Like the WikiLeaks documents on the war in Afghanistan, a new study on the relationship between the funding sources of drug trials and their results doesn’t tell us much in general that we didn’t already know. The study found that—surprise!—industry-funded trials are more likely than trials with other funding sources to publish outcomes favorable to the drugs being tested. But the interesting information is in the details, such as delays in publishing the results.
The study looked at 546 clinical trials of five drug classes registered in ClinicalTrials.gov, a public database, from 2000 to 2006. The drug classes were anticholesterol drugs, antidepressants, antipsychotics, proton-pump inhibitors, and vasodilators. Sixty-three percent of the trials were funded primarily by industry, 14 percent by federal government sources, and 23 primarily by nonprofit or nonfederal government sources.
Eighty-five percent of the industry-financed trials had positive outcomes for the test drug; compared with half of the government-funded studies and 72 percent of those underwritten by nonprofits or nonfederal government sources. Some of the latter trials also had some industry funding, and they were more likely to have favorable results than trials with no industry funding.
It’s easy to imagine that one reason that industry-funded trials had favorable results is that companies were selective about which trials they underwrote – they placed their bets with studies that were likely to make their products come out ahead. And most of the industry-sponsored studies were phase 3 or 4 trials, advanced enough to give some measure of certainty about how the drugs would fair. But when the researchers adjusted for trial phase, the studies with industry backing were still more likely to have positive results.
Another key revelation is that the industry-funded trials were least likely to be published within two years of completion. Delays in publication raise a red flag, suggesting the possibility that a company is withholding unfavorable information about a drug or manipulating the study in some way to cast the findings in a better light. Previous studies have found big discrepancies between published and unpublished data, with the unpublished data containing more negative information. This was the case with Avandia, a diabetes drug that a panel of the Food and Drug Administration voted to restrict last month because of evidence that the drug increases the risk of heart attacks. These were the results of a study conducted over a decade ago by SmithKline Beecham, but the manufacturer did not publish them.
Industry-funded trials also had the lowest rate of registration in ClinicalTrials.gov before publication. ClinicalTrials.gov was started in 1999 to address concerns that there was too much secrecy in the clinical trials process, and therefore potential for drug companies to manipulate data and make their products look safer and more effective than they really are. “Without impartial, a priori registration of trials,” the authors wrote, “planned data analyses and study outcomes may be altered or omitted in favor of the experimental treatment after trial data are available.” Registration was voluntary for most of the period that this study covers.
In 2005, the International Committee of Medical Journal Editors started requiring that trials be registered before they begin, or else they won’t be considered for publication. Since 2007, the FDA has required all trials to be registered before they begin. These changes should help avert bias going forward, but they can do nothing about the hundreds of published studies on medications that are influencing patients’ treatments right now.
And given that so many of the trials were funded by drug companies, Florence T. Bourgeois, the lead author and an emergency medicine physician at Children’s Hospital Boston, toldReuters, those companies have had a “very large impact on medical decision-making.”
Susan Gilbert is the staff writer for The Hastings Center and managing editor of Bioethics Forum.