Survival Analysis

Abstract

This is the third edition of this text on survival analysis, originally published in 1996. As in the first and second editions, each chapter contains a presentation of its topic in “lecture-book” format together with objectives, an outline, key formulae, practice exercises, and a test. The “lecture-book” format has a sequence of illustrations and formulae in the left column of each page and a script in the right column. This format allows you to read the script in conjunction with the illustrations and formulae that highlight the main points, formulae, or examples being presented. This third edition has expanded the second edition by adding one new chapter, additional sections and clarifications to several chapters, and a revised computer appendix. The new chapter is Chapter 10, “Design Issues for Randomized Trials,” which considers how to compute sample size when designing a randomized trial involving time-to-event data. We have expanded Chapter 1 to clarify the distinction between random, independent, and noninformative censoring assumptions often made about survival data. We also added a section in Chapter 1 that introduces the Counting Process data layout that is discussed in later chapters (3, 6, and 8). We added sections in Chapter 2 to describe how to obtain confidence intervals for the Kaplan–Meier (KM) curve and the median survival time obtained from a KM curve. We have expanded Chapter 3 on the Cox Proportional Hazards (PH) Model by describing the use of age as the time scale instead of time-on-follow-up as the outcome variable. We also added a section that clarifies how to obtain confidence intervals for PH models that contain product terms that reflect effect modification of exposure variables of interest.