Ensuring Data Quality and Identifying Potential Fraud in Clinical Trials

• Activity Number: 217
• Type: Topic Contributed
• Date/Time: Monday, August 4, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Nancy Geller*+
• Companies: NIH
• Keywords: fraudulent data ; research misconduct ; data integrity ; detecting fraud

Abstract

Research misconduct has been formally defined. After reviewing this definition, we are concerned with two aspects of research misconduct, falsification and/or fabrication of data which we loosely call "fraud". Several examples of fraud or allegation of fraud will be given in which we examine why fraud was alleged, how it has been detected and how long it took. Inquiries into publication fraud have been set off by an allegation (e.g. graduate students alleging data disappearance) or a strange and unexpected coincidence (e.g., consent forms missing). Statisticians are in a unique position to aid in the detection of fraud, but should play a more active role than attempting to detect fraud after the fact. One should be especially wary of results that appear to be "too good to be true." It is wise to undertake certain routine simple comparisons in all of our clinical trials to be assured that data or individual centers don't show unusual patterns. Making clinical trial data and computer code available to others promptly will allow verification of reported results.