Stroke Trials' History Shows Bias Toward Good News

Negative findings are less likely to be published, researchers say

MONDAY, Sept. 25, 2006 (HealthDay News) -- Negative and even neutral findings from stroke trials are less likely to be published than positive findings, according to new research.

What makes the new study especially noteworthy is that stroke research in general has a more negative track record than other areas of medicine, the researchers noted.

"Publication bias is well-demonstrated in other areas of medicine and science," the lead researcher, Dr. David Liebeskind of the UCLA Stroke Center in Los Angeles, pointed out. "This is the first study showing such bias in stroke research."

Liebeskind's team analyzed 182 published and unpublished acute ischemic stroke clinical trials conducted from 1955 to 1999. They found that three out of the four studies not published (75 percent) had produced negative or harmful results, while 94 percent of the published studies had produced positive results.

The research, published in the Sept. 26 issue of Neurology, also found that negative studies took longer to be published than positive studies (an average of 2.3 years versus 2.0 years), particularly when they were sponsored by pharmaceutical companies (2.8 years versus 2.1 years). And negative studies were significantly more likely to be published in an abbreviated form.

The imbalance suggests that other negative studies never see the light of day, depriving researchers, physicians and patients of potentially valuable information, Liebeskind added.

"But it's difficult to put a number on it, because what we're essentially doing is looking for holes and gaps in the literature," he said.

"There are phase III clinical trials for neuroprotection in acute ischemic stroke that were never published, large studies that never made it, studies that were never presented at scientific meetings because of various influences," Liebeskind said. "That data is important."

"During the past half-century, we've studied a lot of neuroprotective drugs that failed to help patients with this devastating condition," he said. "So, when you find a modest degree of bias against publishing negative stroke studies, it's a little more striking."

Despite the evidence, Liebeskind said it was difficult to assess whether such publishing bias harmed or benefited patients. "It could go in either direction," he said, adding, "What publication bias does to patients, the public and the research community is leave us in the dark."

The findings reflect an earlier study published in May in the Journal of the American Medical Association, which showed that clinical trials funded by drug companies and other for-profit entities were more likely to report positive findings than similar trials funded by nonprofit groups.

The problem is a research culture that emphasizes "positive and winning" accomplishments, said Adil Shamoo, a professor of biochemistry and bioethics at the University of Maryland in Baltimore, who was not part of the study.

"The key reason for publication bias is a lack of incentives," Shamoo added. "Funding agencies very rarely fund or encourage negative publication. Journals do not want to publish negative results. Finally, no one gets promoted for publishing negative results. Have you seen a researcher win a Nobel Prize for finding that a certain drug does nothing?"

"The implications of this study are enormous because of the lost amount of knowledge," he said. "Because so many resources are lost or wasted, science has to do it again and again."

When negative data are not published or not presented at scientific meetings, that can adversely affect medical practice and patient outcomes because doctors may continue using drugs that are potentially hazardous, Shamoo said. As an example, he said, "Part of the data on Vioxx (the popular arthritis drug that was withdrawn from the market after it was shown to increase the risk of heart attacks) was not published or given to the FDA."

To ensure that research data is preserved for future researchers, Shamoo has long argued that there needs to be a federal repository for all government-sponsored studies.

Liebeskind takes that a step further. He supports the registration of all clinical trials -- including corporate-sponsored ones -- in a public trials registry.

But even that, Liebeskind said, may not be enough. "I would almost take the radical position that there needs to be a data clearinghouse subsequent to trial completion and that, at a certain point, full data sets are made available to the public," he said.

Even when studies are published, their full data sets, or protocols, often remain confidential. "There's a lot of reluctance to share this information because of the feeling that this data is proprietary and belongs to whoever sponsored and paid for the work," Liebeskind said.

"This has to be considered from the patient level," he said, noting that the research community has a special obligation to the thousands of patients who enroll in the clinical trials.

More information

For more on clinical trials, visit the National Institutes of Health.

SOURCES: David Liebeskind, M.D., UCLA Stroke Center, Los Angeles, Calif.; Adil Shamoo, Ph.D., professor of biochemistry and bioethics, University of Maryland, Baltimore, and cofounder, Citizens for Responsible Care and Research; Sept. 26, 2006, Neurology
Consumer News