Course summary - Advanced survival analysis

Survival analysis is the study of time-to-event outcomes which is widely applied in health research.  Special methods are required to analyse such data due to issues of censoring and truncation. This two-day workshop focuses on advanced methods for the analyses of these data. It will commence with a short review of the basic concepts and methods, to provide the background required to tackle the three main topics discussed. First, we will discuss regression models for survival data beyond the Cox model, including additive hazard models and the accelerated failure time model. Second, we will discuss incorporation of time-dependent covariates in survival analyses, that is, of variables that change over time (e.g. blood pressure, body mass index, drug intake). Finally, we focus on the topic of competing risks, which are events that preclude observation of the event of interest (e.g. death from injury prevents observation of death from cancer recurrence). Specifically, we show how to analyse survival data with competing risks using the recommended multi-state model framework, discuss the pitfalls of some approaches, and introduce other types of useful multi-state models for medical research. The lectures will focus on interpretation, with example studies used to illustrate the different concepts. Computer practicals will provide participants with the tools to implement the approaches. Syntax will be provided in both Stata and R. Basic knowledge of at least one of these software packages and familiarity with introductory level survival analysis are required. Participants should bring their own laptop with R or Stata installed.