Pharmaceutical companies, academic researchers, and government agencies such as the Food and Drug Administration and the National Institutes of Health all possess large quantities of clinical research data. If these data were shared more widely within and across sectors, the resulting research advances derived from data pooling and analysis could improve public health, enhance patient safety, and spur drug development. Data sharing can also increase public trust in clinical trials and conclusions derived from them by lending transparency to the clinical research process. Much of this information, however, is never shared. Retention of clinical research data by investigators and within organizations may represent lost opportunities in biomedical research.
Despite the potential benefits that could be accrued from pooling and analysis of shared data, barriers to data sharing faced by researchers in industry include concerns about data mining, erroneous secondary analyses of data, and unwarranted litigation, as well as a desire to protect confidential commercial information. Academic partners face significant cultural barriers to sharing data and participating in longer term collaborative efforts that stem from a desire to protect intellectual autonomy and a career advancement system built on priority of publication and citation requirements. Some barriers, like the need to protect patient privacy,
1The planning committee’s role was limited to planning the workshop, and the summary has been prepared by the workshop rapporteurs as a factual summary of what occurred at the workshop. Statements, recommendations, and opinions expressed are those of individual presenters and participants, and are not necessarily endorsed or verified by the forums, the roundtable, or the Institute of Medicine, and they should not be construed as reflecting any group consensus.
present challenges for both sectors. Looking ahead, there are also a number of technical challenges to be faced in analyzing potentially large and heterogeneous datasets.
Despite these barriers, there is increasing acknowledgment among researchers of the importance and potential benefits to sharing clinical research data at various stages of the research, discovery, and development pipeline. Precompetitive collaboration models promote the sharing of resources and risk among competitors at early stages of the research process, with the goal of providing benefit to all parties. A number of collaborations, including public-private partnerships, have formed to overcome these barriers in order to advance clinical research and accelerate the discovery and development of therapeutics and diagnostic tools.
On October 4-5, 2012, four groups within the Institute of Medicine—the Forum on Drug Discovery, Development, and Translation; the Forum on Neuroscience and Nervous System Disorders; the National Cancer Policy Forum; and the Roundtable on Translating Genomic-Based Research for Health—collectively hosted a workshop to examine and advance this pressing issue. The workshop explored the benefits of sharing clinical research data, the barriers to such sharing, and strategies to address these barriers to facilitate the development of safe, effective therapeutics and diagnostics. Box 1-1 provides the objectives of the workshop. The workshop was designed to provide a neutral venue where stakeholders from government, academia, industry, foundations, public-private partnerships, patient groups, and the public could meet to discuss issues of mutual interest. It is part of a larger effort to build partnerships and enhance collaboration within and among sectors on research, development, and assessment of pharmaceutical products. While acknowledging the importance of other kinds of clinical data, the workshop organizers focused on issues relating to sharing of data from preplanned interventional studies of human subjects. This summary of the workshop presents the observations, viewpoints, and suggestions made during both the presentations and discussion sessions as a way of informing the public, the press, and policy makers about the major issues surrounding the sharing of clinical research data.
In the final session of the workshop, the moderators of each workshop session identified the major points that emerged during their sessions
This public workshop focused on strategies to facilitate sharing of clinical research data in order to advance scientific knowledge and public health. While the workshop focused on sharing of data from preplanned interventional studies of human subjects, models and projects involving sharing of other clinical data types were considered to the extent that they provided lessons learned and best practices. The workshop objectives were to
• examine the benefits of sharing of clinical research data from all sectors and among these sectors, including, for example:
— benefits to the research and development enterprise and
— benefits to the analysis of safety and efficacy;
• identify barriers and challenges to sharing clinical research data;
• explore strategies to address these barriers and challenges, including identifying priority actions and “low-hanging fruit” opportunities; and
• discuss strategies for using these potentially large datasets to facilitate scientific and public health advances.
along with issues that warrant more focused attention. These points are presented in the next section of this first chapter of the workshop summary as an introduction to the themes of the workshop. Chapter 2 of this report summarizes the benefits of sharing clinical research data and suggests why and when increased data sharing can improve both scientific knowledge and public health. Chapter 3 considers the barriers to data sharing and examines changes that could overcome those barriers. Chapter 4 describes different models of data sharing to demonstrate best practices and lessons learned from each project. Chapter 5 looks at the standardization of clinical data, for both data collected in the past and future data collection efforts. Chapter 6 contains a discussion of mechanisms and incentives to enhance data transparency and sharing across all sectors. Chapter 7 concludes this workshop summary by gathering key take-away points from the workshop that were identified by speakers and other participants as a way to highlight and elaborate on next steps for advancing the sharing of clinical trials data.
Many journal articles are at most a synopsis, and in many ways more like an advertisement, for immense quantities of data from which they are derived, said John Ioannidis, C.F. Rehnborg Chair in Disease Prevention at Stanford University. Much of the underlying data are not made available to other researchers or to the public, and are eventually discarded. As Ioannidis put it, the equivalent of several Libraries of Alexandria disappears every day. “How can we regain that information before it is too late?”
The benefits of sharing research data have been amply demonstrated in areas such as cardiovascular disease, where death rates have fallen 40 percent in recent decades, pointing toward the great potential of data sharing to improve human health, said William Potter, co-chair emeritus of the Neuroscience Steering Committee for the Biomarkers Consortium of the Foundation for the National Institutes of Health, in his summary of major messages from the first session of the workshop. Building and sharing datasets within and across the public and private sectors are practices that should be widely emulated. Advances in treatment for the complex illnesses faced by society today are not likely to result from the data from a single study that has been analyzed one time, he said; increased transparency and sharing at the participant level are needed to tackle these challenging diseases.
Several challenges have severely inhibited such sharing, said Jeffrey Nye, vice president for Neuroscience Innovation and Partnership Strategy at Janssen Research & Development, LLC, a Johnson & Johnson pharmaceutical company. For example, data holders have refrained from making data available to others due to privacy concerns. The process of de-identifying and standardizing data in some past cases of data sharing has been expensive and time consuming. Industry is interested in protecting proprietary information that can contribute to commercial products. Academic researchers have incentives to withhold data for their own use so they can advance professionally.
But a cultural shift is occurring, said Potter. Measures such as finding ways to credit academic researchers for sharing data are helping to reduce barriers to sharing. Industry has realized that it needs to collaborate to overcome the major obstacles it faces in developing new drugs, Potter continued. The sharing of data can correct mistaken conclusions and lead to new discoveries that would otherwise go undetected. Transparency in the use and dissemination of data can strengthen public trust in the biomedical research enterprise. Finally, he said, regulatory agencies are also recognizing the importance of facilitating this process and are working with researchers in academia and industry to identify paths forward.
Many stakeholders are involved in the sharing of clinical data, including participants in a trial, researchers, private companies, regulators, and the public, and each has particular interests and expectations. Effective communication and mutual understanding will be essential to identify common values and to take full advantage of current opportunities.
Models of Data Sharing
Successful models have demonstrated the value of sharing clinical trials data, said Jeffrey Nye, in his summary of the session on best practices and lessons learned from past experiences with data sharing. These models provide concrete examples of a vision of data sharing that can motivate action and lead to progress.
These models also have demonstrated some of the challenges of data sharing. The partners in data-sharing initiatives can have different cultures, practices, expectations, and rules. These differences need to be resolved, or at least accommodated, for sharing to occur. Sharing also can build trust among the participants in a collaboration, which in turn can provide the foundation for future initiatives.
One important message from the models presented at the workshop, Nye said, is that summary data can be inaccurate. Examination of participant-level data from clinical trials can be essential to draw correct conclusions from shared data.
Standardization to Enhance Data Sharing
Disclosure does not equal transparency, said Frank Rockhold, senior vice president for global clinical safety and pharmacovigilance at GlaxoSmithKline Pharmaceuticals Research and Development, who moderated the session on standardization and governance at the workshop. Data need to be understandable and analyzable if they are to be useful.
The need for data standards in clinical trials being conducted today and in the future is clear, Rockhold said. Standards can improve data quality, enable separate studies to be combined, and facilitate regulatory review. The application of data standards retrospectively to trials conducted in the past involves additional considerations. For example, the use of retrospective data to answer specific questions may be preferable to standardizing data and depositing the results in a repository for future use.
To date, most standards development has been done on a volunteer, ad hoc basis. Greater recognition of the value of standardization may lead to more cooperative efforts and more sustainable standards development models, Rockhold concluded.
Changing the Culture of Research
Data sharing is a public good, and the actions of the biomedical research enterprise should reflect that good, observed Robert Harrington, Arthur L. Bloomfield Professor of Medicine and chair of the department of medicine at Stanford University. But a variety of disincentives today create a culture that works against sharing. Academic researchers are afraid of losing credit for the work they have done to generate data. Industry is concerned about the loss of proprietary information and potential liability. Patients are worried about privacy. Essentially, every stakeholder associated with clinical trials faces difficulties in moving toward a more open system.
Good will and altruism go only so far in changing a culture, he said. Additional incentives are needed for substantial and enduring change to occur. For example, several factors came together to create powerful motivations for investigators to register trials at ClinicalTrials.gov. New
standards and tools could create new reasons to share data. Funders could consider an investigator’s previous experience and future data-sharing plans in making grant funding decisions. Journals could agree on standard practices that authors must follow. A culture of data sharing could be built into the education of the next generation of clinical researchers.
The advent of organizations outside the traditional biomedical research enterprise offering data analysis and medical advice over the Internet has introduced a new force of cultural change, Harrington observed. If the system does not change from the inside, change may be imposed on it from the outside.