Competing Risks and Multi-state Models

Overview

What you will study

Competing risks models play an increasingly important role for predicting absolute risks of disease and prognosis using time to event data. An overarching goal of the Competing Risks and Multi-state Models course offered by University of Bern is to provide a solid introduction to important concepts when performing a competing risks analysis (e.g., which quantities can be estimated and what they do represent) as well as practical aspects of estimation. 

Multi-state models provide an extension of competing risks models that also enable modelling of complex disease profiles. By modelling transitions between disease states, accounting for competing events at each transition, we can gain an improved understanding of a patient’s prognosis and how risk factors impact over the whole disease pathway. 

Throughout the course we will place emphasis on the use of flexible parametric survival models that incorporate restricted cubic splines on the log hazard or log cumulative hazard scale. This will include models with time-dependent effects (non-proportional hazards). We will focus on obtaining clinically useful and directly interpretable predictions, which are particularly useful for more complex models, but also describe the challenges and various approaches to calculating them. We will also discuss assumptions of the models, including the Markov assumption and how this can be relaxed. The course will be taught using Stata making use of the multistate and merlin packages.