Methods for mid-course corrections in clinical trials with survival outcomes

Interim monitoring of randomized controlled clinical trials often requires that the assumptions under which the trial was designed be evaluated and that appropriate design modifications be made. Furthermore, if there is little hope of ultimately demonstrating benefit, the study may be terminated on either ethical or economic grounds. Design modifications may be made based on either blinded or unblinded analyses. Since observed patterns of recruitment and patient outcomes may differ from those required by standard methods, specialized tools may be required to perform the necessary computations. In this paper we demonstrate how the Markov chain model of Lakatos can accommodate arbitrary observed patterns of recruitment, hazard rates, and other design parameters for performing mid-course corrections in trials with survival outcomes.