Abstract
<title>Abstract</title> <p> <bold>Purpose:</bold> Real-world data (RWD) quality assurance is essential for oncology research and regulatory decision-making, yet validated, multi-stage frameworks for heterogeneous, multi-source oncology RWD remain inconsistently documented. This study describes and empirically validates a regulated, four-stage data quality (DQ) framework for integrating two independent oncology real-world data sources into a HIPAA-compliant platform harmonized to the OMOP Common Data Model, supporting a pharmaceutical sponsor's Phase I proof-of-concept evaluation. <bold>Methods:</bold> A four-stage methodology—Mapping File, Structural Mapping, Conceptual Mapping, and Data Synchronization Validation—was applied across eleven OMOP CDM domains using pre-specified, rule-based SQL validation scripts under documented test protocols, aligned with GCDMP principles and the Kahn et al. DQ taxonomy. Five testing phases (Connectivity, Sanity, ETL, System Integration, Regression) were independently executed to formal sign-off criteria. <bold>Results:</bold> All five testing phases achieved a Passed outcome for both sources. Ten of eleven domains in Data Source A and all eleven in Data Source B met structural mapping criteria; four domains per source were not applicable for conceptual mapping and synchronization. The framework identified 354 patients in Data Source B with a visit record dated after death, with direct implications for survival analysis validity. <bold>Conclusion:</bold> A regulated, rule-based, four-stage DQ framework provides a reproducible, auditable mechanism for certifying multi-source oncology RWD before research or regulatory use. Detection of a clinically material temporal defect underscores the necessity of formal, pre-specified DQ assurance as a precondition for trustworthy real-world evidence generation. </p>