Advanced Survival Analysis and Prognostic Modelling

Course aims and objectives

The aim of this course is to give an understanding of and practice in implementing parametric survival models and their use in prognostic modelling, and to introduce the concepts of competing risks and relative survival in modelling time-to-event data.

By the end of the course students should be able to:

• analyse data using Poisson, Cox and parametric (e.g. Weibull) regression models;
• understand the links between these approaches;
• use flexible parametric survival analysis to improve model fit to the data;
• develop prognostic models;
• use relative survival models that incorporate expected survival to overcome lack of specific cause of death data;
• analyse survival data with competing outcomes.

Who the course is intended for

Participants should have a knowledge of survival analyses using Poisson and Cox regression models and their implementation in Stata to at least the level achieved in the 'Introduction to Rates and Survival Analysis' short course. Familiarity with Stata is a pre-requisite for this course.

Course outline 

• Comparison of Poisson, Cox and parametric survival models. 
• Parametric regression using the Weibull and other distributions.
• Flexible parametric survival models and their use in prognostic modelling. 
• Validation of prognostic models.
• Time-varying hazards.
• Analysis of competing risks data using cumulative incidence function and an adapted version of the Cox regression model.  

Please note: Practical sessions of this course will be held in a computer lab, so you will not need to bring a laptop. During the course Stata 14 will be used.