Hospital-level variation and predictors of admission after ED visits for atrial fibrillation: 2006 to 2011

Background: Outpatient management of atrial fibrillation can be a safe alternative to inpatient admission after emergency department (ED) visits. We aim to describe trends and predictors of hospital admission for atrial fibrillation and determine the variation in admission among US hospitals. Methods: We analyzed ED visits and hospital admissions for adult patients with a principal diagnosis of atrial fibrillation or atrial flutter in the Nationwide Emergency Department Sample 2006 to 2011. We identified patient and hospital characteristics associated with admission using hierarchical multivariate logistic regression. We analyzed admission rates overall and for patients at low risk of thromboembolic complications (CHA(2)DS(2)-VASc score 0). We compared hospital-level variance with residual variance to estimate the intraclass correlation in models with and without hospital characteristics. Results: From 2006 to 2011, annual ED visits for atrial fibrillation and atrial flutter increased by 30.9% and admission rates decreased from 69.7% to 67.4% (P=. 02). Admission was associated with setting (metropolitan teaching vs nonmetropolitan, odds ratio = 1.93 [1.62-2.29]) and region (Northeast vs West, odds ratio = 2.09 [1.67-2.60]). Among patients with 0 CHA(2)DS(2)-VASc score, the national average admission rate was 46.4%. The intraclass correlation was 20.7% adjusting for patient characteristics and hospital clustering, and 19.2% after additionally adjusting for hospital variables. Conclusions: From 2006 to 2011, ED visits for atrial fibrillation in the United States increased by almost a third, with a minimal change in ED admission rates. One-fifth of variation in admission rates is due to hospital site and not explained by hospital characteristics. Hospital-specific practice patterns may identify opportunities to increase outpatient management. (C) 2016 Elsevier Inc. All rights reserved.