Over the past year the European Parliament and the House of Commons Public Accounts Select Committee added their support for transparency in clinical trials. These petitions for openness have not gone unheeded by the pharmaceutical industry - with big players in the market now beginning to open up their data.
A year ago, an alliance of pharma companies set up the online portal clinicalstudydatarequest.com. Through this website, researchers can make requests to access clinical trial datasets which the companies have made available.
Since then, the alliance has grown to eleven companies who now offer access through the web portal. To learn more about this industry led initiative, Robert Cuffe from Viiv Healthcare and Rebecca Sudlow from Roche talked to us about how the portal will work with researchers.
Can you give some background as to how clinicalstudydatarequest.com came about?
There is a growing societal demand for access to data and information, to see the evidence behind treatment decisions and to hold companies and other trial sponsors to account. Many people working on this project, and some of the patient groups with whom we speak feel that analysing the data from a clinical trial only once is a failure to make the best use of the time and effort of study participants and trialists. We collect so much data in clinical trials, and put so much effort into cleaning and documentation, that to have them mouldering unused in an archive seems like a waste.
GSK were the first pharmaceutical company to step out publicly with commitments on clinical trial data sharing in October 2012. As their long term vision was to provide tools and processes that could be used by all types of clinical trial data holders (pharma companies and academic research groups), they were very keen to engage with companies who were starting to think about these issues too.
Once GSK took the lead in setting up the technical systems that simplify the process of requesting data and that provide data in a secure way to researchers, it made the implementation much easier for like-minded companies. Discussions kicked off between Roche and GSK in spring 2013 and quickly evolved into a partnership. It keeps growing: currently eleven pharmaceutical companies are on the site.
We are all proud of the portal - it is a great resource of industry quality data and documents, ready for analysis, with an independent review process that enables researchers to access data from many companies with a single application.
Is there any precedent for a project like this that combines so many datasets from different pharma companies?
Companies and researchers have collaborated in the oncology setting with Project Data Sphere. This is an independent, not for profit initiative designed to enable the sharing and analysis of comparator arm data from phase three clinical trials in oncology with the goal of advancing future research worldwide.
Our web portal adds to this initiative by providing data from many therapeutic areas and by providing the full data (both comparator and investigational arm). We hope that simplifying the process of asking for data from multiple sponsors, and creating a single analysis environment where researchers can access the data from those sponsors will encourage reanalyses and pooled analyses of the data from clinical trials, sparking further research in many different fields of medicine.
A review panel will consider research requests before access is granted to the data, how is the panel chosen and what are the key attributes they look for in a research proposal?
The panel contains a mix of clinical, statistical and legal expertise with representation from patient advocates.
The panel assess the scientific quality of the research proposal and its publication plan. They also examine the qualifications of the team (looking in particular for the inclusion of a qualified statistician on the research team), and the team’s ability to manage any conflicts of interest.
How does a researcher go about requesting data from a trial that is not listed on the site?
If a researcher cannot see the trial that they are interested in within the lists provided on the website, then they can submit an enquiry to the majority of sponsors from within the website. The company concerned will assess whether the study is in scope for sharing (based on their published data sharing policy). If it is in scope and available, the study details will be added to the list on the website and the researcher can then include it in their research proposal.
How does the trial data made available on the site by each company differ?
On the website, each sponsor has a ‘study sponsor’ page that contains their criteria for listing studies and other relevant information. The key pre-requisite for a study being in scope for patient level data sharing is that the drug and indication in question has completed the regulatory review process.
Some companies are sharing data from drug approvals on or after 1 January 2014, whilst others are willing to share data and documentation from older studies. Sponsors differ in terms of the phase of trials that are in scope - most focusing on phase two and three ‘registrational studies’, whilst others (GSK for example) are more broad in their remit and include trials from every phase of development including post marketing studies. Studies from terminated development programs can also be shared.
Datasets will generally not be shared until the primary publication for a study has been published - this is to enable the researchers who have worked on the study the opportunity to publish first. Examples of trials that may not be shared include phase one studies or trials conducted in a foreign language.
Researchers will need to agree to a data sharing agreement. What effect will this have on any insights gained by researchers from their analysis?
One of the key purposes of the data sharing agreement is to ensure that any insights gained from accessing these data are made available to all via the publication of the work when completed. The other key principles captured within the agreement are to ensure the protection of patient privacy and a commitment to analyse the data as predefined in the research proposal.
There are some other commitments in the agreement that should not impact the academic freedom of a researcher but are also important. For example, if a research project uncovers a safety signal, this should be communicated to the sponsor and relevant regulatory authorities immediately.
When the data is released to researchers, will datasets from different companies be in formats that can be easily combined in a meta-analysis?
Each sponsor will be supplying annoymised copies of the datasets that were used for the original clinical study reporting. More and more sponsors now use the CDISC data standards for their datasets set-up and so recently conducted trials should be more consistent in their dataset structure. However older studies will be in a variety of structures and formats.
We are well aware of the amount of effort required by a researcher to combine datasets from a variety of sources and work hard to avoid the undocumented data experience parodied on YouTube by Karen Hanson. We hope that the regulatory standard documentation we provide (variable names and labels, blank case report forms, analysis plans, data specifications and documentation of variable derivations and clinical study reports) will make it easier for researchers to navigate their way around our data and ultimately to generate new information and insights.