Group sequential tests when the primary endpoint is a delayed response
Lisa Hampson (University of Bath)
22/01/08
Group sequential tests propose that the data accumulated continuously
throughout the course of a clinical trial or epidemiological study are
analysed after each group of subjects has been observed. There are both
ethical and economic advantages associated with sequential monitoring.
Current group sequential designs assume that the primary response from
each patient is available immediately after the treatment has been
administered. In practice however, a delay is inherent in many endpoints
where the response can only be measured after an extended observation
period has elapsed. In this talk we explain how we can change the
structure of a GST to formulate an efficient and systematic approach to
handling delayed data in a sequential setting. Using Bayesian
sequential decision theory and dynamic linear programming, we find
tests which are
optimal in the sense that they minimise, under certain scenarios, the
expected number of subjects recruited on termination of a trial and
conclude that for sufficiently small delays there are certainly good
efficiency gains to be made from using GSTs.
Back to the seminar homepage
LSE Home Page |
Departmental Home Page |
Baurdoux Home Page
[Last modified: Jan. 9th 2008 by Erik Baurdoux]