Retracted papers originating from paper mills: a cross-sectional analysis of references and citations.

ABSTRACT

OBJECTIVES: The aims of this study are (1) to analyze the references cited by retracted papers originated from paper mills; (2) to analyze the citations received by retracted papers originated from paper mills; and (3) to analyze the potential relationships existing between paper mill papers and their references and their citations. STUDY DESIGN AND SETTING: This study was a cross-sectional study. All original papers retracted in 2022 identified as having originated from paper mills and had been published at least 12 months before their retraction (hereinafter "source-retracted papers") were included. The Retraction Watch database was used to identify the source-retracted papers and Web of Science was used to identify the references contained within them and the citations received by them. We described the characteristics of the papers and journals. Additionally, 2 networks of source-retracted papers mutually interconnected via their citations and references were built 1 with only retracted references and retracted citations and the other with all references and citations (retracted or unretracted). RESULTS: A total of 416 paper mill papers retracted in 2022 (sourced retracted papers) were identified, with a median of 1247 (interquartilic range, 907.8-1673.5) days between publication and retraction. Of all authors identified, 92.3% were affiliated with Chinese institutions. There were 14,411 references contained in the source-retracted papers and 8479 citations received by them; the median number of references and citations was 35 (29-40) and 16 (9-25), respectively. In total, 473 references and citations had also been retracted for being paper mill papers. Among the 416 sourced-retracted papers, 169 (41.9%) and 178 (42.8%) were referenced or were cited by at least another retracted paper, the majority of which also originated from paper mills. The first network analysis, which included source-retracted papers along with their retracted references and citations, found 3 clusters of 53, 48, and 44 retracted papers that were mutually interconnected. The second network analysis, with all references and citations (retracted or unretracted) identified a large cluster of 2530 interconnected papers. CONCLUSION: Retracted papers originating from paper mills frequently reference and are cited by papers that are later retracted for having originated from paper mills, displaying inter-relationships. Detecting these inter-relationships can serve as an indicator for identifying potentially fraudulent publications.