This paper proposes a causal decomposition framework for settings in which an initial regime randomization influences the timing of a treatment duration. The initial randomization and treatment affect in turn a duration outcome of interest. Our empirical application considers the survival of...