Elsevier Has Four Outcomes for a Flawed Article, and Only One Is a Retraction

Four words get used loosely in conversation and mean four specific, different things in Elsevier’s own policy.

Correction (Corrigendum or Erratum). Used when there is an error or omission that does not affect the article’s results or conclusions. “A Corrigendum will be published where it is required to correct an error or omission”, drafted by the authors and approved by the editor; where the publisher caused the error, an Erratum is issued instead. An author-list change may be posted if the editor finds it justified on supporting documentation.

Retraction. Applies to both published articles and articles still in press, for errors too extensive to correct, or for policy violations: plagiarism, fabricated data, compromised peer review, undisclosed major conflicts of interest, unverifiable authorship, or a breach of Elsevier’s generative AI policy. “The retraction of an article by its authors or the journal Editor under the advice of members of the scholarly community” is how the policy describes it. The notice is titled “Retraction: [article title]”, signed by the editor and, where appropriate, the authors, and appears in a later issue, linked back to the original. The original PDF survives with a “retracted” watermark; the HTML version is removed.

Withdrawal. This one is the narrowest and the most misused in conversation: it applies only to articles-in-press, never to a published article, and only where the article was released early due to an editorial or production error, or duplicates another published article because of such an error. The HTML and PDF are both replaced with a withdrawal notice.

Removal. Reserved for narrow legal grounds: the article is defamatory or infringes legal rights and retraction will not fix it, it is subject to a court order, or leaving it up poses a serious health risk. The title and author list stay visible; the text is replaced with a notice explaining the legal removal. Where the concern is a serious health risk specifically, authors may instead retract the article and publish a corrected replacement, with the retraction notice linking to it.

Separating withdrawal from retraction by timing — before formal publication versus after — matters because the two carry completely different weight on a researcher’s record. An article-in-press withdrawn for a production error was never finished publishing; a published article retracted for misconduct is a permanent, citable record of a finding that failed. Conflating the two in conversation, as happens constantly, makes an administrative fix sound like a misconduct finding or the reverse.

Keeping the original retracted PDF accessible, watermarked rather than deleted, protects the scholarly record’s continuity: anyone who already cited the article, or needs to see exactly what was claimed, can still find it, with the retraction clearly marked rather than the paper simply vanishing.

Routing retraction decisions through an editor or ethics committee, consulting a dedicated Research Integrity & Publishing Ethics Center of Expertise, rather than leaving it to a single handling editor’s judgement, is what keeps the decision defensible when an author disputes it — which, for a retraction specifically, is the normal case rather than the exception.

How it reaches a specific journal. The policy is published at group level and implemented by each journal’s own editorial team, with Elsevier’s Research Integrity & Publishing Ethics Center of Expertise available to every title for complex cases.

A journal may receive a correction request faster than a retraction one: the policy describes correction as something an editor can approve directly, where retraction routes through an ethics committee and a specialist team, which is a structurally slower process by design.