70 Research misconduct can promote overtreatment. A multi-institutional case study from japan, with implications for osteoporosis management with vitamin K

Objective To examine the impact of delays in investigating and retracting randomised controlled trial (RCT) reports informing guidelines for osteoporosis treatment with vitamin K. Methods In 2007, concerns were raised with a leading medical journal about implausible results of 17 RCTs from a Japanese group from three universities, including 3 RCTs in a 2006 systematic review of vitamin K to improve bone mineral density and prevent fractures. No action was taken by the journal. In 2013, our systematic review of 33 RCT reports by this group indicated widespread concerns, but was not published until 3.5 y later after several journal submissions. By April 2018, 21 of the 33 RCTs had been retracted, for reasons including data fabrication, concerns with data integrity, honorary authorship and self-plagiarism. To investigate the impact of the RCTs on systematic reviews and guidelines, we undertook citation searching in Scopus and Web of Science. Here we focus on the three RCT reports of vitamin K supplementation, and their subsequent impact. Results The 2006 systematic review of vitamin K included only 7 small RCTs with fracture outcomes, including these 3 RCT reports, all published by one journal. To date 2 of the 3 RCTs have been retracted. The 3 RCTs had been cited 157 times by April 2018. The systematic review has been cited 190 times. Particularly of note is that the review’s meta-analysis plot for fracture outcomes was reproduced as the main evidence to support vitamin K for the prevention of fractures for the 2011 Japanese osteoporosis guidelines (replaced in 2015). The systematic review originally presented an odds ratio of 0.23 (95% CI: 0.12 to 0.47) for hip fracture, but a recent correction, omitting these 3 trial reports, amends this to 0.30 (95% CI: 0.05 to 1.74). This is still mentioned as a ‘large effect’ even though based on only 3 events. Two subsequent trials in Japan with 6361 participants have failed to demonstrate that vitamin K prevents fractures. Conclusion 11 years have passed since concerns were raised about these RCT reports, which strongly influenced 2011 Japanese guidelines. Much prompter investigation of concerns about research misconduct is needed. We have found that journals and publishers are averse to flagging up expressions of concern, slow to investigate concerns, unwilling to accept publication of detailed statistical investigations, or retract articles. All of which spread the pernicious influence of research misconduct leading to inappropriate or ineffective treatment.