Year: 2026
Type: article
Abstract: Leveraging external control data has been used to enhance the efficiency of clinical trials, especially in rare diseases where recruitment is often challenging. However, directly pooling data from different studies without appropriate adjustments can lead to biased results when populations differ across these studies. In addition to limited sample sizes, trials for rare diseases commonly assess treatment efficacy through multiple clinical endpoints using composite endpoints. The win ratio has gained attention for composite endpoint analysis, as it enables prioritized comparisons that account for the relative clinical importance of each endpoint. Motivated by these two challenges, we proposed novel propensity score (PS)-integrated win ratio methods to incorporate external control data. Specifically, two PS-based weighting approaches, PS-ratio and PS-difference, are proposed to adjust for between-study baseline covariate differences, thereby mitigating the risk of potential bias. Simulation studies and real-world case analysis demonstrate that the proposed methods consistently improve statistical power while maintaining proper Type I error control. This framework offers a robust and practical solution for composite endpoint analysis using external controls, with particular relevance to rare disease trials and regulatory decision-making.
Institutions: Baylor University, Eli Lilly (United States)
Language: English
Subfield: Statistics and Probability
Field: Mathematics
Domain: Physical Sciences
Open Access status: bronze
OpenAlex ID: W7204248104