72 - A Landscape Survey of Generalist Data Repositories to Facilitate Researcher Selection and IRB Review

Background:  Data sharing for secondary research is an important catalyst accelerating biomedical research. Understanding the landscape and features of available externally-operated data management and sharing platforms (repositories) promotes good data stewardship for investigators and their institutions. Meanwhile, updated guidance for IRBs helps navigate ethical concerns, as well as data security, privacy, and attribution obligations.

Methods:  We conducted a landscape survey studying ten commonly used generalist repository platforms. We analyzed eight traits to synthesize guidelines for IRB and institutional review. Key areas of focus included the protection of sensitive data through access gating features, as well as institutional concerns such as data security, privacy and attribution requirements. IRB review guidance, definitions of repository attributes, and useful public resources were compiled as a handout to supplement the survey.

Conclusion: The ten repositories surveyed have unique feature sets that should be weighed when assessing each repository’s suitability for a particular study’s data. Key insights include: Several repositories support some form of access gating, while others are strictly open access. Few repositories support data containing HIPAA identifiers. Most use cloud storage, with Amazon AWS being the primary choice, which carries several industry-recognized data security certifications. Nearly all repositories include citation tools, and all mandate attribution to data contributors. All ten repositories are appropriate for human data, though Vivli focuses on clinical research. Only a few offer cloud computing environments, which would reduce the need for data downloads to local systems. Most require a range of user credentials, though some permit unrestricted open access.

Limitations: This landscape survey and analysis only addresses generalist repositories. Domain-specific repositories are options for researchers, but limit the scope of data sharing, and can reduce cross-disciplinary collaboration. Only repositories that allow open evaluation of their features were included in the survey. Additional factors to consider for the future include the use of AI or machine learning on shared datasets.

Discussion:  Robust data sharing is an important catalyst for accelerating biomedical research discovery and fostering collaboration. Data availability for secondary use ensures that the contributions by individual research subjects yields the most value as a social good. However, considerable ethical and privacy evaluations are crucial to ensuring that individual subject wishes are upheld. This landscape survey identified a list of ten generalist repositories outlining the presence, absence, and nuances of data stewardship elements. This survey and the accompanying resources handout can be used to inform IRBs, researchers, and institutions of available repositories and address some of the considerations needed to ensure ethical and compliant data sharing.