Articles may be flawed due to honest errors, native mistakes, or research misconduct. If errors are serious enough to invalidate a paper's results and conclusions the paper should be retracted to correct the literature. Redundant publication, plagiarism, peer review manipulation, reuse of material or data without authorization, copyright infringement or some other legal issue (like libel, privacy, illegality), unethical research, and/or a failure to disclose a major competing interest that would have unduly influenced interpretations or recommendations may warrant retractions.

The article’s author(s), an institution, readers, or the editor may ask for retractions. However, authorship disputes without doubts regarding the validity and reliability of the data should not lead to retraction. Authors of the retraction should ideally be the same as the retracted article but other responsible persons or the editor may also be authors of the retraction. The reasons for the retraction, the person asking for a retraction, and possibly how the journal found out about it would be mentioned in the retraction notice (names would be mentioned only with permission). It would also completely cite the article. Both retractions and all forms of the retracted articles would be clearly labeled and would also be linked in both directions. The retraction would appear on all online searches for the retracted publication.

Publications would be retracted as soon as possible after the editor is convinced that retraction is required; however, when there isn’t enough evidence, the editor could consider publishing an expression of concern.

In rare cases when the article is defamatory, violates personal privacy, is the subject of a court order, or might pose a serious health risk to the general public the article would be removed. Still, the metadata (title and authors) would remain and the retraction notice would clearly state why the full article has been removed.

An author may republish some of the work if not all of the content is found to be unreliable. Authors would notify the editors of the new journal of the prior retraction. It is likely appropriate to cite the retraction, indicating why the work was flawed and what has been corrected in the new article. Permission to republish also needs to be agreed with the copyright holder of the retracted work. ‌

In addition, if the error in the paper is considered unintentional, the underlying science appears valid, and the changed version of the paper survives further review and editorial scrutiny, then retraction with republication (or “replacement”) would be considered with an explanation. In such cases, the journal will show the extent of the changes in supplementary material or an appendix, for complete transparency.