Handling Intercurrent Events Through Hypothetical Strategy in Delayed-Start Designs

The medical community has been keen on developing disease-modifying (DM) products that can delay the underlying pathological or pathophysiological disease processes when treating chronic progressive diseases such as Alzheimer’s disease, Parkinson’s disease, and multiple sclerosis. The DM effects can be established through the delayed-start design, which first randomizes patients to new treatment and control (first phase), and at some timepoint, crosses over the control to new treatment (second phase). If patients initially on control then switched to treatment fail to “catch up” (after a reasonable long time) with patients who received new treatment for the entire trial duration, a DM effect is confirmed. As the delayed-start design has a long duration and includes two phases, many patients can have the intercurrent events during the trial that result in missing observations for the first and/or second phase (e.g., missing data more than 30% of the total enrollment). Therefore, how to handle the intercurrent events becomes an important and inevitable topic when testing the DM effects in the delayed-start design. In this article, we focus on the hypothetical strategy for dealing the intercurrent events and discuss how to implement a few methods for handling missing observations when examining a key hypothesis for establishing the DM effect. Statistical properties of these methods are explored through simulations that are set up based on the real trial experience.