Aims and background: The aim of the study was to describe the extent of variability among Italian cancer registries in data managing practices that may affect differences in incidence and possibly in survival estimates. Methods: a self-administered questionnaire was sent to each participating registry. The definitions of the disease, of the start point and of the end point of survival computation were investigated. Moreover, information on the proportion of histologic confirmation, of ill-defined sites and of DCO (death certificate only) was also considered. Results: There were some differences in cancer registration techniques among Italian cancer registries. As regards disease definition, the most relevant problems arose for urinary bladder. Skin melanoma should also be considered with some caution, due to variability among registries in coding in situ cases. For the CNS and meninges, the proportion of cases that could be differently considered was so limited that no effect on survival is expected. For female breast, colorectum and cervix uteri, the effect of early diagnosis services (which are active only in some areas) may lead to better survival estimates. The variability in incidence date definition was high among registries and sites, but its effect on survival was very limited. There was a wide variability in the proportion of DCOs end of DCls (initially known from death certificate), which should be considered in survival comparisons. All the registries stated that they carried out an active follow-up of their patients. Conclusions: In general, quality standards of the registries are good and allow comparability of survival data. The variability of rules adopted by Italian registries may affect geographic survival differences only in a limited number of cancer sites, so that results should be interpreted with caution.
