Waiting for Trial Evidence or Using Real-World Evidence: Trade-Offs for Coverage with Evidence Development.
TL;DR
A decision-analytic framework is developed that extends value-of-information methods by incorporating an explicit bias parameter to capture structural distortion in RWE treatment effect estimates and quantifies the probability of RCT-based CED dominance.
Abstract
Coverage with evidence development (CED) is used to manage uncertainty in new health technologies, but its implementation varies across settings. Some evidence-development arrangements restrict access to patients participating in randomized controlled trials (RCTs), whereas others provide broader conditional coverage while real-world evidence (RWE) is generated concurrently. RWE may be generated more quickly than trial evidence, often using larger samples and at lower cost. However, it is vulnerable to bias such as unmeasured confounding, and both the magnitude and direction of such bias are typically unknown. Decision-makers therefore lack an ex ante framework for determining when RWE-based CED is preferable to RCT-based CED. We develop a decision-analytic framework that extends value-of-information methods by incorporating an explicit bias parameter to capture structural distortion in RWE treatment effect estimates. Drawing on quantitative bias analysis, the framework defines policy indifference bias thresholds as the primary decision metric, identifying the maximum tolerable bias for RWE-based CED to remain preferred. The framework also quantifies, as a complementary measure, the probability of RCT-based CED dominance, defined as the ex ante probability that the RCT-based strategy is more favorable even when RWE is unbiased. A case study illustrates how these two metrics are determined.