Researcher takes aim at clashing drug data
A witness in the 2008 case Pfizer lost for making false presentations about epilepsy drug Neurontin, Kay Dickersin, published a piece in open-access journal PLOS Monday that takes the drug maker, and by implication the industry, to task over a failure to ensure that published information reflects corporate knowledge.
At issue are clinical trial documents Dickersin reviewed with a group of colleagues that cross-referenced clinical trial data for off-label use of the drug against internal documents about these same trials. They found a lot of daylight between the corporate version and the public one and a lack of consistency across them, reporting a mishmash of information that left them “unable to compare the internal research report with the main publication”—for variables such as the number of participants or even a consistent form of analysis. On the in-house communications side, Dickersin's team also found that Pfizer's documentation showed that the published trials “did not always accurately reflect what was actually done in the trial.” Dickersin had access to the internal documents—which she noted are not generally available—because of the lawsuit.
Among the points of contention: internal documents and published documents used two different definitions for the term “premature discontinuation” among three of 11 trials that were examined. They also found that in six out of 10 trials researchers couldn't make a match between company data and published data regarding just how many patients were used to assess efficacy because the type of analysis wasn't defined or a primary outcome wasn't established.
Patient numbers are a critical piece of data, and varying the definition, for example, can mean a drug maker can cherry-pick who is being counted. A similar issue surfaced during the May 2012 FDA committee review of Xarelto, in which a panel member was exasperated by the data which made it impossible to understand the patient population—such as enrollments, drop-outs, exclusions—because it made the drug's pros and cons that much more opaque.
Dickersin noted that although journal editors tried to tamp down on discrepancies when they issued a Consolidated Standards of Reporting Trials in 1996—which she said it was a good start to increasing transparency—more has to be done.
Among her recommendations:
- Make trial protocols public
- Make research reports—internal documents, also known as clinical study reports—public
- Standardize definitions for types of analyses
- Require clinical trial registration and related data and protocols for both approved and off-label uses