CERwatch
NewFirst production run: 5 of 5 findings matched Crossref ›

Your CER, watched every night.

It was right when you signed it. CERwatch tells you the morning after any paper it cites is retracted, corrected or flagged, and exactly which claims rest on it.

PER-0087 · SARS-CoV-2 RT-PCR assay · Rev BIllustration
Nightly check · 03:14 UTC · Crossref · PubMed · Retraction Watch
Source register
  1. [1]Corman VM, et al. Euro Surveill. 2020;25(3)Checking…Expression of concern · not a retraction
  2. [2]Wang W, et al. JAMA. 2020Checking…Clear
  3. [3]Wölfel R, et al. Nature. 2020;581:465–469Checking…Correction · 11 Dec 2020
  4. [4]Mehra MR, et al. N Engl J Med. 2020;382(25)Checking…Retracted · 4 Jun 2020
Claims in this report
  • 4.1Primer design follows the assays first described in [1].
  • 4.3Viral load is highest in respiratory specimens [2, 3].
  • 5.2Outcomes in hospitalised patients are summarised from [4].
Retracted. 1 claim cites this source: 5.2. You decide what it means for the report.
Each paper's status is its real public record in Crossref on 30 Sep 2026. The report is fictional.
Notices checked against
72,790

retraction, correction, concern and reinstatement notices in the Retraction Watch database, checked against your sources every night. A paper can be retracted years after your CER cites it. The reference list doesn't change. The evidence does.

67,363 retractions3,727 expressions of concern1,538 corrections162 reinstatementsas of 30 Sep 2026
How it works

Three steps. Then it watches.

01

Send the reference list

An export from EndNote or Zotero, or the Word file of the CER itself. Every citation is matched to its paper. Anything that can't be matched is listed for you, never dropped and never guessed.

.ris.bib.docx
02

Every source, every night

Each paper is checked against Crossref, PubMed and Retraction Watch. Its status comes from the published notices themselves, and the most serious one decides.

CrossrefPubMedRetraction Watch
03

See what depends on it

When a status changes, you see which claims cite that paper. You record your determination, and sign an Evidence Integrity Statement for the technical file.

FindingsSigned statement
A real run

What it found the first time it ran.

On 1 October 2026, CERwatch's first production run checked a test document citing 30 papers. It raised five findings. Every one matches the publisher's own record in Crossref.

Retracted◷ 5 Jun 2020
Hydroxychloroquine or chloroquine with or without a macrolide for treatment of COVID-19
Mehra MR, et al.inThe Lancet2020
Retracted◷ 4 Jun 2020
Cardiovascular disease, drug therapy, and mortality in Covid-19
Mehra MR, et al.inN Engl J Med2020
Expression of concern◷ 3 Dec 2020
Detection of 2019 novel coronavirus (2019-nCoV) by real-time RT-PCR
Corman VM, et al.inEuro Surveill2020
This paper also has four corrections and an addendum. CERwatch reported the expression of concern, because that's the most serious notice on record. It did not call it a retraction.

Plus two more retractions. The other 25 references carried identifiers that don't match the paper they describe (the test document was built to be awkward). Each one was listed for a person to resolve. None was dropped, and none was guessed.

Why CERwatch

Built for evidence a notified body will read.

Never over-flagged

A false alarm on a compliance document costs as much as a missed one. Every notice is reported as exactly what it is.

Expression of concern→Reported as expression of concern
Correction→Reported as correction
Retraction→Reported as retraction

Notices, not guesses

Status is read from published retraction, correction and concern notices. Nothing is inferred, and no model output is ever reported as fact.

Claim by claim, confirmed by a person

You see which statements in the report rest on which paper. A link can be suggested automatically, but only a named person can confirm one.

Statements you can reproduce

Each signed Evidence Integrity Statement records the exact data it was checked against, so it can be regenerated and compared years later.

Retraction Watch snapshot 3d28d6230 Sep 2026signed · meaning: approval

An audit trail nobody can edit

Every decision is recorded with who, when and why. The database itself refuses edits and deletions.

INSERT finding · reason: status changed at Crossref
UPDATE claim_source · reason: confirmed in the confirmation queue
UPDATE audit_log · 0 rows changed: refused by the database

Part 11 ready

Named logins, signatures that state what they mean (review, approval or responsibility), and an append-only record.

Compliance belongs to your whole validated system. CERwatch is built to fit into one.

“CERwatch never tells you a conclusion is wrong. It shows you what depends on what, and it records what you decide.”

For anyone who signs or maintains clinical evidence under EU MDR 2017/745 and IVDR 2017/746.

Regulatory consultantsMedical writersIn-house RA teamsCER · PER · PSUR

Get a free scan
of your references.

Send us the reference list: the bibliography only, not the CER. We'll reply with every cited paper that has been retracted, corrected or flagged, and the notice behind each one.

Email your reference list
or write to hello@cerwatch.com