Negative Results Left Out of Stroke Literature?

Caroline Cassels

September 28, 2006

September 28, 2006 -- Stroke studies that report negative results are less likely to be published than those reporting positive or neutral results, a new study suggests.

The comprehensive review of all acute ischemic stroke trials reported in English over a 45-year period revealed that 75% of unpublished studies of treatment for ischemic stroke had negative or harmful results, vs 6% of published studies.

"This is evidence of publication bias. Failure to publish negative results deprives doctors, patients, and future researchers of valuable data and intellectual discoveries," the study's first author, David S. Liebeskind, MD, from the University of California, Los Angeles Stroke Center, said in a statement released by the American Academy of Neurology.

The paper appears in the September 26, 2006 issue of Neurology.

The researchers examined trial methodology, quality, outcome, study sponsorship, and timing of publication to identify various forms of publication bias. Potential forms of bias included nonpublication, abbreviated publication, and time lag.

Less Harm Than Good

The analysis included all studies reported in English for the treatment of acute ischemic stroke from 1955 to 1999. In total, there were 178 published trials, which included 73,949 subjects and evaluated 75 agents or nonpharmacologic interventions. The researchers also identified a total of 4 unpublished studies that met their selection criteria.

They investigators found that trials in which the tested agent proved harmful were substantially less likely to be published than those in which the study authors felt the tested agent was neutral or beneficial.

Of the 4 unpublished studies, 3 (75%) demonstrated author-described harmful effect of the study agent, while the remaining study yielded neutral findings. In comparison, of the 178 published studies, 11 (6%) reported harmful effects.

Conspicuous Absence

Further analysis suggested there is an even larger set of nonpositive trials has failed to appear in the literature. "A conspicuous lack of small nonpositive trials was apparent on review of all graphic analyses, suggesting that nonpublication bias affected both single-center and multicenter studies and both corporate and noncorporate studies," the authors write.

In addition, the authors report that, after adjustment for trial quality, larger studies were more likely to result in the publication of positive outcomes.

Of the 178 published studies, 136 (76%) were efficacy studies vs self-designated dose-escalation, surrogate, or safety studies, a finding that "raises concern that the preliminary phases of human stroke research remain unpublished."

Furthermore, trials that were reported only as abstracts or in abbreviated form were less likely to show an author-described beneficial effect than those published as full manuscripts.

Call to Action

Analysis of time to publication did not reveal a time-lag bias in studies of acute ischemic stroke. However, there was a suggestion of different rates of publication based on outcome among corporate-sponsored stroke trials.

"Only a small number of studies were included in this analysis, yet a trend was noted toward more rapid submission of a beneficial trial following trial completion. This finding likely reflects financial pressures driving corporate-sponsored stroke studies," they write.

However, they add, the source of financial support for stroke trials had no relationship to the frequency of author-described beneficial trial outcomes.

In order to ensure that scientists and patients have a complete picture of the stroke literature, the authors highlight the need for prospective reporting of all trials, regardless of outcomes, in centralized stroke registries.

"Investigators should be encouraged to report the results of all clinical trials, fulfilling their ethical obligations to patients and the scientific community. Rapid publication of negative stroke trials may spare patients from exposure to useless and overtly harmful treatments and from participation in futile ongoing trials," they write.

Neurology. 2006;67:973-979.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.