Unlocking Practice Patterns at Scale: A Framework for Developing Clinical Insights Using Epic’s Cosmos

BACKGROUND: Real-world data from vendor-aggregated health information exchanges represent a powerful resource for retrospective clinical research. Epic’s Cosmos platform, a large-scale federated data warehouse, enables querying deidentified electronic health record (EHR) data. OBJECTIVE: We describe the iterative development of a generalizable clinical informatics workflow for retrospective studies using Cosmos. METHODS: We applied the Plan-Do-Study-Act cycle to iteratively refine a collaborative research process using Cosmo’s SlicerDicer interface, focusing on setting, team, technical preparation, and workflow. RESULTS: We identified multiple areas of improvement to facilitate streamlined collaborative studies. Setting: Our Institutional Review Board confirmed Cosmos studies as non-humans subject research. We expanded collaborations across institutions. Team: We identified key insights for team definition across clinical, research, and skill domains. We expanded our team based on technical skills, domain expertise, and educational aims. Technical preparation: We optimized query building and cohort construction, data analysis and validation, and communication processes. Workflow: We clarified and optimized a collaborative workflow across clinical and informatics expertise. DISCUSSION: Our collaborative approach to secondary data analysis in Cosmos supports the development of meaningful clinical evidence to support high-quality, well-evidenced patient care. This study demonstrates the application of the PDSA cycle to a collaborative workflow for clinically focused secondary data analysis. CONCLUSIONS: Our approach allows rapid, reproducible cohort construction and analysis, is adaptable across clinical domains, and scales to multi-organizational collaboration. This approach offers a model for others seeking to develop key clinical insights via retrospective studies within the Cosmos data aggregation tool.