Where’s the Missing Data?

According to Ben Goldacre, the first randomized trial is in the Bible: Daniel was forced to eat the king’s meat in order to become stronger. He refused, saying he only wanted to eat vegetables and proposed a trial in whether the king’s meat was actually making people stronger (It didn’t) This was published in the Bible “a very high impact journal”, so randomized trials are not so new nor complicated.

But in medicine, contrary to popular believe, randomized trails aren’t that old. Decisions used to be made by discussions between doctors based on reputation, “somewhat like people in a psychiatric ward believing to be Napoleon” When randomized trails started, many doctors considered them immoral, since some of the participants were withheld promising treatment.

When you do a randomized trial in medicine, you don’t have to publish it. Doctors can just hide studies of which the results aren’t in favor of your new drug. The logo of the Cochrane Collaboration that aims at improving evidence in health care, represents a number of studies on steroid injections given to pregnant women. The lines represent studies that weren’t conclusive and for years steroid injects were not used. But when the studies were synthesized (into the diamond) This is why, says Ben, combining studies is very important.

Ben furthermore shows that studies are sometimes ‘lost’ if they are small and negative. Especially for anti-depressants, there are several stories in which only positive results were published. A new initiative is AllTrials: Ben is going to try to get the data from all trails that are registered by the FDA, so nothing can go missing again. Great work that’s improving science!