Skip to main content
The Journal of Global HealthInternational Health JournalEvidence for health where it's hardest to deliver
Vol.
XXIV
No.
04
Year
2026
Menu

Quick Links

ArticlesCommunity ForumContact
Language
Appearance
Editorial Policy

Corrections, Expressions of Concern & Retractions

Global Health Equipment Review·Field Series·Evidence Desk·Peer-reviewed in spirit · Educational

Equipment changes, supply chains break, and even a careful assessment can be wrong. This page sets out how we repair the record when that happens — when we correct, when we flag concern, when we retract, and the version-of-record principle that keeps a citation to an old verdict honest.

Read our publication ethicsReport an error to the editors
Plate II
Fig. — field plate
Figure. Engraved reference plate — evaluating equipment on field resilience, not lab-bench ideals.
The principle

Repair the record in the open, never in silence

An assessment on this platform is dated for a reason: it is a snapshot of the evidence as we read it on a given day, not a permanent verdict. When new field-failure evidence, a changed specification, or a documented error of fact reaches us, we re-examine the assessment and put the correction on the record — visibly, dated, and with its reason — rather than quietly editing the page and pretending the earlier state never existed.

Silent editing is the practice we refuse. A reader who acted on last quarter's verdict, or cited it in a procurement decision, is entitled to see what changed and why. So every substantive change leaves a trace. This is the operating arm of the integrity commitments in our publication-ethics statement.

From the review desk

Three responses, matched to severity

Not every change warrants the same action. We distinguish carefully so the response is proportionate to what is wrong.

  1. 01

    Correction

    For a discrete error of fact, an updated specification, or new field evidence that refines but does not overturn a verdict. The assessment is amended and a dated correction note is appended explaining what changed. A change that affects the grade itself is flagged at the top of the review.

  2. 02

    Expression of concern

    For a credible but unresolved doubt about an assessment's basis — an allegation under investigation, evidence we can no longer fully stand behind pending review. We attach a dated notice so readers are warned while we work, rather than leaving an uncertain verdict to stand unmarked.

  3. 03

    Retraction

    For an assessment whose conclusion no longer holds — fundamentally flawed evidence, a conflict that should have excluded it, or a verdict shown to be wrong. The review is marked retracted, the reason stated, and the original text retained and clearly labelled rather than deleted, so citations to it still resolve to an honest record.

Our field-equipment review method

How an error becomes a correction on the record

From the moment a concern reaches us to the moment the record is repaired, the sequence is the same.

  1. 01Step

    Report received

    A reader, contributor, program, or vendor brings a documented error, a changed specification, or contradicting field evidence to the editors. The concern is logged with its date and substance.

  2. 02Step

    Re-examination

    An editor re-checks the assessment against the new evidence and the published rubric, drawing on independent review where the matter is significant or contested. We test whether the verdict still follows from what is now known.

  3. 03Step

    Decide the response

    We choose correction, expression of concern, or retraction by severity: refine, flag, or overturn. The choice is recorded so the reasoning behind the action is itself part of the record.

  4. 04Step

    Publish and date the notice

    The correction note, concern notice, or retraction is published with the affected assessment, dated, and — where the grade moved — flagged at the top. The earlier state is preserved and labelled, never silently overwritten.

Method of review
Plate II
Fig. — field plate
Figure. Engraved reference plate — evaluating equipment on field resilience, not lab-bench ideals.
Version of record

Why we keep the superseded version visible

The version-of-record principle is simple: the published assessment a reader cited must remain identifiable, even after it is superseded. When we amend or retract, we do not erase the prior text — we mark it, date the change, and keep it reachable, so a reference made to a verdict six months ago still resolves to the exact thing that was said and to the notice explaining what changed since.

This matters most for procurement and program decisions, where a verdict may have been acted on with budget and lives downstream of it. Pretending an earlier assessment never existed would hide exactly the information a reader needs to re-evaluate that decision. So the record is cumulative and honest: current verdict, the change history, and the reasons — all open, consistent with our open-access and archiving stance.

Review criteriaBy which we judge field gear
01Criterion
100%
Substantive changes dated and noted on the record
02Criterion
0
Silent edits to a published verdict
03Criterion
3
Graded responses: correct, flag concern, retract
Standards of the desk
Plate II
Fig. — field plate
Figure. Engraved reference plate — evaluating equipment on field resilience, not lab-bench ideals.
Who can report

Anyone with the evidence — including the vendor we graded

The correction log is open at both ends. Any reader, contributor, program, or vendor can bring a documented error to the editors, and the standing of the person reporting it does not decide whether we act — the evidence does. A field logistician with a logged failure, a procurement officer with a changed specification, and a manufacturer whose device we may have graded against an outdated datasheet are all heard on the same terms.

What a report needs is documentation, not authority: the specific assessment, and the evidence behind the concern — a field-failure report, an updated specification, a factual error we can verify against the record. We weigh that evidence under the same rubric the original verdict was reached under, drawing on independent review where the matter is significant or contested. A vendor's report earns no automatic correction, and a critic's earns no automatic retraction; both earn a fair re-examination.

How a verdict is reasoned
Correspondence

Queries to the desk

05 entries on file
Q1.What is the difference between a correction and a retraction?

A correction refines an assessment that remains broadly sound — a factual fix, an updated specification, new evidence that adjusts but does not overturn the verdict. A retraction is for an assessment whose conclusion no longer holds, because the evidence was fundamentally flawed, a disqualifying conflict surfaced, or the verdict has been shown to be wrong. A correction amends; a retraction withdraws the conclusion while preserving and labelling the original.

Q2.When do you issue an expression of concern?

When there is a credible but unresolved doubt about an assessment — an allegation under investigation, or evidence we can no longer fully stand behind while we re-review. It is a dated warning to readers that the verdict is in question, used so an uncertain assessment does not stand unmarked while the matter is worked through. If the concern is resolved, we convert it to a correction, a retraction, or a note that the assessment stands.

Q3.Do you delete retracted assessments?

No. A retracted assessment is marked as retracted with the reason stated, and the original text is retained and clearly labelled. Deleting it would break the version-of-record principle and hide what a reader may have cited. The point of a retraction is to flag the error openly, not to erase the evidence that it occurred.

Q4.How do I report an error in an assessment?

Contact the editors with the specific assessment and the documentation behind your concern — a changed specification, a field-failure report, or a factual error. We log it, re-examine the assessment, and respond with a correction, an expression of concern, or a retraction as the evidence warrants. If the concern is not borne out, we tell you that too. Documented reports are the main way the record stays honest.

Q5.How fast do you act on a reported error?

A clear factual error is corrected as soon as it is verified. A contested matter, or one touching a verdict a program is acting on, goes through re-examination and where needed independent review before we publish a notice — we will not rush a retraction or sit on a concern. We keep the reporter informed of where the matter stands rather than letting it go quiet.

ArticleField Series

The repair stage of a single discipline

Corrections are the fourth checkpoint in one continuous discipline. Our editorial standards set the rubric; our peer-review policy checks a draft before it carries the journal's name; our publication-ethics statement sets the conduct expected of everyone involved; and this policy repairs the record when, despite all three, something turns out to be wrong. The first three guard the verdict before publication; this one keeps it honest afterwards.

That after-the-fact stage matters most because the alternative — a silently edited page — would quietly undo everything the earlier checkpoints achieved. So the correction log is open, dated, and consistent with our open-access and archiving stance: the current verdict, the change history, and the reasons are all reachable by the same readers, with no paywall on the repair. We follow the spirit of widely-recognised guidance on correcting the scholarly record; we state that as our own commitment, not as membership of a body or an accreditation we do not claim.

End of section
From the editorial desk

Spotted an error, or have evidence that changes a verdict?

The correction log only works if the evidence reaches us. If a specification has changed, a device has failed in the field, or an assessment got something wrong, the editors want the documentation — repairing the record is part of the work, not an admission to avoid.

Report it to the editors