COMPSTAT 2018: Start Registration
View Submission - COMPSTAT2018
A0254
Title: Age, time, and gender-specific share component models to predict cancer incidence when mortality is known Authors:  Jaione Etxeberria - Public University of Navarre (Spain) [presenting]
Tomas Goicoa - Universidad Publica de Navarra (Spain)
Maria Dolores Ugarte - Universidad Publica de Navarra (Spain)
Abstract: Updated incidence and mortality measures play an important role in a comprehensive overview of cancer burden. In Spain, cancer mortality figures are routinely recorded by Statistical Offices while cancer incidence is systematically recorded by regional cancer registries. Generally, incidence numbers become available three or four years later than mortality figures. In this context, to predict incidence rates in periods when the mortality is already known becomes necessary in order to provide the most updated cancer overview. According to International Cancer Agencies, realistic predictions of incidence rates should fulfil a list of requirements: 1-They should be stable over time, 2-They must be comparable in different populations or regions, 3-Age-specific incidence curves should be provided (including childhood cancer rates) and 4-Mortality-to-Incidence ratios should be taken into account. Considering all these, we propose to use age, time, and gender-specific shared component models for predicting incidence rates in lethal cancers where there exist a high correlation between incidence and mortality. Different models will be considered and their performance will be analyzed using brain cancer incidence and mortality data by gender and age-groups in 27 health units from Navarre and Basque Country (two Spanish regions) during the period 1998-2008. A fully Bayesian approach based on integrated nested Laplace approximations will be considered for model fitting and inference.