|
Abstract:
|
In the setting of cardiac rhythm management, events of interest can occur multiple times (e.g., hospitalizations or arrhythmic episodes), and it is of interest to decrease the event density---such as number, rate, hazard, or intensity. In some settings, time-to-event analysis seems to be the most appropriate, but in the setting of trial design, it presents challenges. In this talk, we will discuss a group sequential design for medical device trials for which the primary endpoint is time to all recurrent events. We will describe a procedure for obtaining the sample size while adjusting for imbalance of event density across subjects. Statistical considerations are given to model robustness, time-varying covariate effects, and patient heterogeneity. Examples of design parameters and monitoring plans for efficacy/harm or futility will be provided.
|
