Contents
On this page
- Distinguishing five related forms of overlapping publication
- Duplicate publication as a threat to the evidence base
- Prevalence estimates and sources of variation
- Measured characteristics of covert duplicates
- Detection by human reviewers
- Detection by software and its limitations
- Duplicate records versus duplicate publications
- Declaring the relationship in a secondary publication
- Limitations of detection
- Key points
- Application to your study
- Frequently asked questions
Academy · Research Publication · Identifying multiple publication
Detection of Duplicate and Redundant Publication: Methods, Prevalence and Limitations
Duplicate publications are identified in three ways: by reviewers familiar with the field, by systematic reviewers screening records, and by text-similarity software. None of these methods is reliable alone; one large text-similarity study reported its own sensitivity as 50.3%. The duplicates that evade all three are those that carry no cross-reference to the original publication.
The practical concern of authors is often not the definition of duplicate publication but whether it will be detected. This question has a specific answer, because the detection methods have been measured. This article reports the findings of prevalence studies, the characteristics of the covert duplicates they identified, how human and software detection operate and where each fails, and what a second paper from a single dataset must declare to be legitimate rather than redundant.
Learning objectives
- The working difference between duplicate, redundant, covert and salami publication, and legitimate secondary publication
- What the prevalence studies found — from 40% of one set of systematic reviews encountering duplicates to under 1% in one specialty audit
- The measured signature of a covert duplicate: authorship overlap, sponsorship, and the delay between the two papers
- How text-similarity detection works, which sections it flags, and the published sensitivity figure that should inform interpretation of its results
- Why duplicate records across databases are a different problem from duplicate publications, and why systematic reviewers conflate them
- What a secondary publication must state, and where, so that it functions as a cross-reference rather than a duplicate
Common misconceptions
- “Publishing the same data twice is always misconduct.” This is incorrect. Secondary publication for a different readership or in another language is explicitly permitted when the relationship is declared. A duplicate is covert because the cross-reference is missing, not because the data overlap.
- “Changing the author list and rewriting the text make duplication undetectable.” These are recognised signatures of duplication. In the analysis by von Elm and colleagues, 64% of duplicates had partly or completely different authorship; differing authorship is therefore a feature reviewers look for rather than one that conceals duplication.
- “Similarity software detects duplicates.” Errami and colleagues reported a sensitivity of 50.3% for their text-similarity method. Software tends to detect the less sophisticated cases; the more careful ones are typically identified by a reviewer who recognises the dataset.
- “Duplicate publication merely inflates an author’s publication record.” It distorts meta-analysis. In the case described by Qi and colleagues, inclusion of duplicated data overestimated the drug’s efficacy by 23%.
- “The duplicates flagged by search software are duplicate publications.” These are usually the same article indexed in several databases, which is a deduplication problem rather than an integrity problem. Conflating the two consumes considerable screening time.
Distinguishing five related forms of overlapping publication
The terminology is inconsistent in the literature, which is consequential because the remedies differ. Alfonso and colleagues, in Revista Española de Cardiología, and DeAngelis in JAMA both addressed this inconsistency directly; Angell and Relman had set out the editorial position decades earlier in the New England Journal of Medicine, and Kassirer and Angell returned to it there in 1995 21,23,27,28.
Duplicate publication is the same study reported again, in whole or substantial part. Redundant publication is the broader category: substantial overlap with something already published, without necessarily being identical. Covert duplication is either of those with no cross-reference to the earlier paper; it is the only one of the three that is consistently problematic, and the one with which the detection literature is concerned 22. Salami publication splits one study into several thin papers, each reporting a slice. Secondary publication is a republication for a different readership or in another language, which the ICMJE explicitly permits subject to conditions, including a title-page footnote giving the primary reference 1.
Yank and Barnes, in the Journal of Medical Ethics, surveyed editors and authors and found consensus on some cases and genuine disagreement on others, which indicates that no single definition should be regarded as settled 24. In practice, the acceptable is distinguished from the unacceptable not by the extent of textual overlap but by whether a reader can identify the relationship. The ICMJE’s own conditions for secondary publication are set out on the ICMJE Recommendations page.
Duplicate publication as a threat to the evidence base
Duplicate publication is usually discussed as a question of credit. Its most important consequence, however, is arithmetical: a meta-analysis that counts the same patients twice gives them twice the weight 26.
Qi and colleagues, in World Journal of Meta-Analysis, report the case study that made this concrete, from Tramèr and colleagues’ work: 17% of the randomised controlled trials and 28% of the patient data in that body of evidence were duplicated, and including the duplicated data overestimated ondansetron’s antiemetic efficacy by 23% 11. Double counting therefore produced approximately a quarter of the drug’s apparent effect.
For this reason, systematic reviewers are the group most likely to detect duplicates, and the detection problem belongs in the methods of a systematic review rather than only in a journal’s ethics policy 22. It also explains why a covert duplicate continues to cause harm long after questions of credit have ceased to matter.
Prevalence estimates and sources of variation
Prevalence depends almost entirely on what is counted and where, and the published range is wide enough that any single figure quoted without its source is uninformative.
The most rigorous study is the analysis by von Elm and colleagues in JAMA, which took 141 systematic reviews in anaesthesia and analgesia and examined what their authors had found 22. Authors of 56 of those reviews (40%) reported identifying duplicate publications. Those reviews covered 1,131 main articles, with 103 duplicates arising from 78 main articles.
Specialty audits give very different numbers. Qi and colleagues summarise several: about 14% of 660 surgical original articles showed some potential redundancy, following the work of Schein and Paladugu in Surgery; under 1% of plastic surgery articles did, from Durani’s audit; and 10% (184 of 1,914) of Chinese Budd-Chiari syndrome articles were covert duplicates, from Qi and colleagues’ own study in the American Journal of Medicine 11. Arrivé and colleagues conducted a comparable audit for Radiology 18.
The fourteen-fold difference between surgery and plastic surgery does not measure the honesty of the two specialties; it reflects two definitions and two detection methods. Each figure should be interpreted as a statement about one literature examined by one method.
Prevalence estimates, and what each one actually counted
| Study / literature | What was examined | Finding | Why it is not comparable |
|---|---|---|---|
| von Elm et al. JAMA | 141 systematic reviews in anaesthesia and analgesia | Authors of 56 reviews (40%) reported identifying duplicates; 1,131 main articles yielded 103 duplicates from 78 main articles | Counts reviews that encountered duplicates, not articles that are duplicates |
| Schein and Paladugu as summarised by Qi et al. | 660 surgical original articles | About 14% showed some potential redundancy | “Potential redundancy” is a broader category than covert duplication |
| Durani as summarised by Qi et al. | Plastic surgery literature | Under 1% of articles | A different definition and a different detection method |
| Qi et al. American Journal of Medicine | 1,914 Chinese Budd-Chiari syndrome articles | 10% (184/1,914) were covert duplicates | A single disease literature in a single language |
| Errami et al. Bioinformatics | 62,213 Medline citations, by text similarity | 1.35% of same-author pairs duplicated, extrapolating to ~117,400 (±40,000) citations | Software detection at 50.3% sensitivity; abstracts only |
| Qi et al. World Journal of Meta-Analysis | Their own reviews searching 3+ databases | 7% to 25% duplicate records | This is bibliographic deduplication — not duplicate publication at all |
Figures as reported in each source. They are deliberately not pooled: the definitions, detection methods and units of counting differ between every row, and the last row counts something else entirely.
Measured characteristics of covert duplicates
The most useful element of the work by von Elm and colleagues is not the headline prevalence but the description of the covert subset, namely the duplicates carrying no cross-reference to the main article. These made up 5.3%, or 65 of 1,234 articles 22. Of the duplicates, 33% were industry sponsored, 64% had partly or completely different authorship, and the median delay between the main article and its duplicate was one year.
This profile functions as a detection heuristic, and editors and systematic reviewers use it as such. The characteristic pattern is a paper appearing roughly a year after a closely related report, with an author list that overlaps but is not identical, reporting what appears to be the same population. The differing authorship is particularly noteworthy: it is commonly assumed that changing the author list conceals the relationship, whereas partial authorship overlap is one of the strongest signals that two papers are connected, and it is precisely what a reviewer notices 22.
The one-year median delay also has a practical implication: detection usually has to operate across a year of literature rather than against a single submission queue, which is why a journal’s own records are a weak detection instrument and a systematic reviewer’s comprehensive search is a strong one.
The measured profile of a covert duplicate, from 141 systematic reviews in anaesthesia and analgesia. Source: von Elm E, Poglia G, Walder B, Tramèr MR. Different patterns of duplicate publication: an analysis of articles used in systematic reviews. JAMA.
Detection by human reviewers
Three human detection routes account for most identifications; they are listed here in approximate order of effectiveness.
The expert reviewer. A specialist in the field recognises the cohort: the unusual sample size, the single-centre setting, the distinctive inclusion criteria, the recruitment window. This mechanism identifies the careful duplicates that software misses, and it is the reason journals send submissions to subject specialists rather than to generalists. It is, however, unreliable in a different respect: it depends entirely on which reviewer was available.
The systematic reviewer. When screening several thousand records against inclusion criteria, with full texts available, a systematic reviewer undertakes comparative reading that no other party performs. The data of von Elm and colleagues exist because this is where duplicates emerge 22. PRISMA, from Moher and colleagues, expects the handling of multiple reports of the same study to be described in the methods, which makes this detection explicit rather than incidental 17.
The editor’s structured checks. Reference lists, author overlap with recent submissions to the same journal, declared prior presentation and registration numbers. The registration number is the most powerful and the most under-used of these: two papers reporting the same registered trial report the same trial, regardless of their titles.
Detection by software and its limitations
Text-similarity detection compares a submission against a corpus and reports overlapping passages. Zhang described the publisher-side implementation in Learned Publishing, and Foltyněk, Meuschke and Gipp’s systematic review in ACM Computing Surveys is the current map of the methods 6,15. Two large studies indicate what it detects.
Errami and colleagues, in Bioinformatics, scanned 62,213 Medline citations 19. About 1.35% of citations sharing authors were similar enough to count as duplicates, which extrapolates to roughly 117,400 Medline citations, plus or minus 40,000 22. About 0.04% of citation pairs with no shared authors were highly similar, extrapolating to roughly 3,500 potential plagiarism cases, plus or minus 1,600 19. The most important figure is the last one reported: the method’s sensitivity was 50.3%, meaning that approximately half of the duplicates against which it was tested went undetected 19.
Sun and colleagues, in PLoS ONE, went further into the structure, analysing 72,011 full-text PubMed Central articles 16. Methods sections were the most frequently repeated section, which is unsurprising and largely benign, since a procedure does not change between papers. However, in 193 manually verified duplicate pairs, the results sections were the most similar, at 0.94, with methods at 0.89 and introductions at 0.82. This inversion is the operationally useful finding: high methods similarity is weak evidence, whereas high results similarity is strong evidence of a duplicate 16. Citron and Ginsparg, in PNAS, found comparable structure in text reuse across a large physics corpus 9.
What software misses: translated duplicates, paraphrased duplicates, papers whose data are identical but whose text was written from scratch, and anything outside the indexed corpus 6. It flags only what it can read. Running a similarity check before submission indicates what the editor’s software will report, which is useful but does not establish that a manuscript is free of duplication.
What text-similarity detection finds. Sensitivity and citation figures: Errami M, Hicks JM, Fisher W, Trusty D, Wren JD, Long TC, Garner HR. Déjà vu — a study of duplicate citations in Medline. Bioinformatics. Section similarities from 72,011 full-text articles: Sun Z, Errami M, Long T, Renard C, Choradia N, Garner H. Systematic characterizations of text similarity in full text biomedical publications. PLoS ONE.
Duplicate records versus duplicate publications
This confusion consumes more of systematic reviewers’ time than any other, and it is unrelated to misconduct.
When the results of searches of PubMed, EMBASE and the Cochrane Library are merged, the same article appears several times with different field formatting. Qi and colleagues, in World Journal of Meta-Analysis, report that across their own systematic reviews searching three or more databases, the prevalence of duplicate records ranged from 7% to 25% 11. Their companion paper in PLoS ONE addressed how to find and remove them 12.
That 7% to 25% is a bibliographic deduplication statistic. It is not a prevalence of duplicate publication and should never be cited as one; this error has appeared in print. The practical distinction is as follows: duplicate records are one study appearing in several databases, and the remedy is software deduplication followed by a manual check. Duplicate publications are one study appearing in several journals, and the remedy is to read the full texts and decide which report to include. PRISMA requires both processes to be described, and conflating them in a flow diagram makes the review’s record count unreproducible.
Declaring the relationship in a secondary publication
For a legitimate secondary or companion paper (a pre-specified secondary analysis, a long-term extension, a translation for a different readership), legitimacy depends entirely on declaration, which requires less effort than is often assumed. The declaration is made in four places: the title, a title-page footnote, the abstract and the cover letter. A declaration in only one of the four produces a paper that reads as a duplicate to any reader who finds the other report.
A secondary analysis that reads as a duplicate, and the same paper cross-referenced
Reads as a duplicate
Title: Efficacy and safety of tenovir in moderate-to-severe disease: results from a randomised controlled trial. Abstract: In this randomised, double-blind trial, 412 patients received tenovir or placebo. The primary endpoint was met, with a mean difference of 2.4 points (95% CI 1.1 to 3.7). Tenovir was well tolerated. Methods: Patients were enrolled at 18 centres between March 2022 and January 2023.
The title is indistinguishable from the primary report’s title. A reader retrieving both would interpret them as competing accounts of one trial.
The primary paper is not cited anywhere a reader will look first: neither in the title, nor in the footnote, nor in the abstract.
The primary endpoint is restated as this paper’s finding, which is the specific move that makes a secondary analysis a duplicate.
No registration number, so the strongest link between the two papers is absent.
The same enrolment window and sample size appear with no explanation, which is precisely the pattern that a reviewer in the field recognises.
Reads as a cross-reference
Title: Health-related quality of life in moderate-to-severe disease: a pre-specified secondary analysis of the ASCENT randomised controlled trial. Title-page footnote: The primary results of the ASCENT trial have been published (Oyelaran K, et al. Journal of Clinical Investigation. 2024;134:e180022). This article reports the pre-specified secondary quality-of-life endpoints from the same trial population; no primary efficacy or safety outcomes are re-reported. Trial registration NCT00000000. Abstract, opening sentence: In the ASCENT trial, previously reported, tenovir met its primary endpoint in 412 patients; here we report the pre-specified secondary quality-of-life endpoints from the same population. Cover letter: This manuscript reports pre-specified secondary endpoints from a trial whose primary results we published in 2024; the primary paper is cited on the title page and in the abstract, and no participant data are re-reported as a primary finding. A copy of the primary paper is attached.
The title names the analysis as secondary and names the trial, so the relationship is visible in any search result.
The title-page footnote gives the full primary reference, which is what the ICMJE conditions for secondary publication require.
The abstract’s first sentence states the primary result as already reported, so the new contribution is unambiguous.
The registration number ties both papers to one trial for any editor or systematic reviewer who checks.
The cover letter declares the relationship to the editor and supplies the primary paper, which removes the discovery risk entirely.
The data are identical in both versions and so is the analysis. What changed is that the four places a reader looks (title, footnote, abstract opening, cover letter) now state the relationship. Given that 64% of the duplicates in the data of von Elm and colleagues had partly different authorship and a median one-year gap, the pattern in the first version is exactly the pattern reviewers are trained to notice; the second version renders that pattern unremarkable.
Limitations of detection
Be clear-eyed about three limits.
Detection is retrospective. By the time a duplicate is identified, both papers are in the literature and both are being cited; a retraction removes one record but not the citations already made to it. Fang, Steen and Casadevall showed in PNAS that misconduct accounts for the majority of retracted publications, with duplicate publication among the recurring grounds; retraction, however, is a late and partial remedy 13.
Detection has no agreed threshold. There is no percentage of overlap at which a paper becomes redundant, and the contention Yank and Barnes documented among editors and authors has not been resolved since. Text recycling work by Moskovitz, in Science and Engineering Ethics, and the text-analytic study by Anson and Moskovitz in Accountability in Research, both argue that reuse in STEM writing is normal in some sections and that blanket thresholds misdescribe practice 5,7. Any single acceptable similarity percentage reflects a local policy, not a standard.
Detection is also weakest where the stakes are highest. A covert duplicate written from scratch, in a different language, with a changed author list, a year later, in a journal outside the publisher’s similarity corpus, will not be found by software. It will be found, if at all, by a systematic reviewer reading two full texts side by side. That is the only safeguard, and it is why the proper declaration of companion papers matters more than any detection tool.
Key points
Key points
Covert duplication, not overlap, is the problem: in von Elm and colleagues’ JAMA analysis, 5.3% of articles were duplicates carrying no cross-reference, 64% had partly different authorship and the median interval was one year. Detection relies on expert reviewers, systematic reviewers screening full texts and text-similarity software, whose published sensitivity in one large study was 50.3%; software therefore tends to identify the less sophisticated cases and human reviewers the more careful ones. Similarity in the results section is a strong signal, whereas similarity in the methods section is weak. Duplicate-record counts reported by search software are a deduplication statistic, not a measure of the prevalence of duplicate publication.
Application to your study
Application to your study
- All outputs from the dataset should be inventoried before a second paper is drafted. These include the primary report, congress abstracts, preprints and companion analyses. Outputs that have not been listed cannot be cross-referenced.
- The relationship should be stated in four places rather than one. These are the title, a title-page footnote with the full primary reference, the opening sentence of the abstract and the cover letter; the primary paper should also be attached.
- The trial registration number should be included in every paper arising from the trial. It is the strongest link that an editor or systematic reviewer can follow, and its omission may appear deliberate.
- A primary endpoint should not be re-reported as a new finding. It should be stated as previously reported, followed by the new findings.
- A similarity check should be run and the report examined section by section. High overlap in the methods is usually benign; high overlap in the results is the meaningful signal.
- Reviewers and screeners should compare sample size, number of centres and recruitment period across candidate papers. A duplicate with a modified author list still reports the same cohort.
- In a systematic review, deduplication and the handling of duplicate publications should be described separately in the methods. The two processes and their counts should be reported distinctly, as PRISMA expects.
- Where two reports of one study both qualify, reviewers should state which was included and why, and how many patients were thereby excluded, since double counting distorts the pooled estimate.
Frequently asked questions
The estimates range so widely that any single figure requires its source. Von Elm and colleagues, in JAMA, examined 141 anaesthesia and analgesia systematic reviews and found authors of 56 (40%) reported identifying duplicates, with 1,131 main articles yielding 103 duplicates from 78 main articles; covert duplicates were 5.3% of articles. Specialty audits summarised by Qi and colleagues range from about 14% of 660 surgical articles showing potential redundancy to under 1% in plastic surgery, with 10% (184 of 1,914) covert duplicates in one Chinese disease literature. The spread reflects differing definitions and detection methods, not differing honesty.
Sometimes. Errami and colleagues, in Bioinformatics, reported their method’s sensitivity as 50.3%; roughly half the duplicates against which it was tested went undetected. Software compares text within an indexed corpus, so it misses translated duplicates, rewritten duplicates with identical data, and anything outside the corpus. It is effective at detecting careless reuse and ineffective at detecting deliberate concealment. Reading similarity reports by section is informative: Sun and colleagues, in PLoS ONE, found methods sections most frequently repeated across 72,011 full-text articles, but in 193 verified duplicate pairs the results sections were the most similar at 0.94, against 0.89 for methods.
Yes. Secondary publication in another language or for a different readership is permitted under the ICMJE Recommendations, subject to conditions including agreement between both editors, faithful reflection of the primary data and interpretations, a title indicating the secondary status, and a title-page footnote giving the full primary reference. Pre-specified secondary analyses, long-term extensions and companion papers from a single trial are also normal and legitimate. What makes a second paper unacceptable is not the overlap but the absence of a declared relationship, termed covert duplication in the detection literature.
No; it signals the relationship. In von Elm and colleagues’ JAMA analysis, 64% of duplicates had partly or completely different authorship, which means that partial author overlap between two similar papers is a recognised detection heuristic rather than a disguise. Reviewers in a field also recognise the cohort itself: the sample size, the number of centres, the recruitment window and the distinctive inclusion criteria remain with the data regardless of the byline. The median delay between main article and duplicate in that dataset was one year, which is the other element of the pattern that reviewers look for.
Because it distorts pooled estimates. Qi and colleagues, in World Journal of Meta-Analysis, describe the case study from Tramèr and colleagues in which 17% of the randomised controlled trials and 28% of the patient data in a body of evidence were duplicated, and including the duplicated data overestimated ondansetron’s antiemetic efficacy by 23%. A meta-analysis that counts the same patients twice gives them twice the weight, so the error propagates into guidelines and clinical decisions. This also explains why systematic reviewers, rather than journal editors, are the group most likely to detect duplicates.
Duplicate records are one article indexed in several databases, so merging PubMed, EMBASE and Cochrane results produces the same study several times. Qi and colleagues report that across their own reviews searching three or more databases, duplicate records ranged from 7% to 25%. That is a bibliographic deduplication figure and should never be cited as a prevalence of duplicate publication. Duplicate publications are one study reported in several journals. The first is solved by deduplication software plus a human check; the second requires reading the full texts and deciding which report to include. PRISMA expects both processes to be described separately.
There is no agreed threshold; any specific figure reflects a local journal or publisher policy rather than a standard. Yank and Barnes, in the Journal of Medical Ethics, surveyed editors and authors and found genuine disagreement about which overlaps count as redundant. Moskovitz, in Science and Engineering Ethics, and Anson and Moskovitz, in Accountability in Research, both argue that some text reuse is normal in STEM writing, particularly in methods, and that blanket thresholds misdescribe practice. In practice, overlap should be judged by section and by whether the relationship is declared, not by a percentage.
The usual outcomes are a correction, a notice of redundant publication, or a retraction of the later paper, depending on the journal and the extent. Fang, Steen and Casadevall showed in PNAS that misconduct accounts for the majority of retracted scientific publications, with duplicate publication among the recurring grounds. The remedy is partial, however: a retraction removes a record without removing the citations already made to it, and any meta-analysis that pooled the duplicated data has already been published. This asymmetry supports the proper declaration of companion papers at submission rather than reliance on detection.
References
- International Committee of Medical Journal Editors. Recommendations for the Conduct, Reporting, Editing, and Publication of Scholarly Work in Medical Journals. ICMJE. 2026. read 9 October 2026. icmje.org
- International Committee of Medical Journal Editors. The New ICMJE Recommendations (August 2013). ICMJE. 2013. read 9 October 2026. icmje.org
- Committee on Publication Ethics. Core Practices. COPE. read 9 October 2026. equator-network.org
- DeTora LM, Toroser D, Sykes A, Vanderlinden C, Plunkett FJ, Lane T, Hanekamp E, Dormer L, DiBiasi F, Bridges D, Baltzer L, Citrome L. Good Publication Practice (GPP) Guidelines for Company-Sponsored Biomedical Research: 2022 Update. Annals of Internal Medicine. 2022;175(9):1298-1304. doi:10.7326/M22-1460 read 9 October 2026. doi.org/10.7326/M22-1460
- Anson IG, Moskovitz C. Text recycling in STEM: a text-analytic study of recently published research articles. Accountability in Research. 2020. doi:10.1080/08989621.2020.1850284 read 9 October 2026. doi.org/10.1080/08989621.2020.1850284
- Foltynek T, Meuschke N, Gipp B. Academic plagiarism detection: a systematic literature review. ACM Computing Surveys. 2019;52(6):1-42. doi:10.1145/3345317 read 9 October 2026. doi.org/10.1145/3345317
- Moskovitz C. Text recycling in scientific writing. Science and Engineering Ethics. 2019;25(3):813-851. doi:10.1007/s11948-017-0008-y read 9 October 2026. doi.org/10.1007/s11948-017-0008-y
- Battisti WP, Wager E, Baltzer L, Bridges D, Cairns A, Carswell CI, Citrome L, Gurr JA, Mooney LA, Moore BJ, Pena T, Sanes-Miller CH, Veitch K, Woolley KL, Yarker YE. Good Publication Practice for Communicating Company-Sponsored Medical Research: GPP3. Annals of Internal Medicine. 2015;163(6):461-464. doi:10.7326/M15-0288 read 9 October 2026. doi.org/10.7326/M15-0288
- Citron DT, Ginsparg P. Patterns of text reuse in a scientific corpus. Proceedings of the National Academy of Sciences of the USA. 2015;112(1):25-30. doi:10.1073/pnas.1415135111 read 9 October 2026. doi.org/10.1073/pnas.1415135111
- Qi X, Deng H, Guo X, Fan D. Covert duplicate publications in Budd-Chiari syndrome articles in China. The American Journal of Medicine. 2013;126:633-639.e2. read 9 October 2026. wjgnet.com
- Qi XS, Bai M, Yang ZP, Ren WR. Duplicates in systematic reviews: a critical, but often neglected issue. World Journal of Meta-Analysis. 2013;1(3):97-101. doi:10.13105/wjma.v1.i3.97 read 9 October 2026. doi.org/10.13105/wjma.v1.i3.97
- Qi X, Yang M, Ren W, Jia J, Wang J, Han G, Fan D. Find duplicates among the PubMed, EMBASE, and Cochrane Library databases in systematic review. PLoS ONE. 2013;8:e71838. doi:10.1371/journal.pone.0071838 read 9 October 2026. doi.org/10.1371/journal.pone.0071838
- Fang FC, Steen RG, Casadevall A. Misconduct accounts for the majority of retracted scientific publications. Proceedings of the National Academy of Sciences of the USA. 2012;109(42):17028-17033. doi:10.1073/pnas.1212247109 read 9 October 2026. doi.org/10.1073/pnas.1212247109
- Das N, Panjabi M. Plagiarism: why is it such a big issue for medical writers? Perspectives in Clinical Research. 2011;2(2):67-71. doi:10.4103/2229-3485.80370 read 9 October 2026. doi.org/10.4103/2229-3485.80370
- Zhang H. CrossCheck: an effective tool for detecting plagiarism. Learned Publishing. 2010;23:9-14. read 9 October 2026. pmc.ncbi.nlm.nih.gov
- Sun Z, Errami M, Long T, Renard C, Choradia N, Garner H. Systematic characterizations of text similarity in full text biomedical publications. PLoS ONE. 2010;5(9):e12704. doi:10.1371/journal.pone.0012704 read 9 October 2026. doi.org/10.1371/journal.pone.0012704
- Moher D, Liberati A, Tetzlaff J, Altman DG; PRISMA Group. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Medicine. 2009;6:e1000097. read 9 October 2026. care-statement.org
- Arrive L, Lewin M, Dono P, Monnier-Cholley L, Hoeffel C, Tubiana JM. Redundant publication in the journal Radiology. Radiology. 2008;247:836-840. doi:10.1148/radiol.2473070969 read 9 October 2026. doi.org/10.1148/radiol.2473070969
- Errami M, Hicks JM, Fisher W, Trusty D, Wren JD, Long TC, Garner HR. Deja vu - a study of duplicate citations in Medline. Bioinformatics. 2007. read 9 October 2026. ori.hhs.gov
- Durani P. Duplicate publications: redundancy in plastic surgery literature. Journal of Plastic, Reconstructive & Aesthetic Surgery. 2006;59:975-977. read 9 October 2026. wjgnet.com
- Alfonso F, Bermejo J, Segovia J. Duplicate or redundant publication: can we afford it? Revista Espanola de Cardiologia. 2005;58(5):601-604. doi:10.1016/S1885-5857(06)60739-1 read 9 October 2026. doi.org/10.1016/S1885-5857(06)60739-1
- von Elm E, Poglia G, Walder B, Tramer MR. Different patterns of duplicate publication: an analysis of articles used in systematic reviews. JAMA. 2004;291(8):974-980. doi:10.1001/jama.291.8.974 read 9 October 2026. doi.org/10.1001/jama.291.8.974
- DeAngelis CD. Duplicate publication, multiple problems. JAMA. 2004;292:1745-1746. doi:10.1001/jama.292.14.1745 read 9 October 2026. doi.org/10.1001/jama.292.14.1745
- Yank V, Barnes D. Consensus and contention regarding redundant publications in clinical research: cross-sectional survey of editors and authors. Journal of Medical Ethics. 2003;29:109-114. read 9 October 2026. revespcardiol.org
- Schein M, Paladugu R. Redundant surgical publications: tip of the iceberg? Surgery. 2001;129:655-661. read 9 October 2026. wjgnet.com
- Tramer MR, Reynolds DJM, Moore RA, McQuay HJ. Impact of covert duplicate publication on meta-analysis: a case study. BMJ. 1997;315:635-640. read 9 October 2026. wjgnet.com
- Kassirer JP, Angell M. Redundant publication: a reminder. New England Journal of Medicine. 1995;333:449-450. doi:10.1056/NEJM199508173330709 read 9 October 2026. doi.org/10.1056/NEJM199508173330709
- Angell M, Relman AS. Redundant publication. New England Journal of Medicine. 1989;320:1212-1214. doi:10.1056/NEJM198905043201812 read 9 October 2026. doi.org/10.1056/NEJM198905043201812
Every source above is a peer-reviewed article, a publisher’s record of one, or a standard issued by the body that owns it. Each carries the date it was last read. Where a standard moves, this article is revised rather than left up with a note.
How to cite this article
Pick your referencing style and copy it. Your university or target journal may vary the punctuation — check its guide if it has one.
Pubrica Academy. (2026, October 9). Detection of Duplicate and Redundant Publication: Methods, Prevalence and Limitations. Pubrica. https://pubrica.com/academy/concepts-definitions/identification-of-multiple-publication/
Pubrica Academy. “Detection of Duplicate and Redundant Publication: Methods, Prevalence and Limitations.” Pubrica, 9 Oct. 2026, pubrica.com/academy/concepts-definitions/identification-of-multiple-publication/.
Pubrica Academy (2026) Detection of Duplicate and Redundant Publication: Methods, Prevalence and Limitations. Pubrica. Available at: https://pubrica.com/academy/concepts-definitions/identification-of-multiple-publication/ (Accessed: 15 October 2026).
Pubrica Academy. “Detection of Duplicate and Redundant Publication: Methods, Prevalence and Limitations.” Pubrica. October 9, 2026. https://pubrica.com/academy/concepts-definitions/identification-of-multiple-publication/
Pubrica Academy, ‘Detection of Duplicate and Redundant Publication: Methods, Prevalence and Limitations’ (Pubrica, 9 October 2026) <https://pubrica.com/academy/concepts-definitions/identification-of-multiple-publication/> accessed 15 October 2026.
Pubrica Academy. Detection of Duplicate and Redundant Publication: Methods, Prevalence and Limitations. [Internet]. Pubrica; 2026 Oct 9 [cited 2026 Oct 15]. Available from: https://pubrica.com/academy/concepts-definitions/identification-of-multiple-publication/
Pubrica Academy, “Detection of Duplicate and Redundant Publication: Methods, Prevalence and Limitations,” Pubrica, Oct. 9, 2026. [Online]. Available: https://pubrica.com/academy/concepts-definitions/identification-of-multiple-publication/
@misc{identification_of_multiple_publication_2026,
author = {{Pubrica Academy}},
title = {Detection of Duplicate and Redundant Publication: Methods, Prevalence and Limitations},
howpublished = {Pubrica},
year = {2026},
month = {oct},
url = {https://pubrica.com/academy/concepts-definitions/identification-of-multiple-publication/},
note = {Accessed: <span data-cite-acc="iso">2026-10-15</span>}
}This article is written to be learned from. Cite it if you are referring to what it says; if you are using the evidence behind it, cite the sources listed above instead.
Companion-paper cross-reference check
This checklist applies whenever a second paper will arise from one dataset (a secondary analysis, an extension, a translation or a subgroup analysis). Part one should be completed before drafting and part two before submission; if part one is empty, part two cannot be completed in good faith.
COMPANION-PAPER CROSS-REFERENCE CHECK
This manuscript: ______________________
Study / trial: ______________________ Registration: ______________________
PART 1 - OUTPUT INVENTORY (list everything already in the public domain)
Primary report ........... ____________________________________________
Other full papers ........ ____________________________________________
____________________________________________
Congress abstracts ....... ____________________________________________
Posters / presentations .. ____________________________________________
Preprint + DOI ........... ____________________________________________
Plain-language summary ... ____________________________________________
Registry results posting . ____________________________________________
PART 2 - DECLARATION IN FOUR PLACES
[ ] TITLE names this as a secondary / extension / subgroup analysis AND names
the parent study
Draft title: ______________________________________________________
[ ] TITLE-PAGE FOOTNOTE gives the full reference of the primary report,
states what this paper adds, and states what is NOT re-reported
Draft footnote: ___________________________________________________
___________________________________________________________________
[ ] ABSTRACT opening sentence states the primary result as already published
Draft sentence: ___________________________________________________
[ ] COVER LETTER declares the relationship, and the primary paper is attached
[ ] REGISTRATION NUMBER appears in this manuscript, same as the primary
PART 3 - DUPLICATE-RISK SELF-CHECK
[ ] No primary endpoint is reported here as this paper's own finding
[ ] Sample size and recruitment window are explained as the parent cohort
[ ] Methods text reused from the primary paper, with the primary cited
[ ] No discussion or conclusion text reused
[ ] Similarity report run; reviewed BY SECTION
results-section overlap ....... ____ (high = investigate)
methods-section overlap ....... ____ (high = usually expected)
[ ] Where a secondary publication in another language: both editors informed,
primary version supplied to the second editor, interval agreed
PART 4 - IF YOU ARE SCREENING RATHER THAN WRITING
For each pair of candidate reports, compare:
registration number ....... same / different / absent
sample size ............... ____ vs ____
number of centres ......... ____ vs ____
recruitment window ........ ________ vs ________
author overlap ............ none / partial / complete
gap between publications .. ____ months
[ ] Decision recorded: which report included, and why
[ ] Patients excluded by that decision: ______ Pubrica AcademyWritten and maintained by Pubrica’s subject matter experts. Articles are revised when the underlying standards change, and each article states the date on which it was last reviewed.
How Pubrica can help
Pubrica compiles an inventory of a programme’s outputs before the second paper is drafted, and prepares the title, footnote, abstract and cover-letter wording that records the relationship between companion papers in all four places. For systematic reviews, Pubrica separates record deduplication from the handling of duplicate publications in the methods and reports both counts, so that the flow diagram is reproducible.