The problem of multiple publication |
||
|
Anyone undertaking a systematic review will soon meet the problem of the same
study being published more than once. Sometimes it is just that a study
published, for example, in Welsh, is also published in a Serbo-Croat journal,
though most often dual publication is English and another language. Sometimes it
is a case of salami-slicing, where a little bit more information is given in a
second publication, but 90% or more of the words are the same. Sometimes the
duplication is obvious, but information on the same patients has been shown to be
published more than once, and sometimes in papers with completely different
authors [1].
|
Whatever the circumstances, it is really annoying, and never more so when the duplicate publications don't refer to one another. The poor reviewer has then to check things like demographic data to show that these two sets of patients are the same.
Systematic reviewers know that duplication is a big problem, but just how big is it? Previous estimates have suggested that about 15% or so of reports were duplicates, with up to 28% of patient data being duplicated [1]. There is a perception that much repetitive publication involves publication in several languages, but this is not always the case.
A survey of the surgical literature [2] suggests that one in every six original articles published in leading surgical journals is some form of multiple publication.
Study
Original articles published in 1998 in Surgery, British Journal of Surgery and Archives of Surgery formed the basis of the study. Names of first, second and last authors were combined with a few key words to search PubMed for possible duplication. When an abstract addressed the same topic as another abstract or paper, and shared methods or results and conclusions, it was suspected of duplication, and was retrieved in full for detailed examination.
A grading system for possible duplication was used:
- Dual publication was defined as papers having identical material, methods and conclusion.
- Potentially dual publication was defined as papers having almost identical material, methods and conclusion.
- Salami slicing was defined as studies representing part of, continuation of, or partial repetition of material found in the original article.
Results
The three journals provided 660 original articles, of which 92 (14%) led to 147 suspected duplicates, about 1 in 6 of the original articles. Of the suspected duplicates, 113 (77%) were not cited by the original article. About two-thirds of the suspected duplicates were published in surgical journals.
Twenty of the 147 suspected papers were defined as definite dual publication (14%), 50 as potential dual publication (34%) and 77 as salami-slicing (52%, Figure 1). Of the 660 original articles, 70 (11%) had actual or potential dual publications involving identical or almost identical studies.
Figure 1: Type of repetitive publication of surgical papers
|
|
Multiple publication was not confined to a second publication of a single article. Some papers had two or more suspected duplicates (Figure 2), and one had the distinction of having six suspected duplicates. Figure 2: Multiple duplication of surgical papers |
|