Track: Drug Utilization Research -> Trends and comparisons
(PO-4153) Identifying Cohorts Of Patients With Type2 Diabetes Mellitus Initiating Dapagliflozin In Three Data Sources
Monday, September 14, 2020
J. Bradley Layton, Cindy K Zhou, Heather E Danysh, Daniel C Beachler, Ryan Ziemiecki, Jade Dinh, Ruihua Yin, Brian Calingaert, Phillip R Hunt, Alicia W Gilsenan and Catherine B Johannes
Background: A multi-database study of dapagliflozin, a sodium-glucose cotransporter 2 inhibitor, utilized United Kingdom- and United States (US)-based data sources to identify historical cohorts of patients with type 2 diabetes mellitus (T2DM) initiating dapagliflozin. While the data sources differ in patient populations, data systems, and health care delivery and practice, a common protocol and statistical analysis plan were applied with data source-specific adaptations.
Objectives: To describe differences in patient characteristics and initiation patterns among new users of dapagliflozin with T2DM in three data sources.
Methods: Dapagliflozin initiators were identified in US-based administrative claims databases—Medicare (aged ≥ 65 years, years 2014-2017) and the HealthCore Integrated Research Database (HIRD) (aged 18-64 years, years 2014-2017)—and the Clinical Practice Research Datalink (CPRD), United Kingdom-based general practice records (aged ≥ 18 years, years 2012-2018). Clinical and demographic characteristics of the resulting study cohorts were described.
Results: We identified 46,801 dapagliflozin initiators (CPRD, 12,051; HIRD, 16,671; Medicare, 18,079). Across data sources, the proportion of patients using insulin at the time of dapagliflozin initiation was similar (CPRD, 13%; HIRD, 13%; Medicare, 17%), but the CPRD had higher proportions of dapagliflozin initiators adding dapagliflozin to other diabetes treatments (CPRD, 89%; HIRD, 66%; Medicare, 72%) and lower proportions initiating dapagliflozin as monotherapy (CPRD, 2%; HIRD, 13%; Medicare, 10%). The older Medicare cohort had higher baseline prevalences of most indicators of diabetes severity, comedication use, and comorbidity, including coronary heart disease (CPRD, 12%; HIRD, 9%; Medicare, 31%) and other cardiovascular disease (CPRD, 9%; HIRD, 3%; Medicare, 14%). The CPRD had lower recorded diagnoses of hypertension than either US data source (CPRD, 54%; HIRD, 70%; Medicare, 90%).
Conclusions: Characteristics of patients with T2DM initiating dapagliflozin vary based on the source population and data source, potentially due to differences in age, data coding and record structures, and practice patterns. The Medicare cohort had higher levels of most comorbidities, as expected due to age. The patterns of dapagliflozin initiation in the CPRD differed from the US samples, which could be due in part to differing diabetes treatment approaches in the two countries. Using patient information from a variety of populations and practice settings is helpful in evaluating the safety of dapagliflozin.