CMStatistics 2017: Start Registration
View Submission - CMStatistics
Title: Modelling the early menarche and late menopause in breast cancer screening through CGAMLSS models Authors:  Elisa Duarte - Universidade de Santiago de Compostela (Spain) [presenting]
Carmen Cadarso Suarez - Universidad de Santiago de Compostela (Spain)
Bruno de Sousa - Universidade de Coimbra (Portugal)
Giampiero Marra - University College London (United Kingdom)
Rosalba Radice - Cass Business School (United Kingdom)
Vitor Rodrigues - Universidade de Coimbra (Portugal)
Abstract: It is a fact that in the breast cancer etiology the risk of the disease increases with reproductive factors as early menarche and late menopause. It is believed that longer is reproductive lifespan (difference between menopause and menarche ages), higher is the risk of breast cancer. This is such an important issue that there are several studies addressing the trend of menarche and menopause ages along the years. The data provided by the Portuguese Cancer League, Central Branch sponsored by the Breast Cancer Screening Program in 78 municipalities refers to the breast cancer screening program over a period of twenty years. The dataset has information about 212,517 postmenopausal women, born between 1920 and 1965 who have attended the breast cancer screening program in central region of Portugal. For this analysis we employ Bivariate Copula Additive Models for Location, Scale and Shape considering the menopause and menarche ages as binary outcomes. The CGAMLSS models extend the scope of univariate GAMLSS using a copula approach. They allow modeling the marginal distributions of different families and types, and the dependence structure using predictor equations that can be flexibly specified using smoothers with single or multiples penalties, hence allowing for several types of covariate effects. All the models parameters are estimated simultaneously. The inference is carried out using the R package GJRM.