OBJECTIVES: To evaluate the fitness of three databases ─ A (large national health insurer), B (regional integrated delivery system), and C (national independent claims-only database) to emulate clinical trials that assessed interchangeability between originator biologics and their biosimilars across multiple therapeutic areas.

METHODS: We assessed eight trials evaluating the efficacy, safety, and immunogenicity of switching between biosimilars and originators in Type 1 diabetes mellitus, rheumatoid arthritis, plaque psoriasis, neovascular age-related macular degeneration, and metastatic colorectal cancer. Variables from each trial were categorized (assessment, behavior, demographics, diagnostic, laboratory, procedure, treatment, vitals), and their data richness and consistency were assessed across sites. The overall data richness was evaluated using (1) the mean observed-to-expected (O/E) ratio across trials, calculated for each category within each trial based on available variables versus an ideal complete dataset, and (2) the average percentage of contributing categories, determined by counting categories with at least one variable present out of the total possible categories across all trials.

RESULTS: Database B demonstrated the highest overall data richness (mean O/E: 99%), consistently providing the most complete data relative to expectations. B and C showed comparable overall data richness (mean O/E: 88% and 85%, respectively) but differed in category coverage. B and A provided broader category coverage (average 75% and 71%) compared to C (58.0%). All databases provided demographics, diagnostic, and treatment data. Database B included measures available in medical records (e.g., laboratory data) but was the smallest, with approximately 2 million patients. Database C had fewer contributing categories and was nationally representative and the largest, including over 160 million patients.

CONCLUSIONS: All databases had strengths and limitations in completeness, variety of available data, and population size that could influence database selection depending on the research questions of interest. Future research should focus on enhancing data capture, standardization, and database interoperability to improve RWE generation.