As mentioned, the previous sections were written prior to and then updated in 2021 with the intention of them being published, but we didn’t get around to it. In any case, it’s more exasperating than usual trying to get work in this area published. What has happened since then, now that 11 years have passed since we first notified JAMA?
Retractions have continued to trickle in, with 16 over the past 3 years, but they will stop altogether soon.
We have continued our academic approaches. We have published further statistical techniques to evaluate publication integrity [29, 30], and collated them together into a freely available package for the statistical program R. A review of these techniques has just been published [31].
We have also continued to highlight our experiences in raising concerns and the problems with the current system, perhaps so that even if we are not able to bring about change, we are able to “bear witness” to the current problems. We assessed retraction and expression of concern notices for more than 115 publications by this group [32]. The median time from raising a concern to an editorial notice was 22 months. Somewhat surprisingly this time did not decrease as retractions from the research group accumulated. Instead, in the face of more and more evidence of compromised integrity, the timelines for retractions lengthened. The retraction notices were desultory: <10% of the concerns raised were mentioned in the notices, and little of the information that is recommended to be included in a retraction notice was included in any of the notices.
We also assessed how expressions of concerns (EoCs) were used for this body of publications and compared it with another large group of RCTs we have raised concerns about [33]. It won’t be any surprise to read that they were slow (median time to publication was 10.4 months), and they were inconsistently used. There was little difference in the time to publish an EoC and the time to retract a publication without a preceding EoC. After an EoC was published, the resulting retraction (in 30% of papers with an EoC) took another 5 months. So EoC then retraction is considerably slower than retraction alone. For the remaining 70% of EoCs, none were lifted so that the EoCs, which are supposed to be used as a “placeholder” while investigations are undertaken and definitive decisions made, remained in place for more than 18 months and counting. It seems very likely that these journals and publishers do not intend to ever make definitive decisions. Among different journals and publishers, use of EoCs, and their visibility, varied dramatically. The results do question what value an EoC, as currently applied, adds – they take a long time to be published, are not used as “placeholders”, so why use them at all? We think they are a useful tool but only if applied quickly after valid concerns are raised, and not left to linger for years later.
We did another RCT, randomising 88 systematic review authors and journal editors that had cited retracted research in their review [34] to receive emails pointing out the citation(s) to the retracted paper(s). Consistent with our previous experience, about half of the contacted authors replied, but after a year only 1 in 10 publications had published a notification about the issue, even though we judged that about half the systematic reviews’ findings would change if the retracted publication was removed from the review, and in most cases, the change would be substantial.
We have written a substantial review of the current situation around publication integrity [35], and hope to shortly publish a critique of COPE where we make constructive suggestions for improving its guidance and oversight role. We also have a couple of other ideas for projects, including systematically describing the non-response of journals and publishers to our emails.
But apart from these few things, the saga has largely run its course and remains in basically the same unsatisfactory state it has been since 2013. Of >300 publications by the lead authors, fewer than 50% have been retracted, and most of the remaining publications have no public notification that strong concerns exist about the work by these authors. Only one of the co-authors (K Satoh, the President of Hirosaki University at the time) has ever been publicly sanctioned for their part in the publications. He said that his role in the publication was limited to correcting the English and the sanction received was to give up 10% of his salary for 3 months. In other words, he didn’t contribute meaningfully to the paper and didn’t meet criteria to be an author. But his statement conflicted with the contributor statements in some journals which stated that he had much greater involvement (see Section 2). We suspect that all of the publications are similarly affected by authorship misconduct, where authors who had little or no involvement with the publication were ‘gifted’ authorship- but it is unclear how much knowledge individual co-authors had of the situation. To our knowledge, no other co-author has made any public statements or had any negative consequences arising, even though there are 50 co-authors on these retracted publications, include 1 co-author on >50 retracted publications, 5 on >25 publications, 11 on >10 publications, and 19 on ≥5 publications. The co-authors must have been aware of (at least some of) the publications and the problems with them, yet chose to take public authorship and, presumably, the benefits that arose from that, but have shied away from taking any responsibility for correcting the literature once the publication integrity was questioned and shown to be compromised.
It is difficult to see any positive developments arising in the future from this case, and that seems a fitting way to end this piece.
Concerns about publication integrity 