Importance: Retraction is a tool that journals can use to deter research misconduct and alert their audience to erroneous content published in the journals. However, retracted articles may continue to damage science if they are still cited as legitimate articles.
Objective: To characterize patterns of postretraction citations, particularly in microRNA biomarker research, a field with one of the highest rates of retraction.
Evidence review: Retracted scientific articles on microRNAs were retrieved from PubMed, Web of Science, and Retraction Watch between database inception and July 17, 2021, according to preestablished search strategies. Control articles with characteristics in common with retracted articles (ie, published in the same journals in the same years and months and with the same number of authors) were matched and retrieved from PubMed. Citation metrics of retractions and control articles were collected from Web of Science. PubPeer was referenced to examine the public response or comments on included retractions. Data were analyzed from September 2021 through March 2023.
Findings: A total of 10 461 articles were analyzed, with 887 retractions and 9574 articles as controls. Among retracted articles, which were published from 1999 to 2021, there were 756 articles (85.23%) written by researchers affiliated with Chinese institutions. Retracted articles were cited 6327 times after retraction. Of 792 retracted articles that were cited, 621 articles (78.41%) were cited at least once after retraction and 238 articles (30.05%) were cited more often after retraction than before retraction. Overall citations (comprising citations before and after retraction) and postretraction citations accumulated over time (eg, the median [IQR] number of postretraction citations was 1 [1-2] and 23 [9-44] citations at the first 6 and 66 months, respectively, between retraction and citation retrieval). A random sample of 87 retracted articles (9.81%) recorded 478 citations after retraction, with 208 citations (43.51%) in articles published 12 months or longer after retraction. Of these citing articles, 19 articles (3.97%) mentioned the retractions. Compared with the control group of 1620 nonretracted articles, no significant differences were found in overall number of citations or citations after retraction. Among 478 articles citing retracted articles, 414 articles were found on PubMed and had matched control articles; these articles had higher odds of being subsequently retracted than 7954 matched control articles (odds ratio, 6.57; 95% CI, 3.39-12.72).
Conclusions and relevance: In this study, retraction was not associated with a reduction in citations of retracted articles, but articles that cited retracted publications had higher odds of later retraction. These findings suggest that journals may need to implement mechanisms for detection of postretraction citations.