Science and Technology Committee - Minutes of EvidenceHC 104

Back to Report

Oral Evidence

Taken before the Science and Technology Committee

on Monday 22 April 2013

Members present:

Andrew Miller (Chair)

Stephen Metcalfe

Stephen Mosley

David Tredinnick

Roger Williams


Examination of Witnesses

Witnesses: Dr Catherine Elliott, Director, Clinical Research Interests, Medical Research Council, Sharmila Nebhrajani, Chief Executive, Association of Medical Research Charities, Professor Peter Johnson, Chief Clinician, Cancer Research UK, and Nicola Perrin, Head of Policy, Wellcome Trust, gave evidence.

Q37Chair: Welcome, everyone. I apologise to those who are standing for the proceedings. We will try to cope and make it as comfortable as possible for everyone. Can I welcome the witnesses? Thank you for coming this afternoon. For the record, I would be grateful if you would introduce yourselves.

Nicola Perrin: I am Nicola Perrin, head of policy at the Wellcome Trust.

Dr Elliott: I am Catherine Elliott, director of clinical research interests at the Medical Research Council.

Professor Johnson: I am Peter Johnson, the chief clinician at Cancer Research UK.

Sharmila Nebhrajani: I am Sharmila Nebhrajani, chief executive of the Association of Medical Research Charities.

Q38Chair: Thank you. We have got quite a lot to get through in a fairly tight time frame, so I am going to try to rattle through some questions. First, how much medical research do you fund annually? What proportion of this relates directly to clinical trials of drugs, devices or other types of therapies?

Professor Johnson: Cancer Research UK funds at any one time a portfolio of more than 240 clinical trials, of which about a third are recruiting internationally as well as within the UK. Around 80% of those involve an investigational medicinal product of one description or another, and around 5% involve testing novel techniques, such as radio therapy or surgery. We have a large portfolio of trials, which gives us a unique insight into the process and the regulatory impediments to it.

Q39Chair: About a third are non-UK-based.

Professor Johnson: A third recruit internationally as well as in the UK. They all recruit in the UK.

Sharmila Nebhrajani: Perhaps I should go next representing the medical research charities, some of whom are also at the table. Medical research charities fund about £1 billion of medical research each year. I am here representing 125 members. About a third of those-41 of our 125 members-fund clinical research of some description, including clinical trials but also non-medicinal interventions, new diagnostics and so on. AMRC members funded 32% of all studies taking place in the NIHR. That is about 2,600 studies, which are by their definition clinical studies. Of those, 14%-or about 360 studies-are clinical trials in phase I to phase IV of investigational medicinal products, so that is the scale of it across the charitable sector.

Dr Elliott: The Medical Research Council currently has about 120 live clinical trials. As with CRUK, those are both in the UK and are also part of our global health portfolio. Our current commitment is around £105 million. Our annual spend over the past five years has been, on average, about £20 million on that specified portfolio of trials. We also spend about £50 million every five years on intramural units that deal with clinical trials, methodology and statistics. Our portfolio has changed due to NIHR taking over the larger phase III studies in UK through their health technology assessment scheme, and we now fund more in the early phase trials, and also phase II trials, which are administered by NIHR through their EME scheme.

As with other funders, the majority of the trials would be of medicinal products, but we also have a large portfolio dealing with surgery or interventions. Through the NPRI initiative, we also fund prevention initiatives, such as interventions for weight loss or stopping smoking.

Nicola Perrin: The Wellcome Trust funds approximately £600 million a year of biomedical research. Of that, a relatively small proportion of our funding involves clinical trials; it is less than 5% of the total funding that we provide each year. Within the UK, most of the trials that we fund are not for medicinal products; they are more often for behavioural interventions and psychological therapies-for example, cognitive behavioural therapies. A significant proportion of the clinical trials that we fund are global. For example, we fund in partnership with MRC and DFID the global health trial scheme. We also fund a number of earlier stage trials in low and middle income countries, including, for example, the recent trial of a new vaccine for TB. We do not run a dedicated trials unit or programme and we are not a sponsor of trials; we are a funder.

Q40Chair: In your case, there are some crossover cases where you are funding projects here that are working in parallel elsewhere. Are the data a little complicated in the case of Wellcome?

Nicola Perrin: The data are, as you say, very complicated, particularly since a number of the awards that we fund are large strategic awards, or for our major overseas programmes we give a block grant and a number of different trials take place as part of that. We are now trying to improve our reporting systems to have a better figure of the number of the actual trials that we are funding, so it is not as easy as we would like.

Q41Stephen Metcalfe: In your opinion, what is the greatest barrier to conducting a larger number of trials here in the UK? Is it EU legislation, or are there specific issues that are challenges here in Britain?

Dr Elliott: We have produced statements across the funders; maybe we have similar views. There is no doubt that the European regulations have added complexities and delays to conducting clinical trials in the UK. However, in the UK, the trials regulations are overlaid on to an already complex regulatory framework. The report of the Academy of Medical Sciences and our own experience would be that that home-grown complex framework contributes to the complexity. Probably the most difficult bit relates to conduct across multiple NHS sites and the requirements for that, and, even more particularly, probably issues relating to patient data, access to data from health records and the longer-term follow-up of records.

Sharmila Nebhrajani: I will not repeat the points that Catherine has already made, but in addition there are two other issues, one of which is the capacity in the UK to conduct clinical research. We know of members-for example, Action on Hearing Loss-who would like to fund UK trials but cannot find the capacity to do so and are funding about £1 million worth of research overseas for that reason.

The other issue is access and knowledge. For example, we know that almost three quarters of people surveyed want to be involved in research. We also know there is a demand for that opportunity but there is a supply problem. There was recently an NIHR mystery shopper survey that went around trusts to ask, "Where are the opportunities to fund research?" Patients want to do it, but they have no idea how to get on it. There is definitely a knowledge and access barrier also.

Professor Johnson: I have personal experience because I conduct trials, both first-in-man and large randomised trials, and have been involved with both academic and pharma company-sponsored studies. My general view is that we have a very vibrant clinical research environment in the UK and have made enormous progress over the last 10 years with concerted investment in this area. Moves such as the embedding of research into the NHS constitution have been extremely positive and helped to move things forward and build capacity. The regulatory hurdles are well documented, and certainly when the European Clinical Trials Directive was first implemented it was a major impediment. We have made a lot of efforts collectively as a research community since then to improve things. Looking at our data and the information we have about the length of time it takes us to get trials started, there is evidence of success, but I have a sense that that has plateaued recently. We still need to put continuous emphasis on improvement, and, very importantly, also to be careful of avoiding the unintended consequences of additional legislation and regulations.

Nicola Perrin: As well as agreeing with the previous comments, the creation of the Health Research Authority has been a very important step forward. We particularly welcome its feasibility study to try to simplify assessments and have a single assessment that all NHS Trusts can accept and trust. We look forward to seeing its pilot study, which is coming shortly.

Q42Stephen Metcalfe: Do you see the EU clinical trials regulations draft proposals that are coming forward adding a greater degree of complexity, or are you hopeful that they will build on the progress you have already made?

Dr Elliott: I am hopeful that they will build on the progress. The UK-particularly the MHRA, which is now working with the HRA-has led a lot of the work on reducing this complexity in the regulations. We strongly welcomed the initial draft adopting a more risk-proportionate approach and allowing greater harmonisation.

Sharmila Nebhrajani: It is a development in terms of trying to find a regulatory framework that is both a bit more agile as well as more proportionate, but we must not forget that science is advancing more quickly than that. You have to look at things like stratified medicine. Doing work on smaller sample sizes is going to require a greater degree of cross-working across boundaries, so, in a way, our clinical trials regulatory framework needs to be even more agile ahead of the way in which science is advancing.

Q43Stephen Metcalfe: Do you feel those particular issues are being listened to by those drafting the regulations? Who is driving this process? Is it being put in place to help conduct safe and proportionate clinical trials, or do you get the sense that it is there to control you? Who is driving this?

Sharmila Nebhrajani: Others may comment. It is clear that the philosophical basis is correct. It is trying to be properly regulatory and protective, but, if the unintended consequence is that these new scientific advances cannot be developed into meaningful trials, then we have a problem.

Professor Johnson: The way the regulation was initially drafted looks very promising in terms of harmonising the establishment of trials-for example, across the different countries of the European Union-to make it much easier to get trials up and running. There is a good deal of detail that we need to see to know how it is going to work in practice. In general, the shape of the thing and the proposal to take a risk-adapted approach, for example, to the regulation of trials is very positive, but we need to be mindful of the detail that is going to come through.

Nicola Perrin: One of the problems with the previous directive is that it had a disproportionately negative effect on academic clinical trials, and we have been very pleased to see that, in the new draft regulations, that is one of the things that has, hopefully, been addressed. There are still some questions about scope and we would welcome greater clarity about some of the key definitions, but it is moving in the right direction.

Q44Stephen Metcalfe: The other side of conducting clinical trials is recruiting patients, which you have touched upon. Is there a particular challenge in recruiting patients in this country? Is there something we can do through better use of NHS data, and, if so, what would that look like?

Nicola Perrin: The biggest problem with recruiting patients in respect of using patient data is the so-called consent for consent issue. To identify whether or not people are approved for trials, researchers may need access to identifiable information-for example, date of birth, postcode and so on. Initially, they would not have a legal basis to access that information, so either you have to rely on the clinicians, who clearly have other understandable priorities and do not have time to do it, or you have to contact everyone to ask if it would be okay to look through their information to re-contact them to ask them to consent. That is a hugely complicated process for something that everyone wants to happen-and patients want it to happen.

We have been talking about this as a barrier for a number of years. We are hopeful that we are beginning to see improvements in it. The recent revisions to the NHS constitution, hopefully, will be very helpful but need to be accompanied by much better information to patients to help them understand what it means and to explain it better. The Caldicott review will be reporting shortly and we hope it will address this issue to some extent.

The other way it could be improved is through the CPRD-the Clinical Practice Research Datalink. It has some very exciting technologies that make use of non-identifiable information; there are identitiers behind the scenes that are looked at only by computers. If that provides the solution that it promises, it will be enormously exciting.

Q45Stephen Metcalfe: Does anyone want to add to that?

Dr Elliott: I completely agree with all that Nicola said. The steps in the NHS constitution are very helpful, but we probably could go still further in helping patients to be aware of clinical research-and for those patients who would like to take part to know what is available. CRUK has done a great job with CancerHelp in terms of making that available, and the UK Clinical Trials Gateway has also taken us down that line, but we could do more.

Sharmila Nebhrajani: I agree; I think there is more to do. There are a lot of resources out there, but they are not necessarily well found by patients. There is definitely a marketing and access point. The UK Clinical Trials Gateway is great, but in a recent survey 80% of people had not heard of it. So you have demand for trials but no way that people can understand how to get on to them. The NHS constitution is also a good development here. There is something about, culturally, clinicians being trial-friendly and understanding their obligations to make opportunities available to patients, and patients having a right to be informed about trials, particularly vulnerable groups. The elderly and ethnic minorities are much more poorly represented in those populations than the norm. There is also something about the kind of practical help that the NHS or charities can give to facilitate access to trials. We have good examples of some of the charities working with Alzheimer’s and Parkinson’s trying to make the practical logistics of access to research work for patient groups.

Q46Stephen Metcalfe: It strikes me that there is not very much public awareness of the whole issue of clinical trials until you arrive at a point where you might need to be involved. Do you think we should do more to raise general awareness among the well, not just among the sick, that clinical trials are things in which they may take part some time in their lives, similar to the way we have raised the profile of organ donation to the point where people now are much more accepting of it, so that it does not come as a shock when perhaps they are invited to take part in a clinical trial? Is there some way we could do that?

Professor Johnson: That is a very important point. Different groups of people have different levels of enthusiasm to take part in research. In general, people diagnosed with cancer have a very high level of motivation to take part in research. More than 37,000 people a year take part in one of Cancer Research UK’s trials, which is about one in five of people diagnosed with cancer. People with a diagnosis find it easier, but it is a very difficult time to be making those kinds of decisions and to get on the internet to try to find the information. A much higher level of awareness among the general population would be enormously helpful and it is something on which we have been doing a lot of work. Our CancerHelp website is a means by which we promote the clinical trials both to the well and people with cancer.

Dr Elliott: That would apply particularly to clinical trials in this context but also, as has been alluded to, the use of tissue for research as well as organ donation and lawful use of data. Again, more awareness might make this research easier.

Q47Chair: In some cases where trials are conducted, the patient is told about what is going on but in a fairly negative way. I have seen a document that says, "We are carrying out this procedure on you. This is how to opt out of participating in sharing that data," rather than first explaining why the data is being collated. Is that a managerial or regulatory issue?

Professor Johnson: There is an art to the construction of a properly put patient information sheet. Increasingly, we draw on help from patients and carers to help us write information sheets in a way that is readily comprehensible. Unfortunately, in investigational medicinal product-type trials, the volume of information required is quite high.

Chair: I am afraid there is a division just now. I am going to suspend the sitting for 15 minutes, assuming it is only one division; if there are two, with apologies to the witnesses, I will suspend it for 25 minutes.

Sitting suspended for Divisions in the House.

On resuming-

Chair: Can we resume, ladies and gentlemen? Apologies for the delay. There were two votes. We will go straight on, and I ask David Tredinnick to ask a question.

Q48David Tredinnick: What proportion of the trials that you fund are pre-registered and published in peer-reviewed journals?

Dr Elliott: We looked at our portfolio of trials from 2006 to 2012. Of the studies that we identified correctly as trials-it sounds a bit long-winded, but doing that has been slightly complicated-and those funded before 2012, 94% have been confirmed to be registered. Of the total portfolio, 86% are either registered or are soon to be registered, because some of them have been funded quite recently.

Q49David Tredinnick: Are you happy with those levels?

Dr Elliott: Our requirement is that all trials that we fund are registered, but we are looking at the gap of the trials that are not registered. We know that for some of them it is because they have only quite recently been funded. We are doing further work to look at the reasons why trials might not be registered from that group. The most common reason seems to be a difference in definition of trials-whether we consider them as trials in our portfolio or whether the researchers considered them as trials-because these are all trials, not just investigational medicinal products. So the definitions can be quite variable. We would like all of those that we consider to be trials to be registered, so that should be 100%.

Professor Johnson: It is a condition of funding by Cancer Research UK that any trial is registered. The way we police that is to check at the first annual renewal of the grant that the trial is indeed registered, so we think we have 100% take-up rate for registration and very high rates of publication. Inevitably, especially large international trials may take some years to complete and analyse the data. Of the rather more than 300 trials that we think have been completed over the last five years, we know that somewhat over 200 have already been published and the remainder are in the process of being analysed.

Sharmila Nebhrajani: From the perspective of charities that do not have their own clinical trials but are funding them, which is the point of your question, 80% that we know are funding clinical research have a requirement in the terms and conditions of their grant that the output must be published, and it is long-standing advice and guidance from the association that all charities should do that.

Q50David Tredinnick: Does it worry you that some of the drugs that are approved go on to be prescribed as packages that have never been assessed? You get three or four drugs being used at the same time.

Professor Johnson: Do you mean combinations of drugs? Certainly, in cancer treatment, combination chemotherapy is very much the standard approach to treatment. So, often, we will have three or four drugs in combination routinely being used, and it is those combinations that are subjected to clinical trials as a general rule. I think we are comfortable with the data that we have on those combinations.

Q51David Tredinnick: How do you monitor and enforce compliance with your stated transparency policy?

Nicola Perrin: From the Trust’s perspective, we require all our funded trials to be registered and we expect publication as well. It is not an area that we have actively policed until now, but we have just updated our policy and, as a result of that, we are giving it greater attention and priority. We are implementing new reporting systems so that we can track trials through our e-val mechanism, but at the moment we do not have a figure of compliance.

Q52David Tredinnick: The National Institute for Health Research health technology appraisal programme has achieved an extremely high level of transparency by withholding funds prior to trial registration and maintaining its own dedicated log. Do you see potential for this approach to be more widely used across the non-commercial research community?

Dr Elliott: Like Wellcome Trust, we have also implemented new systems to monitor outcomes in publications from research, so we also require that the results are published and that data is shared with other researchers. From the sample we have looked at, about 89% of the trials have been published to date and there is an ongoing trail of publication. We do need systems-which we call Researchfish, which is a very similar system to e-val-to monitor annually the outcomes of research and the publication rates.

Sharmila Nebhrajani: There is collaboration between the MRC and the AMRC to roll out that Researchfish project across charities so that you start to have a common language among non-commercial funders to track impact, not least because a lot of charity funders are funding in partnership with others. That will be a helpful process.

Q53Roger Williams: Could I ask your views on whether trial registration should be limited to randomised control trials for investigational medicinal products, or should all forms of trials for all sorts of interventions be registered?

Dr Elliott: From the preliminary review we have made, we have seen the difficulty of defining trials and tracking them through these systems. Clearly, the RCTs of IMPs that you mention have to be registered under statutory requirements, but we would agree that all clinical trials should be registered. It is probably more difficult sometimes for researchers to do that and find the appropriate registry, and we could do more to help identify the appropriate registries and simplify that registration process.

Professor Johnson: In general terms, it is a sound principle that all trials should be registered if patients are giving informed consent to take part in them and the data is being collected. CancerHelp UK, which is our information database, collects and puts up information with good lay summaries, not just for the trials that Cancer UK funds but for as many cancer trials as we can find. We continue to update that on a rolling sixmonthly basis so that the information is as current as we can possibly make it and, furthermore, so that we capture the outputs of trials as they are either presented at international meetings or get published in the literature. That is the way in which we maintain a high level of transparency and visibility of both the trials activity and also the results.

Nicola Perrin: Registering trials is an essential first step towards transparency, so we would certainly agree that it should be much wider than just CTIMPs; it should be all clinical trials, and that is why we have that as our policy. As I mentioned before, a lot of the trials that we fund are not for new medicinal products.

Sharmila Nebhrajani: It has been said before that transparency is a bigger agenda than just registration of trials. There are some trials that do not lend themselves to randomised placebos-for example, surgical interventions.

Q54Roger Williams: What trial registries do you recommend that your grant-holders use, and how well suited are these registries to nonIMP studies?

Dr Elliott: Our grant-holders could use any of the range of trials registries. The trials that come under the European directive will be on the EUdraCT database, and the American US trials have to be on the dataset. We also have the ISRCTN register. Under the WHO Clinical Trials Registry Platform there are 14 other registries. There is a plethora of registries, and it may add to the complexity that there are these different resources. We do not mandate or stipulate that a particular one must be used, but one must be used.

Nicola Perrin: On the other hand, we do specify that our researchers should use a subset of the metaRegister of Controlled Trials. The benefit of it is that it covers all types of clinical trials, not just medicinal products. It also means that they get a unique identifier-the ISRCTN. There are benefits to it. We are looking at whether it is still fit for purpose. It is not a user-friendly register by any means. If you know what you are looking for, you can just about find it; if you are a patient looking for a trial, you would not go to that site to find it. The CancerHelp registry is excellent; it is a really good resource; our current one is not. We have provided extra resource to help them improve the lay summaries. We need to do much better to make it more accessible, and we are very keen to work with other funders and users of registries to try to develop a broader and more accessible approach.

Sharmila Nebhrajani: I would support that.

Q55Roger Williams: Professor Johnson, where do you get your information from for the CancerHelp UK register?

Professor Johnson: The information that we put out there is written in-house by a dedicated team. We invest a certain amount of the resource that we collect to make that available. About six whole-time equivalent staff are continuously trawling the different databases looking for trials and the information, and then we write the information sheets for the trials that we put up there, usually in concert with the researchers. We check the accuracy of what we are putting up with the researchers conducting the trials to make sure that we have got it absolutely right and that we put it in terms that are readily understandable to the people arriving on the website.

Q56Stephen Mosley: In response to earlier questions, you said that you produced reports and published 89%-in some cases all-of the results from clinical trials. Are the reports you produce full ICH-compliant clinical study reports, or are they of a different format?

Dr Elliott: When we refer to "publication", that would normally be peer reviewed in an academic journal. That would not normally be in the full ICH CSR format, because that is a very specific format required for marketing authorisations. We require that authors and researchers follow the CONSORT guidance. MRC and others fund groups to provide guidance as to what should be in a publication so that other researchers and the public can see the protocol being followed, and there is a set format in which a trial should be published. We mandate that our trial should be published in that format, but it would not include all of the information that would necessarily be in the very tight definition of the CSR in the ICH.

Q57Stephen Mosley: It is that tight definition in which I am specifically interested here.

Dr Elliott: A lot of academic trials are not directed towards a new marketing authorisation for a drug, so that would not be produced as a matter of course. Our MRC clinical trials unit, which does a lot of trials and sponsors on MRC’s behalf, estimates that it takes about three months’ additional work to produce one of those reports, over and above all of the statistical analysis and data that it will produce for an academic publication.

Professor Johnson: It is also important to dispel any misconception about how useful this data is. This data in its unedited and unanalysed form is not necessarily particularly useful. Despite the fact that it may take a lot of time and resource to produce, there is a risk of an adverse consequence of putting out a lot of that kind of data, which is that we are simply swamped in largely meaningless data. In general terms, our clinical trials units, of which Cancer Research UK core funds eight, has not felt that this would be a good use of its time. There is an opportunity cost to deploying resources to produce this data, which is difficult to interpret and analyse, versus doing more trials and doing them faster.

Sharmila Nebhrajani: I wanted to make the link back to the access point we were discussing earlier. From a charity funder’s perspective, they require the outputs of their research to be published in peer review journals and, in many cases, also require that they are in open access models, like UK and Europe PubMed, which is very important, because that information is then freely available to researchers and the public. Alongside that, on the point about access, many charities also publish a lay summary of the research on their own website. That is not a report for the hard of thinking or the stupid-but an insightful lay indication of what the research says, why it is important and what its shortcomings are, not least because charity funders are funded by the public. So it is a kind of feedback loop that says, "This is what your money funded; this is what we did with it; and this is the next iteration of research that it will fund."

Q58Stephen Mosley: It would be standard practice for you to produce this lay report.

Sharmila Nebhrajani: It is our guidance that this may be a helpful way of ensuring the outcomes of research are accessible to patients and the public.

Nicola Perrin: We do not specifically require CSRs to be produced unless they are needed for marketing authorisation. We require our researchers to make their research findings freely available through open access-both positive and negative findings.

Q59Stephen Mosley: No doubt you will be aware of the January 2013 report on the draft EU clinical trials regulation and the proposed amendment that, within one year from the end of a clinical trial, the sponsor must submit to the EU database a clinical study report, including a lay summary of the clinical trial. I guess from your previous answers that there is some scepticism about that, but could you just put your thoughts on that on the record, please?

Nicola Perrin: Where a CSR has already been produced, we see no reason why it should not be made available, although we would clarify that by saying that, if the EU database is publicly available, we would have to redact any patient-identifiable information, which means that, potentially, you lose some of the value of the CSR. In any event, it would introduce a huge additional financial and time burden on non-commercial researchers, so from that perspective we certainly do not agree with that amendment. The other important point is what you are trying to achieve by it. If you are trying to increase access to underlying research data for the purposes of validation and scrutiny, the CSR is not the best way of doing that. You need the data in a reusable form that can be fully analysed, and a PDF is not the way of doing that.

Dr Elliott: We completely agree with that. We have been having some discussions about what would be the purpose of publishing a CSR or making that a blanket requirement across all trials, and not just those that already have a CSR produced. We would agree with Wellcome’s position on where CSRs are already produced. The key thing is: what is the end purpose? The academic trials will also have statistical analysis, plans and details of the analysis conducted, and the datasets that have been used, for those that could be made available. Again, we would have concerns about the individual level data being made available, but there are ways that the same information could be made available and the ends could be met, both to allow the researchers to do further work or to allow the public greater insight into what has happened in a trial without mandating CSRs to be produced for all trials.

Professor Johnson: Our feelings are very similar. Proportionality here is very important. There is a sense of history repeating itself. The original Clinical Trials Directive set out to improve safety reporting and quality, but because it did not take a proportionate approach there were enormous difficulties with the running of trials, and safety and quality were not greatly changed. We run the risk of doing the same thing with this requirement for blanket data publication in this way. This is not to say that individual patient data is not already being used in analyses, and all our clinical trials units are continuously making individual patient data available for meta-analysis for further studies by different groups, so the data can be obtained. The question is how much resource one would want to divert into doing that for every single trial to no particular benefit in many cases.

Sharmila Nebhrajani: I would echo the comments made by others. There are already good examples of data sharing between researchers. Tissue banks are a good example of where those things are made available freely. So we do have models that work. The philosophy of sharing data is a good one, but one does not want to do that without regard to the overhead and the cost to research funders, because the opportunity costs of that are significant.

Q60Chair: Is cost the only downside risk, if you like, associated with creating better anonymised datasets?

Dr Elliott: Cost is certainly one issue, and, particularly when we are talking about producing a clinical study report, that is probably one of the key concerns. Perhaps you are talking more about releasing the identifiable or individual datasets, even if they are anonymised. There is, again, an issue of cost in there. There are also the concerns that appropriately to anonymise that dataset would take a lot of resource and may not achieve the end of ensuring that there is appropriate privacy. In order to do that, the data may become much less useful for other researchers by the time the measures have been taken to ensure privacy. We think there are better ways to ensure that data is shared between researchers. We mandate that data should be shared from MRC studies, but there may be better ways to do that than releasing datasets in a publicly-available situation, where they have to be very stringently deidentified.

Nicola Perrin: We would agree with that. We would prefer more limited controlled access to a fuller dataset. In particular, if you are doing research into rare diseases, it is very difficult not to have the identifiable data or data that could potentially be identifiable. We would suggest it would be much more appropriate to have a tiered access model, where researchers can access the full dataset, but with independent review and a sort of access committee to approve that research first.

Q61Chair: Who provides the independent review? Is there a sort of gatekeeper role to be played here? If so, who is that?

Nicola Perrin: There needs to be a gatekeeper role. We need to explore models to deliver that. There is not a solution at the moment, but we need to look at a consortium that involves industry, academia, funders, Government and regulators. We held a workshop at Wellcome Trust last week with a cross-sector group of stakeholders, including all those different parties. There was agreement that we need to explore workable, practical solutions to this, but the key conclusion was that whatever is decided needs to be a global approach. The UK could and should take a leadership role, but we cannot go it alone. Whatever is the mechanism for accessing data needs to be a global solution.

Dr Elliott: I completely agree with the gatekeeping role, but, if this is done properly, it is facilitating research and in some ways it is also a gate opening, because it is making sure that the research teams requesting data get the right data in the form they need, appropriately annotated and linked back to the original research, while maintaining patients’ privacy and their additional consent requirements. It is not simply to delay access and scrutinise it-but also to make sure the right and most informative datasets are released.

Professor Johnson: I entirely concur with that. We would support a requirement to make the data available on suitable request, provided that request is accompanied by a protocol of what is going to be done with the data, how the analyses are going to be conducted and, very importantly, how those results will subsequently be made public by the people conducting the analyses.

Chair: Finally, is there anything else you want to add before we close this session, as we were slightly disrupted? No-silence all round. Can I thank you very much for your attendance? I apologise again for the disruption.

Examination of Witnesses

Witnesses: Dr Bina Rawal, Director of Research, Medical and Innovation, Association of the British Pharmaceutical Industry, Dr James Shannon, Chief Medical Officer, GlaxoSmithKline, William M Burns, Member of the Board of Directors, Hoffmann-La Roche, and Dr Ben Goldacre, Wellcome Research Fellow in Epidemiology, London School of Hygiene and Tropical Medicine, gave evidence.

Q62Chair: Can I ask the second panel formally to introduce themselves?

Dr Rawal: I am Bina Rawal. I am research, medical and innovation director at the Association of the British Pharmaceutical Industry.

Dr Shannon: I am James Shannon. I am chief medical officer at GlaxoSmithKline.

William Burns: I am Bill Burns. I am a non-executive director on the board of HoffmannLa Roche in Switzerland, and for the years 2000 to 2009 I was the CEO of the pharmaceutical division of that company.

Dr Goldacre: I am Ben Goldacre. I am a doctor and academic. I write about problems in science, and I am the co-founder of AllTrials.

Q63Chair: I thank all four of you for coming. First, I put a couple of specific questions to Dr Rawal. Could you provide us with a brief overview of the current state of the health of the pharmaceutical industry in this country, focusing on the UK’s position as a location for clinical trials?

Dr Rawal: I will certainly give it a try. The UK has suffered in recent years as a choice destination for clinical trials. A number of factors fed into that, one of which was the implementation of the EU Clinical Trials Directive, with the increase in bureaucracy and complexity that that brought. Overlaid upon that over the last decade or so were bureaucratic issues specific to the UK. The UK, being devolved in terms of NHS trusts, each having its own processes for getting clinical trials started and off the ground and contracts negotiated, meant there were perhaps increased hurdles here compared with some other countries.

Having said that, over the last two or three years, particularly with the investment put in by the National Institute for Health Research and the Office for Clinical Research Infrastructure that has been active, and also with the creation of the Health Research Authority, there now appears to be an upturn. The factors that are feeding into this are not only the excellent scientific base here in the UK; there is now a lot of impetus both at the level of early stage translational research but also in terms of moving forward with late stage clinical trials. Investment in the Clinical Practice Research Datalink and other initiatives to draw on the richness of NHS data and diversity of the population in the UK, and to facilitate that by putting in place mechanisms to do so, has helped enormously.

The NIHR NOCRI office, as well as HRA, have worked very closely with the industry, and the ABPI has helped streamline on several fronts, including the development of model clinical trial agreements and so forth.

Q64Chair: The second question is to Dr Shannon and Mr Burns. In each of your companies, what proportion of your company’s global clinical trials is conducted in the EU and the UK?

Dr Shannon: For GlaxoSmithKline, I cannot specifically answer the proportion in the EU as a whole, but, as far as the number of patients is concerned, overall about 4% of our global patient population comes from the UK. A lot of that is driven heavily by a single study using electronic health records and the health system in Salford. If you exclude that single study, it comes to about 1% in total of our patients.

As to what drives us in the location of trials, all of our trials are run globally. Approximately 25% of our trials in total have UK representative centres. As to those things that drive us in the location of trials, the first is quality of the data and the science, where the UK scores highly. Secondly, it is the ability to compare against the most modern medicines available and the best medicines of standard of care, where sometimes the UK does not score highly. Thirdly, it is the complexity and time to initiate the study, where other countries may be faster to initiate and less bureaucratic and complex-and, hopefully, the new legislation will help with that. Fourthly, we come to the cost of running the clinical study.

William Burns: It is remarkably similar from the Roche window. The majority of trials are run globally. As an order of magnitude, but not a precision, roughly 40% US, 40% Europe and 20% rest of world would be a rather typical profile to run a major clinical trial. The one thing that one always admires within the United Kingdom is intellectual curiosity, and that seems to play particularly to translational medicine and some of the early medicine. It is probably a real strength of the UK, and it is certainly where I know many of the people in OSCHR have put a particular focus.

For Roche, the proportion actually within the United Kingdom is 2% to 3% of patients. This is strongly influenced by many of the points that Dr Shannon has mentioned, particularly the very large proportion of the work of the company in the cancer field, the complexity of cancer itself and the cocktails required as a standard of care. The adoption rate of new treatments in the UK in the cancer field particularly has lagged behind. With Mike Richards’ help and a number of other initiatives, that has improved, but there is still a difficulty if you are looking at an investigational drug and a tougher standard of care. If standard of care is not widely used in that country, it becomes more difficult, more costly and slower to conduct the clinical research.

Q65Chair: Can we try to home in a little more on the UK? How many clinical trials have you got going today in the UK, and what are the main therapeutic areas?

William Burns: For Roche at the moment, to frame it from a global perspective, there are over 320,000 patients actively involved within a clinical trial programme of 2,280 trials, in 35,720 health centres around the world. I do not have the equivalent numbers for the UK, but, out of a global research and development expenditure of approximately 8 billion Swiss francs, 400 million is spent through the UK.

Dr Shannon: For GSK, the overall total patient numbers and number of centres would be roughly similar. We have 59 active trials ongoing in the UK, which represents about 25% of our trials in centres worldwide and 1% of the patients or 4% of the patients.1

Q66Stephen Metcalfe: You have covered some of what I wanted to talk about. Can I talk about what would make the UK a more attractive place to conduct clinical trials? In the opening gambit, we heard about the fact that there is a slightly toxic mix of EU legislation and UK bureaucracy that has put people off. What could we do to improve that situation to be able to attract more trials here?

Dr Shannon: As we look at starting a clinical trial, in many cases speed is of the essence in the ability to get a trial from the point of deciding to do a trial to the first patient being enrolled, and then patient enrolment at a rapid rate. We have to step through a number of hoops. In the UK, in the past, we had to go through the EU Clinical Trials Directive for the individual country, followed by the MHRA, ethics committees and individual trusts, and those individual trust discussions took a long time. I believe the HRA has started to help on the complexity as far as ethics committees and the initial portal are concerned but really has not dealt with the individual trust contract negotiations, which sometimes remain longsome and difficult.

William Burns: Interestingly, the European Clinical Trials Directive has been implemented in a variety of different ways, depending upon which country you are in. For example, the Dutch took the directive and implemented it with a complete lightness of touch because they felt that much of the prevailing legal framework, predominantly Roman law, addressed all the principles. The view of the Belgian authorities was that the current legal framework met the need and they did not feel the need for an additional framework. We, with our Anglo-Saxon legal system, try to cover every eventuality-positive or negative-and outcomes, and that does result in a more complex legal framework. If something is not covered, you can get some freezing in the system as to, "Can we navigate our way through here?"

The new directive coming from Europe is probably an opportunity with the new portal. Offline with Dr Goldacre there was some discussion. I agree it is somewhat vague at the moment, but it is an opportunity for a simplification and streamlining as we bring it from a directive into UK national law. That would be one area where we could get improvement.

A second area would be a great initiative through the review that has led to the HRA being established. I hear back from colleagues as to whether they have the resources to do the task that we have empowered them to do-and it is probably the same with NIHR-to make sure that we do not start something well but the logical follow-through is not there. Probably one of the practical consequences of the question would be that I would look at whether the HRA has the resource to help simplify the trial framework for the United Kingdom.

Dr Rawal: I understand that the HRA is currently undertaking a feasibility exercise to look at a single process for getting ethics review and R and D approvals, with only the specifics having to be determined at the local trust level-for instance, whether there is a fridge in the pharmacy. But, for other matters and certainly a lot of the contractual matters, they are trying to undertake a feasibility study and will have some idea of how best to streamline the whole process by about the summer time frame.

Just in terms of putting numbers on it, from the previous question, in 2011 and 2012 there were approximately 500 clinical trial approvals issued by the MHRA, with industry-sponsors. The total number was about 200 higher

Q67Stephen Metcalfe: Is patient recruitment an issue in the UK?

Dr Rawal: The first focus for the HRA will be on time to start. They have made it clear that they are going to aim for a 70-day target from receipt of a validated application. That will be their primary focus. In terms of recruiting to time and target efficiently and smoothly, yes, it will be important that they can demonstrate that and collect the metrics that can carry some weight in terms of UK affiliates talking to their global colleagues with the metrics to show that it is being achieved. In terms of the outlook being good, at our conference last November, a big CRO was able to present information that supported that there was now an upturn.

Dr Shannon: Patient recruitment depends on many factors, including the interest or novelty of the active agent under investigation. The more interesting, novel or greater patient benefit that may be derived, frequently patient recruitment is better. If we look at how clinical research is conducted around the world-for example, in the United States-the process of clinical research is heavily driven almost by a business within the clinical site, where there are administrative staff focused on driving patient recruitment, managing those patients’ records and so on. In the UK, it tends to be an add-on to a regular clinical job and it is less professional in the way it is conducted compared with some other countries.

William Burns: I do not sense from my colleagues any hesitancy from patients themselves. It is much more to do with the timeline for ethics committee clearances. Is the infrastructure there to facilitate the process for a patient? People are people, and any of us as citizens would want access to what we believe is the best standard of care we can get. I do not think there is any hesitancy from the British public; it is an issue of process.

Q68Stephen Metcalfe: To what extent is the culture of the NHS a barrier to conducting more trials? I know you have covered some of that, but do you see anything else?

Dr Rawal: As has been alluded to earlier, clinical research is not embedded sufficiently as part of the work of NHS trusts in general, although there are increasing signs that it is changing, with the creation of the Academic Health Science Networks and the onus on them to undertake research and collaborate across the sectors to do clinical research. Testament to the fact that it is changing is the information that Dame Sally Davies, chief medical officer, presented at our conference last year that 99% of NHS trusts had entered a single patient into a clinical trial last year, and approximately two thirds or 60%-I do not know the exact figure-had entered a patient into a commercial trial. It will take time for the culture to change, but it is evolving and the outlook is good.

Q69Chair: To push you a little further on that, since Sally Davies made those comments, we have a new structure in the system. Does that strengthen or worsen the engagement of people in the trials?

Dr Rawal: Optimistically, it ought to strengthen it. I have mentioned the Academic Health Science Networks, but the focus on information, sharing of data and the open data platforms that are being created-the HSCIC and CPRD, to name just a couple of sources of large datasets-will also facilitate more clinical research happening.

William Burns: To be eminently practical, this is not driven from the national health service, but, in running an organisation, if you make structural changes to it, you disrupt the natural rhythm of the place, and it takes a while for the rhythm to re-establish itself. Until it is clear what the rhythm is of the new world, given that trials are not a mainstream activity, undoubtedly I believe it will take a little while for this to show through in numbers that we can all quote to you.

Q70Chair: I thought you were indicating you had something to say.

Dr Goldacre: The General Medical Council has recently revised its guidelines for best clinical practice and removed the obligation for doctors to work to reduce uncertainties about the effects of treatments wherever possible, which is a concerning backward step, especially since at the moment there is such a huge amount to be gained in the UK from using the electronic health records of 60 million patients and only a minority of general practices participate in the Primary Care Research Network. In the trial that I am involved in, which is a low intervention trial comparing two statins, we have been able to recruit only about 10% of the general practices that we have approached.

Q71Stephen Mosley: Can I move on to regulatory reporting? We saw a study last year in the US that suggested that only 22% of clinical trials had been properly and fully reported after 12 months. I know that the ABPI has also done a study that suggests it is a lot better in Europe. Even so, I think you suggested that 79% had fully complied, which still means that 21% had not. In terms of your companies, ABPI- and then I will come on to Dr Goldacre-in what proportion of your clinical trials in the recent past, let’s say five years, have you reported in advance and reported the results within 12 months?

William Burns: We are in a similar bandwidth. You have different datasets coming through at different times. The first priority that we give to the dataset is to satisfy a dossier for the regulators, which will include all the data. For that data, which goes down to individual patient data, as far as the United States is concerned, the FDA wants to see all the patient records, having previously discussed the prospect of statistics that will be applied to the study. Other Governments want to see only summary reports, so there is a major effort. That is our first obligation, if you like. That data has been given to us and generated to try to see whether this should or should not be a medicine. Does it pass first licensure, or not? Hot on the heels of that is the translation of that into publications, reportage at medical congresses and so forth. We are within that bandwidth of reportage. I have seen the figures getting better in recent years. This year it is about 90%. I am still not satisfied. We would want to strive towards 100%.

Dr Shannon: We report at a number of different levels. If we start with clinical trial registration, we require that all trials where we require informed consent are registered on a registry, namely the GSK Clinical Trials Register or and/or the EUdraCT database. We are close to 99% to 100%. You may ask how I know these numbers. I monitor this on a monthly basis. We have a transparency council. I have a transparency report, and I watch it on a monthly basis personally. On registry, I know we do that.

As soon as a trial of a medicine is complete, we require that within eight months of the end of it we will publish a result summary on the GSK clinical trials results database. We are 95% to 97% compliant with that. We require that all studies are published or submitted for publication within 18 months, and I know, again, that at the end of 2012 we were at 97% compliance on that record.

As far as concerns submitting clinical study reports to health authorities, that depends on whether we have so far submitted an application for a product licence. We supply all safety data as part of the clinical study reports, and all serious adverse events are reported within 15 days,2 if they are unexpected, or again within the clinical study report. That is where we stand.

Dr Rawal: To clarify, reporting within the BMJ, of 22%-only a fifth-of trials reporting their results as per the FDA’s mandatory reporting requirements. At a conference last November, the FDA presented a re-analysis of the information that related to those trials that completed in 2009 that were the subject of publication in the BMJ. Their preliminary review suggested that, instead of four out of five trials being overdue, more like 21% were overdue.

At the moment there has not been widespread, complete and comprehensive monitoring of compliance with either the FDA Amendments Act, which requires certain applicable trials to be posted and the results filed on, or with the joint positions of the international and European trade industries and the ABPI. The ABPI commissioned a piece of research to look at all the new active substances approved by the EMA in 2009, 2010 and 2011. These are all new products, excluding vaccines and combination products, which are 53 new active substances. We looked to track all the trials that were part of the development of those drugs, from European public assessment reports and publicly accessible registries, international as well as the companies’ own registries, and tracked the results either to a publication in scientific literature or a publicly accessible database.

Our preliminary review of the aggregated information for all three years’ products-we are talking of approximately 1,000 trials-shows that about 87% of those trials had their results disclosed as of January this year. If a strict 12month line was drawn at the time of completion of the study, about three quarters would have been reported within that time frame either in the scientific literature or on a database.

This review, which examines the evidence base for the medicine, takes into account all the trials that were done. So, for products approved in 2009, the clinical development would have been started perhaps in 2002, 2003 or 2004. Those trials were not necessarily subject to any requirements-mandatory or otherwise-for reporting, so it is a complete evidence base.

Q72Stephen Mosley: On that point, does it include trials conducted outside of Europe in the US as well?

Dr Rawal: Yes. It includes all the trials that we could locate on the companies’ registries, on European public assessment reports and, because the companies are mandated to file there. As far as we could tell, it included the trials that formed part of the submission for marketing authorisation and any subsequent applications for extended indications. So what we were looking at were only industry-sponsored trials, and we were also looking at any trials in patients-not the healthy volunteer trials.

Q73Stephen Mosley: If a trial is not submitted when they apply for marketing authorisation and it is not one that they choose to put forward, who makes the decision whether or not it goes forward? Is it the case that all the trials go forward, or is there an element of choice? Who makes that choice, and for what reasons would they be omitted?

Dr Rawal: All trials that are relevant to the assessment of a medicine must be submitted to the regulator. The regulator is entirely able to ask for any trials. You have a clinical trial approval when you first want to embark on a trial. The regulator knows that you have asked for approval for those trials. They can request those, but, in essence, when you want to submit for marketing authorisation, all trials that are relevant to that authorisation- even if it is in a different indication and the trial failed-have to be submitted.

Dr Shannon: If a phase III trial-the end stage or any part of the programme-were to fail and there was no evidence or no evidence of efficacy versus a comparator, most likely we would not submit an application for a product licence; we would terminate that programme. The clinical study reports in those cases would not necessarily be submitted to a regulator because we were not applying for a product licence. However, all of those studies would be submitted for publication in a peer review journal.

Dr Goldacre: That may be true for GSK, but I am not sure it is universally true that trials that were not part of a marketing authorisation package would be made publicly available. I think, quite commonly, they would be withheld.

Q74Stephen Mosley: Dr Goldacre, I know you have made the comment that, even if compliance is 100%, many times these registries would be incomplete by design. Could you expand on that and explain why?

Dr Goldacre: To practise medicine in a safe, informed fashion, we need the results of all of the trials for all of the uses and treatments that we are currently prescribing, and we do not have that at the moment. There is no legal requirement for all results of all trials to be shared with doctors, researchers, the public and commissioners of health services. Similarly, there is no legal way we can identify all of the trials that have been conducted on a treatment.

When it comes to recent figures on better compliance, the devil is in the detail. We need to know exactly what list of trials you are looking for publication of. It is ironic, in a sense, that the figures we get from Dr Rawal are for unpublished analyses of the extent of missing data, because we need a fuller description of where the trials came from, how they were looked for in the public domain and what was and was not missing. Even if we had complete reporting of all trials starting from today-2013-or from 2008 under the FDA Amendments Act, that still would not improve the evidence base very significantly for medicine today, because about 85% of the medicines prescribed are generic, which means they came on the market more than 10 years ago. We need the results of clinical trials that completed in 2006, 2003, 1999 and 1994 in order to practise medicine safely, because those are the trials that inform the decisions for the treatments we are giving today.

Q75Stephen Mosley: That sounds fair enough. What would you two guys say about that?

Dr Shannon: What Dr Goldacre says is absolutely true. Most of the data that originates from medicines on the market today are from the early 2000s, or even 1900s. There are two ways to look at this. One is that clinical trials as such are a rather small piece of the data once a product becomes available, and most of the data on a medicine that has been available for 10, 15 or 20 years would have been captured on the marketplace. One of the things that we miss-I think Dr Goldacre would agree with this-is to look at electronic healthcare records and try to maintain records of the products and product performance on the market. I know that was one of the ambitions of Sally Davies in trying to drive electronic health records and make sure that every patient who was on a clinical study was on electronic health records.

As far as we are concerned, we are trying to go back to the year 2000 from the formation of GSK to make all of those trials available,3 which will capture many of the clinical trials of our medicines that are on the market today. But, before that, it becomes very difficult when we get into a situation where most of those reports could have been produced by a typewriter on paper. It is very difficult to make those available. Electronic databases prior to that time were in different data standards and are difficult to combine, and the usefulness of that data is difficult to ascertain and get hold of. As we have looked at it, we have tried to deliver what it is possible to deliver, even with difficulty, and for us even before 2000 it is very difficult to do.

William Burns: The bit of the jigsaw that comes to mind here that, maybe, we are missing is that this is a highly regulated industry, and the rules and regulations are predominantly put in place by society through an established regulatory authority. What seems to be happening is that, in recent years, there are more questions about whether the regulatory authorities are doing their job properly as well. It is not just whether the data is available and in what form, but also how we empower as a society our regulatory authorities.

We have seen an increasing requirement over recent years for more stakeholders to have more access to the data. If society wants that to happen, we have to respond; indeed, that is one of the reasons why James and I are here, because we believe this is a correct response to what society is looking for. But we should not underestimate that we should be working with a multiplicity of regulatory authorities. Any significant international pharma company is dealing with probably 80 different principal country authorities, and it goes from individual patient data with the Food and Drug Administration through to summary reportage in Europe. But do not underestimate that FDA and the European, Australian and Canadian authorities are themselves in a network and swapping notes about data and dossiers; they are also networking.

We are seeing here a requirement that is not just new today; it probably started 10 or 15 years ago. You saw more use of and the reportage mechanisms that are there. Clearly, society wants us to go further, and probably the two companies sitting here are the first two out of the gate saying, "We will have a new policy going forward, with greater disclosure than we have ever had in the past." That is one of the reasons we are with you, because you tell us how far society wants to go.

Interestingly, sitting from a perspective outside the UK but being a UK citizen, one point I have heard James make is that, in different societies like Scandinavia, the societal benefit outweighs the rights of the individual, all the way through to the United States where the individual’s privacy outweighs the relevance to society. As international companies, we see different countries drawing a dividing line between the individual and society. We need to find our way here, with guidance from various countries, as to how you want us to go, because our natural partner in a highly regulated world is the regulatory authority.

Dr Goldacre: It is not satisfactory to say that the results of trials should be reported only to regulators. We are talking about very difficult problems. What are the true benefits of a drug? What are the risks of a drug? If we look at some of the biggest problems that have been spotted in the evidence base for medicines over the past few years-Vioxx, Rosiglitazone/Avandia and Tamiflu-those were not spotted by regulators but by independent doctors and academics reviewing data, and their fight to get hold of that was often hard won. Those problematic signals were not missed by regulators because they are not clever. They are very clever people indeed, but these are difficult scientific questions and they benefit from having many eyes on them. It is a central tenet of science that we are open about our data, methods and results. It is odd that, in medicine, there should ever have been a presumption that it was enough only for regulators to see this information.

It is also important to be clear that we have known about the problem of publication bias since 1986, and people have been calling for complete registration and publication since then. We still do not have it today.

Lastly, we are not demanding individual patient datasets to be posted recklessly online. We are talking about very brief summary results in some cases. We cannot get hold of the simplest results about a trial from the 1990s saying that drug x was 22% better than a placebo, or drug x was 15% better than drug y. I cannot see any legitimate reason why doctors, patients and academics should not have access to the results of those clinical trials in order to make fully informed decisions about which treatment is best.

Q76David Tredinnick: Would Dr Shannon or Mr Burns like to comment on what they have just heard?

Dr Shannon: We agree with Dr Goldacre 100%. The more eyes that are put on data the better, and that is why GSK has taken the lead to commit both to patient level data transparency as well as clinical study reports. We believe that, in a world of very complex, data the more eyes we can put on it will benefit patients, science and medicine, so we are 100% aligned in that objective.

William Burns: The same is true for Roche. We have come out publicly with a revised incrementalism in the policy, which will take the study reports to the level being looked for. If that is what society wants to deliver, that is fine. We just need to be mindful of the prospect of analyses of the data and that it is in the interests of patients and healthcare. Are there ways in which we can avoid measles outbreaks with a whole generation of children who were not given the measles vaccine because of a wrong piece of science?

Q77David Tredinnick: That leads me neatly to the question that I had prepared, which again is for Dr Shannon and Mr Burns. Could you summarise your internal policies across all trial phases regarding trial registration, publication of clinical study reports, academic publication and the provision of third-party access to patient level data? I am aware that you have covered some of this ground already.

Dr Shannon: GSK has a requirement that all studies will be registered on the GSK Clinical Study Register, US and/or the European Clinical Trials Register. We monitor that and follow it up. As a study is complete and finished, we require that within eight months of the last patient’s last visit a summary of the results is published on the GSK clinical results registry, and that is a short summary of the information. We will write a clinical study report on all trials that are conducted as per the ICH guidelines. Within 18 months of the last patient’s last visit, we require that all studies are written up and submitted to a peer review journal for publication. How long it takes to get published is not always in our hands, but all must be submitted within 18 months of that time.

At the end of last year we have taken one step further, which is to say that we will make anonymised patient level data available for all trials that have been conducted globally post-2007 within a clinical database. Scientists externally can request that data to be made available to them, subject to review by an independent panel as to the validity of the research. Initially, we started using trials post-2007 because we standardised the data at that point. It is easier for us to anonymise the data and we used standardised informed consent, so we know that. We will go back and make the data available for all trials since the formation of GSK. In February 2013, we went further and signed up to the AllTrials initiative, which said that we will make clinical study reports available for all studies since the formation of GSK in 2000.

For clinical study reports, it is somewhat more difficult to make sure that we anonymise the patient level data and the patient’s identity, because those are on paper. While we can do that in a database by changing the patient code, we either have to redact the patient code in the study report or change it in some way, and we need to work out how to do that.

William Burns: There are some small differences, but in general principles we are on a similar journey. The use of is the principal instrument that we use for the announcement of all the trials that are running. We have a similar policy towards publication and the time periods after completion of the studies within which they should be met. We have also announced this year moving to clinical study reports that will give redacted information on patients. It gets more difficult the smaller the patient numbers, or the more unique the disease. If you had cystic fibrosis but you give a postcode, there may be only one child in that postcode that has the disease. As you get down to the more orphan diseases, it needs a little more thoughtfulness about how to protect the interests of the patient.

We have not yet signed up to AllTrials, but you will find that our policy as written meets all the principles in AllTrials that Dr Goldacre has initiated. We are still working on the roadmap that we want to undertake to implement this. Once we have got through that, and also had discussions at European level-because there is a similar initiative on data transparency at that level-we will be publishing our roadmap. But, in terms of the principles as such and transparency, there is nothing to test between two independently worked-up proposals.

Q78David Tredinnick: Do your policies vary from country to country or region to region?

William Burns: No. We want to have a policy that would apply to the corporation and which is there worldwide.

Q79David Tredinnick: How will you monitor and ensure compliance with your new policies going forward?

William Burns: That will also be part of the implementation roadmap. We recognise, as Dr Rawal showed, that performance has not been as good as we might have liked it. Speaking on behalf of my company, it has not been as great as we would have wanted. However, we have seen the performance move dramatically over recent years. We want it to move further.

Q80Roger Williams: We have heard tonight that GSK and Roche have very broadly similar approaches to trial registration, access, publication of results and proposed access to patient level data, and yet Dr Goldacre is critical of one company and congratulates another. Perhaps you can explain how you come to that conclusion.

Dr Goldacre: Certainly. GSK has been on a very long journey from being rather badly behaved, I suppose. It was the subject of a $3 billion fine last year for criminal and civil acts of fraud in the US, coming up to as recently as 2007. However, for many years it has made a series of commitments to greater transparency on which it has begun to deliver, whereas, for Roche, particularly in the case of Cochrane’s attempts to get clinical study reports on the drug Tamiflu, on which the UK Government spent £500 million in 2008 alone-that is 5% of the total NHS drugs budget in one year-the experience of Cochrane has been very different from the aspirations described here today. They were blocked at every turn, and there is a very long and troublesome story, which I understand the Public Accounts Committee-or at least the National Audit Office-is looking into in some detail. If you wish to call Cochrane, it would be very happy to go into that in detail, but it was asked to sign non-disclosure agreements that also had clauses in them that said it was not even allowed to discuss the existence of those.

In December 2009, Roche made a commitment to share all the clinical study reports on Tamiflu, and that promise has still not been paid out on as yet. My concern is that Roche is certainly making exactly the right kind of noises that we would like to hear about sharing clinical study reports, but in its specific interactions with Cochrane so far it has fallen well short of those aspirations, albeit those may be very novel and announced only recently and only in press releases.

It is important to be clear about what we want and why. Clinical study reports contain much more information about the methods and results of a trial than a brief report in a publication, or a brief summary report on a clinical trials database. By comparing clinical study reports on Tamiflu against brief published reports, Cochrane has already found discrepancies. For example, things that were described as placebos in the academic journal article turned out to be, first, a different colour from the active treatment, and, secondly, not to be inert placebos at all; in fact they had active ingredients. It has also found episodes where the case mix was not as described and was flawed in a way that might be expected to over-exaggerate the benefits of Tamiflu, and that was clear only from the full clinical study report, not the brief trial report. Lastly, they found at least one case where in an academic journal article describing a trial there were no adverse events, but in the clinical study report describing the same trial there were adverse events. It is to resolve discrepancies like that that we need better access to clinical study reports.

The second question is about what to do with individual patient data that is in these full clinical study reports. That poses some very interesting questions. That is the only area where there is any reasonable area for disagreement about how much sharing there should be, and to what extent that should be behind an approval committee. It is important to be realistic here. You find bits of individual patient information even in academic journal articles. For example, you might find a sentence that says one patient developed pneumonia and was admitted to hospital, but it should be borne in mind that the patient was immuno-compromised and had HIV. That is the sort of sentence that one might expect to read in a published academic journal article as much as in a clinical study report. A lot of what you see in narrative descriptions of adverse events in CSRs is not the kind of information that people would necessarily be very worried about protecting, but at individual patient data level, when it comes to very long and complete databases with postcodes, dates of birth and so on, we have to be thoughtful about how we control that.

You asked-and I will finish in a moment-about the difference between GSK and Roche. One concern that I and many others have about both companies is that they say there is an independent review panel that will look at the analytic strategies being proposed by independent researchers to be conducted on the datasets that the companies have collected. First, I am concerned that there should be pre-approval. It is reasonable to ask for pre-registration in order to prevent people going on a statistical fishing trip. For pre-approval, if that is to be acceptable at all, it would have to be with a very low bar. Secondly, there is the matter of who is on these independent panels. As I understand it, we do not yet know who is on the GSK one, but for the panel called MUGAS by Roche there are four independent researchers. As I understand it, three of those have been paid consultants for Roche and one has had a trial research grant from Roche. We need to be cautious about whether it is okay for individual companies to set up their own boards.

It is not surprising that those disagreements will arise very early on in society’s negotiations about what should be shared and how we should do it. We need to look at having standard protocols perhaps nationwide across both independent academics and also industry, and ideally internationally, setting down standards on what it is reasonable to ask, who should be on the review panels and also making sure that all the processes are transparent so that everybody can see what has been rejected.

Q81Chair: Right of reply.

William Burns: Right of reply, but I will try to keep it short in the interests of your time. I would put before the Committee that, on Tamiflu, we have shared data with 80 world health authorities, including individual patient data. We have had a multiplicity of independent reviews from the World Health Organisation, CDC Atlanta and a number of specific Government-established Departments or initiatives looking at pandemic planning, which all came left field as a potential utility of the medicine. We have had problems with one independent review group. When I look back on that, there are faults on both sides and miscommunications. It is difficult when you ask for a way to protect patient confidentiality and there is pushback on the signing of it. That is fine. There is an offer and we should sit and discuss it, but it ends up in e-mail exchanges and there is not a meeting and clarity of mind, and, sadly, we start to see that the emails are being copied in to the media. We know there is another agenda at work here.

There are probably faults on both sides on one independent review, but in the holistic look at the total medicine, and indeed in some of the UK Government’s reviews as to whether or not it is appropriate to use a neuraminidase inhibitor in pandemic planning, clearly there is a role for this medicine, and I do not think that is in any way in dispute. I recognise that the Committee wants to think higher on the total transparency, but please do not look at the company on the basis of one independent interaction on one product-but on a broader basis.

As to the other points you raised, on individual patient data, as a company we would agree with a lot of what you say. We should not be so prescriptive on the way in which people want to review that data, but there needs to be enough intellectual rigour and that it is not transparency for its own sake but for a purpose that needs to lead to improvements for patients. It is not just a matter of intellectualising for the sake of it, but everyone should be looking at whether the dataset is going to improve patient outcomes.

The third question you raise is to do with the independence of an advisory panel. Dr Shannon can speak on his own behalf, but, if you are looking for world experts, you will find they are on top of their subject, probably have worked with a multiplicity of companies and have had experience in running trials on a multiplicity of new chemical entities. If anything, it enhances their independence rather than saying that because they did some work for one company they are in the pocket of that company. That is almost insulting to the academics.

Dr Shannon: As far as concerns the independent review panel, I may sound like a broken record but I agree with Dr Goldacre. We also believe that a review panel should be independent. We have our system up and ready to go.

Q82Chair: If it has to be independent, answer Mr Burns’ point. Where you are getting narrower and narrower expertise, he is arguing, as I understand it, that the experts are bound to be the people you would want on your review panel.

Dr Shannon: That is absolutely true.

Q83Chair: How do you solve it?

Dr Shannon: We are planning step two before we plan or initiate step one. At the moment our system is up and ready to go. We have available a panel of independent experts some of whom have worked with other pharma companies and are ready to act as our initial panel, but we are also in negotiations with a completely independent clinical research organisation to have them nominate a panel that is completely separate from GSK and where we have no involvement whatsoever in the nomination of that panel. It is clear that a panel nominated by GSK will not be acceptable to Roche, Novartis, Pfizer or any other company. We wish and hope that a panel nominated by an independent body will be acceptable to other companies and encourage them to join up to our initiative.

William Burns: Probably the key thing is the transparency with which it is done and that this should be declared.

Dr Goldacre: I certainly would not say that anybody who has received money from a company is inherently biased, but we need to be aware of conflict of interest, and we know it is a risk factor.

William Burns: I can agree with that-absolutely.

Dr Goldacre: I do not think that people on a review panel necessarily need to be people with familiarity with one particular drug or one company’s drug, so there will be statisticians and clinical trial methodologists with no ties to any one company, or indeed any company, who may be willing to participate on such a project.

I come back very briefly to Tamiflu, not because I think it is important as a particular drug but because it is the most well documented case where we are aware of discrepancies between trials being reported in different ways in different territories. It is important to be clear that, although it has been reviewed by several different regulators around the world, they have come to very different perspectives and judgments on it. For example, in Japan, Tamiflu has a black box warning on it for use in 12 to 18 year-olds, which it does not have anywhere else. The FDA did not permit the claim that it reduces the risk of complications, which is a medical euphemism meaning pneumonia and death after influenza infection, whereas elsewhere in the world the company is permitted to say that it reduces the rate of complications.

These discrepancies in final conclusions between regulators reflect perhaps two different things: one, that they have seen different data, and there is evidence collected by Cochrane to show that different regulators around the world have been shown different data on Tamiflu; and, secondly, that different regulators come to different conclusions even based on the same information, because this is, as we have already discussed, complex data and different people will hold legitimately conflicting views on it. It is important to be clear that regulators do not necessarily have all of the information and they are not necessarily infallible, and that is just another reason why we need to ensure that as many people as possible have access to the data in order to make their own judgment.

William Burns: In that case it is the second and not the first.

Q84Roger Williams: Can I take it forward and ask Dr Rawal, in your opinion and that of your organisation, how GSK and Roche compare with the rest of the industry in their approach to these matters?

Dr Rawal: It is an evolving discussion. To set the scene, the ABPI has 68 members who are either full members or research affiliates, so in some way they engage in clinical research work. Of those 31 that I have called smaller companies, their turnover would be less than 75 million. We represent the whole spectrum from smaller to larger companies.

To set the context for this discussion, the disclosure of clinical trial results can be looked at at different levels. First, the disclosure of summary results is already a requirement as per the ABPI code of practice. Indeed, the EU clinical trials database is being upgraded to be able to accept the summary results of all the clinical trials conducted in the EU going back to 2004, so there will be visibility and line of sight to the summary results for those studies going back to 2004. Prospectively, of course, it is a requirement, and ABPI has already announced an intent to put in place measures to monitor the rates of disclosure of clinical trial results. That is at the level of summary results.

It is also important to look at what the EMA is doing. The EMA embarked on a process at the end of last year to increase transparency in the reporting of clinical trials and, in particular, launched five working groups that are looking at the specifics of how to share clinical study reports, for example. The whole clinical study report can extend to thousands of pages, if you include the appendices. That is very different from disclosing summary results, which are generally based on a synopsis of the whole clinical study report.

The five working groups are looking at issues such as protecting patient confidentiality, which is absolutely crucial, because there is sensitive personal information threaded through-particularly in the appendices of the clinical study reports-that would need to be anonymised. They are looking at clinical trial data formats and standardisation of formats; they are looking at the rules of engagement as to how to make those accessible to everybody, or should the requester register and say what they are going to use it for? All those questions are being debated. A fourth group is looking at good analysis practice, and a fifth one at the legal aspects. That shows how complex it is when you get into the discussion about disclosing full clinical study reports.

It is also important to note when that disclosure happens. Up to now, the summary results are required to be disclosed after first marketing authorisation of the product. We need to know in timelines whether we are talking about pre-approval or post-approval of the product. It is important to make that clear.

The third level of transparency is about individual patient datasets-i.e. the raw data that underpin all of what we have just discussed. Those datasets are what I think GSK and Roche are talking about in terms of making access available to independent researchers. Those requests by researchers will be reviewed by an independent panel so that there is rigour and transparency in those analyses and so that medicine and science can be advanced. Those are very complex issues.

ABPI ran a workshop a month ago to talk about clinical study reports. As Nicola Perrin mentioned in the previous session, the Wellcome Trust had a workshop last Friday to talk about disclosure and sharing of datasets. There is general consensus that we need to move forward at an international level here. It is no good the UK just deciding what to do because, by and large, clinical trials are a global enterprise, and, in the UK, clinical trials regulation is governed at EU level. It is really important that this is done in the context of those two facts.

To create data warehouses or repositories, there still needs to be a lot of discussion at international level and across sectors, because clearly this affects academia and other researchers and public and charitable funders, just as much as it affects industry. We all want to advance research and get the maximum benefit, and exploit the data that we are collecting on patients, because that is what serves the patients’ interests best-if we can make the most of every single trial patient and their data, and get absolutely what we can out of it to serve for future product development, future trials and future medicines.

Q85Chair: Very briefly, I want to follow that up. In that perfect world you have tried to describe, will that stop the kind of litigation we have seen just recently, or do you think that is an inevitable consequence of the crazy world we live in?

Dr Rawal: I am not clear which litigation you mean.

Q86Chair: There was a case in March of this year in which AbbVie sought an injunction to prevent the EMA from releasing raw data in relation to its arthritis drug. It is not a very satisfactory situation to be in, is it?

Dr Rawal: I cannot comment on the specific case. There are provisions under freedom of information to request any part of a marketing authorisation application. It depends on whether we are talking about clinical study reports as opposed to other information in the marketing authorisation application. Issues are there that still need to be addressed around what constitutes commercially confidential information. When is regulatory data protection crucially important in order not to disincentivise the innovative pharmaceutical industry? There will be examples of regulatory data protection that rely on that, as opposed to intellectual property, for the protection of the interests of the developer of those drugs.

Dr Goldacre: The European ombudsman has been clear that there is no commercially confidential information in clinical study reports, and where there is patient-confidential information it is easy to redact that. It is important to be clear that every one of the systems Dr Rawal has set out has holes, and it is great that the ABPI has come very late to aspiring to greater transparency here. For example, the ABPI code of practice, as I understand it, two months ago, does not cover Roche because they are not members of the ABPI.

William Burns: Roche fully subscribes to the ABPI code of practice on a voluntary basis. We have said we will absolutely comply with the code of practice.

Dr Goldacre: That is different from the information I was given by the ABPI, since you are not members of the ABPI.

On the issue of the European Union Clinical Trials Directive sharing results, we are not yet certain what will happen. We are not yet certain what will happen with the EMA’s changes for sharing individual patient data and whether that will be retrospective. If it is only trials completing after 2014, it will do nothing to improve medicine until most of us are dead.

The European Union Clinical Trials Register demonstrates more clearly than anything else how incomplete these systems are by design. The European Clinical Trials Register promised that it would carry results by 2012, and it still shows no signs of doing so. More importantly, it covers only trials conducted after 2004 and only those conducted within the borders of Europe. That is no use at all. We need the results of all of the trials conducted on all of the medicines currently being used in Europe. It does not matter if they are trials submitted as part of the marketing authorisation package or if they are trials done after a drug came to market in Brazil, Russia, India and China. It does not matter if they are for unlicensed uses. If there is a trial on the use of an antidepressant drug to prevent post-traumatic stress disorder or anxiety and we know that currently it is being prescribed for that, even though it does not currently have a marketing authorisation, all of the trials for those indications should be shared as well.

Every single system we have at the moment, by ambition, is incomplete. The very simplest thing that we could do to change all of this would be for the European Medicines Agency to say, as part of their approval process, "Congratulations. Your drug is for sale in Europe. As you know, we hold the trials register that contains all results of all trials for all medicines being used in the European Union. Here are the forms. Give us the results of all of those trials." There is no reason why that could not be done. We have known about this problem for 27 years now at least. I do not think it is industry’s fault that we have not moved forward on this; there has just been a failure of ambition on everyone’s part.

Q87David Tredinnick: Following on your theme, Dr Goldacre, as a clinician, how do you think increased access to individual patient data might ultimately impact on individual clinical decision making in public health?

Dr Goldacre: The best currently available information on the extent of missing trial results comes from the NIHR HTA review in 2010, which estimates that about half of all trials do not go on to be published, and trials with positive results are around twice as likely to be published. That may have improved for very recently conducted trials, but that does not reflect the decisions that we make on a day-to-day basis. We do not know what we are missing; we find out only in a piecemeal fashion.

I can give individual stories. For example, doctors who prescribed Reboxetine before we had all of the information in 2010, after IQWiG-the German equivalent of NICE-demanded that all trial results were made publicly available, were not able to have all of the results. I prescribed this drug myself. If I had had all the results, I would not have prescribed it. The company Pfizer was withholding lots of results. Similarly, for all antidepressants, the best available meta-analysis by Cipriani cannot take account of publication bias, so we do not know which antidepressant is best. I do not think we are very commonly prescribing drugs that do more harm than good, but we are probably being misled about which of the many available treatments in one class is the best.

Q88David Tredinnick: How do you think the right balance should be struck between increasing the transparency of raw trial data and protecting the confidentiality of those involved-we touched on this earlier-many of whom provided their consent many years ago when the modern culture of data sharing could not even have been anticipated?

Dr Goldacre: It is important to be clear that we are breaching the ethical pacts that we forge with patients, and breaking the bond made with them in the consent form, when we fail to publish the summary results of trials. That happens very commonly, so we already trample over the obligations that we have reciprocally between triallists and trial participants. However, it is extremely important that we respect patient confidentiality. None of our requests for summary results have anything to do with patient confidentiality and there is still a problem with getting summary results. As to the details of how we get individual patient data more widely available, there is a huge, long and complex technical discussion to be had on that, but it can only be had among those people who have already committed to the principle that that is something we all want to do. There are many people in industry who reject that proposal.

Q89David Tredinnick: I have a final question for Mr Burns and Dr Shannon. If your commercial competitors were to make patient level data freely available, would you access it? If so, how would you use it?

Dr Shannon: Would we do it, and how would we do it?

David Tredinnick: If your commercial competitors were to make patient level data freely available, how would you access it and, if so, how would you use it?

Dr Shannon: I can tell you that, by accident, our clinical trial registry-the share system that we talk about here for patient level data-went live last Wednesday for about four hours. In those four hours, even though there was no guidance to that system, three other companies accessed the data, plus Johns Hopkins hospital in the United States.4 If companies make data available, other companies will access it.

There are many ways in which we can use patient level data on both active drugs and terminated drugs. The history of the pharma industry is that many of us, in pharma, have chased a target for a specific disease, not knowing that another company had failed in their trials for that target. If we understood that other companies had failed in that target and it did not work in the disease, we could either change the trial design to make the trial more appropriate and get better results or drop the target and not move forward. We could also understand better the types of patients in whom the drug works or does not work. We look here at the best drug for a patient. What we show in clinical trials is the average result of the effect on a population and what is the best drug for an individual patient, or group of patients, whose specific characteristics may be different. We do not know today from many clinical trials. We could identify which are the right drugs for the right, specific patients, for example.

William Burns: Another way you see this data also playing out is that, since the advent of and the publication of the work that is under way, this helps to navigate for doctors who are not in trials. Is there a trial running for a patient with a certain disease? Would they qualify? They can put them in touch with them. It also helps other clinical triallers to say, "I was wondering about comparing it with drug x or looking at sub-population y." Then they see that there may be three other trials running, and they do not start a clinical trial unnecessarily exposing more patients to it until they know what the results are of the trials running. There are elements of this transparency that can enrich and inform.

On the point about legal interventions, I would not be at all surprised if we did not see more tort lawyers in America scouring through the data to see if they can find a basis for some cohort or class action in some way. I would not be at all surprised to see generic companies trying to enrich their dataset in lesser regulated countries out of the public domain. In countries where the files are available for a very modest price, historically some generic companies have had paper copies they have been able to enhance. This is not in the interests of public safety. There is an exposure; we just have to be ready for it. I am not saying it is any reason not to go forward. Absolutely, as an optimist in life, you see two companies here who have said that we are prepared to go down this route and we are prepared for that degree of risk.

Q90David Tredinnick: Isn’t sunlight the great disinfectant? Really what we are talking about is more information, for the right reasons, being out there to stop duplication and improve the condition of the people.

William Burns: Yes, but I think you are speaking to the converted there.

Chair: The unconverted are not with us today. Can I thank all four of you for your contribution this afternoon? We went on somewhat longer than we had anticipated, but thank you very much indeed for coming.

[1] The witness later clarified that, these data cover pharmaceutical studies in patients. The percentage of patients recruited from the UK in 2012 was 4%. This falls to 1% if one large single-country COPD study in Salford is removed from the analysis.

[2] The witness later clarified that, f or fatal or life-threatening serious unexpected serious adverse reactions (SUSARS) the timeframe is 7 days.

[3] The witness later clarified that, GSK will make available Clinical Study Reports (with personal information removed) from published studies of approved or terminated medicines. CSRs for all clinical outcome trials going back to the formation of GSK in Dec 2000 will also be made available in a stepwise manner, with priority given to the most commonly prescribed medicines. For GSK’s initiative to provide access to patient level data, studies are listed after the medicine studied has been approved by regulators or terminated from development and the study has been accepted for publication. Global studies conducted since 2007 will also be included and over the next two years global studies going back to the formation of GSK will be added. All studies (including local studies) started in and after 2013 will also be included.

[4] The witness later clarified that, the draft Share website was temporarily available at other times. For information: the Share website was formally launched shortly after the evidence session on Tuesday 7 May.

Prepared 16th September 2013