Full dates (day, month, year) should be used in population-based cancer survival studies
Menée au Royaume-Uni, cette étude plaide pour la prise en compte des données complètes de datation (jour, mois, année) pour estimer correctement la survie des patients atteints de cancer dans les études en population
Accurate survival estimates are essential for monitoring cancer survival trends, for health care planning and for resource allocation. In order to obtain precise estimates of survival, full dates (day, month and year) rather than partial dates (month and year) are required. In some jurisdictions, however, cancer registries are constrained from providing full dates on the grounds of confidentiality. The bias resulting from the use of partial dates in the estimation and comparison of survival makes it impossible to determine precisely the differences in the risk of death from cancer between population groups or in successive calendar periods. Important operational arguments also exist against the use of incomplete dates for survival analysis, including increased workload for cancer registry staff and the introduction of avoidable complexity for quality control of survival data. Cancer survival is one of the most widely known outputs produced by population-based cancer registries, and it is a crucial metric for the comparative effectiveness of health services. The bodies that set data access guidelines must take a more balanced view of the risks and benefits of using full dates for the estimation of cancer survival. © 2012 Wiley-Liss, Inc.
International Journal of Cancer , résumé, 2011