In 2010, the European Medicines Agency (EMA) put into place the world’s first freedom of information policy that also covered clinical trial data (EMA Policy 0043). Five years later, it remains the only regulator in the world to freely and unconditionally release clinical trial data from its holdings.
We have previously documented how this revolutionary policy has released millions of pages of regulatory data. But these analyses only showed us how the system was working from a 30,000 foot perspective. Today we shed light on how the system is working on the ground.
What our study reports
Our paper published today in Trials reports a case series of 12 requests to EMA for regulatory data. The 12 requests were for 118,000 pages relating to 29 medicines and biologics. This includes our 2011 request for Tamiflu data that enabled our Cochrane review to begin to move beyond publications and instead consider the far more detailed clinical study reports that regulators routinely use to understand and assess trials.
Given the growing recognition of regulatory data as the key to unlocking reporting bias in the scientific literature, we report that using the EMA’s new system is more complex and convoluted than one might hope.
Given the growing recognition of regulatory data as the key to unlocking reporting bias in the scientific literature, we report that using the EMA’s new system is more complex and convoluted than one might hope—and that there are signs of a system that must throttle its output to cope with demands.
Response times are getting longer, the output seems to be slowing down and bureaucracy is gaining the upper hand. Yet the output of this system–clinical trial data otherwise inaccessible–is all the more important because of signs that the EMA’s separate data access policy (Policy 0070) may not deliver on its promises due to excessive redaction of documents.
Ideally we should have carried out a systematic survey of requests for EMA data. However this is not feasible, as EMA does not disclose the identity of requestors. We had to resort to the oldest design known in medicine: the case series.
Selected, biased by our experience and views perhaps, but we think closest to the truth than anything else we have seen. In short this is the way it has been for us for the last five years. We know from informal contacts with other researchers that things for some of them are even worse, bewildered by a new lexicon and showered with letters written in EMA bureaucratese.
The way forward
We think it is important to show what things are really like not because we want to damage EMA in any way but because we believe the EMA policy is the way forward. In fact at present it is probably the only way forward for Evidence Based Medicine to regain its credibility dented by the increasing impact of reporting bias.
It is worth fighting to improve the EMA practice, perhaps with more resources and regular open audits. And we think there may be lessons here for the emerging clinical trial data access scheme in Canada under the new Vanessa’s Law.