[Reliability of administrative databases in epidemiological research: the example of end-stage renal disease requiring renal replacement therapy in patients with diabetes]

Abstract

Health and administrative databases are widely used in epidemiology, mostly in studies of chronic diseases, but biases may undermine the external validity of the results. In nephrology, the use of these data sources is in its early days and needs to be validated. The aim of this study was to compare the data on the incidence of ESRD and death in a cohort of patients with type 2 diabetes (DM2) obtained from administrative databases with the results of a traditional, well-performed cohort study. The study was conducted in the Health District of Venice (Italy) on a cohort of 18,416 DM2 patients on hypoglycemic drug therapy enrolled from 1 January 1998 to 31 December 2002 from administrative databases.Comorbid conditions were recorded from hospital discharge records, the database of death certificates was used to identify patients who died within 31 December 2004, and the database of the Dialysis and Transplantation Registry of the Veneto Region served to identify patients who started renal replacement therapy within 31 December 2004. Record linkage was performed using the unique personal identification codes (fiscal number) of Italian citizens. The cumulative incidence of ESRD was estimated using Gray's method for competing risks. The mortality rate was 50.95 per 1000 person-years, the ESRD incidence was 0.68 per 1000 person-years, with a relative risk of 2.62 with respect to all other causes of ESRD. The crude cumulative incidence of death was 22% and that of ESRD 0.33% at the end of follow-up. The results were similar to those obtained in traditional cohort studies. The results of our study prove the external validity of the administrative database approach in epidemiological studies in nephrology.