Redundancy can survive a complete rewrite
Duplicate publication often describes nearly identical work appearing twice. Redundant publication is broader: a new paper may add a small element while republishing a substantial scientific core. Text-matching software can reveal copied wording, but it cannot decide whether two differently written papers claim the same result.
The question is not simply how many words overlap. Review what readers are being asked to regard as new. Reuse of a cohort may support a distinct analysis; reuse of the same main endpoint and interpretation is harder to distinguish. Journal policy, field practice and publication agreements add venue-specific requirements.
Compare stable research objects, not impressions
Place both manuscripts beside the protocol or registration, dataset IDs and related abstracts. Extract the primary question, population, recruitment period, outcomes, analytical model, main tables, figures and conclusion. Record publication status: draft, conference abstract, preprint, accepted manuscript or version of record.
A similarity percentage is only one clue. Standard methods may carry recurring language while results differ. Conversely, paraphrased results can conceal complete redundancy. Translations require the same scrutiny: a second language may broaden access but does not transform an old study into new original research.
Secondary publication has explicit conditions
A secondary version can have a legitimate public purpose, for example making an important guideline available to another audience. ICMJE describes safeguards including editorial approval, respect for priority, faithful presentation, a different readership and a prominent reference to the primary publication. The actual journals may impose further conditions.
A plain-language summary is not automatically a second research article. The risk rises when the later item presents itself as a fresh original report. Rights are separate from attribution: citing the first article does not grant permission to reproduce a protected table, and permission does not remove the need to state that findings appeared earlier.
Give editors a concrete map of the relationship
In the covering letter, name every materially related output with citation, status and link or attached manuscript. State which data overlap, what the new question is and which figures have appeared before. “There are related papers” is not enough for an informed decision.
Cite prior work in the manuscript where methods, datasets or findings connect. Disclosure to the editor and disclosure to readers serve different functions; neither substitutes for the other. If peer review is anonymised, follow the journal’s instructions rather than quietly removing information the editor has requested.
The salami-slicing matrix tests whether outputs from one dataset carry independent questions. The text-recycling matrix examines passage-level reuse. The publication ethics hub joins these checks. The single foundation route is the evidence-review guide.
Include dissertations, registered reports and public research reports in the inventory, not only journal articles. State whether their tables or conclusions entered the manuscript. A change in author list may reflect new work, but cannot erase shared data. Ask both author groups to review the comparison and preserve any unresolved disagreement for the editor.
Respond early when unreported overlap is found
Before acceptance, notify co-authors and the editor, supply the comparison, and propose clearer separation, merger or withdrawal. Do not withdraw from one venue while concealing that the same work remains under review elsewhere. Keep correspondence and the exact versions considered.
After publication, an editor may link, correct, label or retract an article depending on the evidence and policy. Authors should cooperate and avoid destroying the history needed for assessment. Record discovery date, people notified and any amended files.
- Inventory all related research outputs.
- Compare questions, data, analyses and conclusions.
- Review textual and visual reuse separately.
- Write the new contribution in one exact sentence.
- Check journal rules, contracts and licences.
- Disclose related work to editors and readers.
- Archive decisions and approved wording.
A defensible record does not promise zero overlap. It makes necessary overlap intelligible and ensures that previously reported work is not presented again as an undisclosed first result.