Case text (Anonymised)
A paper submitted to Journal A was rejected after critical peer review. Although the data and methods were sound, the data in the paper were not new and had been described, at least in part, in previous publications. The authors could also have combined the outcomes in the current paper with previous papers, thereby avoiding salami publication. The methods section was opaque, making it very difficult to decipher which subjects were new and which were already counted in previous studies.
The authors appealed against the referee’s comments saying that the data were new, and the method of data collection was entirely different from their previous studies. They argued that there were too many outcomes for all of them to be included in one study. The journal that published one set of outcomes had actually asked them to divide them. All four papers referenced each other so that readers could see what the overlap was. The authors also stated that it was only through analysing the previous papers that the hypothesis for the latest study came about. But they agreed that the method section was not clear and agreed to rewrite it to clarify the relationship with the previous papers.
Journal A considered the appeal but did not feel able to recommend or reject the paper because they did not have the depth of knowledge about the dataset that the reviewer did. There were also discrepancies between the author’s rebuttal and the paper itself. The rebuttal was therefore sent to the original referee who re-analysed the data and concluded that the overlap between the current and previous studies amounted to about two thirds of the data.
There were major discrepancies within the current manuscript in terms of the number of participants analysed. Even he could not be sure what the degree of overlap was, but he was not surprised that the editorial team were confused, and he suggested that readers would never be able to work this out. He therefore recommended that it would not be ethical for Journal A to publish the paper.
The editors can’t be absolutely sure that this is a case of salami publication without asking the authors to provide a much more detailed paper that they have very little intention of publishing.
- What should the editors do?
- Is salami publication a heinous enough crime to warrant all the extra effort and bad feeling?
- A distinction needs to be made between salami and redundant publication: where there is a two thirds overlap, it is redundant publication.
- If the hypotheses were completely separate questions then it is acceptable for them to be posed in two separate papers.
- If they are related questions, or very closely related, then they should be published as a single paper.
- Salami publication is where papers cover the same population, methods, and question.
- Splitting up papers by outcomes is not legitimate.
- It is an editorial decision as to whether to publish or not: there is no ethical problem here.