Ben W. Mol,1,2 Jim Thornton,3 Wentao Li1
When data integrity concerns are raised about a published article, the Committee of Publications Ethics (COPE) recommends that editors investigate. It is unknown how effective this process is.
A prospective cohort study on published articles with concerns about data fabrication was performed. Concerns could be multiple per study and included implausible timelines, implausible effect sizes, discrepancies between publication and trial registration, copying of tables from other articles, plagiarism, authors having published fabricated studies elsewhere, and wrong statistics. Editors of the involved journals were contacted by email with a summary of the concerns for each article and asked the editor to investigate according to COPE recommendations. The email suggested that the editors request original data and included an offer to help investigate data sets if needed. The editors’ responses, final decisions, and reactions are reported herein.
Between March 2017 and May 2022, editors were contacted about 546 articles (356 randomized clinical trials [65%]) by 105 authors from 6 countries (11 in 2017-2019; 56 in 2020; 222 in 2021; and 257 in 2022). Articles were mainly from the field of obstetrics/gynecology but also from urology, pediatrics, and infectious disease (COVID-19) and were published in 74 different journals (1 to 55 concerns per journal) by 5 different publishers. A total of 271 concerns (79%) were answered with confirmation of receipt. Most editors respected confidentiality, but some editors copied the complainant while writing to authors and 1 published an Editorial naming the complainant. Some suggested a Letter to the Editor. Five editors (7%) shared original data provided by the authors with the complainant for a total of 18 data sets, 14 (78%) of which showed signs of possible fabrication, mostly repeated strings of numbers in the database. By June 2022, 55 of 546 investigations (10%) had been concluded; 47 concluded with a retraction (n = 31) or expression of concern (n = 16), of which 41 were published, with 6 planned retractions not yet retracted 9 months after the initial decision. In 8 investigations, editors stated there was insufficient evidence of wrongdoing. In 1 case, the journal limited its investigation to the data set that was provided in the review process and did not consider a problematic data set provided by an author who stepped down after publication. In another case, the journal did not assess available original data themselves but relied on assessment by an expert designated by the author. There were large differences between journals in terms of more vs fewer problem articles, shared vs not providing insight in the assessment process, or being more vs less responsive.
Procedures recommended by COPE for investigating concerns about data integrity are not always effective, and many editors appear not to know how to handle them.
1. Committee of Publications Ethics. Homepage. Accessed July 1, 2022. https://publicationethics.org/
1Department of Obstetrics and Gynecology, Monash University, Clayton, Australia, email@example.com; 2Aberdeen Centre for Women’s Health Research, Institute of Applied Health Sciences, School of Medicine, Medical Sciences and Nutrition, University of Aberdeen, Aberdeen, UK; 3Department of Obstetrics and Gynecology, Nottingham University, Nottingham, UK
Conflict of Interest Disclosures
Ben W. Mol is supported by a National Health and Medical Research Council Investigator grant (GNT1176437) and has received consultancy fees from ObsEva and Merck and travel support from Merck.