either not published or published in a form that conveyed a positive outcome. It is not unusual for a published paper to shift the focus from the drug's intended effect to a secondary effect that seems more favorable.
  • Many drugs that are assumed to be effective are probably little better than placebos, but there is no way to know because negative results are hidden. One clue was provided six years ago by four researchers who, using the Freedom of Information Act, obtained FDA reviews of every placebo-controlled clinical trial submitted for initial approval of the six most widely used antidepressant drugs approved between 1987 and 1999—Prozac, Paxil, Zoloft, Celexa, Serzone, and Effexor. They found that on average, placebos were 80 percent as effective as the drugs. The difference between drug and placebo was so small that it was unlikely to be of any clinical significance. The results were much the same for all six drugs: all were equally ineffective. But because favorable results were published and unfavorable results buried (in this case, within the FDA), the public and the medical profession believed these drugs were potent antidepressants.


Dave Rowley said...

As you know, publishing negative results is fairly rare! No reason to think it would be different in drug trials. The same problem bedevils meta-analysis - since these reviews rely primarily on published work and negative results are seldom published, it's likely that clinical effectiveness is over-emphasised.

Anonymous said...

Hi Tom,
I think everybody underestimates the power of the placebo effect. So strong is the placebo effect, that drug trials cannot be trusted, no matter how thorough they are. The following blog describes the placebo effect very nicely: