Retractions, Expressions of Concern
5.4.5 Retractions, Expressions of Concern
After receiving confirmation from the author or authors and/or a report from the author’s institution or other agency indicating that fabrication, falsification, or plagiarism has occurred, the journal should promptly publish a retraction. Preferably this retraction will be a signed letter from the corresponding author and all coauthors. If none of the authors will agree to publish a signed retraction, the editor may request such a retraction from the investigating institution, or the editor may issue a retraction on behalf of the journal. In each case, the editor should inform the author(s) and institutional authority of the plan to publish a retraction. See Boxes 1 and 2 for examples of retraction notices.
A retraction should include a complete citation to the original article and should indicate the reason for retracting the original article. The retraction, whether a formal letter or notice, should be labeled as a “Retraction,” be listed in the table of contents, and be published in a prominent section of the journal on a numbered page in print versions and in a citable format in online versions so that it can be identified easily by indexers and included in bibliographic databases (see also 3.11.14, References, References to Print Journal, Retractions). The US National Library of Medicine will index the retraction as long as it clearly states that an article in question is being retracted or withdrawn, whether in whole or in part, and is signed by an author, the author’s legal counsel or institutional representative, or the journal editor.28 Online versions of journals and bibliographic databases should provide reciprocal links to and from the notice of retraction and the retracted article. Retractions should be made freely available and accessible on a journal’s website (ie, readers should not have to pay an access fee to see the retraction notice).27 A retracted article should be properly labeled or watermarked as retracted in online versions of journals and should not be removed from the online journal or archive. Such labeling may include the words “Retracted Article” or “This Article Has Been Retracted” placed prominently at the top of the online article and on each page of a PDF file of the article. These labels can be hyperlinked to the published retraction.
If an author of a fraudulent article, or any institutional authority, refuses to submit an explanation for publication as a retraction, the editor can leverage the authority and influence of his or her position and that of the journal to compel an appropriate response, keeping in mind the journal’s obligation to publish a retraction.27,29 If, however, the editor is unable to receive a satisfactory or timely response from an author or the investigating authority on the merit of the allegation, the editor may publish an “expression of concern” to alert readers, librarians, and the scientific community that there are concerns that an article may include fabricated, falsified, or plagiarized work, and follow this later with a formal retraction. This notice of concern should follow the same publication format as recommended for notices of retraction. If evidence of misconduct is sufficient and the editor cannot obtain a retraction letter from the author and is awaiting the results of an official investigation, the editor may choose to publish an expression of concern and follow this with a formal retraction once the institution has completed its investigation.
The validity of other work published in the journal by the offending authors should also be questioned. The ICMJE recommends that editors ask institutions to provide assurance of the validity of earlier work published in their journals or to retract those as well. If this is not done, editors may chose to publish a notice or expression of concern stating that the validity of such previously published work is uncertain.22
Box 1 shows examples of retraction notices from authors, an institution, and an editor and a listing in the table of contents. Examples of recent retractions in the literature are shown in Box 2. Some authors may not want to explain the reason for the retraction in a forthright manner. Editors should work with authors or their institutional authority to make these notices as accurate as possible. In some cases, publishing an author’s evasive or incomplete statement might be better than publishing nothing from the author; in such a case, the journal can also publish an explanatory note from the author’s institutional authority or the editor.
When an article is retracted, the original article should not be physically removed from a journal’s website or other online archival publication. However, it should be made clear to all users of online archival material that the article has been retracted and should not be used or cited. This requirement includes clear labeling of retracted articles and 2-way linking between retraction notices and the original articles. The National Library of Medicine does not remove the citation of a retracted article; the citation is updated to indicate that the article has been retracted, and links between the original citation and the citation to the retraction notice are added.28
Retractions may also be used for articles that are seriously and pervasively flawed because of honest error that is not a result of fabrication, falsification, or plagiarism. However, retraction of an article because of serious and pervasive errors should be used cautiously. Indeed, Sox and Rennie27 have called for retractions to be reserved solely for cases of scientific misconduct. Retractions should never be used for typical errors; in these cases, a correction is appropriate (see also 5.11.9, Editorial Responsibilities, Roles, Procedures, and Policies, Corrections [Errata]). A study of 395 articles retracted during the years 1982 through 2002 found that 107 (27%) reflected scientific misconduct and 244 (62%) represented unintentional errors (another 44 [11%] represented other issues or provided no information about the reasons for the retractions).30 The National Library of Medicine cites examples of such serious and pervasive errors as “conclusions based on faulty logic or computation” and data obtained after inadvertent contamination of cell lines or through poor instrumentation.28 If the errors in an article are substantial and pervasive (eg, incorrect data throughout the text, tables, and figures), the journal may choose to publish a retraction notice from the original authors as well as a replacement article.31 In this case, online versions of journals and bibliographic databases should provide reciprocal links to and from the notice of retraction, the retracted article, and the replacement article, and the retracted article should be labeled as retracted.