[Termination rules in therapeutic trials. Application of sequential methods in 2 clinical trials in hematology]

In haematology, randomized clinical trials often use survival or disease-free interval as response criterion (censored data). For ethical reasons, interim analyses are usually performed, but these interim analyses lead to an increasing Type I error. Sequential methods allow for repeated testing. Three sequential methods are well-adapted to censored data: group sequential analysis, the sequential probability ratio test and the triangular test. The triangular test has the best properties: it allows an early termination of the trial whatever the difference between the treatment groups and without the risk of long duration trials. Sequential analyses can be performed at the same frequency as the usual interim analyses. These three sequential procedures have been applied to the analysis of two cooperative trials in haematology. It appears that the use of sequential methods has to be recommended in the design of randomized clinical trials with a censored response criterion.