Validation of diabetes case definitions using administrative claims data

Abstract

Our aim was to validate three diabetes case definitions for children and adolescents aged <20 years in Canada using administrative and clinical data in the province of British Columbia. We identified potential cases of diabetes from physician claims, hospitalizations and prescription drug records between 1992/1993 and 2007/2008 using the three different case definitions, which included a national standard as well as two regionally developed case definitions. Each case definition used a different combination of administrative data; however, only one definition used prescription drug records. The sensitivity of each definition was calculated against the 'gold standard' of diagnosed cases recorded in British Columbia's Children's Hospital Endocrinology and Diabetes Unit clinical database. During this time period, 2611 patients were seen at the British Columbia's Children's Hospital. The sensitivities (95% CIs) of the national and two regional case definitions were 0.95 (0.941-0.958), 0.97 (0.964-0.977) and 0.82 (0.800-0.830), respectively. Our results highlight the benefit of regional case definitions that exploit the availability of different data sources, but also support that a nationally derived definition is sensitive among children and adolescents.