Syllabus database for doctoral courses

    Startpage
  • Syllabus database for doctoral courses

SYLLABI FOR DOCTORAL COURSES

Print
Swedish title Konkurrerande risker och multi-state models: begrepp, metoder och programvara
English title Competing risks and multi-state models: concepts, methods and software
Course number 2597
Credits 1.5
Responsible KI department Institutionen för medicinsk epidemiologi och biostatistik
Specific entry requirements Statistical modeling and inference using likelihood (course 2261) and Biostatistics III: Survival analysis for epidemiologists (course 1685) or equivalent courses/practical experience (e.g., a masters degree in mathematical statistics combined with application of survival models in medical research).
Grading Passed /Not passed
Established by The Board of Doctoral Education
Established 2012-02-24
Purpose of the course
Intended learning outcomes After successfully completing this course students should be able to:
- describe the quantities that can be estimated in a competing risks framework, how they are interpreted, and the assumptions that must be made.
- understand the relationship between the cause-specific hazard and cumulative incidence and describe why it is no longer one-to-one in a competing risks framework.
- propose a suitable statistical model for assessing a specific research hypothesis in a competing risks situation, fit the model using standard statistical software, evaluate the fit of the model, and interpret the results.
- recognise when a multi-state model is appropriate, fit a multi-state model using standard statistical software, evaluate the fit of the model, and interpret the results.
Contents of the course Competing risks and its extension to multi-state models play an increasingly important role in the analysis of time to event data. For competing risks models, there is a lot of confusion with respect to the proper analysis. The most important reason for the confusion is conceptual: which quantities can be estimated and what do they represent. Especially the consequence of violation of the independence assumption has often been interpreted incorrectly. Once the concepts are understood and the proper type of analysis has been chosen, most analyses can be performed with standard software for survival analysis. For multi-state models with exactly observed transition times, estimation is reasonably straightforward; the real challenge is in (dynamic) prediction. Panel-observed data occur if the state is only known at a finite series of observation times, giving rise to interval censored transition times. A further complication arises if states are misclassified. With the recent availability of software we are seeing a marked increase in applications of multi-state models in the medical literature.

The first part is devoted to competing risks. We explain the main concepts: the independence assumption; cause-specific cumulative incidence; cause-specific hazard and subdistribution hazard; competing risks as a multi-state model. We discuss regression models on both cause-specific and subdistribution hazard. We show how analyses can be performed with standard software. In the second part of the course, the extension to multi state models is discussed. Concepts like transition intensities and transition probabilities are explained. Nonparametric estimation and regression models are considered, as well as methods to obtain predictions of future events, given the event history and possibly covariable values of a patient. With right censored and/or left truncated data, we show that it is possible to perform many types of analyses using standard software, using the same techniques as in the multi-state representation of the competing risks model. Finally, we discuss some issues in the analysis of multi-state models for panel data.
Teaching and learning activities Lectures, exercises focussing on analysis of real data using statistical software, exercises not requiring statistical software, group discussions, literature review.

The course will include practical exercises using the R statistical computing environment. We will provide laptop computers (running Windows in English and configured with R) for all participants, but participants are welcome to bring a laptop with them for use during the course if they prefer. The mstate and the dynpred R-packages will be used during the course and all participants who will bring their own laptops are encouraged to install these packages prior to the course.
Compulsory elements
Examination The course grade is based solely on a written examination.
Literature and other teaching material There is no required literature. Annotated lecture notes will be provided at the start of the course. The following are recommended:

Putter H, Fiocco M, Geskus RB. Tutorial in biostatistics: Competing risks and multi-state models. Statistics in Medicine 2007;26:2389¿2430.

Andersen PK, Geskus RB, de Witte T, and Putter H. Competing risks in epidemiology: possibilities and pitfalls. International Journal of Epidemiology 2012.
Course responsible Paul Dickman
Institutionen för medicinsk epidemiologi och biostatistik
0852486186
0704382333
Paul.Dickman@ki.se

Nobels väg 12A

112 44
Stockholm
Contact person Camilla Ahlqvist
Institutionen för medicinsk epidemiologi och biostatistik
0852483869

Camilla.Ahlqvist@ki.se

Marie Beckeman
Institutionen för medicinsk epidemiologi och biostatistik
+46 8 524 82471

Marie.Beckeman@ki.se