Check the status before copying the reference
AMA Citations for Retracted or Corrected Articles
Follow one original paper to its correction, concern notice, and retraction. Choose the record supporting your sentence and check the current AMA correction guidance.
Before using an AMA retracted article citation, decide why the paper belongs in your discussion. A history of a retraction needs a different source statement from a claim that a treatment works. We checked one Lancet paper’s PubMed record and its archived notices, then compared ICMJE, Crossref, and the AMA Manual’s public updates. The original paper, its correction, and its retraction have separate identifiers. Copying the original DOI alone leaves that relationship unexplained.
Check the publisher and PubMed status
Open the article’s publisher landing page and inspect any correction, concern, or retraction links. Check the current online text as well as the PDF you saved earlier. A library copy can predate a notice, so its title page alone cannot settle the article’s present status.
Next, open the PubMed record. ICMJE identifies PubMed as the authoritative source for retraction information for MEDLINE-indexed articles. NLM distinguishes the publication type Retracted Publication, attached to the original paper, from Retraction Notice, attached to the notice. Follow the links in both directions to confirm that they identify the same work.
For papers outside MEDLINE, start with the publisher and follow its notices. Record the notice type and date in your notes. A successful DOI lookup retrieves a bibliographic record; read its status information before deciding how to use it.
Map one original paper to its notices
The worked example is Mehra and colleagues’ 2020 multinational registry analysis of hydroxychloroquine or chloroquine. Its original PubMed identifier is 32450107. That record links to a Department of Error notice, an expression of concern, and a retraction. The table shows selected relationships from that record, rather than every related publication.
The retraction notice is by Mehra, Ruschitzka, and Patel. It reports that an independent review could not proceed because the required material was not transferred, and that the authors could no longer verify the primary data sources. That notice supports an account of the withdrawal. It provides no basis for treating the withdrawn analysis as dependable treatment evidence.
| Record | Identifiers | Relationship to the original |
|---|---|---|
| Original registry analysis | PMID 32450107; DOI 10.1016/S0140-6736(20)31180-6 | Work identified as retracted |
| Department of Error | PMID 32485145; DOI 10.1016/S0140-6736(20)31249-6 | Erratum for the original |
| Expression of concern | PMID 32504543; DOI 10.1016/S0140-6736(20)31290-3 | Concern about the original |
| Retraction notice | PMID 32511943; DOI 10.1016/S0140-6736(20)31324-6 | Retraction of the original |
Distinguish correction, retraction, and concern
A correction identifies a change to published information. Read what changed and whether your sentence relies on the affected passage. ICMJE allows correction when errors leave the direction and implications of the results and conclusions intact. A corrected article can remain usable, provided you consult the relevant corrected version.
An expression of concern alerts readers to a problem under investigation. Follow it to any later outcome. In this example, the concern notice described serious questions while an audit was ongoing; the later retraction changed the status readers needed to report. An older concern notice alone would give an incomplete account.
A retraction removes the paper’s standing as reliable support for its conclusions. Read the stated reason without inventing a misconduct finding. NLM’s retraction labels do not distinguish honest error from misconduct. If a work is retracted and republished, inspect the replacement and the relationship notice before assuming the old and new analyses are interchangeable.
Choose the record supporting your discussion
ICMJE advises against citing retracted articles except when discussing the retraction. A research-history sentence can identify what was published and explain its subsequent withdrawal. Cite the original when you need to identify that historical object, and the retraction notice for the withdrawal and stated reason. Keep the status visible in the sentence.
If your argument needs evidence about treatment benefit or harm, find suitable unretracted research and evaluate it on its own merits. Merely attaching a retraction notice to an old finding does not restore that finding’s evidential value. This example teaches source selection and makes no clinical recommendation.
Illustrative sentence about publication history
The registry analysis by Mehra and colleagues was retracted after three authors reported that they could no longer verify its primary data sources.
Cite the retraction notice for this account. Identify the original separately if your discussion examines its published content.
Use the current AMA correction guidance
The AMA Manual’s May 28, 2020 update to section 3.11.14 says that published-correction wording and an additional citation need not be appended to the original article’s reference. An older template that always demands that attachment can therefore give you unnecessary instructions.
For an AMA corrected article reference, check the corrected text and cite the version supporting your statement. If you discuss the correction itself, identify its notice separately. Follow any more specific submission requirements from your journal. The AMA update addresses reference formatting; it does not tell you to ignore what the correction changed.
Reference identifying the correction notice itself
Department of Error. Lancet. Published online May 30, 2020. doi:10.1016/S0140-6736(20)31249-6
PubMed lists no authors and identifies this item as an online-ahead-of-print erratum. This example identifies a notice in a retracted paper’s history; it does not endorse the paper’s findings.
Refresh metadata and verify the final reference
Crossref recommends a separate document, DOI, and metadata for an editorially significant update, with a link back to the affected work. It also recommends marking a retracted work’s status in its title metadata. Treat those relationships as information to inspect, not as proof that every citation export displays them correctly.
Before finishing an AMA correction citation, compare the imported record with the item you opened. Check whether the authors and page belong to the article or the notice. Refresh an old saved citation when necessary, and reread the corrected passage before leaving its associated claim unchanged.
Preprints require their own version check. Follow a later journal publication without silently transferring its DOI or publication status into the earlier record. The preprint guide covers that distinction.
What a DOI or PMID import can establish
AMA Cite formats bibliographic fields from an identifier, but imported records may omit a later status update. Open the publisher and PubMed notices yourself. To cite a retraction in AMA, use the notice’s identifier for its reference and compare the result with the notice before copying it.
