Presentation (CTI)

OBJECTIVES: Regulatory agencies including FDA, EMA, DKMA, PMDA, and NMPA increasingly require or recommend that real-world data (RWD) submitted in regulatory packages be transformed into compliant data standards such as CDISC. This is particularly true for effectiveness studies. However, regulatory guidance on RWD transformations remains limited, resulting in potential variation in implementation. Here, we evaluate performance of data transformations and present a framework on key considerations for implementing RWD transformations.
METHODS: We conducted mappings of synthetic real-world data into CDISC data standards including SDTM and ADaM. Various real-world data formats were evaluated for mappings, including HL7, OMOP, and PCORnet. Study types evaluated included retrospective cohort and test-negative control design studies. Therapeutic area considerations, such as transformation of domains specific for oncology/hematology, imaging, and genomics were evaluated. Strengths, gaps, and challenges of mapping and transformation processes were assessed.
RESULTS: We enumerated a framework for best practices when transforming HL7, OMOP, and PCORnet data to CDISC SDTM and ADaM datasets. We present a decision tree for handling design choices including type of observational study design (cohort, case-control, external control arm), whether a trial component is present, assignment of index dates, assignment of treatment arms, and maintaining investigator blinding. We evaluated the integrity and robustness of the data conversion of source data with respect to data loss, along with conformance issues. We outline key considerations, such as approaches to address potential investigator bias when transforming to domains containing outcomes.
CONCLUSIONS: RWD transformation should align with design choices and disease area for optimal performance. We describe a practical framework that enables tailored RWD transformation into CDISC SDTM and ADaM datasets. The framework highlights key design and implementation choices that affect data integrity and conformance with regulatory submissions using RWE. These considerations are relevant when RWD is used in lieu of, or in addition to, traditional clinical trial data.