“Computable Phenotypes”— Understanding Their Importance in Regulatory Submissions of RWE

PURPOSE: Growing interest in using real-world evidence (RWE) to inform regulatory decision making has given rise to significant refinements in analytic methods to elevate the quality and reliability of RWE. Many of the advances have addressed bias and confounding in analyses of secondary real-world data (RWD) sources (predominantly claims & electronic health records [EHRs]). However, the importance of selecting the right patients, classifying them into the right treatment groups, and assessing the right outcome measures in RWD cannot be overstated. This is the domain of computable operational definitions, or “phenotypes”, which are prominent in the field of clinical informatics but less well established in health economics & outcomes research (HEOR). The objective of this workshop is to evaluate the criticality of computable phenotypes in the development of valid and reliable RWE.

DESCRIPTION: A computable phenotype is a machine-executable algorithm to identify patient cohorts, exposures, and outcomes from variables in RWD. Guidance issued by the US Food & Drug Administration in 2021 for analyses of EHRs and claims data to support regulatory submissions describes the importance of computable phenotypes as a means of standardizing key design elements across different analyses. However, challenges associated with implementation remain a concern. This workshop will address these challenges head on. It will begin with an introduction to computable phenotypes based on examples from different disease states and types of RWD. The next segment will present a case study quantifying efficiency gains associated with use of a technology platform to standardize computable phenotypes across alternative data sources. This will be followed by a description of the use of computable phenotypes in analyses of EHR data at the US Department of Veterans Affairs Health System. The final segment will provide an FDA perspective on the topic. Real-time polling will be used to solicit audience feedback on the issues discussed.