Fraud Found in GSK's Avandia Trial Reveals FDA's Lack of Oversight

If even a poorly designed trial can’t demonstrate that your drug is safe, then hiding data is the next-best trick for getting approval. Both techniques are commonly used by Big Pharma to get their products approved. When the FDA doesn’t carefully review study design or check the data in trials, such methods work. Both of those tricks worked for GlaxoSmithKline’s Avandia study in 2007. Fortunately, a researcher decided to look more closely in advance of the drug’s review next week.

Avandia, made by GlaxoSmithKline (GSK) has been coming under criticism for severe and deadly adverse effects, including heart attacks. In response, the FDA has scheduled a safety review for 13-14 July. However, the FDA reapproved this drug in 2007, and a primary piece of evidence used was GSK’s RECORD trial. They simply accepted the documentation given to them, without seriously looking at the methodology used or checking out the data’s validity.

Now, though, shortly before the Advisory Committee Meeting (ACM) on whether to continue its approval of Avandia, an FDA insider, Dr. Thomas Marciniak, has reported that he actually reviewed the trial’s design and demanded to see records that hadn’t been reviewed the first time around. What he found was shocking, completely negating GSK’s claims of safety and showing much the same results as another trial, which the FDA tried to suppress.

The FDA Briefing Document for the ACM includes specific examples of the missing data under the heading, “Extreme mishandling of events”⁽¹⁾. The first sentence in that section reads,

These four cases represent what we judge to be the worst mishandling of events in RECORD, mishandlings that we judge should not be found even as single occurrences and that suggest serious flaws with trial conduct.

These were only four of the 70 “mishandlings” found after reviews of 549 patients. 12.75% of the records were found to have been misrepresented in RECORD’s reporting! These misrepresentations include changes in diagnosis, such as changing heart failure to atrial fibrillation⁽²⁾ and deleting the fact that one patient had suffered an intracerebral hematoma⁽³⁾.

Even before looking at the missing data, Marciniak noted that RECORD’s design was deeply flawed. In the Briefing Document report, he wrote,

We did not review the protocol prior to study implementation. If we had, we would have judged it to be unacceptable for the reasons detailed in Appendix 2.⁽⁴⁾

The design itself was so flawed that, even if data had not been hidden, the study should never have been accepted by the FDA! Yet, this is the same document that the FDA accepted at face value three years ago.