End-of-life : insight from administrative data

The purpose of this study is to develop and test the validity of an algorithm allowing the classification of the decedents as were dying of, died of and died with BC, using administrative data, for the study of service delivery to the terminally ill.Validation was carried out through a chart review of a sample of BC decedents extracted from a tumor registry. The algorithm was then applied to the decedents of a cohort of women with BC. The three groups were compared for their distributions of age and place of death.The validation showed an excellent crude agreement (0.96). Of the 3334 deaths, 2293 were classified as were dying of, 142 as died of, and 949 as died with. The comparison of age showed that women who died with were older than women who were dying of. These two groups differed in their distributions of place of death