Editorial Withdrawal of an Article Posted on the Journal Web Site

1. In March 2006, a paper was accepted and published on a journal’s Web site in the articles-in-press section. This paper compared the results of a pivotal population-based randomised clinical trial with official cancer and mortality statistics in the same region and found, with reasonable assumptions, that it was highly likely that many cancers and deaths were missing from the trial report.

2. On 29 March, the editor informed the authors that the journal had received comments from a number of sources regarding some of the claims made in the article and that as a result of these comments the article had been removed from the Web site pending further discussion and clarification. The editor did not forward to the authors the comments he had received,  but asked for some clarifications.

3. On 7 April, the authors addressed the comments from the editor in a resubmission.

4. On 25 May, the editor informed the authors that his decision to withdraw their paper was final. The editor wrote that his decision was made after release of new information concerning the randomization process and the trial’s opening and closing dates and new anonymous peer reviews that seemed to originate with people closely connected with the trial and with a vested interest in the area.

5. The authors firmly believe they can respond to the new peer reviews and that the trial in question is still likely to be problematic, even after receipt of the new information about the trial that seems to have been unpublished, and that, furthermore, seems to be in contrast to published information on this trial.

6. On 29 May, only four days after the second letter from the editor, the authors note that their paper is officially listed as a withdrawn paper in PubMed. The authors inform the editor
that they will have to publish their results elsewhere and ask the editor to forward the comments he has not yet forwarded to the authors.

7. On 2 June, the authors appeal to the editor to reverse his decision and write that if they have not heard from him before 9 June, they will assume that he will not reconsider his decision.

8. The authors did not hear from the editor.

I believe it is a very serious step to withdraw an already published paper, and such a step is usually only taken after due process and usually only if scientific misconduct has been established. I also believe there was no due process in this case; in fact, the whistle-blowers ended being the victims, as has happened so often, and the people who are trying to protect this trial from fair criticism can now say that the authors’ paper was so problematic that it had to be withdrawn by the editor. Even worse: Withdrawn papers are vere rare and they invariably give the impression that the authors committed scientific misconduct.

Neither the journal, nor its editor, are members of WAME and there seems to be no ombudsman or ethical committee at the journal.

What should the authors do? Should they submit the paper again, to another journal, describing in the background section of the paper the process—or rather, lack of process—that took place with the journal that retracted the research? I think so, as cases of misconduct should be exposed.

Peter Gøtzsche
Director, The Nordic Cochrane Centre

The proposed decision—submit article to another journal with description of the article history in the background section—is the right one. To name the journal and the article and the trial right now will be of interest. Not only because the misconduct must be named, but because this story encourage me to think that the original trial was really a trouble, if it need to be covered in such a way.

The only reason not to disclose details is that it may reduce the probability of the publication.
 

Vasiliy V. Vlassov
Editor in Chief, MJMP

1. The article is published on the journal's Web site, not in the journal. One must verify whether it is mandatory for every article on the Web site to be necessarily published. It is not. Then the editor is within his right to withdraw the article. Although, when he does, he and his journal suffer reduced credibility with authors. For a disgruntled author will obviously squeal.

2. The authors can contest editorial decisions, can answer peer reviewers, old and new. And would surely want to, for they want publication. But if the editor decides to close communication, they have no option but to seek publication elsewhere.

3. As far as the propriety of the editorial decision goes, we are assuming it amounts to editorial misconduct. And other than scientific forces have played a role. It may be so, if any such evidence comes to light. It may also be the editor does not want to enter into any further communication, having once made up his mind.

4. The authors concerned cannot pursue this matter beyond a point. They must call it quits with this journal, find another, disclose what happened to the new journal, and expect better sense to prevail there.

5. Let this not be considered an justification for editorial highhandedness. Editors are tried for their acts of omissions and commissions in the long run. Their own credibility suffers with acts of impropriety. So does that of the journal owners who tolerate such editors. Hence, justice does get dispensed, though in more subtle ways. So editorial highhandedness, or misconduct, does not really go unpunished. Ever.

Ajai

Ajai has raised an interesting and important question, which is when publication is deemed to have taken place. I always assumed that publishing on the journal's Web site did constitute publication.

Does WAME have a consensus on this? If not, should it have one?

Tim Albert

For our journal, publication Online Early is the publication/embargo date. It is officially published as of that time.

Bill Tierney
Co-Editor-in-Chief, Journal of General Internal Medicine

One distinction I would make here, however, is that this was listed as "in press"—this implies to me "not yet officially published." Our journal has, on a rare occasion or two, pulled manuscripts out of page proofs (essentially the same thing) when an unforseen problem has arisen. I don't know if I see any real editorial misconduct here, any more than there would be "misconduct" in retracting a paper in print because it was later discovered that the authors engaged in research fraud. It is the right of editors to make such decisions.

This would make an interesting discussion, though, because authors who are so keen on getting online early face the very real possibility of this kind of peer-review turning up very real problems with their articles—something they perhaps didn't bank on in their eagerness to bolster their CVs.

However, once the "in press" notice is removed, I would say the article/research report would officially be part of the record (if the point of the Web site is to duplicate the print journal, or if there indeed is such an in-press to in-print transition online) and then the steps the editor would take would likely be different, if only for practical reasons: He or she would probably need to do a retraction linked to the article rather than simply remove the article from the site. But again, I don't see how any of this involves misconduct on the part of the editor. 

Lisa

An important distinction must be made here.

1. Publication on a Web site is publication all right. But it is 'electronic' publication. It usually predates a 'print' publication in those journals that post their material online before it is printed. While electronic publication amounts to publication, it is not the same as print publication. In most cases, matter in both remains the same. But in those where the electronic publication is modified/retracted for whatever reason, it doesn't.

2. There is another important difference. It is easier to modify/retract electronic publication by simply replacing it with new matter on the Web site. In the case of print publication, this is not possible. Hence a number of print publications, even while accepting the utility of electronic publication, are wary, and not unjustifiably so.

3. There are journals that post their matter online before print publication and even invite responses, out of which they reserve the right to publish some in their print edition. Here the print and electronic versions (as far as responses go) can be widely different.

4. Moreover, if and when they decide to correct/retract the electronically published matter in the light of fresh evidence before print publication (which evidence maybe supplied even by electronic letters), they may find themselves in a real embarrassment. What is electronically published may never see print publication at all.

This is both the strength, and weakness, of early online publication for essentially print publications. Strength, because it allows them to revise, review even accepted MS before printing. Even print responses earlier. Weakness, because it allows for meddling before printing by extra-scientific forces.

Ajai

It is disappointing that there were so few comments on the important issues raised by the disturbing case described by Peter Gotzsche—perhaps this was because other topics were being discussed on the list at the same time.

To summarise: A paper was accepted for publication and the pdf file posted on the journal's Web site under a label "in press". It was then unilaterally withdrawn by the journal by being removed from the Web site, and it was labelled as "withdrawn" on PubMed. These actions were seemingly based on comments received after the article appeared on the Web site, including the "release of new information" (ie, information not in the public domain when the article was written). 

I share the view of Tim Albert and others that once an article appears on a journal's Web site it should be considered published. There should not be any ambiguity about this. (And this suggests that the journal's label "in press" is an inappropriate synonym for "yet to appear in print".)

Once an article has been published, I don't consider that there can be any justification either to modify it or to remove it from the Web site, with the possible exception of cases of scientific misconduct. The published research record should not be modified retrospectively on the basis of post-publication peer review nor in relation to the publication of information not previously available. The correct place for concerns to be aired is in the journal's correspondence pages. (Likewise, factual or typesetting errors in a paper discovered after publication should be addressed by published corrections, not by changing the document.)

However, I know of a case where an article was published with a Web supplement, where the Web supplement was subsequently amended in the light of correspondence received by that journal. Such an action is not acceptable in my view.

(Conversely, I know another case where the authors detected an error in their paper after returning corrected proofs but before the pdf file was posted on the Web, yet the journal refused to allow them to change the text, preferring instead to post the 'incorrect' version and then issue a correction subsequently.)

Returning to the case described by Peter Gotzsche, as a consequence of the journal's action, this paper is apparently now labeled as withdrawn on PubMed. As Peter wrote, "Withdrawn papers are very rare and they invariably give the impression that the authors committed scientific misconduct." Without serious concerns about misconduct, can withdrawing a paper be justified?

In passing, I note that the paper's appearance on PubMed as 'withdrawn' surely implies that the NLM considers it to have been published, as otherwise surely it could not have been withdrawn!!

So, I see 3 key questions:

1. When is a paper considered to be published?
2. Once a paper is published can it be withdrawn by the journal? If so, under what circumstances?
3. Once a paper is published can it be replaced by a different version? If so, under what circumstances?

These issues are especially important in relation to online publication, for which there is the technical possibility of rewriting history.

There is much advice for authors on what is (un)acceptable behaviour, but arguably a lack of adequate guidance for editors. I feel that WAME could usefully make a clear statement regarding (1), and under (2) give guidance on what is acceptable. 

Doug

Answering Doug’s explicit questions:

1 When is a paper considered to be published? For our journal, when an article appears in our “online early” it is published. That is the embargo date.

2 Once a paper is published can it be withdrawn by the journal? If so, under what circumstances?  Articles are occasionally retracted, for example when the data were shown to be fraudulent. Medline/PubMed has a way of noting this in its citations. It is never appropriate to remove an article once published electronically. Someone may have done something in response to seeing the electronic publication that could not be justified should it disappear. The appropriate thing to do is to leave it there and publish a retraction notice (instead of removing it) or an erratum (instead of editing any inaccuracies found).

3 Once a paper is published can it be replaced by a different version? If so, under what circumstances?  This is never appropriate. See above comment.

Bill Tierney