Methodological advances in regulatory Real World Evidence generation systems: Perspectives from Sentinel and DARWIN-EU

Background: Routinely collected longitudinal healthcare data including insurance claims and electronic health records (EHRs) remain the most important source for real world evidence (RWE) generation including studies of disease epidemiology, medical product utilization, and post-marketing safety. With availability of more granular information from variety of sources including EHRs, biobanks, and linked nation-wide registry data, the breadth of questions that can be reliably answered has increased.

Objectives: To describe the expansion of data sources and resulting methodological innovations that improve national and international regulatory RWE generation systems.

Description: Members of the panel will share learnings related to the following domains: 1) the need for innovation in large scale RWE evidence generation: this section will provide regulators’ perspective on the importance of advanced methodology and newer data sources (Ball, FDA; Arlett, EMA), 2) fit-for-purpose data sources: this section will cover aspects related to data continuity, data granularity, and data chronology and outline various data infrastructure initiatives underway to enhance the US Sentinel System and DARWIN EU® (Schneeweiss, Sentinel; Duarte, DARWIN EU), 3) methodological initiatives in Sentinel: this section will cover various domains with applied examples including use of natural language processing and machine learning for computable phenotyping and expedited endpoint validation, advancing causal inference with ultrahigh-dimensional confounding adjustment, and leveraging partially observed confounder data in EHRs with advanced statistical methodology including generalized ranking and targeted maximum likelihood (Desai), 4) methodological initiatives in DARWIN EU: this section will cover activities to broaden the scope of RWE evidence generation including genomics and nation-wide linked registries in DARWIN EU® to enable pharmacogenomics, evidence triangulation and mendelian randomization, and population-based studies (Prieto-Alhambra). The symposium will conclude with a session dedicated to address questions and comments from the attendees to facilitate discussion of all pertinent issues.