Prevalence and characteristics of data sharing policies across the health research life cycle: Funders, ethics committees, trial registries, journals and data repositories

Preview
Authors
Aidan C. Tan, Sol Libesman, Zijing Yang, Rani R. Chand, Weber Liu, Talia Palacios, Kylie E. Hunter, Angela C. Webster, Anna L. Seidler

Objective: One way to reduce the gap between high in-principle support for the concept of data sharing and low in-practice commitment to sharing data is if major stakeholders across the health research life cycle implement policies to recommend or require data sharing. The primary objective of this study was to determine the prevalence and characteristics of data sharing policies.

Design: This cross-sectional study included the 55 largest private and 55 largest public and philanthropic health research funders by annual health research expenditure, all national ethics committees, all clinical trial registries, the 5 highest impact peer-reviewed scientific journals by Journal Impact Factor for each of the 59 fields of clinical medicine, and all research data repositories in clinical medicine. All official websites, online reports and grey literature information sources of stakeholders were assessed for the presence of a data sharing policy and, if present, its magnitude of support for data sharing.

Results: Overall, 110 funders, 124 ethics committees, 18 trial registries, 273 journals, and 410 data repositories were included. More than half (55%, N=110) of funders recommended or required data sharing, and this was more common if they were a private pharmaceutical and biotechnology funder (80%, 35/44), from a high income country (64%, 56/88), or from North America (71%, 22/31). Only 4 (3%, N=124) ethics committees recommended data sharing, and all were from high income countries. Only 1 (6%, N=18) trial registry recommended data sharing. Almost two-thirds (63%, N=273) of journals recommended or required data sharing, and these had a similar median 2022 Journal Impact Factor (median 8.6, IQR 15.2) to other journals. Few (12%, N=410) data repositories recommended or required data sharing, and these were more commonly from East Asia and Pacific (20%, 5/25) or international (20%, 4/20).

Conclusion: Data sharing imperatives were not met by most stakeholders.