Clinical trials transparency

I'm currently at the EMWA conference in Budapest, where this morning I attended a fascinating symposium on clinical trials transparency (the symposium went on all day, but I was teaching a workshop on critical evaluation of medical literature this afternoon, so sadly I couldn't stay for the afternoon session).

There is no doubt that transparency of clinical trial data is a hot topic at the moment. It is certainly to be welcomed that it is being so widely discussed these days. Transparency is very much on the radar of the pharmaceutical industry and the regulators.

However, progress does seem slow.

We were treated to a fascinating talk by Martin Harvey Allchurch from the EMA, who shared a great deal of information about the agency's new policy on clinical trials transparency. If you follow such things closely, you may recall that the policy was originally planned to come into effect at the beginning of this year, but had been delayed as it turned out to be more complex than the EMA had anticipated.

The EMA have been working hard on the policy, and it is due to be finalised on 12 June. I learned a lot today about what is going to be in the policy (in fact I think today's symposium may have been the first time that some of the information had been publicly announced).

The new policy will not include access to raw data. Although this had been discussed, it was felt too difficult at this stage. It undoubtedly is difficult: there are some very serious problems to be resolved before raw data can be shared, most seriously to do with patient confidentiality. An EMA policy on raw data may be forthcoming in the future, but it's not happening now.

The new policy will also not include access to existing data. It will apply only to new data submitted to the EMA after the "effective date" (a date which has not yet been determined, but which I'm pretty sure is still in the future).

However, for newly submitted data, interested researchers will be able to have a kind of access to clinical study reports. The EMA will not be vetting the purposes for which researchers want to access reports, but users of the reports will have to agree to terms of use. They will not then be able to download or print the reports: they will merely be able to view them on-screen on the EMA's website.

This is certainly better than no access to clinical study reports, but I'm sure many will argue that the policy doesn't go far enough, and I would find it hard to disagree. The lack of access to raw data or pre-existing data, and the inability to download clinical study reports, are all disappointing limitations.

Expect to hear a lot more about this sometime after 12 June.

The next talk was from Hans-Jürgen Lomp from Boehringer Ingelheim, who talked about what the pharmaceutical industry in general, and BI in particular, were doing.

The moves towards transparency from some pharmaceutical companies seem to go quite a way beyond what the EMA is proposing, though of course not all pharmaceutical companies are joining in. GSK has been widely praised for leading the way, and some other large companies, including Roche, Novartis, and BI, are doing similar things.

There are 2 important ways in which the industry (or at least parts of the industry) are going beyond the EMA's initiative. First, they are not limiting access to only new data, and second, they are making raw data available. Well, sort of available. I'll come back to that.

In the case of BI, they are making full clinical study reports and data available to qualified researchers (they will be vetting applications to make sure they are scientifically sound) for studies going back as far as 1998. I believe GSK are going back to a similar extent.

BI, GSK, and a number of other companies will be making raw data available through a joint initiative on a dedicated website, ClinicalStudyDataRequest.com. Although sharing of raw data is really a huge advance in transparency, there are some serious limitations to how the data are shared that I wasn't aware of until today.

The problem is that researchers will not be allowed to download the data. I can completely see the logic in this: patient confidentiality concerns are a serious problem, and restricting access to the data as far as possible is a sensible way of countering this. However, it does mean that the data will not be as useful as if researchers were able to download them.

The way it works is that researchers will have access to the data within a secure analysis environment. You can see how it works here. There are 2 big problems with this that I can see.

The first problem is that you are limited to the software tools supported by the system, which is just SAS and R. Now, both of those are fine pieces statistical analysis software, but many researchers will find the choice limiting. If you want to analyse the data using SPSS, Stata, WinBugs, or anything else, then tough luck.

The second problem is that it is impossible to combine the data with data from other sources. If the purpose of getting the data is so you can do an individual patient meta-analysis of various studies, including some from within the industry system and others that you have obtained elsewhere, then there is no way of combining them.

It is of course to be welcomed that transparency is very much on the agenda. We are having conversations about it that would have been unthinkable just 10 years or so ago. That is undoubtedly progress.

But it is slow progress. We are still a long way from the point where clinical trial data are routinely available to any interested researchers.