Skip navigation
Link to Back issues listing | Back Issue Listing with content Index | Subject Index

Déjà vu all over again

Review
Results
Types of duplication
Comment

Bandolier has reported on a number of different forms of bias found in clinical trials, and briefly referred to problems with duplicate publication. This is the publication of an article whose content overlaps substantially with another article published elsewhere. If the connection is not known and the connection not referenced (covert duplication), we can be misled into thinking that more information is available than is the case, and this can alter our attitude to an intervention. A study not only looks that this in more depth, but presents a taxonomy of duplicate publication [1].

Review

The source was a comprehensive list of systematic reviews in perioperative medicine available on the website of the University Hospital of Geneva. These were screened for any mention of duplicate publication. Where duplication was not mentioned, authors of the reviews were contacted to ask whether they had looked for, and found, duplicate publications.

The oldest of any possible multiple publication was used as the index, and compared with any later publication for authorship, and for clear or unclear cross-referencing. Unclear cross-referencing was so defined if the corresponding article was referenced, but the relationship between the articles was obscured.

Results

There were 141 systematic reviews screened for mention of duplicate publication. Of these, 42 (30%) reported duplication. Authors of the other 99 systematic reviews were contacted, and, of those, 14 reported duplicate publication not mentioned in the original review, 55 did not identify any duplicate publication, and 30 did not respond. Overall, 56 systematic reviews (40%) identified duplicate publication.

The 56 reviews had 1131 main articles with 129,000 patients. 103 articles were duplicates (8% of total) with 12,500 patients (9% of total). Many of the 103 articles were duplicated several times (Figure 1).



Figure 1: Duplicate publishing





There was no cross-referencing (duplication was covert) in 65 of the 103 duplicate articles (63% of the duplicates, 5% of all original articles including duplicates). Some of the duplicates (12%) were translations. Pharmaceutical sponsorship occurred in 34 (33%) of duplicate publications. Most (90%) were published within three years of the original article.

Types of duplication

Six different patterns of duplication were identified, and these are outlined in Table 1. Pharmaceutical sponsorship was very common (81%) in type 1B, and uncommon (4%) in type 2.



Table 1: Taxonomy of duplicate publication



Pattern
Number
Description
1A
21
Reproduction of an already published article using identical sample and outcomes
1B
16
Assembly of two or more articles to produce another article
2
24
Reporting on different outcomes from the same study sample
3A
11
New data added to a preliminary article
3B
11
Reporting part of a larger trial and reporting identical outcomes
4
20
Sample and outcome different from the main article


Comment

There may be good reasons why duplicate publication can occur. For instance, if an important paper were originally published in Welsh, a translated version might be appreciated by the few of us who are not fluent in Welsh. Or a study reporting outcomes at two years might, three years later, report outcomes after five years. In both cases, though, what we would hope is that any subsequent publication would clearly refer to the original. Then we would know where we stood, and not, perhaps, be misled into believing that there was twice as much evidence than we first thought.

This study tells us that about one paper in 12 is a duplicate, and for about 1 in 20 there is no cross reference, so we might be misled if we didn't have our thinking caps on. We should not be surprised, because duplication has been reported before, and in areas other than perioperative medicine.

For example, two analyses of the otolaryngology literature in the last few years come to similar conclusions [2,3], but using different strategies for finding duplication, other than systematic reviews.

One [2] identified all the authors and articles published in one journal over an eight-year period. There were 1,965 authors of 1082 articles. They picked 1,000 at random, found other articles they had published, and checked them for duplication. Of the 1,000 authors, 201 had published 644 articles with some degree of duplication, amounting to a 1.8% duplication rate.

The secondexamined all original articles from two otolaryngology journals in 1999, and then searched on the first, second and last authors for cross-referencing to possible duplicate publication. They found that 40 of 492 articles had 42 duplicate publications – the same 8.5% found in the Geneva study.

Duplicate publication is present, and may mislead us. There is a considerable literature, and these three papers provide some very interesting reading.

References:

  1. W von Elm et al. Different patterns of duplicate publication. An analysis of articles used in systematic reviews. JAMA 2004 291: 974-980.
  2. BJ Bailey. Duplicate publication in the field of otolaryngology-head and neck surgery. Otolaryngol Head Neck Surg 2002 126: 211-216.
  3. EL Rosenthal et al. Duplicate publication in the otolaryngology literature. Laryngoscope 2003 113: 772-774.

previous story