Accusation of Salami publication: the new bane of large database investigations? Young investigators beware!

Abstract

Editor, We were recently contacted by the Editor-in-Chief of the European Journal of Anaesthesiology (EJA) to say that it had been brought to the attention of the Editorial Office that a considerable overlap exists between an article we published in Paediatric Anaesthesia in 20091 and an article we recently published in the EJA in 2010.2 We wish to begin by thanking the Editors for giving us a chance to respond to this accusation of ‘salami slicing’. The advent of anaesthesia electronic databases has created unlimited opportunities for outcome research in addition to allowing for hypothesis-generating studies that can be used to design future prospective studies.3 Unfortunately, the strengths of electronic databases are also their weakest points. Although it is possible to run several queries of a large database system, it is impractical (journal space limitation, or similar but unrelated hypotheses) to include all aspects of a study in one manuscript. Although it is impossible to explore and report all the components of a study in a single publication, it is important to give careful consideration to all subsequent publications from the same database in order to avoid Salami publication. So what is salami slicing? Salami publication is defined as intentional publication in any form of an article that overlaps substantially with one previously published by the same or different authors.4 In some cases, the authors involved attempt to hide the existence of a previous publication by submitting to a different journal, or more commonly by publishing the same study in different languages.5 More frequently, the results of two or more studies are drawn from the data collected from a single study or data collected over the same period. When duplicate publication is the result of re-analysis and clear cross-referencing (overt publication), it may be deemed acceptable.5 On the contrary, when the same data is published again without any cross-referencing (covert publication), this is largely unacceptable for a variety of reasons.6 It is not clear how to categorise longitudinal data collected during routine clinical care with increasingly popular electronic information systems. These data repositories are often the source of hypothesis-driven or hypothesis-generating studies. Very often, one hypothesis may lead to a slightly different but related hypothesis. Due to current strict requirements and page limitations for most journals, it is often impossible to report all the findings of one study in a single manuscript. We explain our ‘apparent salami slicing thus’. The two manuscripts dealt with the subject of paediatric intra-operative hypotension. The first publication explored risk factors for intra-operative hypotension in children undergoing non-cardiac surgery. We identified a number of risk factors as detailed in the publication in Paediatric Anaesthesia.1 Following this publication, and using the same clinical information database, we explored the role of childhood obesity in the occurrence of intra-operative hypotension. Although the two articles dealt with paediatric hypotension, the central question and hypotheses of the studies were different. The two database queries were separated by almost 12 months. Although we used the same methodologies for the two studies, we studied different age groups of patients and reached independent conclusions in the two studies. The first publication was appropriately referenced in the second study published in the EJA. Duplicity was not our intention. With the clear benefit of hindsight, perhaps what we ought to have done was submit a copy of our earlier publication at the time of the initial submission to the Editors of the EJA with a note explaining why the two studies are different.5 We sincerely hope that readers of the EJA will be able to read the two publications for themselves and draw their own conclusions. What is known about the subject In a cohort study of 22 263 children aged 1–17 years, some independent risk factors for intra-operative (pre-incision) hypotension were documented for the first time. We concluded the following observation from this study: Intra-operative hypotension is a common problem in children and the risk factors are not well characterised.1 This study identified some general risk factors as independent predictors of intra-operative (pre-incision hypotension). The main factors explored were baseline hypotension, duration of pre-incision period, patient's age, high American Society of Anesthesiology status and propofol co-induction. What the current study adds This study explored the role of obesity in intra-operative (pre-incision) hypotension in 19 400 children aged 2–17 years undergoing non-cardiac surgery.2 The outcome variable was intra-operative (pre-incision) hypotension. The main predictor variable was BMI category. We concluded the following: Obesity is a risk factor for intra-operative hypotension in children. Obese children are more likely to have a longer pre-incision period, more likely to require propofol co-induction and more likely than their lean peers to be hypotensive following induction of anaesthesia and that baseline (pre-induction) hypotension was less common in obese children than their lean peers. Finally, although we understand and will respect whatever decisions the Editors of the EJA take regarding our publication, we must be emphatic in stating that duplicity (overtly or covertly) was not our intention.