| Abstract: | A two-stage procedure using occurrences of a surrogate endpoint regarding a true clinical endpoint has been proposed (FLANDRE, O'QUIGLEY and BROOKMEYER, 1994). The procedure allows estimation of the survival function S(t) reliably and comparison of treatment groups. The aim of this paper is to investigate the efficiency of such a two-stage procedure compared to a one stage procedure, i.e. a classical study. The efficiency is estimated by the ratio of the standard derivation of S(t) for the classical procedure to the standard derivation of S(t) for the two-stage procedure and represents the precision on the estimation of S(t). Standard deviation of S(t) is approximate by the delta-method. Using the exponential model. some numerical results are presented. |
