Methods for Censored Survival Time Data

Abstract

This chapter deals with survival time data for which the inference cannot be made by any of the parametric and nonparametric methods that we learned in the past chapters.

Typical survival time data type is time to event (e.g., time to death, time to treatment failure, time to disease relapse, time to recovery after surgery). The word “survival time” was originated from the time to death event. The survival time distribution commonly appears as nonsymmetrical and another unique feature of these data is censoring, which is difficult to deal with using the methods covered in the previous chapters.

Some common statistical questions are: What is the probability that a subject in a group would survive longer than t years?; What is the median survival time of the group (i.e., what is the time point when half of the subjects would remain alive)?; and Are the survival time distributions significantly different between the two groups (i.e., did Group A survive longer than Group B on average, etc.)?

This chapter lets you walk through two examples that illustrate the typical survival time data with censoring and how to tackle the data analysis problem.