Real-World Evidence for Crossover Adjustment: Challenges, Opportunities, and Newly Proposed Methods

PURPOSE: 1 – Oncology trials often allow those receiving the control treatment to crossover to the experimental treatment at disease progression. Since participants who switch often have a different prognosis than non-switchers, estimating the treatment effect with regards to overall survival can be challenging. Specialized statistical methods (e.g., rank-preserving structural failure time models (RPSFT), inverse probability of censoring weighting (IPCW), two-stage estimation (TSE)) are often recommended for “crossover adjustment”, but these can fall short when sample sizes are small, and the degree of crossover is high. 2 – Given that real-world evidence (RWE) is increasingly used to support applications for regulatory and HTA approvals, the potential of RWE for crossover adjustment is certainly worth considering. 3 – We review examples of how RWE has been used for crossover adjustment in a number of recent studies (e.g., to validate different crossover adjustment methods (Hawkins et al. 2022); to create an external control arm without any crossover contamination (Morschhauser et al. 2023)). 4 – We discuss newly proposed methods to supplement RCT data with RWE for efficient and unbiased crossover adjustment.

DESCRIPTION: Workshop attendees will develop an understanding of methods for cross-over adjustment and the potential of RWE. The audience will be asked (real-time polling) to consider if RWE has potential to improve cross-over adjustment and if potential biases can be addressed. Ms. Cope will introduce the main ideas and summarize the specialized methods that are often recommended for cross-over adjustment (Part 1 - 10 min). Dr. Jansen will review how RWE has been used in different applications across HTA (Part 2 - 15 min). Prof. Hawkins will then review specific examples of how RWE has been considered for cross-over adjustment (Part 3 - 15 min) and Dr. Campbell will discuss newly proposed statistical methods (15 min). Ms. Cope will conclude leading an audience discussion (5 min).