We consider the data of two studies of patients with node-positive breast cancer. The first, a clinical trial [GBSG-2 study, n = 686 patients, 299 events for event-free survival (EFS)] is used to identify prognostic markers and to investigate the influence of different model-building strategies. This dataset was used previously for prognostic modelling [for example, by Sauerbrei and Royston (1999) and Schumacher et al. (2006)] and is available at http://www.blackwellpublishers.com/rss/. The data of the second, smaller Freiburg-DNA study (n = 139 patients, 76 events for EFS) are exclusively used to validate the results obtained in the GBSG-2 study. Prognostic markers evaluated in both studies were patient's age, menopausal status, tumour size, estrogen, and progesterone receptor, tumour grade, and number of involved lymph nodes. More details with references are given in Schumacher et al. (2006).
Was this article helpful?