Diagnosis of fraud

Serious research fraud is usually discovered during the formal audit or monitoring process of randomised controlled trials when all patients apparently recruited into the study cannot be accounted for from a search of independent medical records.9 Fraudulent clinical trials have also been detected after publication, often when there are some unusual features to the study. Surprising, unexpected results emerging from an apparently, large single-centre study written up by a single author might attract attention. Published work may come into question when an investigator is found to have committed research misconduct and previously published fraudulent work is then revealed following a systematic review of the contents of the author's curriculum vitae. The other important route by which research fraud is detected is through a whistleblower, commonly a colleague in the same or a closely related department. Although there are many examples when whistleblowers have subsequently suffered more than the accused, there is now a firm legal framework to protect people in the workplace who become concerned about the behaviour of colleagues.

One of the most active, contemporary whistleblowers in the UK is Peter Wilmshurst, a consultant cardiologist in Shrewsbury. Wilmshurst wrote an article in the Lancet in 1997 entitled "The code of silence" in which he suggested that there were a number of cases of research misconduct in the UK that had not been adequately investigated or suppressed, perhaps by senior members of the profession.10 One of the cases referred to anonymously in this report was finally brought before the Professional Conduct Committee of the GMC in November 2000; this resulted in the suspension of a Dr Anjan Banerjee, a surgeon from Halifax.11'12 It took almost a decade for this case to be resolved despite the fact that many colleagues knew of the allegations for much of this period, and certain individuals, who had been willing to speak out, were silenced with the threat that their future career would be at stake. Peter Wilmshurst assembled the evidence and eventually the case was heard. Without his tenacity it is unlikely that Banerjee would have been required to face his misdemeanours nor would the medical literature have been corrected.

This sounds like a satisfactory ending to an otherwise sad story. However, a number of issues remain unresolved. As editor of the journal in which Banerjee published his fraudulent work I am responsible for issuing a retraction notice in the journal, a request that came promptly from the author following the GMC judgment. However, it is extremely difficult to effectively erase an article from the published literature since the November 1991 issue of Gut will remain on library shelves for many decades to come without any indication that the Banerjee paper is fraudulent. In addition we know that retracted papers continue to be cited after retraction almost invariably without reference to the falsehoods that are contained therein.13 If publication of scientific material becomes solely electronic, only then will it be possible to erase a paper completely and put the record straight. Another issue is the delay that has occurred in bringing this case to the GMC. Intuitively most of us would believe that it would be preferable to solve a crime promptly to minimise the risk that further misdemeanours might be committed. Banerjee had already been suspended by his NHS Trust earlier in 2000 for, I understand, a totally different reason and has recently resigned from this post. Whether earlier resolution of the research misconduct case would have influenced this issue remains a matter for speculation.

Unfortunately the case does not rest here. Banerjee's supervisor at the time he produced the fraudulent work, Professor Tim Peters, also appeared before the GMC in early 2001, when he was administred for his role in the affair. Finally, and perhaps most important of all, what happened following the internal enquiry of the Banerjee case at King's College Hospital Medical School in 1991 will be determined. If Banerjee's work, as it is alleged, was found to be questionable at that time,10 why did the Dean of the day not refer the case to the GMC immediately? Why was the Gut paper not retracted in 1991 and why did Banerjee's supervisor remain associated with the work when other collaborators withdrew their names from the paper?

We all have a responsibility to be vigilant about the quality of research with which we are directly or indirectly associated. If you have suspicions that a colleague may be producing research findings dishonestly, the most important first step is to ensure that you obtain evidence of your suspicions before reporting this to a senior colleague. In some instances it may be entirely appropriate to discuss this with the person concerned, particularly if you feel that the main problem is "trimming and pruning", which the individual may be pursuing out of ignorance. Similarly, if there are potential misdemeanours regarding publication ethics, such as dual submission or redundant publication, then these should be dealt with by open discussion. However, if the misdemeanours are of a more serious nature, such as fabrication or falsification of research data, then you must try to provide evidence of the misdemeanours before reporting to a higher authority. I have been approached by potential whistleblowers who have waited weeks or even several months to ensure that their suspicions are well founded. At this point you should discuss this with your head of department, although, if that individual is directly involved with the research, it may be more appropriate to discuss your concerns immediately with the head of the institution. At this point a preliminary inquiry should be performed along the lines described in the Royal College of Physicians Working Party Report14 or as described in the Medical Research Council's document, which describes a procedure for inquiring into allegations of scientific misconduct within MRC units.15

Journal editors and reviewers also detect research and publication misconduct. In my experience as an editor of a specialist journal, sharp-eyed reviewers have drawn my attention to the majority of cases of misconduct that I have come across in the past three years. However, in almost every case there has been an element of serendipity about the discovery. For example, in one case of plagiarism that we have seen, I just happened, by chance, to send the manuscript to a reviewer whose papers had been plagiarised. How many other manuscripts have passed through the system and escaped detection because the reviewer did not have such an intimate knowledge of the text of related papers? Redundant publication and "salami-slicing" are again usually detected by expert reviewers who know their subject well. On several occasions, we have by chance sent a manuscript to a reviewer who had been sent a closely related and in some cases an identical manuscript by the same authors from another journal, simultaneously. In this case publication can be stopped but if the manuscript had been sent to another reviewer, who had not seen the closely related manuscript, then both manuscripts could have entered the public domain with inevitable redundancy. Thus, the detection of publication and research misconduct seems to carry with it a large element of chance and many editors of biomedical journals suspect that only a small proportion of misdemeanours are detected. It would appear that crime does pay! Diagnostic approaches would thus seem to be inadequate and, given the obvious limitations that a whistleblower, editor, or reviewer might have, perhaps we should focus our attention more on treatment and prevention.

Was this article helpful?

0 0

Post a comment