Our journal published a manuscript as part of an editors’ forum which, as an invited forum paper, received reviewer feedback but did not follow our usual double-blind peer review standard for regular submissions (the reviewers were aware of the author’s identity but the author did not know the identity of the reviewers).
Following the publication of this article, the editor-in-chief received correspondence from a reader (hereon referred to as the “scholar”), who indicated he (used generically for simplicity without implication of the author’s or scholar’s gender) had contacted the original author directly and was concerned about the data used in the paper. The author also emailed the journal to advise us of his correspondence with the scholar, which had made him aware of oversights in data collection.
The original author provided a revised paper and drafted a corrigendum which was edited and published as an erratum in a subsequent issue of the journal. The scholar was given the opportunity to read the erratum prior to publication. After publication of the erratum, the scholar continued to vocalise his concern that it still misrepresented the author’s data and offered readers misleading conclusions, and he remained frustrated over the original author’s unwillingness to share the dataset despite his request for it.
The scholar complained to the editor-in-chief of the journal that the author ignored his many email messages and asked the editor-in-chief to pass on his concerns. After consulting with another editor of the journal and deciding that it was not the journal’s place to be involved in such disputes, the editor-in-chief talked with the author by phone and asked the author to correspond with the scholar directly about this issue. The editor-in-chief also told the scholar that the journal could not be the go-between in passing messages from one to the other but would consider a written response to the published erratum so that the debate would be in the public domain. They were encouraged to talk to each other directly.
We received the response article from the concerned scholar outlining his points of contention, which was intended for publication with the understanding that the original author would be given an opportunity to reply if he chose (but the scholar would not be offered space for a further rejoinder). We provided the original author with the scholar’s response article, and he prepared a reply. The scholar was then given an opportunity to read the original author’s reply to point out any factual errors in need of correction, and he again felt the reply article would be misleading to readers as well as misrepresentative of his own claims. He further raised concerns about a flawed review system, not only in regards to the original article not being reviewed double-blind, but also to the editing process for his versus the original author’s latest reply articles (both were edited—primarily proofreading in nature—by the editor-in-chief and another editor).
At this time, the editor-in-chief has contacted both the scholar and original author simultaneously in the hope of encouraging them to work together with shared data to produce a new paper (if, indeed, a re-analysis leads to new results). Should they not choose to pursue this, the journal intends to publish the scholar’s reply and original author’s response, likely alongside a final word from the editor-in-chief. Any response or outcome along these lines is still pending.
How might editors ensure the integrity of data beyond the safeguards built into the normal manuscript review process? What role should the editor play when authors refuse to share their data? More generally, how should editors address concerns of readers who remain aggrieved by authors’ responses to readers’ commentary on an erratum provoked by questions raised by said readers in the first instance?
Following discussion with the officers of COPE, the summary of advice is as follows:
1. Double blind review is not always possible. For example, in small fields, most people can easily know the author of a piece of work. Also, there is no evidence that double-blind review is better than single-blind or non-blind reviews. 2. The editor cannot possibly double check the accuracy of the data in every paper and thus it is a good policy for the editor to allow readers a forum to voice their concerns about the published work. 3. The editor has the right to decide what gets published in the journal in light of the mission of the journal, its style, and the space constraints. 4. There is a point where a critique/concern becomes an opinion. The editor has the right to decide when the disputants have had the chance to voice their views and when it has reached the point for the readers to judge the merit of the grievance or responses from the published comments. 5. I have given the reader the courtesy and opportunity to voice his/her view.
Does the Forum agree with the officers’ advice and is there anything more I should do?
The Forum agreed with the officers’ advice. All agreed that the effort on the part of the editor to publish the debate was sufficient. The editor raised the issue of data sharing and the Forum were told that at the COPE council meeting, council had discussed this issue and were considering drafting a discussion paper to outline the issues. The editor was told that some journals (eg the BMJ) now publish a data sharing statement on all papers, which states whether or not the authors are prepared to share their data, with links to the data as appropriate.
One of our journals has published several articles describing use of a particular cell line X, which belongs to company Y. The authors included employees of company Y. A reader at a university, Dr Z, wished to gain access to cell line X, and requested it from company Y. He was informed by the director of science of company Y that ‘...it is Y’s intention to keep control of the integrity of the X materials as much as possible, thereby taking care of our licensees’ interests. Hence, Y does not widely disseminate the X material to academic institutions.’
The researcher complained to the editor-in-chief of the journal that this violated the general principle that materials described should be made available for further work, and for independent testing of the reproducibility of the published results.
The journal requires authors to accept that “By publishing in [the journal], authors agree that any viruses, plasmids and living materials, such as cell lines or bacterial strains, that are newly described [my italics] within the article are available without unnecessary delay and at a reasonable cost to members of the scientific community for non-commercial purposes”. As cell line X was not first described in our journal, the authors have clearly not broken the letter of the undertaking, but there is a more generic principle to ask the committee to discuss.
A search on Google Scholar yields over 800 articles that describe cell line X, in a number of different journals. Some of these have less qualified author undertakings, stating that reasonable access to reagents should be available for non-commercial work. However, examination of a sample of the 800 papers showed that all had authors from commercial companies—presumably licensees of the X cell technology—and that none were from purely academic institutions. This was confirmed by a further letter from the director of science of company Y, in which he states that “Licensees and potential licensees have to meet a number of requirements that can guarantee the correct use of the X cell technology from a technical as well as a legal perspective. Based on this consideration one of the policies company Y applies in licensing out the X cell technology is that we exclude academic institutions”. In response to the researcher’s point that the technology is therefore not accessible to independent scientific validation, the director of research of company Y wrote “In view of the fact that our X cell technology is fully approved by the FDA and that the technology is in use for preclinical and clinical research by a substantial number of licensees, we are confident that independent scientific validation is not an argument to share our X cell line with Dr Z”.
The editor-in-chief may wish to consider his response the next time a paper using cell line X is submitted to the journal.
The Forum were unanimous in their conclusion that this was clearly unacceptable behaviour on the part of the authors. If a journal has a condition of publication that reagents should be available for further work, then the journal has every right to see this implemented. The editor in this case told the Forum that his journal’s instructions to authors state only that “newly described” cell lines or bacterial strains should be available to the scientific community. The Forum suggested changing the instructions to authors to be more specific. The Forum also suggested rejecting future papers from these authors. The advice was to encourage the complainant to write a letter which the editor should publish in the journal, along with an accompanying editorial on this issue. Other advice was to consider contacting some of the other 800 titles informing them of the situation.
In April 2007, an original scientific article was published on line (ahead of print—it is now published in print, September 2007). In July 2007, the editors received the following request from a scientist who read this article: "Since I am interested in this subject and I already work with it, I need to know some technical information from the authors. I have called the group five times and wrote them several emails, but unfortunately there was no response. I have decided to write to the editor to help me get the information I need".
What should be the responsibility of the editors to try to influence the authors to comply? Note that the Instructions to Authors include a statement that the journal adheres to the COPE guidelines.
The Editors have already replied as follows:
"We agree that manuscripts should provide all information regarding the technical details to allow repetition of the scientific work. In addition, it is generally accepted scientific practice for investigators to share materials and reagents that are relevant to the experiments and are not publicly available to allow parallel work, once the research has been published. It is generally accepted that "publish-ahead-of-print" manuscripts are published work. However, the editors consider it beyond the scope of their activities to intervene in discussions or interaction between investigators regarding work published in the journal."
Is this response appropriate or should we have done more?
The Forum considered that the editor has a duty to write to the authors and remind them that they are obliged to provide technical details to any interested party. The editor might wish to try to establish whether the authors are just being uncooperative or if there is any more ominous reason as to why they are withholding the details. If the editor receives no response from the authors or the information is not available, then the scientist could be encouraged to write a letter to the journal with his complaint. In this way it will highlight the fact that the authors are not following good practice. The editor should ensure that the instructions to authors states that all data should be available to interested parties. COPE is generally not in favour of sanctions and would not recommend them in this situation
In response to the advice from COPE, the editor had made the following addition to the Instructions to Authors to make it clear what is expected: “Authors should fully describe their methods so they could be reproduced by others. Additional technical information, unique reagents, antisera, cell lines, and genetically modified animals necessary to replicate the work should be made be available to interested parties”.
Our journal has received a submission regarding clinical trial results. The authors wish to include the “raw data” as an appendix to the manuscript. The study was completed several years ago and was controversial at the time. The authors wish to publish the raw data to allow the public to view the findings and make their own decisions about the trial. We do not know the best way to handle this situation.
Does anyone already publish anonymised raw data in their journal and, if so, how does one handle the ethics of such a publication:
Do you assume that the original ethics approval covers this new publication?
Do you require the authors to go back and obtain ethics approval to publish raw data? If so, does the author have to go back to the original ethics committee?
Also, does one allow the authors to publish only “relevant” data or does one require all of the data generated from the trial to be published?
The discussion centred on the definition of anonymised data. If there are three or more patient identifiers, strictly speaking the data are considered pseudo-anonymised, as theoretically patients could be traced. The committee acknowledged that ethics committee approval should be sought. If this is not possible from the original ethics committee, the author’s current ethics committee should be contacted. Regarding the question of publishing only “relevant” data, the committee agreed that this decision could be made by the editor.
The data were never submitted by the authors, and the manuscript was not published.
In this case, an international organisation wished to study the use of various regimens for medical termination of pregnancy in a developing world setting where termination of pregnancy is not supported by the state. They have performed an audit and have obtained data which will be of considerable value in other similar settings around the world. However, they are not willing to state the identity or location of the clinics supplying the data because these clinics fear reprisals and may even be shut down. Moreover, their staff might be put at risk (see the recent editorial in the NEJM on this topic). They are prepared to tell us the location of the clinics provided we do not publish it.
Is it acceptable to publish the data in the interests of women’s health without publishing the source of the data or indeed any evidence of local ethics approval for the study (as the clinics are not officially sanctioned there is no formal institutionally affiliated ethics committee)?
The general consensus of the committee was yes, it is acceptable to publish the data without naming the identity or location of the clinics. The editor stressed that it would be possible to name the continent and the type of setting, but for reasons of security the identity of the setting could not be named. The committee felt that such a decision could be legitimately made by the editor. The situation could also be explained to referees if needed. The editor and/or author could act as guarantor for the validity of the data.
The paper has been accepted and publication will be going ahead. The authors of the paper are willing to tell the editor the names and addresses of the centres involved with the research, but the editor has declined so that there is no risk of giving away their identity inadvertently. However, the authors have agreed to tell the journal in the future if asked.