Published online 15 June 2010 | Nature | doi:10.1038/news.2010.299


Companies pledge to make more trial data public

Voluntary agreement by drug firms calls for all large clinical trial results to be published.

Doctor with clipboardMore clinical trial data may go public if the pharmaceutical industry follows voluntary guidelines announced last week.Punchstock

An international organization of pharmaceutical companies has adopted new guidelines governing the publication of clinical trial results. The policy, announced on 10 June, calls for the industry to publish the results of late-stage, 'phase 3' clinical trials in peer-reviewed journals, regardless of the trials' outcome.

The new guidelines from the International Federation of Pharmaceutical Manufacturers and Associations address mounting criticism of pharmaceutical companies for not fully reporting the results of clinical trials — particularly those that fail. As a result, researchers looking to analyze all clinical trial results for a given therapy often run up against publication bias: the tendency to publish positive results but not negative ones. One recent study found that less than half of completed clinical trials registered in publicly accessible databases were published in peer-reviewed journals (see 'Publication bias continues despite clinical-trial registration').

The guidelines also address concerns over the use of 'ghost-writers' — writers or statisticians hired to work on the paper without being credited as an author. Nearly 8% of papers published by either industry or academia in six prestigious medical journals may have used a ghost-writer, another recent study found (see 'Ghosts still present in the medical machine'). The new policy calls for such writers, as well as any potential conflicts of interest they may have, to be acknowledged.

“It leaves a lot of space open for bias to creep in.”

"I personally think this represents a major development," says spokesman Matt Cahill of Merck, a pharmaceutical company headquartered in Whitehouse Station, New Jersey. "It's important to have broad guidelines so that everyone has a base set of standards for publishing clinical trials data."

But others worry that the guidelines do not go far enough. "Overall, this is a move in the right direction," says John Ioannidis, an epidemiologist at the University of Ioannina School of Medicine in Greece. "But it leaves a lot of space open for bias to creep in."

Bias cut

The policy calls for submission to a peer-reviewed journal no later than 18 months after a clinical trial is completed. Furthermore, all industry-sponsored clinical trials should be submitted for publication, whether the results are positive or negative, the policy states. But the guidelines focus in particular on phase 3 clinical trials — ones that use large groups of patients to check the results of previous small trials, stating: "At a minimum, results from all phase 3 clinical trials and any clinical trial results of a significant medical importance should be submitted for publication."

At present, the US Food and Drug Administration requires drug companies to register clinical trials of FDA-regulated products in a US National Institutes of Health database called, but does not require submission of trial results to peer-reviewed journals.

The new guidelines' focus on phase 3 trials leaves out many important earlier-stage trials, says Ioannidis. Further, he notes that the guidelines do not ask for trial protocols to be deposited up front in full detail in public databases. "There is still room for manipulating outcomes and analyses," he says.


Meanwhile, Cahill says that Merck routinely submits results from phase 3 trials, and often earlier-stage trials, for publication in peer-reviewed journals. But when a trial fails, many journals lose interest in publishing it, he adds. "We often get rejections of negative trials," Cahill says. "The industry does as a whole have a difficult time finding a journal in which negative results from clinical trials can be published."

Ultimately, the guidelines may have little effect simply because they are voluntary, says Deborah Zarin, director of Many companies had voluntary guidelines calling for releasing clinical trial designs and results, but rarely followed them until legally required to deposit the information in publicly accessible databases.

Overall, Zarin says, the guidelines outline behaviour that should already have been the status quo. For example, the policy notes that publications should provide "an accurate report of the clinical trial findings, including adverse events". "That's great," she says. "But it simply reiterates a very basic, fundamental principle that all authors should adhere to." 

Commenting is now closed.