Fig. 3From: Augmenting the logrank test in the design of clinical trials in which non-proportional hazards of the treatment effect may be anticipatedRMST analysis of the PATCH1 trial. All statistics are displayed as functions of time in years since randomization. a RMST in each treatment group; b RMST difference between treatment groups with 95 % pointwise confidence interval; c Chisquare statistics corresponding to (b); d P-value derived from values in (c)Back to article page