|
Abstract:
|
In some medical device trials, the device under investigation is the first of a kind and "best medical therapy" is used as a control. If "best medical therapy" is not perceived by patients to be very effective, then they may be reluctant to enroll in a randomized clinical trial out of fear of randomization into the control arm. Hence the planned randomized enrollment is too slow to finish within a reasonable time frame. A possible mid-course adaptation is to add some non-randomized sites in the trial, allowing investigators to enroll and assign patients to treatment arm using their medical judgment. This will result to two different patient cohorts in the trial, a randomized cohort and a nonrandomized cohort. In this talk, different approaches to analyzing the data from the two cohorts are proposed and compared, and some simulation results are presented.
|
