Por Luís Antunes (Instituto Português de Oncologia do Porto).
Abstract: In the context of population-based cancer survival analysis, cause of death is seldom available or is unreliable precluding the use of cause-specific survival. The disease-relate survival must be obtained indirectly, assuming that the observed hazard can be decomposed in two additive parcels: the excess hazard (disease related) and population hazard (other causes related). The survival directly calculated from the excess hazard is the net survival. The modelling of the excess hazard allows not only the evaluation of the covariates effects, the description of the excess hazard variation with time but also the prediction of net survival. In this seminar the use of flexible parametric excess hazard models to estimate age-standardised net survival will be presented.