To check whether a paper you cite has been retracted, open each reference at its DOI or journal page, look for a retraction notice, and confirm biomedical papers in PubMed. Recheck before you submit. For a long reference list, RefGuard, Proofig AI’s reference integrity product, flags citations to retracted papers, with the reason, date and type of notice.
What does it mean when a paper is retracted?
A retraction tells readers that a paper’s findings can no longer be relied on. The COPE retraction guidelines, last reviewed in August 2025, describe retraction as a way to correct the literature and alert readers to seriously flawed or erroneous content, whether the cause was honest error or misconduct. Its purpose is to protect the record, not to punish authors.
A retracted paper does not disappear. The ICMJE recommendations ask that retracted articles stay publicly available, be clearly labeled as retracted in every form, and be linked in both directions to a retraction notice that explains why.
Two other notices are easy to confuse with a retraction:
- A correction (an erratum or corrigendum) fixes part of a paper that otherwise stands.
- An expression of concern is an editor’s warning that questions about a paper are still open. COPE suggests one when an editor cannot yet reach a conclusion.
A full retraction is the notice that withdraws the findings as a whole.
Where do retraction notices appear?
Retraction notices appear in several places, but not in a PDF you saved earlier: that file is a snapshot and will not show a retraction issued since. Check the live record instead.
- The publisher’s article page. This is where the retraction label and the link to the notice should appear.
- The DOI metadata and Crossmark. Crossref’s guidance for publishers asks them to issue a separate retraction notice rather than edit the original, to record the update in the article’s metadata, and to add “RETRACTED:” in front of the title. Where a journal shows the Crossmark button, it opens the article’s current status, including corrections and retractions.
- PubMed. The National Library of Medicine assigns the publication type “Retracted Publication” to retracted articles and links each record to its retraction notice with a “Retraction in:” line.
How do you check each reference, step by step?
A manual check is slow but simple. Work from the reference list in the manuscript, not from your library: that is the list the editor will see.
- Give each entry an identifier. Add the DOI, plus the PubMed ID for biomedical papers. Entries without one take longest to check, so mark them.
- Open each DOI. A DOI resolves to the publisher’s current page. Look for a “Retracted” banner or watermark, a “RETRACTED:” prefix in the title, or a linked notice. If the page has a Crossmark button, click it.
- Search PubMed for biomedical references. The ICMJE treats PubMed as the authoritative source on retractions for MEDLINE-indexed journals and suggests searching for “Retracted publication [pt]”.
- Read any notice you find. Note whether it is a retraction, a correction or an expression of concern, its date, and the reason given. The type of notice decides what you do next.
- Keep a short log. Record the date of each check and what you found, so the next pass is faster.
- Check again before you submit and at each revision. Retractions can land during drafting or review; Octym’s post on citing retracted papers explains why that timing catches careful authors.
What should you do when you find a retracted reference?
The ICMJE places the duty on authors: they are responsible for making sure that no reference cites a retracted article, except when the reference is about the retraction itself. In practice, you have three options.
- Remove it when other references in your list already support the statement.
- Replace it with a sound source that supports the same statement. If no other source does, revise or soften the claim, because the evidence behind it has been withdrawn.
- Keep it on purpose when the retraction is the point: for example, when you discuss the retracted study, its influence on the field, or the problems that led to its withdrawal. Cite the retraction notice next to the original, and say in the text that the paper was retracted and why.
Then look past the citation. If the retracted paper supplied a method, a dataset or a comparison, review those parts of your manuscript too. If the manuscript is already under review, tell the editor what you changed in your response letter.
What if a cited paper is retracted after you publish?
A retraction of work you cited does not retract your paper. Whether you need to act depends on how much your conclusions rest on it.
The risk is that readers follow your citation without knowing. A study of retracted biomedical papers, published in Quantitative Science Studies, found that only 722 of 13,252 citation contexts written after a retraction (5.4%) acknowledged it. A BMJ Open investigation followed 12 clinical trial reports that were later retracted for misconduct. By 2016 they had been cited in 1,158 publications, and by 2018 only one of 68 citing reviews, guidelines and trials appeared to have reassessed its conclusions.
If you learn that a paper you cited has been retracted:
- If it was background, you may need no action, but note the retraction when you cite the topic again.
- If a result, a method or a conclusion depends on it, contact the journal’s editor and ask whether a correction is appropriate.
How can RefGuard automate the retraction check for a reference list?
RefGuard is Proofig AI’s reference integrity product. It checks a manuscript’s reference list for retracted papers in one pass. You upload the manuscript as a PDF or Word file (text-based, up to 150 pages), and RefGuard:
- flags citations to retracted papers, using a proprietary retraction database;
- shows the reason, the date and the type of notice for each flagged citation, so you can read what the notice says before you decide what to do.
The same pass covers other reference problems. RefGuard flags references that cannot be matched to any real publication, which is where hallucinated citations surface. It flags citations that do not fit the claims they are attached to, and it points out incomplete or broken references, so they stand out.
RefGuard is available on its own, and it powers the reference integrity check inside PubShield, Proofig AI’s hub for publishers and research institutions, and inside Octym, which reviews the entire paper with 26 checks. Its results are signals for a person to review: you still decide whether to remove, replace or keep each citation. See what RefGuard checks and what its report shows.
Frequently asked questions
Is citing a corrected paper the same as citing a retracted one?
No, a corrected paper still stands, while a retracted paper has been withdrawn. The National Library of Medicine treats corrections, corrigenda, addenda and partial retractions (for a single figure, table or statement) as errata linked to the original article. You can usually keep citing a corrected paper. Read the notice first: if the correction changes the result you cite, cite the corrected version and the notice, or choose another source. RefGuard, Proofig AI’s reference integrity product, flags citations to retracted papers with the type of notice.
Which version should I cite if a retracted paper was later republished?
If a retracted paper was later republished, cite the republished version. Journals sometimes retract a paper because of errors in the science and then publish a corrected version. PubMed links the two records with “Retracted and republished in” and “Retracted and republished from” lines, so you can reach the current version from either record. Refer to the earlier version only when the difference matters to your argument, and say that it was retracted. RefGuard, Proofig AI’s reference integrity product, flags citations to retracted papers in your reference list.
Can a perfectly formatted reference still cite a retracted paper?
Yes, a correctly formatted reference can still point to a retracted paper. Formatting checks how a reference is written, not the status of the work it cites. A retraction is recorded on the publisher’s page, in the DOI metadata and in PubMed, not in the citation itself, so a reference can follow the journal’s style exactly and still cite withdrawn work. Status needs a separate check, by hand or with RefGuard, Proofig AI’s reference integrity product.
