Building a Better Clinical Trial System

Companies don't publish all their results. How can we get more information into the public domain?

It's always been a bit of a puzzle to me that the government has pharmaceutical companies conduct clinical trials for new drugs. Oh, it's not really that puzzling--clinical trials are very expensive, and having the companies do it keeps the cost of the trials off the government's books. But this seems like false economy. No matter how ethical the pharmaceutical companies are--and I, for one, do not think that they are comic book supervillians eagerly poisoning their customers--the incentives are skewed. Pharmaceutical companies have a strong personal interest in not finding out that their drugs don't work, or don't work as well as competitors. That's not their fault; it's just the hard fact of human nature.

One of the biggest complaints that critics have is that companies have an incentive to selectively publish data. Say you commission a head-to-head trial of your drug against an older generic. If the trial shows that your drug does better, smashing--publish away! And if your drug doesn't do so well? Just file that study away in a locked filing cabinet stuck in a disused lavatory with a sign on the door reading "beware of the leopard".

A group of researchers is attempting to tackle this problem head one. Here's their plan:

Endorsed by both the BMJ and PLOS Medicine, the Restoring Invisible and Abandoned Trials (RIAT) initiative is the brainchild of Peter Doshi, a postdoc in comparative-effective research at Johns Hopkins University (JHU), who wanted to do something about the fact that only half of all clinical trials are published. Doshi and his colleagues have already gathered 178,000 pages of previously confidential trial data on several drugs that came into the public domain thanks to legal battles or the policies of the European Medicines Agency, which has been releasing trial data in response to requests since 2010.

If other researchers have access to similar amounts of data, Doshi and his colleagues have set out a plan of action for those willing to participate in RIAT. First, they should notify the company who sponsored the research and provide 30 days for the company to commit to a new analysis based on the newly unearthed findings. The company would then have a year to come up with goods. If it does not comply, researchers should contact the BMJ and PLOS Medicine, which have agreed to consider publishing the data. The researchers may also include a critical appraisal of the trial they report in the discussion.

It's not a bad plan, as these things go. But I've got a rather simpler one: have the relevant governments take over the clinical trial process. Finding out which drugs work, and how well, is a true public good, one that no individual has incentive to provide on his own. Perhaps the government could spend more of its budget on this, and less on, I dunno, paying farmers too much for various agricultural products.