Who notifies a journal that a publication must be retracted, and when does that notification occur if an allegation of research misconduct has been made? As the Council of Scientific Editors White Paper notes, journals receive notices of correction or retraction from many people, including authors and institution officials (www.councilscienceeditors.org/wpcontent/uploads/entire_whitepaper.pdf).

Recently, two Harvard University professors, senior authors on certain journal publications, sued Harvard Medical School, Brigham and Women’s Hospital, and the relevant institution officials when the journals were notified that specific publications should be retracted at the conclusion of an institutional research misconduct inquiry.¹ The case underscores the tension between providing notice to a journal of a flawed publication early in the research-misconduct investigation and waiting until the end of a research-misconduct investigation (either institutional or federal) to provide such notice. Here, we explore the tension and benchmark current practices with respect to journal notice and action.

The Complaint

An element in the foregoing lawsuit, filed in the US District Court for the District of Massachusetts, asserts that in March 2014, a dean, acting on behalf of Harvard and Brigham and Women’s, notified two journal editors that an investigation had commenced regarding 2011 and 2012 publications and recommended retraction of the articles. The complaint asserted that such journal notification was “contrary to established practices” and that “papers are rarely or never retracted without first exploring the possibility of issuing a less serious correction and without the consent of the authors.” One journal issued a retraction, and one issued an expression of concern. The complaint further asserted that the authors were willing to issue a correction but could not do so without the approval of the relevant institutional review board (IRB).

Who Notifies a Journal of a Problem Publication?

The CSE White Paper notes that numerous individuals can notify a journal that there is significant evidence that a publication contains possibly falsified and fabricated data (White Paper, §3.5). Such persons include an alert reader, an institution official, the senior author, the corresponding author, all the authors, the majority of the authors, and the author accused of an impropriety. In recognition of the authors’ expertise and responsibility to correct the literature and because journals have different policies and procedures for handling retractions or corrections and for determining who has standing to make such a request, many institutions encourage the coauthors of a paper to submit the appropriate correction or retraction of a publication to the journal editors. Typically, only if authors do not submit such a retraction or correction does an institution official provide the journal the notice. Appropriate federal agencies notify a journal about a problem publication only if and when a federal research-misconduct finding is made. On the basis of two informal surveys of members of CSE (see CSE Annual Meeting 2007² and 2012³), most journals, on receiving notice from an institution official, would not retract or correct a published article without further evidence of misconduct, an admission of misconduct, or the institution’s investigation report. Many journals, on notice from an institution official, would ask the authors later whether the article should be retracted or corrected.

When Is a Journal Notified of a Problem Publication?

When an allegation of research misconduct is made, US federal regulations and corresponding institutional procedures prescribe a series of steps:

• Institutions have primary responsibility for evaluating and investigating allegations of research misconduct.
• The first step prescribed by US federal regulations is an inquiry. The purpose of the inquiry is to determine whether there is sufficient evidence of research misconduct to warrant an investigation.
• The second step is an investigation to determine whether research misconduct occurred under the institution’s definition of misconduct, which may be broader than the conduct prescribed by federal regulations.
• If the research is sponsored by the federal government, the relevant agency must be notified of the outcome of the institution’s investigation, and the agency will determine whether the conduct constitutes misconduct under the relevant agency definition.

Despite the foregoing timeline and process for a final misconduct finding, some retraction and correction requests are submitted before a final institutional finding of research misconduct.⁴ With the exception of cases involving an admission, relatively few retraction or correction requests are submitted at the conclusion of an inquiry.⁵ This delay occurs, in part, because the purpose of an inquiry is to determine whether there is sufficient evidence of misconduct to conduct an investigation and because generally no final determination has been made regarding whether an article can be corrected or must be retracted.⁶ In contrast, a review of some 98 cases involving at least one publication and a federal Office of Research Integrity (ORI) finding of research misconduct showed that in about two-thirds of the cases the relevant article was corrected or retracted before a federal finding was announced in the Federal Register. Again, on the basis of the extended time that ORI takes to review institutional findings and propose federal findings, such action is not surprising—few institutions and coauthors are willing to delay taking corrective action pending that review.⁷

Allegations in the Complaint

The complaint alleged that articles are not retracted if a correction is possible. Such an assertion is contrary to CSE survey results that indicate that if a falsification or fabrication occurred, the article will be retracted, regardless of whether it is possible to correct it (see CSE Annual Meetings, surveys 2007² and 2012³ ). The complaint also asserts that because the article is related to a clinical trial, the authors could not correct the literature unless and until such correction was approved by an IRB. However, the complaint asserts that there was no reason to disclose the investigation to the journal until the investigation was completed.

Federal regulations require prompt notice if a misconduct case involves issues of public safety in that institutions seeking approved assurance must have policies that address appropriate interim institutional actions to protect public health. Institutions also must notify ORI of issues that the public and research community should be informed about. It is difficult to reconcile the asserted need for IRB approval but at the same time, to claim notice is not required to journals that published such clinical-trial research. The essence of the complaint is that the notices did not indicate that the senior authors were not liable for the data falsification or fabrication.

Conclusion

Institutions should carefully consider who provides notice to a journal of a flawed publication and when such notice should be issued. Early notice should be balanced against the need to ensure that researchers do not build on flawed research, regardless of whether a flaw was intentional or the result of carelessness, and the need for confidentiality.

References and Notes

1. See Anversa v. Partners Healthcare System, D.C. MA, Civ. 14-CV-14424 (Filed Dec. 16, 2014) (the “Complaint”).
2. Goodell H. Editorial Policy Committee Ethics Clinic (Part 1). CSE Annual Meeting, Austin, TX, May 21, 2007.
3. Parrish DM. What would you do? What should you do? CSE Research Misconduct Survey Results. CSE Annual Meeting, Seattle, WA, May 21, 2012.
4. See, for example, Meleik Goodwill 76 Fed. Reg. 7569 (Feb. 10, 2011) wherein the journal was notified before the investigational report was concluded in January 2008. Note that the associated Federal Register notice was more than 3 years later.
5. But see Bengu Sezen, 75 Fed. Reg. 73084 (Nov. 29, 2010) wherein the coauthor retracted six papers at the end of the inquiry and before an investigation was begun.
6. But see also the case of Tian-Shing Lee, NIH Guide, Vol. 22, No. 23 (June 25, 1993) wherein Harvard provided notice to the journals. In that case, the respondent had left the country.
7. For example, in the case of Bengu Sezen, the institution completed its investigation by February 2007, but ORI did not make those findings until November 19, 2010—3 years and 9 months later.

DEBRA M PARRISH is an attorney specializing in issues of research integrity. Parrish Law Offices are located in Pittsburgh, PA and Washington, DC.