Far East Journal of Theoretical Statistics
Volume 16, Issue 2, Pages 269 - 284
(July 2005)
|
|
A METROPOLIS WITHIN GIBBS SAM PLING IN RELATIVE SURVIVAL
Roch Giorgi (France), Ariane Sahel (France), Jean-Pierre Daures (France) and Joanny Gouvernet (France)
|
Abstract: Relative
survival analysis is a method which provides
an estimate of the effect on survival
corrected for the effect of other independent
causes of death, using the natural mortality
in the underlying general population as the
reference. This method is frequently used when
the specific causes of deaths are uncertain or
unavailable as in some population or
hospital-based registries. We proposed a
Markov Chain Monte Carlo (MCMC) approach to
perform relative survival analysis using a
proportional hazards regression model. We used
gamma and normal prior distributions,
respectively, for the baseline mortality
hazard function and the regression parameters
and we established the likelihood function.
Conditional posterior distributions cannot be
reduced analytically to well-known
distributions and we used a Metropolis within
Gibbs sampling to obtain samples from the
conditional posterior distributions. The
accuracy of the estimates obtained by this
MCMC approach were evaluated in simulations
studies. Data from a population-based of
pharyngeal cancer were used to illustrate our
approach. |
Keywords and phrases: relative survival, proportional hazards model, Bayesian methods, Gibbs sampling, Metropolis sampling, survival analysis. |
|
Number of Downloads: 262 | Number of Views: 747 |
|