gms | German Medical Science

62. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e. V. (GMDS)

Deutsche Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie

17.09. - 21.09.2017, Oldenburg

Misspecification of cause of death in a progressive illness-death model

Meeting Abstract

Suche in Medline nach

  • Michael Lauseker - Ludwig-Maximilians-Universität München, Institute for Medical Information Sciences, Biometry, and Epidemiology, München, Deutschland
  • Christine zu Eulenburg - University Medical Center Groningen, Dept Epidemiology, Med Stat & Decis Making, Groningen, Niederlande

Deutsche Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie. 62. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e.V. (GMDS). Oldenburg, 17.-21.09.2017. Düsseldorf: German Medical Science GMS Publishing House; 2017. DocAbstr. 137

doi: 10.3205/17gmds037, urn:nbn:de:0183-17gmds0378

Veröffentlicht: 29. August 2017

© 2017 Lauseker et al.
Dieser Artikel ist ein Open-Access-Artikel und steht unter den Lizenzbedingungen der Creative Commons Attribution 4.0 License (Namensnennung). Lizenz-Angaben siehe



Introduction: Cause of death is increasingly becoming a topic in oncology. It is usually distinguished between disease-related and disease-unrelated death. In insufficiently documented data, a frequently used approach is to define death as disease-related when a progression to advanced phases has occurred before, otherwise as disease-unrelated. The data are analysed as competing risks, while the underlying model is in fact a progressive illness-death model. Therefore, the purpose of our work was to analyse if this misspecification leads to any systematic bias in the results.

Methods: Data were simulated according to different scenarios, following a Markov, semi-Markov and a non-Markov progressive illness-death model. This was done by varying the hazard for the transition from progression to death. This hazard was either simulated as independent of the time to progression, dependent on the time since progression or dependent on the time of progression. Censoring was added with the censoring distribution being independent of the events. We compared the cumulative incidences of the events “disease-related death” and “disease-unrelated death” in the competing risks analysis to the state occupation probabilities in the progressive illness-death model with regards to bias and variance.

Results: Although both estimators are not equal when censoring is present, we could not prove any systematic bias. However, variance in the misspecified competing risk analysis was slightly larger than in the progressive illness-death model.

Discussion: Although the use of the competing risk estimator is not recommended in the situation described above, we conclude that the results are still meaningful.

Die Autoren geben an, dass kein Interessenkonflikt besteht.

Die Autoren geben an, dass kein Ethikvotum erforderlich ist.

Beitrag wurde bereits vorgestellt: ISCB 2017