(312) Strategy for identifying data sources in a rare disease through data landscaping and maximizing the usefulness of the results for researchers

Background: Despite ~7,000 rare diseases (RD) affecting ~30 million people in the US and ~36 million in Europe, knowledge of a RD is often limited given the small number of patients, unique patient journeys, and differences in health care systems. Many RDs do not have distinct diagnosis codes, which complicate identification in administrative databases. Disease or product registries provide in-depth clinical and treatment data and do not rely on diagnosis codes or phenotypes. Knowledge and use of these registries often exists in isolated pockets of an organization and is not routinely shared between researchers in the same field.

Objectives: Comprehensive data landscaping can identify and collate available RD data sources. Strategic storage, communication, and continual updating are key for long-term maximum impact of data landscaping results for research.

Methods: In scope of a larger data landscape project for multiple RDs, a comprehensive review of real-world data sources for multiple myeloma (MM) was performed with a special focus on population-based data sources, registries, and data sources in the US. Structured searches of PubMed and Embase as well as unstructured searches of the gray literature were done to identify and characterize suitable data sources. An Excel-repository was developed to store the characterization of different data sources in a structured way and to prepare the transfer to a search tool called “RD Navigator” which allows for an initial fit-for-purpose data source assessment.

Results: Approximately 40 relevant data sources for MM were identified along with information on relevant clinical and epidemiological data available within the data source. Data were transferred to the “RD Navigator” and shared with the RD community of practice within the organization, including MM researchers in Commercial, Medical Affairs, Regulatory, R&D and Safety functions. Researchers identified which databases had been used previously and which ones were newly identified. Plans were developed to use large language models to update data landscaping results biannually.

Conclusions: The “RD Navigator” provides a technology for informing researchers about the availability, contents, and status of data sources for RDs. Making researchers within an organization – independent of function - aware of the available databases helps ensure that the most fit-for-purpose data sources are used and prevent redundant searches. Having all relevant data sources in one place and keeping them updated, will facilitate accelerated collaboration with database owners, quicker responses to regulators and insurers, development of diagnostic tools for RDs, appropriate identification of external comparator arms, and quicker development of federated data networks.