Survival Graphics Software: Current Status, Future Needs and Criteria of Assessment

Abstract

Survival analysis might never have become such a famous statistical method in medicine had there not been the Kaplan-Meier curve (KMC) visualizing the survival time of patients also in the presence of incomplete (censored) data. The Kaplan-Meier estimator satisfies good statistical properties, simplicity, intuitive interpretability and capability to stimulate imagination. Hence, each renowned survival software package allows plotting the KMC. Beyond KMC, survival software is rather heterogeneous in analytical and graphical methods. A comprehensive comparison of survival software although important and welcome to many users seems out of reach, and in face of the many commercial and public domain software systems it might not be manageable at all. Therefore we restricted ourselves to well disseminated and important software packages. Among those, we chose SAS as a widely distributed package with the pretention of general applicability also to less statistically trained persons, and we chose S-Plus as a more open and more rapidly developing system designed to be used predominantly by statisticians. The questions were: Which graphics are available for the analysis of survival data? Which graphics are needed and should be realized? What support is given for obtaining easily good graphical outputs?