Skip to main content

13-12-2012 | Article

Industry-sponsored trials prone to bias


Free PDF

medwireNews: Drug and medical device studies are more biased when the sponsor of those studies is the manufacturing company, according to a review conducted by The Cochrane Collaboration.

The nonprofit organization, which examines medical research and the strength of evidence that supports medicine, updated a review conducted 10 years ago. This time, they compared 48 primary research studies of drugs and medical devices sponsored by industry with those that had other sponsors. The measured outcomes included favorable results, favorable conclusions, effect size, risk size, and whether the conclusions agreed with the study results.

The review found that industry-sponsored studies across a wide range of diseases and treatments had more approving results and conclusions than those sponsored by other parties, and that this partiality could not be explained by standard 'risk of bias' assessments.

"What was very interesting is that there was really strong evidence that the drug that looked better was the one that was made by the sponsor of the study," Lisa Anne Bero, a University of California professor and co-author of the review," explained to medwireNews.

"The major implication is that… an insurance coverage committee really needs to know which is the better, [more trustworthy] study."

Lately, clinical research increasingly gets directly sponsored - or at least partially funded - by the companies that make the drugs and medical devices involved. Yet, as Bero pointed out, there is little incentive for drug companies to provide the data that validates the studies' results and conclusions, which insurance companies and doctors depend on.

"[Manufacturers] are supposed to register [data] in clinical trial registries, but the accuracy of what they put in those registries is pretty poor," Bero said. "There have been reams of evidence that support that negative data still doesn't get out there to the public." One incentive Bero posed is to discontinue human subject approval in trials until accurate data from previous trials is provided.

Bero emphasized that various components of a study can be vulnerable to prejudice. "You can't just say a drug industry-sponsored study is bias[ed] because randomization is bad or it's not blinded," she said. "There are many reasons and ways to bias a study. It's very complicated."

As more people become insured under the Affordable Care Act, the greater the responsibility insurance committees will have when choosing the medicines to provide cover for. "Those committees need to be very skeptical about the evidence to make sure that it's not biased towards any particular products," Bero said. "Those should be evidence-based decisions."

By Peter Sergo, medwireNews Reporter