Basic Statistics

CHAPTER 14

INTRODUCTION TO SURVIVAL

AN ALY S I S

In biomedical applications, survival analysis is used to study the length of time until

an event occurs. For example. survival analysis has been used in studying the length

of time that cancer patients survive. Here, the event is death. For some diseases,

such as multiple sclerosis, the length of time that the disease remains in remission has

been analyzed. Survival analysis has also been used in studying the length of time

that women whose partners use condoms have remained nonpregnant.

What is called survival analysis in biomedical applications is calledfailure time or

reliability analysis in engineering applications. In behavioral science, survival analy-

sis has been used to analyze the length of time a person is on welfare or the time until a

second arrest; it is called event history analysis. Here, we use the terminology survival

analysis whether or not the outcome being studied is death.

In Section 14.1 we discuss how the time to an event is measured and describe why

survival data requires different statistical methods from those given in previous chap-

ters. In Section 14.2 we present graphical methods of depicting survival data. The

death density, cumulative death distribution function, survival function, and hazard

function are described and graphed. In Section 14.3, methods of estimating these

functions are given using clinical life tables and the Kaplan-Meier method. In Sec-

tion 14.4 we compare use of the Kaplan-Meier method with clinical life tables and

Basic Statistics: A Primer for the Biomedical Sciences, Fourth Edition.

By Olive Jean Dunn and Virginia A. Clark

Copyright @ 2009 John Wiley & Sons, Inc.

201