What’s wrong with ordering a few reprints? New issues in publication bias
A recent article published in the BMJ raises questions about the extent and type of publication bias that exists in the literature. Publication bias is the selected publication of studies based on the results, such as only publishing studies that demonstrate a drug works while not publishing studies that demonstrate harms.
The study authors, including Evidence Live Faculty member Ben Goldacre author of the best-seller Bad Science, explore the potential implications of study funding and high reprint orders. They contacted the editors of the top general medical journals (i.e. JAMA, Lancet, NEJM, Ann Intern Med, and BMJ) and requested information on the 20 articles with the highest number of reprint orders. After matching the articles with controls, the authors evaluated whether study funding (i.e. industry, mixed, other or none) was associated with higher numbers of reprints.
The results are telling. The Lancet led the way with a median of 126,350 reprints for the top articles with a range from 24,000 to 835,100. The BMJ was a distant second with a median of 13,248 (range 1,000 to 526,650). Unfortunately JAMA, NEJM and Ann Intern Med did not provide information.
Overall, compared with controls papers with high reprint orders were considerably more likely to be funded by the pharmaceutical industry (odds ratio 8.64, 95% CI 5.09 to 14.68). In addition the cost for reprint orders ranged from £4,002 to £1,551,794: reprints are evidently a lucrative source of supplementary income for journals.
While not designed to detect publication bias, the article highlights the importance of thinking outside the box. Evidence-based medicine is filled with cutting edge issues that are continually evolving and emerging. Do you think that a paper with potentially high reprint orders may affect an editor’s decision to publish? Should journals disclose the number of reprints for each article?
What do you think are undiscovered sources of publication bias? Here's your chance to tell the experts. Tweet to us @EvidenceLive with your thoughts, and come to the conference at the University of Oxford, 25-26 March 2013.