Retractions: How Scientific Papers Die and How to Check Before You Share

A retraction is a journal's formal withdrawal of a published paper as unreliable or invalid. It is the most serious correction in scientific publishing: not a revision or an expression of concern, but a statement that the paper should never have entered the record as it stands. The volume is larger than most readers assume. The Retraction Watch database, which tracks the practice across publishers, recorded on the order of 10,000 retractions in 2023 alone — a record year driven partly by the mass retraction of paper-mill products, in which commercial outfits sell authorship on fabricated studies. The 2020 example that remains most familiar to the public: the Lancet's retraction of a large hydroxychloroquine study after the underlying patient data could not be verified, within two weeks of publication.

This article is information, not medical advice. Publication status is one fact about a study; clinical decisions should rest on a clinician's reading of the whole evidence base, of which retractions are an important but single component.

Why papers get retracted

Surveys of retraction reasons consistently find a long tail of causes, with overlap between them: duplicated data or images, plagiarism, unreliable methods and analysis, undisclosed conflicts, ethics-approval failures, and outright fabrication. Honest error — a statistical mistake or contaminated sample discovered later — accounts for a meaningful minority, and the classic retraction reasons also include author disputes that sometimes have nothing to do with validity. The fastest-growing category is different: paper mills producing fabricated manuscripts at industrial scale have forced publishers to retract thousands of articles, particularly in certain open-access journals, and motivated editors to screen submissions with image-forensics tools.

• Misconduct-adjacent: fabricated or falsified data, manipulated images, plagiarism — the categories that make headlines.
• Honest error: analysis mistakes, wrong samples, coding bugs; journals correct these via retraction when the findings cannot stand.
• Process failures: missing ethics approval, unconsented patient data, or compromised peer review, including reviewer-identity fraud.

How the retraction process works

Anyone can raise a concern: a reader, a rival lab, an institution, or the authors themselves. Journals investigate, often deferring to the authors' institutions for misconduct findings, and the process can take months to years — which is where interim labels come in. An expression of concern flags serious unresolved doubts; a correction handles fixable errors; retraction removes the paper's findings from reliance. Retracted papers are not deleted: they stay online, typically watermarked, so that the scholarly record remains complete and honest about its own errors. Authors can dispute a retraction, and a small number are later retracted-with-replacement or reinstated after appeal, but the default is permanence.

Related stories: Preprint vs Peer-Reviewed: What the Study Behind a Headline Actually Is · “50% Higher Risk”: Relative vs Absolute Risk in Health News.

The zombie-paper problem

The problem for the public is not the retraction — it is the afterlife. Studies find that retracted papers continue to be cited for years after withdrawal, often without acknowledgment of their status, because PDFs are downloaded, hosted, and summarized far from their journals. The COVID-19 era supplied vivid cases: the retracted Lancet and New England Journal studies of 2020 continued to shape treatment debates for weeks, and a 1998 retracted paper on vaccines and autism — fraudulent, and retracted in 2010 after more than a decade of investigation — is still cited in anti-vaccine material today. Retraction reverses the record for careful readers; it does not chase down every copy.

How to check a study in two minutes

1. Search the DOI or title in the Retraction Watch database, now integrated into the Crossref system that many reference managers use, so flags can appear automatically in scholarly tools.
2. Look at the journal page, not the PDF: retraction notices are posted alongside the article, and the article page carries the current status, including expressions of concern.
3. Check the citation trail: if later papers cite the study approvingly years after its retraction, treat their conclusions with matching suspicion; if it was retracted and superseded by a corrected study, find the successor.

What retraction is not

Retraction is not proof that the whole field collapsed or that science is unreliable in general — the mechanism exists precisely because the system corrects itself, publicly, on the record. Nor is it proof the opposite conclusion is true: when a study claiming X is retracted for fabrication, the honest state of knowledge is usually “unknown,” not “X is refuted.” Media coverage often makes both errors at once, treating retraction as either a scandal about all science or as vindication of a pet alternative.

Who watches the watchers

The retraction system itself has critics and referees. Retraction rates vary enormously between publishers, and the growth of pay-to-publish outlets has produced journals that retract rarely and poorly — some studies of editorial practice have found large numbers of journals that never retracted anything despite publishing thousands of papers, which is arithmetically implausible for honest science at scale. Institutional investigations, meanwhile, can drag for years while authors keep publishing. The emerging counterweight is transparency infrastructure: post-publication peer review platforms, image-forensics screening at major journals, and the integration of retraction flags into reference managers and grant systems. For a reader, the takeaway is to weight the outlet as well as the paper: a journal with a functioning correction culture is more trustworthy than one that claims never to have erred.

What to do with a health claim resting on a retracted paper

Before sharing any striking health finding, run the two-minute check above. If the paper is retracted, look for what replaced it: in most cases subsequent better studies exist, and their results — not the zombie paper — are what your family's decisions should rest on. And if a retracted study underpins advice you received somewhere, from a practitioner or a supplement seller, that is a fair question to put to them directly, with the retraction notice in hand.