Michael Farthing, MD
This lecture and the summary paper were developed from a keynote presentation at the 3rd World Congress on Research Integrity, Montreal 2013; an account is ‘in press’ in the congress proceedings. A full version of this paper will appear in the Journal of Gastroenterology and Hepatology in 2014.
Despite a multitude of guidance documents aimed to promote the responsible conduct of research (RCR) 1, the frequency of research misconduct appears to be on the increase 2. Promoting the RCR as a sole primary goal may not be enough to prevent research misconduct in the future; I will propose additional measures that may be worth considering as a way of preventing this rising trend.
Serious research misconduct includes one or more of the triad of activities namely fabrication, falsification and plagiarism, often denoted as ‘research fraud’. There is another type of misconduct that is generally regarded as being ‘less serious’ collectively known as questionable research practices (QRP) 3. These involve a broad spectrum of misdeameanors such as selectivity in data analysis and reporting, disputes about authorship, inadequate supervision, inappropriate image manipulation and reporting errors.
How common is research misconduct? The truthful answer is we just do not know. Reported cases almost certainly represent the tip of an iceberg and account for about 0.01% of reported research studies. Researcher survey data suggest that between 0.3 and 0.8% of research studies include examples of serious research misconduct and 5-15% have evidence of QRP’s. From about 1980 there has been a dramatic increase in retraction rates from less than 5/100,000 publications to about 35/100,000 publications in 2011 4. This increase had been most evident in the last years. The majority of these retractions are because of discovered misconduct with only about 11% attributable to genuine errors.
While researchers are the primary perpetrators of research misconduct there are other players that also breach the fundamental rules of publication and research integrity. Editors, journal owners and publishers have been criticized for manipulating impact factors, for sloppy review processes and bias against publishing negative studies which will inevitably skew the published literature. There is also a concern that further bias is introduced by authors and sponsors when they select data that is chosen to include in the publication, data which selectively enhances the outcome of the study.
There is also a powerful campaign to insist that the pharmaceutical industry places all of the information that it has about a drug or other new treatments in the public domain.
I believe we need to do more to discourage and detect research misconduct. An increasing number of institutions are requiring researchers to confirm that they have read guidance documents on the RCR prior to taking up an appointment and that they will comply with their contents. But this will never be enough. The argument might be extended to include all major research studies whether they are publicly or privately funded. It would follow that such an approach might go a long way to prevent the publication of large numbers of fabricated studies as editors might be encouraged to ask why a major study that had been submitted to the journal had not been registered at its inception.
There is now a major campaign to insist that all clinical trials should be registered so as to track their progress and ensure publication of the results whatever the outcome 5. There is good evidence that negative trials are less likely to be submitted for publication or be published leading to bias in the published literature. The argument might be extended to include all major research studies whether they are publicly or privately funded. It would follow that such an approach might go a long way to prevent the publication of large numbers of fabricated studies, as editors might be encouraged to ask why a major study that had been submitted to the journal had not been registered at its inception.
I would suggest there is also a need for enhanced measures to monitor research practice. This might include random spot audits of research outputs, incorporating the use of digital technology to check for plagiarism, image analysis to determine whether unacceptable manipulation of data images has taken place and examination of numeric datasets to confirm their validity. This proposal will not be welcome by many researchers. However before we protest too much it is important to put research into the wider picture of activities that are undertaken by research intensive universities and research institutes. Every university is required on an annual basis to have procedures in place for internal and external audit of its finances and its financial processes. In addition, in the United Kingdom (UK), the Quality Assurance Agency audits the teaching and learning in universities on a regular basis. Why is there no equivalent process for research which in the research intensive universities can account for between 20 – 50% of total annual turnover?
Some research however is routinely audited in a formal way, notably the large multicenter clinical trials conducted by the pharmaceutical industry. It is now increasingly difficult for investigators to fabricate patients in such trials because of the requirement to match clinical records with the study report for each patient and further assurances provided when the results are analyzed for individual centers to look for any outliers. Lead investigators know that this is the case and I believe it is a strong incentive for them to conduct the study honestly.
‘Whistle-blowers’, remain one of the most important contributors in the process to reveal misconduct. In the past there has been a practice to ignore complaints from anonymous ‘whistle-blowers’. The emergence of websites that place concerns in the public domain and the use of the Internet to make contact with large numbers of individuals to express concerns about individual researchers and their institutions, have progressed rapidly during the last five years 6. The use of digital media both to make allegations about possible research misconduct, notably image manipulation, and to broadcast this widely to the research community is unprecedented and probably unstoppable. Organizations such as COPE and UKRIO have taken a more moderate stance, although clearly there has to be a balance particularly if there is concern about the allegation being vexatious.
There is sometimes a dilemma as to what is the appropriate response when misconduct is discovered. Are we sure that the punishment always fits the crime? Often we do not know the outcome of an investigation as the findings remain confidential within an institution; in some situations sanctions appear lenient whilst in others they may seem harsh. I am uncertain as to whether we are doing enough to encourage a culture where mistakes can be acknowledged, and then followed by a period of support and remediation. This could complement a move by employers to require all new research staff to make a declaration on appointment that they have no active investigations of alleged research misconduct under way and they could also be expected to disclose whether there have been any allegations in the past and what was the outcome of any investigations.
For the future it will be important to collect and collate cases of research misconduct both nationally and internationally. This has proven to be difficult except when an agency representing research integrity is nationally recognized and has authority to demand that research institutions under its aegis should make annual returns on the cases that they have dealt with during the period.
There is also a need for better communication between research institutions regarding their experiences of research misconduct. There appear to be major disincentives for institutions to place their reports on research misconduct cases into the public domain 7. In some cases compromise agreements have constrained both the employer and the employee to make available a full declaration of what went wrong to the wider research community. Only when an external regulator is involved, such as the General Medical Council in the UK, will the outcome of the case be made public as a routine. Since the majority of university researchers are not subject to the rules of conduct of a professional body their name will only enter the public domain if a paper is formally retracted and even then the reasons for the retraction are not always evident.
The danger of this practice is that it can allow serial offenders to move from university to university largely unimpeded8. Ways must be found to allow institutions to exchange information of this nature across national borders without fear of litigation.
It has been suggested that there should be a register for laboratory scientists and other researchers and that maintenance of registration would be an indication of a researcher’s integrity 8. The concept of the ‘research passport’ has already been entertained and might go some way to affirm the importance for a researcher to have a clean record with say a relevant professional body or learned society. For medical and dental researchers in the UK for example, a finding of serious research misconduct could put their registration in jeopardy, and could limit the right to work in the UK as a practitioner. Might it be reasonable to put similar stipulations on other researchers who currently escape this sanction by not being under the aegis of a professional regulator?
Finally, I would suggest we need more research to understand better the motivations of those that commit misconduct and why they feel able to go against the high-level principles that are intrinsic to research integrity across all disciplines. How important is the notion that committing research misconduct is worth the risk because the chances of getting caught appear to be slight? There must be ways of detecting aberrant behaviours before they become a 10 year habit 9!