Course 3
Date
20 – 22 January 2025
Faculty
Dr. Michael Crowther
Red Door Analytics, Stockholm, Sweden
Dr. Caroline Dietrich
Division of Clinical epidemiology, Department of Medicine Solna, Karolinska Intitutet, Stockholm, Sweden and Red Door Analytics, Stockholm Sweden
Venue
CH – 3823 Wengen | Bühlstube
Course description
This course will focus on how to analyse and interpret data in the competing risk, and the more general, multi-state model setting. 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 this course is to provide a solid introduction to important concepts in the presence of competing risks (e.g., which quantities can be estimated and what they represent) as well as practical aspects of estimation. Multi-state models provide an extension to the competing risks situation, enabling modelling of complex disease pathways. 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. Real-world examples will be presented and discussed. The course will be taught using Stata making use of the multistate and merlin packages.
Course objectives
By the end of this course participants will have:
- An understanding of how to fit and interpret flexible parametric survival models, including modeling of time-dependent effects.
- An understanding of competing risks models and how to estimate cumulative incidence functions non-parametrically and using parametric models.
- An understanding of how to construct, analyse and interpret a multi-state model.
- An understanding of the variety of useful measures that can be obtained from multistate models.
- Practical experience of fitting the models using Stata®.
Course audience
Course participants should be familiar with standard survival models, such as the Cox model and/or parametric survival models (e.g., Weibull) and interested in extending their knowledge to the more complex issues of competing risks and multistate models. The course will include theory, but emphasis is placed on applying and interpreting the methods. A pre-course survey will be sent out prior to the winter school and some aspects of the course will be adjusted to reflect participants background and interest.
Course outline
The course runs over three days and consist of lectures and computer practical sessions.
We start in the morning by reviewing the previous day. During the extended afternoon break, participants review course materials, catch up on email, or ski. We reconvene at 4:30 pm for the computer sessions.
Monday, 20 January 8:00 am – 12:00 pm | 4:30 pm – 6:30 pm
- Brief review of time-to-event data including the Cox model
- Flexible parametric survival models
- Modelling competing risks
- Estimating cumulative incidence functions
Tuesday, 21 January 8:00 am – 12:00 pm | 4:30 pm – 6:30 pm
- Introduction to multi-state models
- The illness death model
- The Markov assumption
- Stacked versus separate models
- Estimating contrasts between groups
Wednesday, 22 January 8:00 am – 12:00 pm | 1:00 pm – 3:00 pm
- Expected length of stay in different states
- Resetting the clock and semi-Markov models
- Standardisation in multi-state models
Real-world examples using multi-state models
Credits
1.0 ECTS
Course materials
Bring a portable computer. A course license for Stata® will be available to install before arrival. Participants will receive digital copies of all course material. Onsite University of Bern IT staff provides support upon e-mail () request.
Course fee
PhD Bern: CHF 600
PhD other: CHF 800
Academic: CHF 1000
Industry: CHF 2000
Registration
Accomodation
Book your accommodation separately. Please see recommendations for special prices.