Assessment of Real-World Data Sources and a Hybrid Approach in Real-World Evidence Generation Using Unharmonized Data Sources

OBJECTIVES: Real-world evidence (RWE) is increasingly used for regulatory, reimbursement, and patient access decisions in Asia-Pacific. RWE has the capability to address similar evidence gaps across different countries within a single study through standardized research methodologies, which enables results to be analyzed and presented holistically. However, the heterogeneity of real-world data sources in Asia-Pacific poses a major challenge to conduct multi-country real-world research using a standardized approach.

METHODS: We evaluated data sources across Asia-Pacific countries to support RWE generation and described a hybrid approach to utilize unharmonized data sources of varying levels of data maturity for multi-country research. Data sources were evaluated based on technical and operational elements: E.g. data availability; data format (charts, electronic records); completeness; access (individual, aggregate-level); consent requirements; and capacity and burden on research team and patients.

RESULTS: We adopted a hybrid-Mosaic Model to harmonize data collected using different methods when a single collection method was not feasible for all data sources. In the Mosaic Model, data was directly extracted and enhanced from electronically structured data sources (e.g. electronic medical records, registries), and transformed using a common data model. For paper-based or electronically unstructured data sources, data was abstracted from medical records and subsequently transformed to the same common data model. The use of a common data model harmonizes the data for analysis and overcomes the challenges faced with unharmonized sources and concerns about data privacy, while ensuring consistency and validity of the research under a single protocol, analysis, and timeline.

Adaptations of data collection methods for specific data sources may be required to ensure that data collected is fit-for-purpose while overcoming the limitations of each data source.

CONCLUSIONS: This hybrid approach was successfully implemented in oncology and metabolic disease studies in Asia-Pacific, enabling variables captured across different unharmonized data sources to be standardized for analysis.