Questionable practices in data and code sharing policy in high-profile medical journal and research

Abstract

AbstractBackgroundThe spurious and unavailable data/code sharing actions are crashing open medical sciences. In this study, we aimed to illustrate how high-profile medical journals are practically carried out their sharing policies and what questionable practices regarding data/code sharing are conducted by authors.MethodsIn this study, we appraised the policy on data/code availability of high-profile medical journals ranked at Q1 according to Clarivate Journal Citation Report (JCR 2021). Furthermore, we recruited post-publications published by four leading medical journals (i.e., The BMJ, JAMA, NEJM and The Lancet) from the issuing of data/code availability policy to December 2022 for the questionable practices in data/code sharing. The appraisal of papers was conducted by the Data/code Availability Statement Practice Evaluation Tool (DANCE), developed by systematically integrating mainstreaming open data/code guidelines.FindingsWe found that less than one-tenth journals (9.1%) mandated authors to share data/code, with an available statement. Among these journals, 70.6% (61.2%) did not consider censoring (restricting) spurious/invalid data/code sharing in publications. Furthermore, though journal impact factor could predict policy stringency on “offering availability statements” (p< .001), it failed to predict ones in “sharing data/code” (p= .73). For publications, even in leading medical journals (i.e., The BMJ, JAMA, NEJM and The Lancet), only 0.5% of the papers (16/3,191) fully complied with their public sharing statements for reaching reproducibility. Lack of availability statement, declining data/code sharing without reasons, and invalid repositories were leading questionable practices conducted by authors.InterpretationWe clarified specific questionable actions of implementing and practicing the sharing policy both in journal and papers, which should be addressed not only by the supportive publication ecosystem but also by crediting authors for taking responsibility and maintaining scientific integrity in data/code sharing.FundingNo funding.