The Data Model Dilemma: How Your Database Choice Shapes Real-World Evidence in Immunology and Pharmacology
A new study in Clinical Pharmacology & Therapeutics reveals that the choice of common data model (CDM) can significantly alter the results of real-world evidence (RWE) studies, a critical finding for immunology and drug safety research. Researchers mapped the same UK primary care database (CPRD GOLD) to two different CDMs—OMOP and ConcePTION—and then analyzed the comparative risk of bleeding and cardiovascular outcomes for direct oral anticoagulants (DOACs) versus vitamin K antagonists (VKAs) in atrial fibrillation patients. While both models showed no increased risk of major bleeding from DOACs, they produced divergent results for cardiovascular disease protection, with OMOP showing a significant protective effect not observed in the ConcePTION-mapped data. This highlights a fundamental, often overlooked variable in pharmacoepidemiology and immunotherapy safety monitoring: the data infrastructure itself can be a source of variation, potentially impacting conclusions about drug efficacy and adverse immune-related events.
Study Significance: For immunologists and clinicians relying on real-world data to assess novel immunotherapies or vaccine safety, this study underscores a critical methodological pitfall. The observed discrepancies likely stem from how different models handle cohort construction, phenotype definitions, and imputed variables like drug exposure duration—factors directly relevant to tracking cytokine release syndromes or other immune-mediated adverse events. This work mandates a more rigorous, transparent approach to database methodology, ensuring that safety signals for emerging biologics or monoclonal antibodies are not artifacts of the data model but true reflections of patient outcomes.
Source →Stay curious. Stay informed — with Science Briefing.
Always double check the original article for accuracy.
