TABLE OF CONTENTS Interim analysis has been discussed in Chapter 3. The reasons given there for its use apply equally well to trials in which the main response variable is time-to-death, time-to-relapse, etc. Tsiatis (1981) demonstrates a number of important results that show the distributional structure of the log-rank test, computed over time, is similar to that for instantaneous outcome measures, allowing the theory developed for such measures to be used in the more complex survival analysis setting. For details, readers are referred to the two papers mentioned and the East manual (see Appendix). Here, we simply give an example.
DeMets and Lan (1994) describe the use of group sequential methods in a randomised, double-blind, placebo-controlled trial designed to test the effect of propranolol, a beta blocker drug, on total mortality. In a multicentre recruitment, 3837 patients were randomised between propranolol or placebo. The study used the log-rank test for comparison of the survival patterns of the two groups and adopted the group sequential boundaries suggested by O'Brien and Fleming. Seven interim analyses were planned and the relevant O'Brien-Fleming boundaries are shown in Fig. 8.10. The results of the log-rank test are also shown as the trial progressed. On the 5th interim analysis, the log-rank test approached but did not exceed, the critical value. On the 6th interim analysis, the logrank statistic was 2.82 and exceeded the critical value of 2.23.
