An Elsevier journal has denied the efforts of a group of researchers — well, most of them, anyway — to reverse a retraction after having agreed to the move in the first place.
The dispute centers on a 2018 paper in Preventive Medicine Reports titled “Association between low-testosterone and kidney stones in US men: The national health and nutrition examination survey 2011–2012” — which, as the title implies, found that:
men with low-testosterone had 41% lower odds of KS as compared to men without low-testosterone after multivariable adjustment (OR: 0.59, 95% CI 0.40-0.86).
After publication, Emre Yucel, the first author of the article and the manager of global health economics at Amgen, notified the journal that the paper was flawed.
Here’s the notice, which we encourage you to read in its entirety:
Following publication of the aforementioned article, the first author requested that the paper be retracted and disclosed to the editors information that prompted the following retraction notice to be published in ScienceDirect alongside the journal record for the article. All authors were informed of the retraction and retraction notice in advance of publication.
That notice reads:
“This article has been retracted at the request of the authors. This study was based on data from a large US survey that used weights to permit overrepresentation of certain population groups. After the paper had been accepted and appeared online, the authors disclosed to the journal that they did not appropriately incorporate the sampling weights at the statistical analysis stage, which led to biased estimates of effect. They regret the inconvenience to reviewers and readers.”
Except, well, those regrets were evidently soon regretted.The notice continues:
Subsequent to the retraction, the senior co-authors of the paper contacted the editor and alleged that the first author had acted in haste and had reported scientific errors that did not exist.
In email exchanges involving all authors, they demonstrated to the editors that the statistical analysis had appropriately incorporated the weights used in the survey design, and thus they stood by the technical correctness of the results and interpretation. In consequence, they requested that the paper be restored without the associated retraction notice and label. As evidence of the soundness of their statistical analysis the authors provided Stata scripts and corresponding output.
In a subsequent lengthy consultation with the publisher’s ethical and legal teams, we reached the conclusion that the paper cannot be ‘unretracted’ from the journal’s record. Despite the authors’ volte-face regarding the validity of their paper, we maintain the decision to keep this paper as retracted. However, we encourage the authors to submit their paper elsewhere to reconstruct the public record of their study de novo, without the regrettable circumstances that followed the publication of their article in Preventive Medicine Reports.
Yucel told us that he conducted the work as a graduate student at the University of Texas, where he received a doctorate in public health. After publication, he found what he believed at the time to be an error that was preventing him from replicating the results.
He said he attempted to contact his co-authors, but didn’t hear back. So he went to the journal — which told him he needed to retract the article.
“[T]wo wrongs don’t make a retraction unretracted”
However, once the retraction notice appeared, Yucel said he heard from the biostatistician on the research team who told him that the analysis had in fact been correct. But the journal would not consider reversing its decision.
Yucel attributes the episode to his inherent aversion to conflict and inexperience with scholarly publishing:
I had nobody to wise me up. I had no experience. … I shot at my own foot with my own gun. I tried to do the right thing. But two wrongs don’t make a retraction unretracted.
He added that he:
learned from this episode to be more careful in dealing with manuscript submissions and retractions or withdrawals, and [I recommend] all authors and investigators to be very patient and slow in reaction, carefully thinking over and over in the time of manuscript submission.
Although PMS and Elsevier might look askance at volte faces, we’ve seen at least one case in which a journal published by Cambridge University agreed to un-retract a retraction.
Like Retraction Watch? You can make a tax-deductible contribution to support our work, follow us on Twitter, like us on Facebook, add us to your RSS reader, or subscribe to our daily digest. If you find a retraction that’s not in our database, you can let us know here. For comments or feedback, email us at [email protected].
For once, a case where an Expression of Concern would have been more appropriate.
I get that a journal can’t just decide “well, the author(s) wanted to retract it, now they want to un-retract it, let’s do so” without looking at the concerns upon which the original retraction request was based, nor do I think they should.
But if the reason for those concerns ends up demonstrably proven to not have existed after all, then that should in my opinion be cause for an un-retraction. Perhaps not of the kind the authors would have preferred, without any associated label or note, but un-retracting it with an EoC makes more sense to me than simply leaving a paper retracted when the issue warranting the retraction turns out not to exist.
All I see in this decision is yet *another* reason for authors to be hesitant requesting retraction of their paper if they believe to have found issues in it after publication.