The School of Medicine is safeguarding against research misconduct with a new set of guidelines.
A faculty committee has established measures to improve the quality and integrity of translational medicine research. The multidisciplinary Translational Medicine Quality Framework Committee began examining the issue of reinforcing research integrity last Spring. The committee’s report, released May 2011, outlines four key areas of improvement: bioinformatics support, biostatistical collaboration, research accountability and far more rigourous evaluation of research before moving to clinical trials. Some changes have already gone into effect, though administrators will continue to implement other measures in the next several years.
The changes promote an environment where researchers will feel comfortable questioning one another and suggesting alternative interpretations of data, said Dr. Nancy Andrews, dean of the School of Medicine and vice chancellor for academic affairs.
“Overall, we are ahead of most of our peer institutions in thinking about these issues,” Andrews wrote in an email Wednesday. “Research misconduct, while rare, is a problem that all research institutions encounter periodically.”
The examination of research integrity was motivated in part by the alleged research misconduct of former Duke oncologist Dr. Anil Potti as well as a need to adapt to an advanced and increasingly technical field of medical research. The report focuses on genomics research, but the new guidelines apply to all of translational medicine—treatments that go from research labs to clinical trials.
“Some of our research improvement initiatives, particularly involving clinical research, started well before we became aware of possible issues in the Potti case,” Andrews said. “We are an institution that wants to move biomedical science forward through important contributions that will stand the test of time.”
Since the committee’s establishment, members have focused on encouraging scientists to question research methods and findings and fortify data management and validation, Dr. Robert Califf, vice chancellor for clinical research and director of the Duke Translational Medicine Institute, wrote in an email Tuesday. They also sought to increase access to expertise in the design and analysis of experiments as well as to maintain active management of conflicts of interest.
A new framework
The TMQF Committee has taken several steps to toughen the evaluation of translational research. It has established an integrity line that allows researchers to report concerns anonymously over the phone, as well as more stringent controls to monitor ethical practices and scientific merit in studies involving human subjects, said Sally Kornbluth, vice dean for research at the School of Medicine and vice provost for academic affairs.
A TMQF subcommittee of Duke Medicine leaders is discussing plans for a virtual lockbox that will allow researchers to track what changes are being made to complex data sets and who is making them, Kornbluth said.
“What we’re really looking for is an [information technology] solution that will allow one to determine that the data that are being used to eventually lead to clinical trials are in fact the data obtained at the time of collection and that there haven’t been changes that people are not aware of,” she said.
Although it is unacceptable to alter original data, there are some legitimate reasons for why some changes or comments need to be made, Kornbluth added. The virtual lockbox will allow anyone to view edits throughout the research process and to contact those involved. The online database will also help prevent honest errors.
The subcommittee is searching for a well-adapted and simple IT solution that fits the kind of work being done and does not hinder the research process.
“It is very easy, if you’re working with dozens of Excel spreadsheets to make an error,” Kornbluth said. “What you want is a much more robust system than that, which will enable detection of errors that are either unintentional or intentional.”
Instances of research misconduct in an academic institution are rare but also impossible to prevent entirely, Andrews said, adding that if a person is highly motivated to cheat, they will likely find ways to do so.
Although it is impossible to completely eliminate research misconduct, Califf added that he is confident these new measures will help safeguard research integrity and improve the quality of Duke research.
“These changes will make our research better across the board,” he said.
Cracks in a system
The committee’s framework report recognizes that genomic science has vastly increased the amount of data scientists can collect. With more data and research requiring peer review and increased pressure on researchers to move their work into clinical trials, room for error is significant. According to the report, these developments put the previous research framework to the test, as illustrated with the recent Potti case.
A redflag in the Potti research was the discrepancy between the data he presented and the original data set, Kornbluth noted.
“It makes it very difficult why that is if you have no way of tracking exactly when changes were made and by whom,” she said.
Potti, who published peer-reviewed research and conducted clinical trials on a genomic technique for predicting chemotherapy responses, was questioned in 2006 by Duke’s Institutional Review Board after Texas researchers took issue with some of the data. Potti was allowed to continue until his credentials came under fire and the trials were terminated in 2010.
The multiple levels of institutional review did not discover the flaws in Potti’s genomic research until after the research had been applied to patient treatment in clinical trials.
To date, Potti has retracted 10 papers, and there is an ongoing investigation into alleged research misconduct. Two lawsuits have been filed against Duke, Potti and others on behalf of patients who participated in the clinical trials.
Get The Chronicle straight to your inbox
Signup for our weekly newsletter. Cancel at any time.