Spread out over seven tables, attendees at the Ethics Clinic discussed issues surrounding retractions. Each table included at least one member of CSE’s Editorial Policy Committee as moderator. Two hypothetical cases provided springboards for discussion. Moderator Robert L Edsall periodically brought the room together for large-group discussion.

Case 1, Part 1: An editor receives a short missive from the first author of a paper published in the editor’s journal several months earlier. It states:

I have recently had the opportunity to review some of the raw data used for this paper, and it is clear that the data as represented in this paper cannot be relied upon. Therefore, I request that you retract this paper.

The editor has no further information.

The participants generally agreed that, confronted with such a vague request, they would ask for more information before proceeding. The ambiguity of the letter sparked a discussion about possible underlying malicious intent and its implications. It was generally agreed that the paper should be retracted if indeed its data could not be relied on but that it would be important to have a fuller explanation before publishing a retraction. One participant noted that the editor in the case might never be sure what the problems with the data are; she suggested publishing “an expression of concern” and contacting the authors’ institutions and sources of funding.

Participants said that their journals had different policies about what to do in such situations. Some said that if all authors and their institutions agreed to retract the paper and if the authors said that the data could not be simply corrected, the editors would retract it. Others said that although it would be good form to notify the authors’ institutions about the retraction, the institutions’ permission was unnecessary. The major point of consensus: Don’t publish a retraction without a cohesive explanation from all authors—in other words, until you think you know what happened.

Case 1, Part 2: Several months after receiving the letter in Part 1, the editor learns through an unofficial source that the letter was part of an agreement that the lead author had made with the US Office of Research Integrity (ORI), which had found him guilty of “scientific misconduct”, specifically “falsifying samples and constructing falsified figures”. The lead author maintains his innocence and claims that a “systematic error” made the data unreliable.

Those further details led to a discussion of the ORI’s role in scientific publishing. Some argued that if the ORI finds against an author, the editor should publish a retraction. Discussion ensued over how well the ORI informs journal editors of its findings. Some participants talked about their frustrating experiences in trying to obtain case documents from the ORI under the Freedom of Information Act.

If a retraction is published, what should it say? Generally, it should be as “brief, objective, and as clean as possible”, said one participant. Another suggested directing readers to the ORI for further information.

Do editors have a duty to spread the word about authors who have been found guilty of misconduct? If so, when should they contact the authors’ institutions? Generally, participants seemed to agree that it was appropriate to send a note to other editors who had published articles by the author in question and who had published articles that cited the now-retracted article. However, thoughts on when and how to contact the institutions were mixed.

Case 2: The second and third authors of a published paper write to the editor claiming that the first author had falsified some of the data. These two authors request a retraction or, barring that, that their names be removed from the paper.

Limited time prevented full discussion, but it was generally decided that if agreement exists among all authors (including the author accused of falsifying data), it would be appropriate to retract the paper. However, if such consensus is not reached, the editor’s next step should be to contact all the authors’ institutions, which might prompt them to investigate. To avoid legal difficulties, the phrasing used is important; for example, phrasing the statement to say that the coauthors had a “concern” would be appropriate. The participants seemed to agree that the names of the two coauthors could not be removed, because they had in fact contributed substantially to the paper.