FLICKR, STEAKPINBALLAs an unsuccessful whistleblower, I know first-hand that the system to deal with misconduct in science often fails. Details of my 14-year odyssey are on my website and in a November 2013 Nature article. I advocate that, to minimize scientific misconduct, we establish a better system for reporting and adjudicating it.
My story started when I witnessed what I believed—and were supported by a second observer to be—data fabrication. In short, numbers were recorded for samples that we believed did not exist. I followed the guidelines promulgated by my university for reporting perceived misconduct as closely as I could. I was referred to the Campus Committee on Research Integrity (CCRI), a standing committee of representatives of the various schools on the Newark campus of the University of Medicine and Dentistry of New Jersey of which my employer, the New Jersey Medical School, was a part. Nine individuals sat on the CCRI. Some were administrators; there was a nurse, a midwife, a lawyer, and the vice president for research—none of whom had any expertise in the research involved. The wife of the questioned PI had been a postdoctoral fellow working down the hall from the CCRI chairman. The guidelines mandated a two-stage process, but the CCRI assumed both roles and did not call on an independent expert in the field in question.
The CCRI determined there was insufficient evidence of scientific misconduct. I disagreed. I was not given the final report, and there was no mechanism for me, the whistleblower, to appeal, although the respondents would have been granted that privilege were the case not to have gone their way.
It was this first CCRI meeting that proved to be my undoing. The committee interviewed all parties involved, saw no problem with receiving three different numerical answers from the suspected individual to the same query, and took no issue when an undocumented “new cell line” made an appearance to explain the mysterious presence in an incubator of cultures that were reported to have been removed.
So I turned to the U.S. Department of Health & Human Services’ Office of Research Integrity (ORI). I again lost, and again, I was unable to appeal. But the ORI recommended that I resubmit my case to the university based on statistical evidence that it had uncovered. When I did, I was thwarted and once more faced with no mechanism for appeal. The ORI inspector recommended that I file a case for qui tam, a civil law suit in which a whistleblower (relator) alleges that the federal government has been defrauded.
For those who are unfamiliar with such suits, qui tam cases arose in the Civil War and involved bilking the federal government in various ways—none of which realistically apply to scientific research. And indeed, I maintain that the judge who ruled against me paid no attention to the scientific evidence of fraud. His decision was based on his belief that the PI did not know of the fraud the day he submitted the grant application to the National Institutes of Health, even though, at that time, I believe he had plenty of cause for suspicion.
Although in many respects the qui tam case was a waste of time and money, it did prove enormously useful in that expert witness Joel Pitt—who at the time was a statistician at Georgian Court University in New Jersey—and I gained access to data from the accused postdoctoral fellow and nine other members of the same laboratory. We now have compelling evidence for manipulation of results that were used to obtain the original grant, its renewal, and were published in up to eight journal articles. We analyzed several hundred datasets, finding that the alleged fraudster’s results were extremely far from (expected) random compared to the consistently random results of the nine other investigators in the same laboratory.
After losing the qui tam and subsequent appeal, I returned to the CCRI and the ORI with what I considered extremely compelling evidence of data fabrication only to be stonewalled again by both. For the last two years, Pitt and I have tried to get our data analysis published. We have submitted to 10 journals and have only gotten past the editorial offices of two. No one, it seems, wants to publish a statistical analysis of allegedly fraudulent data.
We continue to strive to correct the scientific record. In the meantime, though, there are some things that policy makers, administrative officials, journal editors, and we scientists can do to make correcting the record much more straightforward.
- Scientists should make raw data publicly available.
- Scientists should be obligated to report failed replication attempts to the journal that published the original results.
- All misconduct proceedings involving state and federally funded research should be open to the public.
- The CCRI (or equivalent) ought to be ad hoc, composed of individuals unfamiliar with the complainant or respondent
- The CCRI (or equivalent) should employ at least one expert in the field under examination
- The final report of the CCRI (or equivalent) and/or the ORI ought to be provided to both the complainant and the respondent
- Both complainant and respondent should have the right to appeal, and the appeal board ought not overlap with the original committee
- Journals ought to take a more active role in retracting papers containing false information
- The scientific community should establish a board of last resort for appeal and adjudication of unresolved scientific questions raised regarding data fabrication and falsification.
Helene Z. Hill is a professor of radiology at Rutgers New Jersey Medical School.
Editor’s Note: In reviewing the preceding opinion, The Scientist reached out to Roger Howell, a professor at Rutgers New Jersey Medical School and Rutgers Graduate School of Biomedical Sciences, and the PI Hill claimed was aware of the fraud she has described, who wrote:
“For over 25 years, my colleagues and I have received steady funding for our scientific research, and we continue to make significant progress in our efforts to understand how radiation emitted by radiopharmaceuticals affects tissue. What has been most frustrating is that so many people and organizations have had to spend enormous resources and time revisiting this topic for 14 years, despite the fact that her claims were rejected by each and every entity that she approached. These included scientific journals, the University (at least three occasions), the National Institutes of Health, the Federal District Court and the 3rd Circuit Court of Appeals. As stated by the Federal District Court judge, ‘Plaintiff’s repeated failure to accept the findings of the very oversight committees she sought out has turned this lawsuit into a quest of Quixotic proportions that ultimately must be put to rest.’”
Anupam Bishayee, the accused former postdoc from Howell’s lab, did not respond to The Scientist’s requests for comment.