Retraction

A retraction is a formal notice indicating that a published research article should no longer be considered a reliable contribution to the scholarly record. Retractions are issued by journal editors, sometimes at the request of the authors, and indicate that the findings are unreliable due to errors, misconduct (such as fabrication or falsification), plagiarism, or ethical violations. Retracted articles typically remain accessible online but are marked with a retraction notice.

Retractions serve as a quality control mechanism for the scientific literature, protecting readers, clinicians, and policymakers from acting on unreliable evidence. The number of retractions has increased substantially over the past two decades, driven in part by improved detection of misconduct, greater scrutiny through post publication review, the growing use of plagiarism detection software, and batch retractions of paper mill output. Despite their importance, retracted articles continue to be cited after retraction, sometimes for years, indicating that the current system for flagging problematic research is imperfect. Researchers conducting systematic reviews should check whether included studies have been retracted and exclude or flag retracted papers to prevent contamination of the evidence synthesis.

Retractions can be initiated by the authors (who may discover errors in their work), by the journal editor (based on an investigation prompted by reader concerns or institutional inquiry), or by the author's institution. The retraction process typically involves an investigation by the journal or institution, a notification to the authors with an opportunity to respond, and the publication of a retraction notice explaining the reason. When there are credible concerns about an article that are still being investigated or cannot be resolved, editors can publish an expression of concern to alert readers; it may later be followed by a correction or a retraction, or remain in place.

Retraction notices should clearly state the reason for the retraction, whether it was initiated by the authors or the editor, and whether the retraction applies to all or part of the article. In practice, many retraction notices are vague or incomplete, making it difficult for readers to assess the nature and severity of the problem.

The Retraction Watch Database, acquired by Crossref in September 2023 and now openly available, records retracted papers and helps researchers identify problematic publications. Major databases (PubMed, Web of Science, Scopus) tag retracted articles, though the visibility and consistency of these tags vary. Crossref's retraction metadata, drawn from notices that publishers register and from the Retraction Watch Database, enables automated checking.

The consequences of retraction extend beyond the specific paper. Co authors may face reputational damage, institutional investigations, and loss of funding. Research built upon retracted findings may need to be reevaluated. In some cases, retracted findings have influenced clinical practice or public policy, requiring corrective action.

Check for retractions whenever conducting a literature review or systematic review, particularly for studies that play a central role in the evidence synthesis. When citing important studies, verify their current status using Retraction Watch or the PubMed "retracted publication" tag. If your own published work contains errors that affect the conclusions, initiate the retraction process promptly and transparently. Journal editors should issue retractions promptly when evidence of serious problems is confirmed, and should follow the COPE (Committee on Publication Ethics) retraction guidelines, updated in 2025, for the process.

A common error is failing to check whether studies included in a systematic review have been retracted, potentially including unreliable evidence in the synthesis. Another mistake is continuing to cite retracted papers without acknowledging the retraction, inadvertently perpetuating unreliable findings. Some researchers confuse retractions with corrections (errata), which are issued for errors that do not critically undermine the paper's conclusions, or with expressions of concern, which alert readers to credible concerns that are still being investigated or cannot be resolved. Others assume that every retraction signals misconduct, but honest error is also a reason to retract; COPE states that the main purpose of a retraction is to correct the literature, not to punish authors. Finally, institutions sometimes delay or suppress retractions to protect their reputation, undermining the self correcting function of science.

A widely cited randomized trial of a surgical technique, published in a major journal, is retracted three years after publication when an institutional investigation reveals that the lead author fabricated outcome data for 30% of the enrolled patients. The retraction notice states that the data are unreliable and the conclusions are unsupported. Three systematic reviews that included this trial need to be reassessed. Re analysis of the meta analysis without the retracted study changes the pooled effect from significant (RR = 0.72, 95% CI: 0.58 to 0.89) to nonsignificant (RR = 0.88, 95% CI: 0.69 to 1.12), altering the clinical recommendation.