C. Armero Cervera, F. Llopis-Cardona, G. Sanfélix-Gimeno
Illness-death models are particular multi-state models with three states: a first state associated with an initial condition of health, an intermediate state associated with illness, and an absorbing state associated with death and accessible from the initial state and from the disease state. Time between states and transition probabilities are relevant outcomes modelled in term of baseline, temporal or spatial covariates.
We focus on a spatial illness-model to deal with a cohort study regarding progression after hip fracture which includes patients discharged alive after a hospitalization due to a hip fracture
during 2008-2015 in the Comunitat Valenciana. We used a Bayesian approach through the integrated nested Laplace approximation (INLA) to assess the geographical variation in the risks and incidences of recurrent hip fracture and death as well as in both the risk and the probability of death after refracture.
Keywords: Bayesian statistics; multi-state models; survival analysis
Scheduled
GT21 Bayesian Inference I
June 7, 2022 12:00 PM
Audiovisual room