blog

Christian Ohmann discusses his research article on sharing individual participant data from clinical trials explaining the current challenges in sharing this data and what can be done to overcome these challenges.  

Sharing sensitive data

Sharing detailed result data, as part of ‘open science’, is the norm in many scientific disciplines, but this has only recently been the case with clinical research. Here the data is about individuals and their health status, and so requires specific measures to protect the privacy of study participants.

Nevertheless, funders and publishers are leading a wider cultural shift towards more open science, with various attempts to explore how clinical researchers can best plan for data sharing, so that they can make their ‘raw’ individual participant data available to others, often under controlled access conditions rather than simply being publicly available on-line. A greater use of standards will also be required, to make the shared data more inter-operable with the Individual Participant Data from other studies.

A consensus document

Working in the context of the European CORBEL H2020 project, I and colleagues in ECRIN, the European Clinical Research Infrastructure Network led a large, multi-disciplinary, international task force in 2016 and 2017 to examine the best ways of supporting researchers in these tasks.

The resulting consensus document included 10 principles of data sharing, and 50 more detailed, pragmatic recommendations for action, systems and processes that could help researchers turn data sharing in clinical research into a practical reality.

Tool and Services for practical support

The principal groups involved in data sharing are the researchers who create the data in the first place, the specialist repositories who will (increasingly) store the data in the longer term, and the researchers who wish to access the data for secondary use. All could benefit from an improved infrastructure that provides practical support for data sharing – including:

• Templates and checklists to be used when planning clinical studies, to ensure data sharing has been included in the study planning (and grant application) process.
• Training on the legal issues surrounding data sharing.
• Libraries of standardized data items and tools to easily find and apply them.
• Guidance on consent processes that can better support secondary re-use of study data.
• Services to help with data de-identification and risk assessment.
• Web-based tools to help researchers apply the correct metadata to their datasets.
• Example data use agreements, that help ensure data is used only for the purposes specified.
• Information on the different repositories available, their costs and facilities.

There are many potentially useful tools and services, and those listed above are only examples.

Today, the main challenge is to provide easier routes for optimal ways of sharing data. The difficult task of organizing data, and the lack of time to do so, require readily available ways to organize and share data, which are easily accessible and usable by researchers.

This should be accompanied by increased education and support on good data management, including readily available advice and support about good data practice, raising awareness about the availability of repositories, and understanding of copyright and licensing of research data.

The ECRIN core group began to examine how such a support infrastructure could best be developed and implemented, but with only finite time and resources it was important to start with an assessment of what tools and services already existed; how effective they were, and where the main gaps were perceived to be.

Defining the Processes

First, however, we wanted greater assurance that we were considering all the possible processes that played a part in data sharing, considering the life-cycles of both the data and the associated study in a systematic fashion. Such an analysis was not readily available in the literature, so we set about creating it ourselves, using the CORBEL consensus document as a starting point.

The processes identified were tabulated and are explained in our recent F1000 paper. For each process, tools and services were identified that could be useful in supporting them, and these were then grouped into six major types of support systems. The analysis was aided by the constructive feedback from reviewers. The work provides more concise, systematic description of all processes involved, complementing the consensus exercise.

This work is a further step in what will need to be a long-term programme of analysis, development, and evaluation of tools and services for data sharing, and ECRIN will continue to work with others trying to implement, or at least demonstrate, some of the tools and services required in this area. We hope that many researchers and research infrastructures (including repositories) will also find the programme of interest.