OBJECTIVES: Researchers, healthcare providers, payers, and regulators are becoming increasingly receptive to evidence generated from reliable and applicable real-world data (RWD) sources alongside randomized controlled trials. Therefore, real-world evidence (RWE) studies must be methodologically robust with due attention given to the suitability of the data source. Here, we describe a methodological framework for identifying and appraising RWE data sources.

METHODS: We synthesized expert insights and experience from RWD studies using existing data sources, as well as undertaking a targeted literature review of relevant scientific literature, guidance and tools published by healthcare stakeholders in order to develop a stepwise framework for assessing the availability and suitability of existing data sources for RWE studies.

RESULTS: The following steps were identified as key to implementing a robust data source feasibility assessment:

1) Establishing research scope: Clearly identifying the research question, objectives, disease indications and eligibility criteria including countries of interest and data type such as electronic health records, administrative claims data, cancer registries, and/or specialty data providers and networks.

2) Data source identification: Conducting a targeted literature review of peer-reviewed scientific literature, databases registries and networks in line with a pre-defined search strategy.

3) Data source screening: Comparison against pre-specified criteria to generate a database listing of the most appropriate data sources for further examination.

4) Data extraction: Collation of key information about each shortlisted data source according to a standardized grid, to include details on parameters such as dataset characteristics, data access, coverage, time-lags, and availability of dataset variables.

5) Assessment and appraisal: Consolidation and grading of information to select the most appropriate data source(s) for addressing the research question and objectives.

CONCLUSIONS: This simple framework can be applied and adapted across multiple disease and therapy areas to facilitate the identification of the most appropriate and relevant data sources for the generation of robust RWD.