Session (CTI)

PURPOSE: Pre-specification of analyses in real-world evidence (RWE) studies is challenging due to the uncertainty of the size, completeness, and other factors involved in the generation of such data. At the same time, RWE studies that aim at causal inference, in particular, require a robust statistical analysis plan, defined beforehand, for these studies to provide robust scientific evidence and support claims. Fixing the analysis plan without any knowledge of the data is similar to taking a shot in the dark. The task would become easier if we knew something about the data. The workshop aims to help answer the question, how much of the data can we check before finalizing the analysis plan? DESCRIPTION: Dr. Olson will introduce (5 min) the problem statement and the minimal sample size estimation of eligible patients that is typically conducted. Dr. Mittmann will highlight uncertainties related to data explorations in RWE when used to support decision-making and provide a few examples (10 min). Prof. Wang will then propose boundaries around the kind of data that can be checked and how (15 min). Dr. Karcher will work through an example with the audience with multiple possibilities of data checks (10 min), and the other panelists will chime in and comment on scientific and operational feasibility, as well as regulatory acceptance of the checks.The audience will ideally be composed of real-world evidence researchers, epidemiologists, and statisticians, as well as decision-makers and regulators using and auditing RWE studies. The panel will prompt them to comment, in the case examples shown, on the type of data checks they estimate to be performed before registering the protocol and finalizing the main analysis of an RWE study (use of polling system). Additionally, a discussion on future regulations concerning real-world data checks is scheduled for the end of the session (10 min).