Skip to main content
Back to SentryData study · June 2026

Cited after retraction

We checked the reference lists of 38,350 recent (20242025) biomedical papers — every citation against the Retraction Watch database — and asked a sharper question than “how often does a retracted reference appear.” Two questions, both about whether something was actually catchable: was the cited paper already retracted when the citing paper was published, and which papers lean on many retracted articles at once?

69%
of the retracted/EoC references we could date were already retracted when the citing paper was published — a check would have caught them (1,024/1,485).
42
papers cite three or more retracted/EoC articles; 5 cite five or more — the concentration integrity researchers inspect most closely.
23
clusters of papers lean on the same retracted work; the tightest is 4 papers sharing 15 retracted references.

What this means (and what it doesn’t)

A retraction that lands after a paper was published can’t be a screening failure — there was nothing to catch. So the number that matters is the 69% where the cited paper had already been retracted, often years earlier. Those are the citations a reference check at submission would have flagged. We measured the citation’s presence, not whether each one endorses or merely notes the retracted work; that requires reading each paper. Every pair below is independently verifiable from the two DOIs.

This is a statement of fact about citations, not an accusation about authors. Some fields — microRNA and oncology especially — have a high retraction base rate, so a review there will cite many now-retracted papers simply because the underlying literature is heavily contaminated. Retractions + expressions of concern only; corrections excluded. The figure is a conservative lower bound.

It tracks how contaminated a field’s literature is

The rate spans roughly a hundredfold by venue — not because some authors are careless and others careful, but because some fields’ literatures carry far more retracted work. Tier of journal matters: the paper-mill-exposed microRNA/oncology corner runs 4.0% on average, the prestige clinical journals 1.6%.

JournalSampledWith a flagged refRate
International Journal of Oncology1703822.4%
Oncology Reports2233013.5%
Molecular Medicine Reports489479.6%
Biomedicine & Pharmacotherapy2,4172279.4%
Aging870586.7%
Cancer Cell317175.4%
Cancer Research500255.0%
Frontiers in Immunology2,4671094.4%
The Lancet49440.8%
Science49030.6%
JAMA49130.6%
Nature Genetics49620.4%
New England Journal of Medicine50010.2%

Representative rows from the panel (journals with ≥150 papers sampled). Full table available on request.

Concentration: papers with many retracted references

42 papers cite three or more retracted/EoC references; 5 cite five or more. The worst single case is a 2025 microRNA review in Biochemical and Biophysical Research Communications that cites 8 retracted articles — all 8 already retracted before it was published. Almost all of these heavy citers are review articles in microRNA and oncology journals, where the underlying literature is heavily retracted. And the heavy citers are not independent: 23 clusters lean on the same retracted papers (the tightest, 4 papers sharing 15 retracted references) — the citation-contamination pattern worth inspecting. The full named list of the 42 heavy citers is available to researchers and journalists on request.

In top journals too

The prestige journals sit near the floor, but not at zero. Each row is a recent paper in a major journal whose reference list includes a paper retracted years earlier. Stated as fact, with both DOIs so you can check; we did not assess whether each citation relies on or flags the retracted work.

Neuron
10.1016/j.neuron.2024.01.028
cites 10.1016/s0925-4773(02)00292-7, retracted 2002: “Efficient gene modulation in mouse epiblast using a Sox2Cre transgenic mouse strain
Developmental Cell
10.1016/j.devcel.2024.09.017
cites 10.1128/mcb.23.12.4283-4294.2003, retracted 2006: “Effects of Rho Kinase and Actin Stress Fibers on Sustained Extracellular Signal-Regulated Kinase Activity
Proceedings of the National Academy of Sciences
10.1073/pnas.2415530121
cites 10.1182/blood-2004-03-0954, retracted 2009: “Promoter hypermethylation of cancer-related genes: a strong independent prognostic factor
Molecular Cell
10.1016/j.molcel.2024.06.005
cites 10.1073/pnas.192569699, retracted 2009: “Estrogen receptor-interacting protein that modulates its nongenomic activity
JAMA
10.1001/jama.2025.1883
cites 10.3324/haematol.2010.021923, retracted 2010: “Similar Hypercoagulable State and Thrombosis Risk in Type I and Type III Protein S-deficient Individuals
Nature Medicine
10.1038/s41591-024-03068-6
cites 10.1016/j.exer.2010.03.008, retracted 2011: “Mitochondria: Their role in ganglion cell death and survival in primary open angle glaucoma
Cell Reports
10.1016/j.celrep.2025.116692
cites 10.1126/science.1095001, retracted 2014: “TAF1 Activates Transcription by Phosphorylation of Serine 33 in Histone H2B
Journal of Biological Chemistry
10.1016/j.jbc.2024.107773
cites 10.1073/pnas.1216844110, retracted 2014: “PINK1 regulates histone H3 trimethylation and gene expression

The papers that won’t die

A handful of retracted papers account for a disproportionate share of the citations. These retracted papers were each cited by many recent papers in our panel. (The last is a “lost citation” rather than a catchable one — it was retracted after the citing papers appeared.)

Method

  • Sample: 38,350 recent (20242025) journal-articles via the Crossref /works API by ISSN, with at least 10 resolvable reference DOIs. A two-tier panel: prestige journals (the accountability question) plus higher-volume biomedical venues where concentration is more likely to appear.
  • Check: every reference DOI matched against the live Retraction Watch dataset (retraction notices carry dates). Retractions + expressions of concern only — corrections excluded. Retraction dates are joined to each citing paper’s publication date to classify whether the reference was already retracted (catchable).
  • Discipline: citing papers that are themselves retraction/EoC/correction notices are excluded (they are not heavy citers). We report citation presence, not intent. “Cites N already-retracted papers” is a fact; we never infer misconduct.
  • Lower bound: Retraction-Watch-based, so Crossref-only retractions not yet in the dataset are not counted; hijacked-journal and DOAJ-withdrawn checks (which the production engine also applies) are not included here.

Summary data and the full named heavy-citer list are available to researchers and journalists on request (erik@manusights.com). Please credit the Retraction Watch Database (ISSN 2692-4579) and Crossref. Companion prevalence study: how often papers cite retracted research.

Check your own bibliography

Paste your reference list into the free Manusights reference-integrity checker — no login, no email gate — and it flags retracted, expression-of-concern, corrected, and hijacked-journal references in one pass.

Preparing a manuscript? Our $39 manuscript readiness scan runs the full citation-integrity suite across your entire reference list alongside methodology, statistics, and journal-fit checks. The preview is free.