The Tamiflu story: Why we need access to all data from clinical trials
The BMJ Open Data Campaign has been attracting a lot of attention. Here Dr Tom Jefferson, one of the people whose attempts to provide reliable information on the anti-flu drug Tamiflu kicked the campaign off, tells the story of how we got here.
We started working on a Cochrane review of neuraminidase inhibitors in 1998. Cochrane reviews are studies summing up what is known of the effects of an intervention in healthcare. In this case the intervention was the class of drugs called neuraminidase inhibitors. At the time this comprised two anti-influenza compounds: zanamivir (sold as Relenza by GlaxoSmithKlein) and oseltamivir (Tamiflu, by Roche).
The Cochrane Collaboration is a network of volunteers who do and update reviews. We don’t take money from pharma and we have to follow highly structured protocols which are posted publicly before we start work. Comments by any reader can be posted on the protocol or the full review at any time. We have to respond.
Rightly or wrongly, our reviews are considered the gold standard for evidence-based decision making. It’s hard work, as we have to update our reviews every two years or so.
In 2009, our review was in its third update, the world was in the throws of an influenza pandemic (or so WHO was telling us) and we received a letter from a Japanese paediatrician. He wanted to know how it was possible that in our 2005 update we had included 8 unpublished Tamiflu trials contained in extreme summary form within another review funded by Roche and carried out by Roche staff and consultants. How could we possibly have done that as we had not seen the original studies? We asked the two Roche consultants for the data. They told us to go and ask Roche. We did. They asked us to sign a confidentiality agreement with a secrecy clause. We said no thank you. Once the very powerful medical journal BMJ got involved with ITV Channel 4, they promised us full study reports, but gave us only the first chapter of the 10 trials. In the meantime we discovered many more trials (the list has grown from 26 to 123 – the vast majority Roche sponsored). We asked for all the Roche completed trial reports. Roche gave us a variety of reasons why they would not share the data with us. You can read about those here.
At the end of 2010 the European Regulator EMA accepted a ruling by the European Ombudsman that trial data for drugs on which a regulatory decision had been made should be accessible. They opened their archives. We received incomplete reports for 16 Tamiflu trials, all they had.
We published half of this (and 2000 pages of FDA comments on Tamiflu) in the 2012 version of our Cochrane review. One consequence of our access to this bonanza of regulatory material has been a comparison between the details and broad message of the few published trials and their regulatory much more detailed reports. Apart from discrepancies in reporting harms and some less-than-detailed aspects of study design, we think the mode of action of the drug is not what the manufacturer says and (like FDA) could not find any evidence supporting a number of effects of the drug (including those for which it was stockpiled).
But we do not know for sure because we do not have all the data. The practical result of all this is our refusal to consider published trials (either on their own or as part of reviews) for inclusion in our reviews. There are signs that this distrust of the published word is spreading.
Meanwhile what started as comment from a Japanese colleague has turned into a global campaign for access to data on trials. You can read about that here. The BMJ set up a Tamiflu micro site on BMJ.com with lots of goodies including our correspondence with Roche, WHO and CDC: http://www.bmj.com/tamiflu. The latter are the two biggest promoters of Tamiflu. If you have time, do read the correspondence. Your time will not be wasted, I promise.
And what about GlaxoSmithKlein? Some of the hype recently suggested that, after its record fine in the US, GSK would open its archives to researchers, albeit with the intermediary of an academic committee scrutinizing the worthiness of the analysis plans in the application. Whether this is a genuine breakthrough or a clever piece of marketing remains to be seen. My group is yet to receive any data from them. Despite the plaudits, I remain unconvinced, as I refuse to receive data with any conditions attached to them – such as exclusivity or bans on sharing.
Trials are experiments conducted on human beings. Full reporting of their results (anonimyzed to prevent individuals being identified) should be a right, not a gift. Your doctor should be in possession of all the facts. Think about that next time he prescribes something for you.
Watch Tom tell the story: