Empirical Insights When Defining the Population Burden of Atrial Fibrillation and Oral Anticoagulation Utilization Using Administrative Data

Abstract

Health administrative data are routinely used to assess disease burden, quality of care, and outcomes for atrial fibrillation (AF). Governments, administrators, and researchers define cohorts differently, based on 3 key factors: the case definition algorithm to identify AF, inclusion/exclusion of transient AF, and the lookback period to identify cases. We assessed the impact of varying these key factors on estimates of the use of guideline-indicated oral anticoagulation (OAC). Hospitalization, ED, and outpatient claim databases were linked in British Columbia. AF was defined by ICD-9 or 10 codes 427.3x or I48.x. We examined a specific (1 hospital or 1 ED or 2 outpatient) vs a sensitive (1 hospital or ED or outpatient) algorithm; inclusion/exclusion of AF associated with open-heart surgery; and lookback periods of 1 to 10 years. We found the more specific AF definition increased OAC utilization by 5% (58.7% vs 53.4%); excluding AF associated with open-heart surgery increased OAC utilization by 0.7% to 2.3%; and each additional lookback year identified more prevalent cases but reduced OAC utilization by approximately 1%. In 40 scenarios, generated by varying all 3 key factors, OAC utilization ranged from 52% to 72%. Assuming a ceiling of 90%, the estimated "treatment gap" therefore varied from 18% to 38%. The 2-fold variation in the OAC treatment gap was based entirely on cohort definition. This has significant implications for health policy and quality indicators.