Check Corrections and Retractions Before Citing an Acupressure Paper

Start with the exact paper, not a similar title

Save the title, authors, journal, publication date and DOI or PubMed identifier when available. Then open the publisher’s record for that same item. A search snippet, reposted PDF or summary can omit notices that are visible beside the original citation.

Compare the identifiers, not just a memorable title. A protocol, results report, correction notice and patient summary can concern the same research while being different documents. Their publication dates and roles should remain separate. This guide checks those documents; it does not decide whether an intervention works.

Keep the date you checked the record beside your citation. The statement “no linked notice was visible in these checked records on this date” describes a bounded observation. It cannot establish that there are no notices anywhere or that none will appear later.

Read the direction of a PubMed link

NLM’s policy defines these paired labels. [1] Follow the link to the notice and confirm that it refers to the original paper you saved. “Erratum for” on a correction record points back to the article; it is not the correction’s own new research result.

NLM also distinguishes an article marked Retracted Publication from a notice marked Retraction Notice. A notice and its referent are two citation records, not two independent experiments. A Comment in/on relationship is different again: commentary should not automatically be rewritten as a correction or withdrawal.

Find what the notice actually changes

Read the entire available notice. Does it correct an author name, a figure label, a value, an analysis, or the interpretation? Save the notice’s wording and locator with the affected part of your note. Do not assume that all corrections are trivial, or that every correction retracts an entire paper.

NLM includes corrections, corrigenda, addenda and partial retractions in its errata policy. Its linked-record system does not decide whether an error was unintentional or misconduct. [1] The notice’s actual scope matters more than an accusation inferred from its label.

An expression of concern raises questions recorded by the issuing journal; it is not automatically a final retraction. Preserve that status exactly. If the notice or corrected text cannot be obtained, state what remains unavailable rather than guessing whether a particular number survived unchanged.

Use Crossmark as a route to update information

Where a publisher supplies it, the Crossmark button opens status and update metadata. Crossref explains that it can also be embedded in PDFs, so a reader may reach updates after downloading an article. Members implementing it commit to maintaining content and registering updates. [2]

The information is supplied through that publication workflow. A badge alone does not evaluate the scientific conclusions. If a button is absent, cannot be opened or supplies insufficient information, keep that limitation in your record and check the publisher’s page and PubMed links where available.

Keep a record of the information you actually opened. Do not add “verified current” to a paper merely because it has a familiar icon; identify the update record and any remaining uncertainty.

Work through a fictional correction without inventing results

No DOI, PMID, participant values or clinical finding is supplied for this constructed example. It demonstrates record handling, not a real error allegation. The defensible summary is: “A linked notice corrects the stated table labels; the corrected table was not available in this example.”

In a second fictional case, DEMO-PAPER-B has an expression-of-concern notice but no retraction notice in the records provided. Record “expression of concern,” its date and stated reason. Replacing that with “retracted” would invent a different editorial action.

Keep the citation and the usable answer separate

If a paper was retracted, retain the record of the retraction when explaining its history; do not continue presenting it as uncomplicated affirmative evidence. If you discuss a correction, identify both the original and notice so another reader can find the relevant version.

For a result, the next task is still to check the actual population, comparison, outcome, unit, time point and people analysed. A clean bibliographic status does not supply those answers. Use the result-reading guide for that separate question.

A paper-status check does not assess care, diagnose a condition, verify an intervention’s benefit or authorize personal use. Finish with a traceable citation and a precise uncertainty, rather than a treatment decision.

Use the optional document record

Open the original TXT record.

PAPER STATUS CHECK — OPTIONAL BLANK RECORD
Exact original title / identifier:
Publisher / PubMed records checked:
Check date and timezone:
Original record's exact linked label:
Notice title / identifier / date:
What the full notice says changed (with locator):
Corrected or republished text available? Which version?
Crossmark information seen / unavailable / not checked:
What remains unavailable or uncertain:
Bounded citation note:
No clinical conclusion or personal entries are supplied.