For the use of RWE in HTA and health policy decision making, it is crucial to use valid estimates of the benefits, harms and costs of the interventions of interest. In this session, methodological approaches and frameworks for causal inference and their applicability in light of data availability and quality are presented and discussed.

Barbra Dickerman will detail the design of the target trial emulation framework and describe applications of the approach to reduce bias (e.g., immortal time bias, time-varying confounding) and to generate actionable evidence to inform decision-making. Nicholas Latimer will present the use of hypothetical estimands in clinical trials when standard intention-to-treat (ITT) analysis does not yield an answer to the decision problem, and relate this issue to the analysis of real-world data and cost-effectiveness. Sebastian Schneeweiss will focus on dealing with differential measurement issues frequently encountered in external control arm (ECA) analyses via calibration and hybrid designs.

The goal of this session is to sensitize all stakeholders to potential biases in the use of RWE and to discuss solutions for providing high-quality RWE for HTA bodies and health policy decision making.