[Prognostic factors in survival after cancer: epidemiological model].

Quality of care assessment is one of the principal issues considered when studying the survival of cancer patients. Survival rates based on hospital series ("clinical survival") essentially aim at evaluating the impact of therapy, taking into account the stage of the tumor. When, on the other hand, survival rates refer to a demographically defined population ("epidemiologic survival") they depend not only on quality of therapy but also on differentials in access to health services which in turn condition partly the stage at time of diagnosis and therapy selection. Since these patterns of access can vary according to socio-cultural characteristics, the epidemiologist is tempted to focus the prognostic analysis of survival on these characteristics. In a cancer registry setting, clinical indicators (stage of the lesions, type of health care institution and therapy) must not be neglected. Rather, these must be analyzed both as co-factors of survival and as phenomena to be explained in their own right. Issues relating to this topic are discussed using the example of breast cancer in Geneva.