Recently, the ABPI has worked with its members to explore definitions and guiding principles to enhance transparency in clinical trial reporting. In discussion with members, it is clear that while greater transparency is broadly welcomed, there is a lack of consensus on the degree of disclosure required to constitute transparency. There is a need to determine what exactly is meant by ‘transparency’.
There is no set definition of transparency in the context of clinical trials reporting. Transparency can span everything from publicly registering trials in a timely fashion through to full disclosure of trial findings and of individual patient data. Industry is committed to the principle of greater transparency and it wants to see this implemented in a manner that supports good research, with data being analysed and interpreted appropriately.
One aim is to enable independent researchers to supplement companies’ own analyses of trial data, particularly through pooling of data from multiple studies. Such meta-analyses can provide clinical decision-makers and health policymakers with valuable information about the effectiveness and safety of inovative drugs.
Any framework for improved data transparency must ensure patient confidentiality is preserved, scientific integrity is maintained and intellectual property rights are protected
An important priority is to ensure that clinical decision-makers have access to the most comprehensive data on which to base their decisions. In addition, different stakeholders have different needs, so the over-arching aim is to get ‘the right data at the right level to the right people at the right time’. However, any framework for improved data transparency must ensure that patient confidentiality is preserved, scientific integrity is maintained and intellectual property rights are protected.
There are three factors that limit the wider availability of clinical trial data:
- Patient confidentiality – industry needs to be careful that data released into the public domain cannot be traced to individuals.
- Commercial sensitivity – for example, in relation to the timing of disclosure.
- Technical challenges to data sharing. For example, data may be stored in different formats and in different locations – not all of it will be in electronic format, thereby making data for past trials difficult to access.
The current system of clinical trial reporting
The ICH, an international body with representatives from regulatory agencies and industry, currently provides the framework for clinical trial data reporting. This system was established in the 1990s. This standard (known as ICH E3) provides a structure for the pharmaceutical industry to report their trials. The format includes:
- A summary document, or synopsis;
- The body of the Clinical Study Report (CSR), typically 100–200 pages long;
- Appendices of individual patient data, typically 1,000–2,000 pages long.
The structure of the CSR highlights the need for clarity when discussing ‘transparency’. In theory this could apply to disclosure of:
- A summary of the trial and its results (e.g. a CSR synopsis)
- Aggregated data on efficacy, safety, etc.
- Individual patient-level data.
A key factor when deciding what data is made available, to whom and in what format, is the question of how the information will be used. In terms of patient safety and clinical decision-making and policy, the principal use is in developing systematic reviews, which may or may not include meta-analyses.
Systematic reviews aim to corroborate all available information about an intervention, integrating information from multiple sources. Meta-analyses, generally undertaken within the context of a systematic review, combine actual data and involve additional data analysis. A systematic review may draw upon information in a synopsis, but in general, is likely to find the information in the body of the CSR helpful. A meta-analysis will require aggregated data and potentially patient-level data.
Further discussion is needed on how greater trial transparency can be achieved
The registration of clinical trials is important as it can alert systematic reviewers to the existence of a trial that should be included in their reviews. Confidence in the review will be greatest when it is known that all potentially significant studies and data have been included but concerns have been expressed that results are not being posted for all registered trials. Preliminary findings from the ABPI have found that 10–20% of industry sponsored trials fall into this category.
Access to more granular data raises concerns about patient confidentiality. When aggregated data are shared, redaction can limit any information that could be used to trace an individual. However, there is not general agreement on what kinds of information should be redacted. Processing of patient-level data before release is highly labour intensive, particularly when applied retrospectively to past studies.
Academic groups also carry out clinical trials on investigational medicinal products. However, the reports for these trials are more variable in content, scope and format than industry reports. It could be argued that academic groups should follow the same practice as industry, but as CSRs require considerable resources to compile, this would impose heavy burdens on academic researchers that may not be counterbalanced by a gain in knowledge. With the growth of drug discovery units in academic settings, this is becoming an important issue and is being considered by the main funding bodies that generally require researchers to publish their findings in the scientific literature.
The results of industry clinical trials are also usually published in the primary scientific literature. Historically, dissemination of negative results has presented a problem, as some journals have often been reluctant to publish negative findings. However, some newer journals, such as PLoS ONE, have undertaken to publish all scientifically validated papers, without making any judgment on their importance, and are another potential medium for disseminating negative findings.
Negative findings may be of considerable scientific interest as they shed light on drug efficacy. For example, a drug may be found to have its predicted pharmacological effect but no impact on disease – suggesting that the original hypothesis of disease mechanisms may have been flawed.
Where is trial data to be published?
Many pharmaceutical companies maintain their own registries of clinical trials. The main registries in Europe are provided by the European Medicines Agency (EMA) and on the website www.clinicaltrialsregister.eu; and in the US by the National Institutes of Health (NIH). Both contain summaries of protocols, but only the NIH provides public access to results (aggregated data), as mandated by the US Food and Drug Administration (FDA).
Although the EudraCT database is comprehensive, as all trials in European Union countries must obtain a ‘EUDRACT’ number and register before they can begin recruitment, an upgrade to the database (scheduled for 2013) will be required before summary results can be made publicly accessible. It will cover all clinical trials commencing in Europe from 1 May 2004 onwards.
Historically, on request, the EMA has provided access to CSRs (including appendixes) for registered trials dating back to 1995, with companies having the opportunity to redact commercially sensitive information and personal identifiers before release. Since 2012, the EMA has been discussing ways to proactively disclose information. At the end of 2012, five working groups were set up to examine key issues and their recommendations will be sent out for consultation in June 2013, with a view to implementing the new system from January 2014.
It is notable that the overwhelming majority of requests for information (often through freedom-of-information requests) have come from competitor commercial organisations
Industry has generally been supportive of the EMA’s desire to enhance access to information but recognises that there are some issues. It is notable that the overwhelming majority of requests for information (often through freedom-of-information requests) have come from competitor commercial organisations. Very few have come from academic groups seeking to reanalyse data. It is not clear if this reflects a lack of awareness of the availability of data, a lack of need for such information, or other practical issues.
As well as what data are made available, the timing of information release is also critical. Issues remain over how far back data should be made available and for which medicines. Agreeing cut-off date criteria is arbitrary but convenient; and prioritisation of which classes of medicines to start with might be helpful but agreeing ‘utility criteria’ could be difficult.
As a starting point, it may be useful for companies to compile lists of all the clinical trials they have sponsored on currently prescribed medicines, even if accessing specific data might be difficult for some trials.
How is clinical trial data published?
Technical issues are likely to present a considerable challenge to data sharing, particularly of patient-level data. Data are likely to be held in different formats that may be difficult to combine. Although industry-wide data management standards do exist, they are subject to company-specific interpretation.
Endpoints may vary, hindering attempts to combine aggregated data from different trials. Reuse of patient-level data could potentially be used to overcome this difficulty, or to examine new issues arising after approval. Patient-level data is also extremely important for analysis of rare adverse effects, which are likely to reflect patient-specific contributory factors.
Furthermore, as ‘omics’ technologies continue to develop, there will be greater interest in linking individual physiological and genetic data to drug responses, which will again require analysis at an individual level. This will once more raise issues of patient confidentiality. However, analysis of patient-level data is a sophisticated activity, and measures need to be in place to ensure that data analysis is carried out by those with the appropriate skills and experience.
In conclusion, there is a widespread recognition that greater transparency in clinical trial reporting could provide health benefits, and that the pharmaceutical industry should not be immune from wider social trends towards openness and scrutiny. It is equally important that discussions about how to achieve greater transparency take place.