Our partner Phaik Yeong Cheah, is the coordinator of the Mahidol Oxford Tropical Medicine Research Unit (MORU) Data Access Committee and has contributed to many debates on data sharing. In this blog, she shares her experience in research data sharing and management.
International health-related research funders, regulators and journals want individual-level health data to be shared more widely, but in reality, the volume of data shared and re-used remains low.(1) Health researchers and other data producers are reluctant to share data unless they know that their datasets are of high quality and reliable. That means having good data management process and procedures and well-trained data managers, which in turn requires investment and commitment by research groups, institutions and funders to provide good data management infrastructure, skills training, data management hardware and software.
So how can data sharing and data reuse be promoted?
First, research groups, departments and institutions should establish their own research data sharing policies to maximise the use of its data.(2,3) These policies can serve as powerful internal data-sharing incentives, which have frequently been overlooked in efforts to promote external (funding agency, journal) incentives.
Second, primary and secondary users should share costs of hardware, software, data storage and staff, training, and data curation. Cost sharing could also remunerate time spent by Data Access Committee members in reviewing data requests, or by institutional lawyers drawing up data access agreements.
Third, promoting data sharing should also entail promoting data reuse. This measure could reverse the lack of requests for data reuse, which can discourage researchers from investing in making their data available.
What can funders do to help?
In addition to recent suggestions,(1) funders should cover costs of data management, sharing, training and development of data sharing policies in their grants. They can encourage data reuse by making specific funding available for it.
For their part, researchers should rely not only on external incentives imposed by funders and journals, but imagine a world in which their datasets can contribute to present primary analyses and offer possibilities for multiple secondary analyses.