Duplicate and salami publication: a prevalence study of journal policies

Please download to get full document.

View again

of 8
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Similar Documents
Information Report
Category:

Social Media

Published:

Views: 2 | Pages: 8

Extension: PDF | Download: 0

Share
Description
Background: Duplicate and salami publication are unethical, but are common practices with substantial consequences for science and society at large. Scientific journals are the 'gatekeepers' of the publication process. We investigated journal
Tags
Transcript
  Original article Duplicate and salami publication: a prevalencestudy of journal policies Ding Ding , 1,2 * Binh Nguyen, 1,2 Klaus Gebel , 1,3,4 Adrian Bauman 1,2 and Lisa Bero  2,5 1 Prevention Research Collaboration, Sydney School of Public Health,  2 Charles Perkins Centre,University of Sydney, Camperdown, NSW, Australia,  3 Australian Centre for Public and PopulationHealth Research, Faculty of Health, University of Technology Sydney, Ultimo, NSW, Australia,  4 Centrefor Chronic Disease Prevention, College of Public Health, Medical and Veterinary Sciences, JamesCook University, Smithfield, QLD, Australia and  5 School of Pharmacy, Faculty of Medicine and Health,University of Sydney, Camperdown, NSW, Australia *Corresponding author. 6N55, Charles Perkins Centre (D17), University of Sydney, Camperdown, NSW 2006, Australia.E-mail: melody.ding@sydney.edu.au Editorial decision 29 July 2019; Accepted 1 September 2019 Abstract Background:  Duplicate and salami publication are unethical, but are common practiceswith substantial consequences for science and society at large. Scientific journals are the‘gatekeepers’ of the publication process. We investigated journal policies on duplicateand salami publication. Methods:  In 2018, we performed a content analysis of policies of journals in the disci-plines of ‘epidemiology and public health’ and ‘general and internal medicine’. Journalpolicies were searched, extracted, coded and cross-checked. The associations of disci-plinary categories and journal impact factors with journal policies were examined usingPoisson regression models with a robust error variance. Results:  A total of 209 journals, including 122 in epidemiology and public health and 87in general and internal medicine, were sampled and their policies investigated. Overall,18% of journals did not have any policies on either practice, 33% only referred to a ge-neric guideline or checklist without explicit mention about either practice, 36% includedpolicies on duplicate publication and only 13% included policies on both duplicate andsalami publication. Having explicit journal policies did not differ by journal disciplinarycategories (epidemiology and public health vs general and internal medicine) or impactfactors. Further analysis of journals with explicit policies found that although duplicatepublication is universally discouraged, policies on salami publication are inconsistentand lack specific definitions of inappropriate divisions of papers. Conclusions:  Gaps exist in journal policies on duplicate and salami publication, charac-terized by an overall lack of explicit policies, inconsistency and confusion in definitions of bad practices, and lack of clearly defined consequences for non-compliance. Scientific V C  The Author(s) 2019; all rights reserved. Published by Oxford University Press on behalf of the International Epidemiological Association  1 IEA International Epidemiological Association International Journal of Epidemiology  , 2019, 1–8doi: 10.1093/ije/dyz187Original article D ownl   o a d  e d f  r  om h  t   t   p s :  /   /   a c  a d  emi   c . o u p. c  om /  i   j   e /   a d v  an c  e- ar  t  i   c l   e- a b  s  t  r  a c  t   /   d  oi   /  1  0 .1  0  9  3  /  i   j   e /   d  y z 1  8 7  /   5  5 7  0  8 7 1  b  y  g u e s  t   on1 7  S  e p t   em b  er 2  0 1  9   publication and the academic reward systems must evolve to credit good researchpractice. Key words:  Publication ethics, journal policies, duplicate publication, dual publication, redundant publication, sa-lami publication, salami slicing, smallest publishable unit, least publishable unit Introduction Duplicate publication (sometimes referred to as redundantor dual publication) implies reporting identical or very sim-ilar data in multiple papers. 1 Salami publication (alsoknown as salami slicing and salami publishing) refers to in-appropriate fragmentation of data into the smallest pub-lishable units. 2,3 Both practices are detrimental to scienceand society at large. 2,4–7 Potential consequences include vi-olation of international copyright laws, 5 distortion of sci-entific evidence (e.g. double counting or overweightingcertain data, 2,8 data dredging 4 ) and misleading peer scien-tists, health professionals, decision makers, guidelines andprofessional recommendations. 6 Such practices also per-petuate the flawed academic reward system by emphasiz-ing quantity over quality of publication, 9 introduce unfaircompetition 10 and burden journal editors, reviewers, read-ers of scientific literature and research funding bodies. 11,12 Duplicate and salami publication are not uncommonpractices. For example, in 2001 Schein and Paladugu ex-amined papers published in 1998 in three leading surgicaljournals and found that 14% of these papers involvedsome form of redundancy. 13 In 2004 von Elm  et al. searched a comprehensive list of systematic reviews pub-lished between 1989 and 2002 in anaesthesia and analge-sia, and found a prevalence of 8.3% for duplicate andsalami publication. 14 In 2014, Cheung  et al.  examinedthree prominent journals in otolaryngology and found that7.5% of the indexed articles had redundancy, 15 whichshowed no improvement compared with a 2003 investiga-tion. 16 A recent large-scale analysis of more than 50 000studies in the health sciences found that the average num-ber of published journal articles per study had increasedfrom the mid-1980s to the late 1990s and stabilized there-after, and the trends differed by subdisciplines. 17 Howeverin this investigation, the authors did not distinguish salamipublication from legitimate division of complex studydata. Besides these few case studies, 18 such practices haverarely been systematically studied in epidemiology, so it isdifficult to determine their prevalence.Despite the lack of empirical data, salami slicing in epi-demiological research has been discussed by editors andresearchers, 19 including the recent discussion on salamislicing in obesity and nutritional epidemiology re-search. 20,21 Common practices of salami slicing in epide-miological research may include: (i) excessive isolation of inter-related individual exposures without consideringcomplementary and contextual risk factors, such as nutri-tional epidemiological research that focuses on single nu-trient/food items without taking into account the overallpattern of one’s diet 21,22 ; (ii) dividing studies based on out-comes when research questions are closely related 23 —thisis the particular practice deemed inappropriate by theCommittee on Publication Ethics (COPE); 24 and (iii)  post hoc  subgroup analysis, 25 such as examining the associationbetween the same exposure and outcome within the samecohort stratified by different subgroups. 26–31 Key Messages • Duplicate and salami publication are common practices that are detrimental to the scientific community and society.However, journal policies are not consistently in place to prevent such practices. • Compared with duplicate publication, journal policies on salami publication are particularly lacking: only one in eightjournals has a stated policy, and the existing policies tend to be inconsistent, unclear and without defined consequen-ces for non-compliance. • The prevalence of having explicit policies on duplicate and salami publication did not differ between ‘epidemiologyand public health’ and ‘general and internal medicine’ journals and was not predicted by journal impact factors. • We suggest that journals keep pace with good research practice and reduce duplicate and salami publication byadopting and enforcing explicit publication policies with clear definitions that distinguish between duplicate and sa-lami publication. 2  International Journal of Epidemiology  , 2019, Vol. 0, No. 0 D ownl   o a d  e d f  r  om h  t   t   p s :  /   /   a c  a d  emi   c . o u p. c  om /  i   j   e /   a d v  an c  e- ar  t  i   c l   e- a b  s  t  r  a c  t   /   d  oi   /  1  0 .1  0  9  3  /  i   j   e /   d  y z 1  8 7  /   5  5 7  0  8 7 1  b  y  g u e s  t   on1 7  S  e p t   em b  er 2  0 1  9   Duplicate and salami publication are not new phenom-ena. For around 40years, editors and authors have beenopenly discussing these problems and declaring their disap-proval of such practices. 4,11,32–34 Specifically, duplicatepublication is uniformly considered unethical and unac-ceptable. The International Committee of Medical JournalEditors (ICMJE) recommends that authors not submit du-plicate publications, defined as ‘a paper that overlaps sub-stantially with one already published, without clear, visiblereference to the previous publication’. 5 Salami publicationon the other hand has been less well investigated, possiblydue to its elusive nature as the thin line between appropri-ate and inappropriate practices and is not clearly andconsistently defined. 4,11 Editors’ perceptions of salamipublication range from considering it to be the same asduplicate or redundant publication, 1 to being a uniqueproblem 3 (Table 1). The COPE explicitly recommends adistinction be made between salami and redundantpublication. 24 Reducing and preventing duplicate and salami publica-tion requires interventions involving different elementswithin the academic system. Of those, scientific journalsare the ‘gatekeepers’ of the publication process and play anessential role in publication ethics. 35  Journal editors haveworked together to improve scientific publication byadopting policies regarding authorship criteria (ICMJE), 5 Table 1.  Inconsistent definitions of ‘salami publication’ in published literature Author, year Definition SynonymsBailey, 2002 ‘Artificial, excessive fragmentation of data’ One type of ‘duplicate publication’Beaufils, 2013 ‘Splitting data from the same research into small units, each of which is submitted—and in many cases published—separately’Explicitly stated that ‘dual publication’ is a completelydifferent phenomenon.Henly, 2014 ‘Preparation and publication of multiple, piecemeal, orfragmental reports of findings from a single study ... ’ and‘ ... can be characterised as the attempt to “write too manypapers from one study”’Explicitly stated that it is different from ‘duplicatepublication’Kassirer, 1995 ‘Results from a single study are inappropriately divided intotwo or more papers’‘Redundant publication’, ‘least publishable units’.Norman, 2008 ‘Submitting different manuscripts drawn from data collectedfrom a single research study or a single data collectionperiod. In some cases studies may draw on data that hasbeen reported and analysed previously and largely repeatanalysis with the addition of some new data or analysis notpreviously reported’Considerable similarity between salami and duplicatepublishing, but no clear demarcation.Radford, 1999 ‘Multiple publications presenting results derived from a singleexperimental protocol, often from a thesis’N/ARoberts, 2009 ‘Inappropriate division of study outcomes into several articles,most often consequent to the desire to plump academic vitae.Each article, an unnecessary compartmentalisation oftenreferred to as the minimum publishable unit or leastpublishable unit, provides a minor incremental increase inunderstanding and could reasonably have been publishedwithin one larger article’‘Minimum publishable unit’ or ‘least publishable unit’Schein, 2001 ‘Suspected study represents a part of, continuation of, orpartial repetition of the index article’A form of ‘redundant publication’, but different from‘dual publication’Smart, 2017 ‘To split one set of results, or data set, into many articles’ ‘Redundant publication (or redundancy)’.Spielmans, 2010 ‘Carving a study’s data into multiple pieces and publishingthese similar pieces individually’‘Least publishable units’.Wager, 2015 ‘The practice of dividing data from a single study into severalpublications, especially when this is inappropriate and doneto boost the authors’ publications rather than to present theresearch in the clearest possible way’N/AWaldron, 1992 ‘The results of a single study are parcelled out to differentjournals rather than being published in a singlecomprehensive form’A form of ‘duplicate publication’, in contrast to theother form, where multiple journals may publishidentical papers. N/A, not available. International Journal of Epidemiology  , 2019, Vol. 0, No. 0  3 D ownl   o a d  e d f  r  om h  t   t   p s :  /   /   a c  a d  emi   c . o u p. c  om /  i   j   e /   a d v  an c  e- ar  t  i   c l   e- a b  s  t  r  a c  t   /   d  oi   /  1  0 .1  0  9  3  /  i   j   e /   d  y z 1  8 7  /   5  5 7  0  8 7 1  b  y  g u e s  t   on1 7  S  e p t   em b  er 2  0 1  9   conflict of interest (COI) disclosure (ICMJE) 5 and report-ing criteria (the EQUATOR network). 36 The aim of our study is to describe journal policies onduplicate publication and salami slicing by examining asample of journals in the fields of epidemiology, publichealth, and general and internal medicine. Methods In May-September 2018, we performed a content analysisof policies of journals that publish in any language in thedisciplines of epidemiology and public health (‘Public,Environmental & Occupational Health’ in Web of Scienceand ‘Epidemiology’ in Scopus) and general and internalmedicine (‘Medicine, General and Internal’ in Web of Science, ‘Internal Medicine’ in Scopus). These disciplinaryclassifications were selected to reflect the broad range of journals that publish epidemiological studies. All identifiedjournals were ranked by the latest impact factor (2016–17)into quartiles, and all quartile 1 journals and a 25% ran-dom sample of quartiles 2–4 journals were selected as thestudy sample. We oversampled quartile 1 journals becausewe expected prominent journals to have more comprehen-sive policies or to adopt policies developed by consensus of editors, such as ICJME. 5 One researcher (B.N.) searched the journal websites(main pages, all tabs and links to external websites) for anymention of duplicate and salami publication and recordedspecific statements in a Microsoft Excel spreadsheet. Weperformed a conceptual content analysis to identify con-cepts related to salami publication; we did not code a pre-defined list of specific words. Instead, we defined anystated policies that referred to publishing identical or verysimilar studies, irrespective of the specific wording (e.g.dual/duplicate/redundant/overlapping) as policies regard-ing duplicate publication. Policies that were explicitlyabout fragmentation or division of data, regardless of ter-minology used (e.g. salami slicing/data fragmentation/min-imal publishable units) were considered salami slicingpolicies. A second researcher (K.G.) performed qualitycontrol by checking all data entry following the same pro-tocol, and reached 100% agreement. Journal policy state-ments were coded as the following. ‘No policy’ refers to anabsence of any policy; ‘guideline/checklist only’ signifiesmentioning a generic publication guideline or checklist(e.g. COPE guidelines) on the journal website without anyexplicit mention of duplicate or salami publication; ‘ex-plicit policies on duplicate publication’ denotes mentionsof specific policies about duplicate publication, but notabout salami publication; and ‘explicit policies on dupli-cate and salami publication’ refers to mentions of policiesabout both practices. We did not create a category forpolicies on salami publication only, because all journalswith explicit policies on salami publication also had poli-cies on duplicate publication. We tabulated these policiesand their locations on the journal websites and providedexamples of policies from different journals.We examined whether the prevalence of explicit journalpolicies differed by disciplinary categories (epidemiologyand public health vs general and internal medicine) andjournal impact factors (as a continuous variable), usingPoisson regression models with a robust error variancein two separate models with ‘explicit policies on duplicatepublication’ and ‘explicit policies on salami publication’as the outcomes. Data analysis was performed in IBMSPSS (v24). Results We selected 122 journals in epidemiology, public, environ-mental and occupational health and 87 journals in generaland internal medicine. In both the epidemiology and publichealth, and the medicine categories (Table 2), around afifth of the journals did not state any policies, about a thirdonly referred to a guideline or checklist (e.g. COPE guide-lines), over a third mentioned an explicit policy on dupli-cate publication and only around 13% reported policies onboth duplicate and salami publication.Adjusted for disciplinary category, the prevalence of having explicit policies did not differ by journal impactfactors. The prevalence ratio (PR) and 95% confidence in-terval (CI) for the association between journal impact fac-tor and having explicit policies on duplicate publication(interpreted as the change in the prevalence of having ex-plicit policies per unit increase in journal impact factor)was 0.99 (0.97, 1.01) and that for impact factor and hav-ing explicit policies on salami publication was 0.90 (0.79,1.02). Neither did we find differences in the prevalence of having explicit journal policies by journal disciplinary cate-gory when adjusted for impact factor [PR (95% CI) for du-plicate publication: 0.91 (0.69, 1.21), for salamipublication: 1.10 (0.54, 2.27), both with general and inter-nal medicine journals as the reference].Explicit policies on duplicate publication were nearlyidentical across journals, with wordings such as ‘the manu-script should not have been published previously (partly orin full) ... ’. However, the policies on salami publicationvaried across publishers and journals (Table 3). Discussion We investigated journals in epidemiology, public health,and general and internal medicine and found that the ma-jority of sampled journals did not provide any explicit 4  International Journal of Epidemiology  , 2019, Vol. 0, No. 0 D ownl   o a d  e d f  r  om h  t   t   p s :  /   /   a c  a d  emi   c . o u p. c  om /  i   j   e /   a d v  an c  e- ar  t  i   c l   e- a b  s  t  r  a c  t   /   d  oi   /  1  0 .1  0  9  3  /  i   j   e /   d  y z 1  8 7  /   5  5 7  0  8 7 1  b  y  g u e s  t   on1 7  S  e p t   em b  er 2  0 1  9   policies on duplicate or salami publication. Policies onsalami publication were particularly rare. We also foundthat more than a third of the journals only referred to aguideline or checklist, which requires clicking throughseveral links to find specific statements within a lengthydocument located on an external website. Such ‘hidden’information on journal policies on duplicate and salamipublication may be perceived as too difficult to access andtherefore not informative for submitting authors or peerreviewers. Table 2.  Journal policies on duplicate and salami publication by impact factor quartile  Journal policies Journal impact factor quartile OverallEpidemiology/public health Q1 ( n ¼ 71) Q2 ( n ¼ 16) Q3 ( n ¼ 17) Q4 ( n ¼ 18)  n ¼ 122No policy 10 (14.1%) 2 (12.5%) 5 (29.4%) 3 (16.7%) 20 (16.4%)Guideline/checklist only 23 (32.4%) 6 (37.5%) 3 (17.6%) 12 (66.7%) 44 (36.1%)Explicit policies on duplicate publication 30 (42.3%) 6 (37.5%) 4 (23.5%) 1 (5.6%) 41 (33.6%)Explicit policies on duplicate and salami publication 8 (11.3%) 2 (12.5%) 5 (29.4) 2 (11.1) 17 (13.9%)General/internal medicine Q1 ( n ¼ 49) Q2 ( n ¼ 13) Q3 ( n ¼ 13) Q4 ( n ¼ 12)  n ¼ 87No policy 3 (6.1%) 6 (46.2%) 6 (46.2%) 3 (25.0%) 18 (20.7%)Guideline/checklist only 17 (34.7%) 1 (7.7%) 3 (23.1%) 4 (33.3%) 25 (28.7%)Explicit policies on duplicate publication 23 (46.9) 5 (38.5%) 2 (15.4%) 4 (33.3%) 34 (39.1%)Explicit policies on duplicate and salami publication 6 (12.2%) 1 (7.7%) 2 (15.4%) 1 (8.3%) 10 (11.5%)Total Q1 ( n ¼ 120) Q2 ( n ¼ 29) Q3 ( n ¼ 30) Q4 ( n ¼ 30)  n ¼ 209No policy 13 (10.8%) 8 (27.6%) 11 (36.7%) 6 (20.0%) 38 (18.2%)Guideline/checklist only 40 (33.3%) 7 (24.1%) 6 (20.0%) 16 (53.3%) 69 (33.0%)Explicit policies on duplicate publication 53 (44.2%) 11 (37.9%) 6 (20.0%) 5 (16.7%) 75 (35.9%)Explicit policies on duplicate and salami publication 14 (11.7%) 3 (10.3%) 7 (23.3%) 3 (10.0) 27 (12.9%) Table 3.  Existing stated journal policies on ‘salami publication’ ‘Salami publication’ policy statement Location  n  Publisher/journal a ‘A single study is not split up into several parts to increase the quantityof submissions and submitted to various journals or to one journalover time (e.g. “salami-publishing”).’Instructions for authors/ethicalresponsibilities of authors16 Springer/Springer Nature/ Springer Link‘It is important to avoid  ...  redundant publications (or ‘salami’publications): this means publishing many very similar manuscriptsbased on the same experiment. It can make readers less likely to payattention to your manuscripts’Writing resources/publicationethics5 Biomed Central‘Avoid fragmenting research to maximise the number of articles forpublication’For authors/ethical guidelines/ author code of ethics2 Bioscientifica‘Data should not be sliced to the size of the smallest publishable unit,but must be a complete documentation of a study’ Journal policies 1  Diabetologia ‘Do not submit several small manuscripts; a single comprehensivepaper is preferable. If the authors believe that subdivision isappropriate, or if multiple articles may result from the same study,contact the Editor-in-Chief’Information for authors/ instructions and guidelines/ general editorial, legal andethical issues1 Simulation in Health Care‘Splitting the data concerning one study in more than one publicationcould be acceptable if authors justify the choice  ...  with goodreasons both in the cover letter and in the manuscript.  ...  authorsshould state what new scientific contribution is contained in theirmanuscript compared to any previously published article  ...  derivedfrom the same study. Relevant previously published articles shouldbe included in the cover letter of the currently submitted article’Instructions to authors 1  Minerva Endocrinologica ‘Plagiarism and other publication misconduct ( ...  salami slice ... ) areliable to strict action’Instructions to authors/ethicalconsiderations1  Journal of the College of Physicians and Surgeons a Publisher name is listed where multiple journals were identified from the same publisher with the same ‘salami publication’ policies. Journal name is listedwhere only one journal was identified from one publisher. International Journal of Epidemiology  , 2019, Vol. 0, No. 0  5 D ownl   o a d  e d f  r  om h  t   t   p s :  /   /   a c  a d  emi   c . o u p. c  om /  i   j   e /   a d v  an c  e- ar  t  i   c l   e- a b  s  t  r  a c  t   /   d  oi   /  1  0 .1  0  9  3  /  i   j   e /   d  y z 1  8 7  /   5  5 7  0  8 7 1  b  y  g u e s  t   on1 7  S  e p t   em b  er 2  0 1  9 
Recommended
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks
SAVE OUR EARTH

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!

x