Answers are grounded in this paper and cite their sources.
The presence of treatment switching in randomized controlled trials may bias estimation of causal treatment effects on outcomes such as overall survival if a standard intent-to-treat analysis approach is applied. To address this, techniques from the realm of non-experimental study methodology such as two-stage estimation (TSE) have been incorporated into the clinical trialist’s toolbox. Of particular relevance to this manuscript is research exploring how TSE may be applied to settings where patients in one study arm may switch to alternative treatment so that potential bias associated with following patients after treatment switching is reduced. We contribute to the evolution of this research by extending the TSE approach so that it can appropriately account for two-arm designs where patients in both arms are allowed to switch treatments. In addition to a discussion of methodology, we have included a case study and a diversity of simulation studies supporting the statistical validity of the proposed technique.
No takes yet. Share an insight, caveat, or question.
Bay et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: