CTDL-positive stable frailty model

The non-PH Canonical Time Dependent Logistic (CTDL) survival re- gression model is extended by incorporating a positive stable frailty component into the hazard function within the Bayesian framework. The resulting model is compared numerically with the Weibull-positive stable frailty model, using data from a placebo controlled randomized trial of gamma interferon in chronic gran- ulotomous disease (CGD). Moreover, supremum bounds of the ratio-of-uniforms (ROU) algorithm, used for sampling from complete conditional distributions, are obtained analytically thus yielding a more efficient form of the algorithm.