Survival ratio plots with permutation envelopes in survival data problems.

A new method is proposed, based on a simple ratio plot, which is useful for identifying time intervals or regions in which survival prospects differ between two distinct populations or treatments. In order to investigate whether any observed difference or trend is due to sampling variation or is due to a possible real effect, resampling techniques are used to generate permutation envelopes as reference bands. The method is applicable to the comparison of survival in two populations for both independent and paired survival problems and also to the assessment of the symmetry of a bivariate survival function.