Data fraud in clinical trials

Published on September 29, 2016   28 min

Other Talks in the Series: The Risk of Bias in Randomized Clinical Trials

Please wait while the transcript is being prepared...
My name is Stephen George. I am a Professor Emeritus of Biostatistics in the Department of Biostatistics and Bioinformatics at Duke University. My topic today is "Data Fraud in Clinical Trials".
An outline is given on this slide. There are five different areas. First, we'll talk about some definitions that are important in setting the stage for the discussion, some estimates of prevalence of data fraud, some key examples that have been uncovered, some information on motivation and contributing factors, and end with some statistical detection methods.
First, definitions.
The US Public Health Service has defined "Research misconduct means fabrication, falsification, or plagiarism in proposing, performing, or reviewing research, or in reporting research results... Research misconduct does not include honest error or differences of opinion". That last stanza is very important. In this presentation, I will focus on what I call "data fraud" which is simply the fabrication or falsification of data in clinical trials.
In addition to data fraud, there are of course other questionable research practices in clinical trials. I'm not going to talk about these much today, but it's important to note that these may be at least as important, if not more so, in producing questionable results. And these include inadequate or inappropriate design and analysis, selective reporting of results, inappropriate subgroup analyses, not admitting that some data are missing, ignoring outliers, over-interpretation of results from small trials, post-hoc analyses that are done but not admitted as being post-hoc, and withholding details of methodology, or someone's day-to-day withholding the data.