A PARTIALLY GROUPED LOGRANK TEST

We propose a modified version of the standard logrank test for surviva l data in which the first contribution to the statistic is based on gr ouping of data before a pre-selected grouping time. The grouping is ac complished by artificially constructing the first table based on the p roduct limit estimates of the proportions surviving at the grouping ti me. Remaining contributions to the statistic are identical to that of the standard logrank statistic. The approach has the advantage of bein g uninfluenced by non-proportional hazards differences prior to the gr ouping time while being almost as efficient as the usual logrank test for proportional-hazards alternatives. The statistic is particularly u seful for interim monitoring in situations where early difference betw een treatments are unimportant and/or lagged treatment effects are ant icipated. We report simulation studies to verify and investigate the t est's properties. (C) 1997 by John Wiley & Sons, Ltd.