|
Abstract:
|
Adaptive group sequential design has been very popular in clinical trials over the past decade. Interim analyses (IA) can provide estimates of treatment effect and allow possibility of early stopping and sample size recalculation. In group sequential design, number and timing of the IA are usually determined based on the recruitment milestone. However, for studies with fast enrolment and long-term endpoint, IA at scheduled time point may not provide much value to the trial adaptation, since almost all the enrolment will be completed at the time of interim. In such cases, it is desirable to utilize short-term endpoints, if defined, in addition to available long-term endpoints at the time of the interim, to estimate the treatment effect. For continuous endpoint, mixed model with repeated measures (MMRM) can provide least square means of the long-term endpoint. The decision on future trial can be made based on the estimate of treatment effect obtained from MMRM. We apply this approach in a medical device clinical trial. Prediction power is evaluated with complete long-term endpoint measurements at the end of the study. Pros and cons of long-term vs short-term endpoints are discussed.
|
