Salami publishing: why we need to talk about it

Ding (Melody) Ding, Adrian Bauman, Esther van Sluijs and Klaus Gebel

Have you ever had this experience? You’re looking into the literature on the association between an exposure (such as accelerometer-measured physical activity) and an outcome (such as all-cause mortality) and, you find six papers, not one, written by the same authors based on the same sample. “Are these the same paper?”, you wonder. On careful perusal of the titles and abstracts, you realise that these are different papers on similar and related research questions, only with minor differences: one looks at the association in men, another in women, a third one in older adults, a fourth in a subsample with a pre-existing condition …

Or you may be interested in the effectiveness of behaviour interventions that promote healthy eating among young people – you find a trial that has led to dozens of publications, which individually address short-term behavioural outcomes, long-term behavioural outcomes, short-term determinants, long-term determinants, mediators and moderators of effects in both the short and long terms, cost-effectiveness … and the list goes on.  

In both these cases, by the time you finish reading all the abstracts, your surprise has become confusion. Where should I start? Which papers are relevant, and which to cite? Surely these papers could — and should — have been written as one or, at most, two papers!

This practice of “salami” publishing (also known as salami slicing, least/minimum/smallest publishable unit, or fragmented publication) is harmful at several levels.

First, it wastes the time (and often taxpayer money) of authors who have been funded to do “meaningful research”, peer reviewers, journal editors, peer researchers, policy makers and other stakeholders who have to read multiple papers instead of just one. In addition to the direct costs in time and resources, such practices also incur opportunity costs for producing, interpreting or disseminating more useful or novel research.

Second, it floods the published literature, distorts scientific evidence and misleads decision makers. Opportunities for salami publication could encourage “data dredging”, and covert duplicate publication can bias results from evidence synthesis.

Third, it creates an uneven playing field in the hypercompetitive world of academia, perpetuating a flawed reward system that is about quantity rather than quality.

Such experiences are common, which motivated us to conduct the study recently published in the IJE, in which we examined journal policies on duplicate and salami publishing. What we found was that, in a systematically selected sample of 209 journals in the disciplines of epidemiology, public health and general and internal medicine, only 13% of journals had policies that explicitly mentioned both duplicate and salami publication.

Even among the journals that provided an explicit statement on salami publication, the actual policy was often vaguely defined, without mentioning the consequences of non-compliance. A vague policy statement, such as “it is important to avoid splitting up a single study into several parts to increase the quantity of submissions”, is clearly not designed for implementation.

A bigger challenge in our research on salami publishing was a lack of consensus regarding its definition. In contrast to duplicate or redundant publication, which is more straightforward to define and identify, salami publication has been described as subjective and “not black and white”. Because of this inherent problem, there have been no attempts to quantify specific rates or trends in salami publication, except in narrowly selected fields. Tackling salami publication seems like a Catch 22: there is no universally agreed definition to determine the scope of the problem, so no mandate for journals to raise the issue; and, therefore, salami publication remains underdiscussed, undefined and undeterred. 

We think academic journals could play a major role here.

First, journal editors should work with the scientific community to develop a clear definition. This is particularly relevant for journals of scientific societies. For example, the IJE could involve the International Epidemiological Association in developing a specific and actionable definition of salami publication for epidemiology.

Second, journals should adopt policies with clearly stated consequences for non-compliance (e.g. rejection, retraction or contacting authors’ research institutions or funding bodies) and enforce these policies to deter salami publication.

Third, journals could educate authors and reviewers about the importance of producing truly “unique” publications by making a “salami slicing check” part of the submission and peer review process, similar to assessments for plagiarism and conflict of interest. In an ideal world, if it becomes harder to publish thinly sliced papers, authors would be less likely to try to do so.

Unfortunately, the reality is far from ideal. Salami publication, along with other unethical practices, is largely enforced by increasing competition driving the “publish or perish” culture of modern academia. Therefore, the most “upstream” strategy to prevent these practices is to rethink and redesign the academic reinforcement and reward system. Academic hiring and promotion committees and funding organisations should move away from valuing quantity over quality (including real-world impact).

In a world where the number of scientific publications is growing exponentially, academics need to reflect on the meaning of what is published, and there is a need for systems that reward academics (and not punish them) for doing so. After receiving the 2013 Nobel Prize in Physics, Peter Higgs said that he “wouldn’t be productive enough for today’s academic system”, raising a serious red flag for academia. The problem is embedded in the current system, and this inconvenient truth must be spoken.

Among other unethical academic practices, salami publishing is hindering the advancement and impact of scientific research. In the absence of interventions at journal, institution and system levels, it will continue to proliferate.

Read more:

Ding D, Nguyen B, Gebel K, et al. Duplicate and salami publication: a prevalence study of journal policies. Int J Epidemiol 2019; Sep 17. doi: 10.1093/ije/dyz187.

---

Associate Professor Ding (Melody) Ding is a Heart Foundation Future Leader Fellow at the Prevention Research Collaboration, Sydney School of Public Health, University of Sydney. She works broadly in physical activity, epidemiology and chronic disease prevention.

Professor Adrian Bauman is Emeritus Professor of Public Health at Sydney University. He has an interest in research translation and dissemination of preventive health evidence and programs.

Dr Esther van Sluijs is an MRC Programme Leader at the MRC Epidemiology Unit, University of Cambridge, UK. Her program focuses on understanding and changing young people’s physical activity and diet behaviour.

Dr Klaus Gebel is a Senior Lecturer in Public Health at the Australian Centre for Public and Population Health Research at the University of Technology Sydney. He works in physical activity and environmental epidemiology.