Science and Technology CommitteeWritten evidence submitted by The Regius Professor of Medicine, Professor Sir John Bell, FRS, FMedSci

1. I am writing to provide evidence to the Science and Technology Committee’s enquiry into clinical trials and data transparency. I think this is a very timely enquiry and I hope it will raise the level of discussion about the UK’s position with regard to clinical trials, the advances made by the Health Research Authority, the remaining challenges with European regulation of clinical trials and also the need for consideration of increased data transparency. While President of the Academy of Medical Sciences, I was closely involved in advancing the case for reduced clinical trial regulation and am familiar with clinical trials in both an academic and commercial context. I thought the Committee might find my thoughts on his subject helpful in its deliberations.

2. In my view, the European Clinical Trials Regulation currently being discussed provides a significant advance over the previous Clinical Trials Directive. The MHRA has made a substantial effort to ensure that the views of the UK were heard during the process of re-drafting the directive. I believe many of us still have some concerns about the extent to which a risk-based and proportionate approach to regulation is likely to be described in the regulations. For example, it would be important that existing drugs with a good safety profile which are being used for new indications with relatively little safety risk are not considered to require the same regulatory environment as new medicines. It is not clear yet that these sorts of issues have been entirely resolved by the new clinical trials regulations and this needs to be carefully monitored. In general terms, however, these regulations are moving in the right direction although, if we expect this activity to be pursued with vigour in this country, we need to retain a major focus on attempting to reduce the amount of regulation and bureaucracy associated with undertaking clinical trials.

3. The Health Research Authority has now begun to deliver what was requested in the Academy of Medical Sciences Report published two years ago. As President of the Academy at the time, I was very anxious to ensure that the level of bureaucracy was reduced for those undertaking clinical trials and also that efforts were made to simplify and speed up the process. The Health Research Authority was one recommendation that successful emerged from the Department of Health and I believe its leadership has begun to deal effectively with many of the obstacles that prevent NHS ethical approval from being granted in a timely and efficient fashion. Ideally, this would result from a single sign off, but this is not legally possible, given the independence and responsibilities of independent Foundation Trust boards. However, the HRA and the NIHR seem to be able to improve the speed by which trials are approved by monitoring approval rates against national standards and reporting them back to hospital chief executives. It also appears that single sign off is occurring amongst clusters of hospital trusts that choose to work together and this may end up solving the single sign off problem without the HRA being directly involved.

4. I am encouraged that the committee has chosen this time to consider clinical trial data transparency. This is a complex issue and I have outlined my thoughts on this issue below:

4.1The definition of transparency. One of the key issues in this discussion is the degree of transparency of clinical trial data that is being considered. Some advocates suggest that the optimal level of data release is patient line data released online and available to the general public at the conclusion of each clinical study. Others have taken a more moderate view suggesting that summary data from both positive and negative trials be made available to genuine investigators on request. Other models are also being suggested, including access to trial data through an independent scientific committee structure.

4.1.1The extreme position of making all patient line data available to all comers has not been properly thought through. Such an approach would be associated with many issues including consent, data protection and privacy issues and the need for trialists and participants to understand fully the nature of anonymisation and its limits. If introduced it could have a negative effect on patient recruitment as patients are likely to be uncomfortable with having their details circulated on the internet. Many of the trials of interest for this sort of data release will be part of a set of global studies and a unilateral position of extreme data release in the UK would be certain to drive most of the important and interesting trials to other jurisdictions. Given that the Academy and others, including the National Institute for Health Research, have made a major effort to ensure the UK attracted more clinical studies, such a move would be very disappointing and probably unhelpful to both patients and physicians. If applied forcefully for early stage trials an extreme form of transparency requirement would essentially eliminate the biotechnology sector in the UK, also with serious effects on our ability to discover drugs (half come from this sector) and for the Life Sciences sector of the economy.

4.1.2Other degrees of transparency might prove more helpful. Availability of summary data from all trials associated with drug registration would be sensible and, in my view, all such studies should be published even if they are negative. Various other layers of detail could be considered (including the release of clinical study reports) but it is not clear what this achieves and the more onerous the requirements, the greater the burden on trialists. It might also be possible for an independent committee of trialists to review data that was contentious. This is an approach which has been used in the past and might be established on a more formal basis going forward.

4.2A second major issue relates to the risk of creating new regulatory barriers for clinical research. The field of clinical trials is already the most regulated of any in medical science. Often, in the past, regulations have been introduced with good intent but, in the end, they have greatly impeded the field with layers of unnecessary regulations that are not risk-based and lead to a box-ticking mentality. The ICH-GCP guidelines are a good example of this and the Academy’s report on clinical trial regulation lays out why all this extra regulation is unhelpful. We have spent an enormous amount of time trying to unpick the unhelpful parts of the EU Clinical Trials Directive and the burden imposed by unnecessary regulation in the UK has driven our share of global clinical trials to <3%. We must be careful not to add further layers of regulation unless we are crystal clear what the problem is that we are trying to fix. I believe that clarity around what we are trying to achieve with increased clinical trials transparency does not exist.

4.3Do we think the system for drug regulation is broken? It is worth noting that, for the specific case of drug registration trials, all the clinical data is available for drug regulators. Most new drugs have the data examined by four or more major independent regulators, as well as health technology assessment regulators such as NICE. Patient line data is routinely examined by the FDA. All studies are submitted to the regulators and considered by these independent and rigorous agencies. Failure to submit data or the submission of erroneous data is illegal and subject to drug withdrawal or large fines. Decisions are often finely judged and not all regulators reach the same conclusion about all medicines based on the same data. It is clear that the FDA does not welcome the release of large amounts of patient line data as this would in inevitably lead to often spurious challenges of the decisions made by regulators. Setting aside the validity of such disputes (see below) it is not clear who benefits from continual controversy over regulators’ decisions. The committee will need to decide whether the structure for regulatory approval of drugs based on evaluation of full sets of clinical data is working or not. I see little evidence that it is failing.

4.4It is important to consider both the benefits and the harm that might arise from a change in the current arrangements for making trial data available to a wider audience. The analysis of large data sets is not straightforward and there is always the concern that untrained or inexperienced investigators introduce bias in the results. It is also evident that, given the publicity associated with apparently overturning widely held beliefs about health care practises, atypical analysis—either unintentional or intentional—can attract public attention and a certain amount of fame or notoriety. Large trial analysis can be done using multiple tools and, by parsing the data in a variety of ways, many different conclusions can be drawn. The public and the press are ill-equipped to deal with such assertions and it can take many years before the effects of such analyses are corrected. Of course, it would be helpful if the medical journals were able to weed out these results, but we are well aware these are not always effective gatekeepers for sensationalist stories.

4.5One example of an approach that commonly produces results that are at odds with current practise is a variation on the use of meta-analysis. This approach does not focus on merging data but, instead, eliminates the majority of studies so that the power to detect effects is lost. A good example of this was the meta-analysis of studies of breast cancer screening for women. These were all in the public domain but the example illustrates the consequences of applying meta-analysis in a selective manner. In 2000, a paper which eliminated 8 out of 10 of these studies was published and not surprisingly had insufficient power to detect a benefit. This was widely reported in the lay press and was followed by a book by a now famous scientist laying out the grounds for a conspiracy. This has undermined confidence in a programme of screening adopted by almost all Western public health agencies and may well have cost lives. It has taken 12 years and multiple official reviews, most recently by Sir Mike Richards in the NHS, to rebut these arguments. In this case, I would argue that this sort of reanalysis contributed little to patient well-being. Similar reanalysis techniques were used to stir up the recent debate about Tamiflu, large numbers of trials were excluded from the analysis and unsurprisingly the power to detect effects also disappeared. Do we really want to release a wave of data in an unconstrained way that will fuel such analyses? Much depends on whether the benefits of transparency will outweigh the disadvantages. In how many cases have regulators failed to analyse the data packages presented for registration inappropriately, or failed to act on good new data relating to safety or efficacy?

My sense is that regulators are relatively effective in their role and have acted when necessary in the interests of patients. There may be occasions where questions arise about trials that make further independent review helpful. In this case, it is likely that sophisticated reviewers are best equipped to consider the issues and hence the suggestion of independent expert groups being available would be appropriate. This would avoid the adverse consequences emerging from controversial analyses appearing in the public domain which, in the end, are inevitably balanced by similar independent committees but only after public confidence is undermined.

4.6A new regulatory regime relating to transparency would also impact academic investigators. While widespread data release would be very beneficial for the meta-analysis brigade and those journals which thrive on secondary data, it would not always be welcomed by investigators with large datasets. These scientists can also occasionally be subject to mischievous attempts to undermine the main conclusions of their studies by parties who have been disadvantaged by these results. They also find themselves already trying to operate in an over regulated environment and more regulation, even if targeted at others, will be very unwelcome.

4.7We should all be attempting to create an environment where competent reanalysis of data that is used for clinical decisions should be made easy to achieve, particularly by independent investigators with demonstrated experience with statistical approaches to large data sets. Transparency is, therefore, a very valuable objective. Care must be taken, however, to make this proportionate, to avoid undue regulatory burdens and to be conscious of the outcomes that might ultimately be unhelpful to patients. Consequently, some control should be exerted over who should have access to data. I personally do not see how sensible new legislation could be formulated without again creating obstacles that will severely impede the field and slow progress towards new therapies. A code of practice agreed by all parties might be the right way to proceed.

February 2013

Prepared 16th September 2013