(P-100) Characterization of Real-world Data Sources to Support Observational Studies of Biologics, Including Biosimilars in Oncology

Background: While the use of large US health insurance claims and electronic health record (EHR) databases in postmarketing safety assessments of medical products is well-established, studies of oncology-relevant biologics and biosimilars require different design and analytic considerations compared to typical pharmaceutical products. The Biologics and Biosimilars Collective Intelligence Consortium (BBCIC) – a multi-stakeholder consortium established to support transparent, methodologically rigorous research – seeks to enrich its data network to better enable answering research questions in oncology.

Objectives: To identify and assess the utility of potential real world data (RWD) sources for the comparative safety and effectiveness of oncology biosimilar and reference products.

Methods: BBCIC convened a Workgroup (WG) of researchers and clinicians from payers, industry, and academia. The WG developed recommendations to assess the utility of potential RWD sources for conducting comparative safety and effectiveness studies of cancer care. We organized discussions around a hypothetical comparative study of trastuzumab originator and biosimilars in breast cancer patients. WG members developed and disseminated a questionnaire to potential RWD sources, soliciting information on general characteristics of the RWD source and the availability of specific data elements.

Results: 18 RWD sources were identified from literature review and WG suggestions. A total of 7 (39%) sources responded to the request: 2 EHR-based sources, 1 claims-based source, and 4 with a mix of EHR and claims data. Three responses included counts of breast cancer patients receiving trastuzumab between 2016 and 2020, with populations ranging from 3,669 to 11,653 patients. Respondents described key data elements including: availability of race and body mass index (n=4, 57%), availability of biomarkers (e.g., estrogen receptor +/– from laboratory data, n=4, 57%), ability to distinguish between reference biologics and biosimilars (n=5, 71%), and ability to link to external data sources (e.g. cancer registries, n=3, 43%).

Conclusions: We found important variation in the type and completeness of data capture for critical data elements. In many cases relevant to oncology research, data sources which incorporate linked EHR and claims data, or which have the ability to link to external data sources, will be the most appropriate. This survey provides a framework for assessing potential data sources for oncology-relevant biologics research, and recommendations for enhancing the BBCIC research network.