Citations of Retracted Ophthalmology Papers Persist, Study Warns

A substantial number of retracted journal articles in ophthalmology continued to be cited in the literature after retraction, risking continued dissemination of “false or misleading findings,” authors of a retrospective review concluded.

Overall, 80% of 356 retracted articles had at least one post-retraction citation, and post-retraction citations accounted for almost half of all citations for the articles. The numbers suggest post-retraction citation occurs disproportionately in ophthalmology as compared with other medical specialties, reported Danny Mammo, MD, of the Cleveland Clinic, and colleagues in JAMA Ophthalmology.

“For example, cardiology literature shows approximately 19% of citations occurring post-retraction, whereas ophthalmology demonstrated a proportion of 47.9% in this study,” the authors stated. “Such disparities may reflect inconsistently flagged retractions or limited awareness within the field. Ophthalmology comprises many smaller subspecialty niches and journals, where retraction notices may be less visible.”

“Efforts should be made to prevent the continued citation of retracted works to preserve the integrity of ophthalmic research,” they added.

Potential actions to prevent post-retraction citation include watermarking retracted documents, clearer labeling on PubMed and other databases, and use of artificial intelligence (AI)-assisted reference checks, although AI assistance might have its own limitations, such as hallucinations, the authors concluded.

The findings appear to be consistent with other studies of post-retraction citation in the ophthalmology literature, said Ivan Oransky, MD, co-founder of Retraction Watch, a database for reporting and tracking retracted articles in scientific literature. At least three studies on the topic have been published in the past 5 years.

An analysis of the Retraction Watch database identified 121 retractions in ophthalmology from 1983 to 2021. Focusing primarily on characteristics of the retracted articles, the authors noted that in many cases the reason for retraction made no mention of fraudulent behavior, “highlighting that not all forms of retraction are due to scientific misdemeanor.”

A study spanning the years 1966 to 2023 identified 151 retracted ophthalmology articles. Investigators found that retractions increased substantially after 2020, but the frequency of post-retraction citation decreased significantly over time (P<0.001).

A review of ophthalmology literature for 1994 to 2019 identified 83 retractions that could be analyzed for details associated with the retraction. Retractions occurred more often after 2010, and the retracted articles had a median of two citations.

Oransky found it curious that the current article made no mention of the previous studies.

“It’s a reasonable question,” he told MedPage Today. “To be honest, I would probably blame the peer reviewers and the editor [for the oversight]. Also, their conclusions are a little bit off. If they had looked at the previous literature, not just retractions in ophthalmology, but citation behavior after retraction in the biomedical literature, they would see that the rate of post-retraction citation that they found is actually lower than what it really is in the vast majority of the literature. Well in excess of 90% of citations to retracted papers don’t mention that [the papers] have been retracted.”

With regard to possible solutions to the problem of post-retraction citation, most of the authors’ suggestions have been proposed for at least the past 10 years, Oransky continued.

“A much better solution is reference managers, like EndNote, Zotero, things like that,” he said.

Information on retractions has been incorporated into large databases that are freely available to the research community, such as Crossref. Notices about retractions are flagged and sent to subscribers of services such as EndNote and Zotero.

Mammo and colleagues reported findings from a query of the Crossref database, seeking retracted articles in ophthalmology from 1980 to 2024. The search and subsequent review of articles identified 356 retractions, 86% of which were indexed in PubMed and 2.5% open access.

The articles were cited 5,850 times, and 2,798 citations (47.8%) occurred after retraction. The median number of citations per retracted article was six, and the median number of post-retraction citations was three. The median time to retraction was 18 months (range of 10 to 37), and 190 (53.4%) retractions occurred from 2019 to 2024.

Retina and vitreous articles accounted for 174 (49%) of the retractions, followed by anterior segment (86, 24%), and oncology or pathology (59, 17%). Basic science accounted for 46.6% of the retractions and clinical trials for 13.2%. The 356 retracted articles had a total of 1,666 co-authors, 232 (13.9%) of whom were co-authors for more than one retracted article.

The most common reasons for retraction were methodology (24.0%), publication issues (24.0%), fraud/data manipulation (21.0%), plagiarism (12.6%), and paper mills (6.6%).