Introduction to Survival Analysis

Presented by: John Klein, PhD, Director & Professor, Division of Biostatistics, Medical College of Wisconsin. We examine statistical methods that can be used to provide unadjusted analysis for lifetime data. We present methods for two distinct scenarios; survival data where there is a single event of interest such as death and competing risks where there are multiple events competing to be the cause of failure. For survival data we present the Kaplan-Meier estimator of the survival function defined as the probability the event has yet to occur. We examine how this statistic can be used to find confidence intervals and bands for the survival function and we examine how to estimate the mean survival time with censored data. For competing risks we present the cumulative incidence function as the proper summary statistic. Finally we present the weighted log rank test and show how it can be used to compare survival outcomes between groups. The focus of the presentation is on understanding the results not on theoretical or computational issues.