Table of Contents
Illustration: David Parkins
The problem
Dear Nature,
I work in an oncology laboratory in the United States. After being asked to review a paper, I noticed that in one of the main figures, several images appeared identical to those of a previous publication I had read.
I empathize with how difficult assembling papers can be, and I understand that unintentional errors can be introduced. Accusing authors of misconduct can have significant professional, social and (potentially) legal ramifications. Even worse, it can alert authors to cover their tracks if actual misconduct did occur.
Who do I tell and what should I say? Is there anything that we, as reviewers, should ask to give authors a chance to respond that is respectful, but that also ensures sound work? – A wary oncology researcher
The advice
Paper retractions are growing, along with awareness of the problem of scientific misconduct. From plagiarism, to text generated by artificial intelligence, to authorship for sale, many papers are being submitted to journals in bad faith. If those papers are then published, this can weaken research fields and potentially have tragic consequences if, for instance, oncologists rely on unreliable research when treating people with cancer.
But part of the review process is checking for mistakes made in good faith, too. Not all papers that are flagged are actually fraudulent. As you suggest, errors can creep into papers even when the authors did not intend to take shortcuts. Unfounded accusations can prevent the publication of important research, damage careers and potentially have blowback for the critic.
Peer reviewers play a crucial part, together with journal editors and publishers, in helping to ensure the quality and credibility of published journal articles. But a lack of time and review guidance can limit reviewers’ ability to raise concerns.
An ‘agony aunt’ for working scientists
Nature reached out to three researchers who have navigated this delicate terrain.
Extend the benefit of the doubt
There might be a reasonable explanation for the images’ similarities. Jennifer Byrne, an oncology researcher at the University of Sydney, Australia, suggests checking for two things. First, do the images in the earlier paper relate to the same things (proteins and cell lines, for example) that are described in the later paper? (You can also consider annotations around the image, such as whether new descriptions been added.) Second, do the two papers have authors in common? If both conditions are met, the authors could be reusing their previous data. This might be justifiable or an honest mistake, says Byrne, who also researches scientific integrity.
Even if the authorship teams are different, there might be good reason for reusing the images. “Some images can be provided by repositories as data to be reused for different research questions,” Byrne explains. Or the submitting authors might have obtained permission from the copyright holders to reuse images from a previous study, for comparison purposes. In any case, Byrne says, “any intentional image reuse should nonetheless be clearly described in the manuscript”. It should therefore also be cited in the references, following almost all publishers’ guidelines.
These kinds of concern should be mentioned in the reviewers’ comments to the authors. However, Byrne says that it could be more worrying if the images in the paper under review refer to things other than those in the published paper.
Apart from apparently duplicated images, a reviewer might pick up on other issues in a submitted manuscript. For instance, Brian Jones, a biologist at Murdoch University in Perth, Australia, checks a selection of references to ensure that they back up the statements being made by the submission. Again, divergence could be an honest mistake by an author in a rush. Or it could be a sign of citation manipulation, in which irrelevant references are deliberately inserted into a paper to boost citation scores.
Talk to the editor
Issues should be raised confidentially with the editor of the journal, for instance by e-mailing the handling editor or using the ‘contact the editor’ feature in online submission systems. “It’s important to recognize that manuscripts are provided to reviewers on a confidential basis, so reviewers should not communicate concerns to anyone outside the peer-review process,” Byrne emphasizes.
This communication should include the evidence you’ve gathered, including a link to the previous publication and the figure numbers that seem to be duplicated, says Jillian Goldfarb, an engineering researcher at Cornell University in Ithaca, New York. In this first alert to the editor, she recommends that you “ask for guidance as to whether or not you should continue to review the paper, or if they would like to pause the review process to investigate the matter”.
Check your tone
Whether in communications with the editor or in comments for the authors, it’s risky to suggest that wrongdoing has been committed. Goldfarb has seen first-hand the political complexities of publishing. In 2023, she resigned as a journal editor, partly because of ethical concerns about suspicious papers.
The advice-givers emphasize that, to avoid legal action, it’s not a good idea to use words such as ‘fraud’. Instead, it’s best to express concisely and objectively the reasons for concern. “Where you will get into trouble is if you suggest that the errors were deliberate,” Jones cautions. Rather than mentioning plagiarism, for instance, he suggests writing something along the lines of “paragraph 6 appears to be identical to that in the paper by XXX”. For detailed legal advice, however, you should contact a professional.
This advice applies even in seemingly blatant cases of misconduct. Jones gives the example of a submitted paper that included an author who had died five years previously. Still, he covered his bases: “I checked the e-mail in that case, and the institute confirmed the death and expressed a lack of knowledge of the paper, so it wasn’t a posthumous submission.”
