Abstract:
|
Traditional methods of sample size and power calculations in clinical trials with a time-to-event end point are based on the logrank test (and its variations), Cox proportional hazards (PH) assumption, and comparison of means of two exponential distributions. Of these, calculation based on PH assumption is the most popular as it allows for covariate adjustment. However, there are many situations where the assumption of proportional hazards in designing a trial is not appropriate such as when previously conducted observational studies indicate a violation of this assumption. Simulations can be conducted to show that in such situations, either a very large (and unaffordable) sample size is required to conduct the trial or that the trial based on the PH assumption is not clinically (and practically) meaningful. For such scenarios, a simulation based clinical trial design using the concept of Relative Time utilizing the Generalized Integer Gamma distribution is proposed. Simulation studies identify the situations in which such a design will be beneficial as compared to the standard designs. Further, some real life examples demonstrating the usefulness of this approach are discussed.
|