Up to 35,000 scientific articles may contain image manipulations serious enough to warrant retraction, according to a preprint posted to bioRxiv earlier this week (June 24). The estimate was based on an extrapolation from a thorough analysis of 960 articles published in the American Society of Microbiology’s Molecular and Cellular Biology, which revealed inappropriately duplicated images in around 6 percent—a rate that the authors suggest could be lowered by better image screening procedures at journals before publication.
“The frequent occurrence of inappropriate image duplication in published papers is a major concern, because it reduces the integrity and credibility of the biomedical literature,” the authors write in their paper. “In the present study, we sought to determine whether an investment by a journal to scan images in accepted manuscripts prior to publication could resolve image concerns in less time than was required to address these issues after publication.”
Study coauthor Elisabeth Bik, science editor at microbial genetics company uBiome, looked for any of three red flags when scanning the articles: duplication of the same image panel for different experiments within the same paper; duplication of a panel with a rotation or shift; and duplication of Western blot lanes or other parts of a photo within a single image. Figures she flagged were then confirmed or rejected as problematic by her coauthors on the study.
In total, the team highlighted problems with 59 of the 960 papers. After the researchers brought these concerns to the journal’s attention, 42 papers received corrections and five were retracted—a result that the authors tell Retraction Watch were within expectations given that many image problems are not the result of deliberate misconduct.
“I think we expected that most image problems were the result of error in assembling figures,” study coauthor Arturo Casadevall, editor in chief of mBio, tells Retraction Watch, “so the 10 percent retraction was not surprising.”
The authors note in their paper that while each image issue in a published paper takes around six hours for journal staff to address, those same issues can be identified in just 30 minutes before publication by image specialists.
“Catching these errors before publication is a much better strategy than after publication,” Bik tells Retraction Watch. “I hope that our study will result in more journals following in the footsteps of [the American Society of Microbiology] by starting to pay attention to these duplications and other image problems, before they publish their papers.”
Casadevall offers an additional recommendation for how to reduce the likelihood of manipulated images slipping into the literature. “We suggest that one mechanism for reducing these types of problems is to have someone else in the group assemble the figures,” he tells Retraction Watch. “At the very least that would mean a second set of eyes looking at the figures.”