Skip to Main content Skip to Navigation

Modèles de regression en présence de compétition

Abstract : Survival analysis focuses nn modelling time to failure from a single cause of failure. In many situations, however, individuals may fail from several distinct causes, which defines a competing risks setting. This work aimed at studying regression models in such a framework. Two approaches were considered, one based on the cause-specific hazard and the other one relying on the subdistribution hazard, i.e. the hazard function associated with the cumulative incidence function.
The proportional hazards formulation was assumed in both cases, corresponding to the Cox model and the Fine and Gray model.

We first developed a sample size formula for the Fine and Gray model. Then, we studied the properties of the estimator of the regression parameter in the Fine and Gray model, when the true underlying model was a proportional hazards model for the cause-specific hazard. Next, the inclusion of time dependent covariates in the Fine and Gray model was investigated.
We conclude with a guideline detailing the implications of either choice when analyzing competing risks data.
Document type :
Complete list of metadatas
Contributor : Aurélien Latouche <>
Submitted on : Wednesday, February 7, 2007 - 10:36:30 AM
Last modification on : Monday, December 14, 2020 - 9:52:58 AM
Long-term archiving on: : Tuesday, April 6, 2010 - 9:51:03 PM


  • HAL Id : tel-00129238, version 1


Aurélien Latouche. Modèles de regression en présence de compétition. Sciences du Vivant [q-bio]. Université Pierre et Marie Curie - Paris VI, 2004. Français. ⟨tel-00129238⟩



Record views


Files downloads