Is The Process To Retract Fabricated Randomised Clinical Trials In Reproductive Medicine Working Sufficiently?

Abstract Study question How do journal editors and publishers respond on randomised clinical trials in reproductive medicine that have been identified as fabricated? Summary answer Despite clear proof of fabrication, only a small minority of fabricated RCTs is retracted within 12 months. What is known already Randomised controlled trials (RCTs) are recognised as scientific investigations that have the least potential for bias and are therefore widely used to direct clinical practice. The validity of data in RCTs matters to the accountability of medical practice and the wellbeing of patients. Detection of integrity problems and subsequent action is therefore of imminent importance. Across all fields of medicine, it takes on average 4 years for papers labelled with research misconduct to be retracted. While this is partially explained by the time needed to detect the misconduct, the process of investigation and retraction is also slow and bureaucratic. Study design, size, duration We studied the articles of 4 authors who have published 52 clearly fabricated RCTs in obstetrics/gynaecology. Data fabrication was clear from duplicate baseline and outcome tables in studies on different interventions done in different patients in different periods. The duplications could be from the author themselves, or from other articles. Our findings were published in the public domain for 3 of the 4 authors, with an article on the fourth author being submitted. Participants/materials, setting, methods After detection of the fabrication, we approached authors and their institutes for an explanation. As a satisfying explanation was not given, we notified the editors of the involved 14 journals in February 2020. The universities where two of the authors were awarded a PhD were informed in August 2020. Here we compare the journals’ response to Committee on Publication Ethics (COPE)-guidelines, we report the percentage of retracted papers and other responses of editors and publishers. Main results and the role of chance Two articles had already been retracted prior to our notification. Twelve months after we had notified the editors, 4/50 (8%) (1 journal) of the articles had been retracted, 3 (6%) (2 journals) were formally investigating with notification on their website, 6/50 (8%) (3 journals) were informally investigating (without visible notification), 3/50 (6%) (1 journal) had made an expression concern and stated “caution that clinical practice or guidelines should not be based on this report” without formal retraction, and one (2%) had investigated original data and cleared it (although numerous data were identical to a study published 10 years earlier). For the other 33 articles (11 journals) no visible action had been taken. None of the journals provided feedback to the whistle-blower as required by COPE. The University of Utrecht and the Vrije Universiteit Brussel had not taken any action against the awarded fraudulent PhDs. Among the reactions of editors were the statements “I have been in the business long enough; It exists in all specialties and in every country”, “we receive 80 submissions a week, I am too busy to respond on this” and “we did still not get an answer about the result of the investigation by the local court”. Limitations, reasons for caution This was a sample of four authors who had published 52 articles. There might be investigations ongoing that are not visible for the outside world. Wider implications of the findings Retraction of fabricated studies is seldom happening, and a majority of journals is not following COPE. This not only puts patients at risk, but it also lets whistle-blowers down and it jeopardises the trustworthiness of research. COPE-regulations consider the interests of authors and publishers, but not the interests of patients. Trial registration number Not applicable