There is substantial variation in the management of atrial fibrillation (AF) in the emergency department (ED), particularly whether these patients are admitted to hospital. We sought to identify factors that predict admission and to examine the relationship between AF admission and outcomes.

We performed a retrospective cohort analysis of patients ≥20 years of age who had an index ED visit with a primary diagnosis of AF from between April 1, 2005, and March 31, 2010, in Ontario, Canada. We excluded patients who died during the index ED visit or hospitalization. A hierarchical logistic regression model was used to determine predictors of hospital admission during the index ED visit. A propensity-matched analysis was used to test for associations between hospital admission and 1-year outcomes.

The cohort consisted of 33,699 patients, of whom 16,270 (48.3%) were admitted to hospital. Substantial variation was seen across the 154 hospitals, with admission rates ranging from 3.0% to 91.0%. Admitted patients had higher rates of comorbidities compared to discharged patients. Mortality rates at 1 year were significantly higher in matched admitted versus discharged patients (hazard ratio 1.45, 95% CI 1.33-1.57, P < .001), as were all-cause hospitalizations (hazard ratio 1.18, 95% CI 1.13-1.22, P < .001).

Wide practice variation was observed between hospitals in terms of the proportion of patients admitted. Our data suggest that selected patients when discharged have similar or improved outcomes compared to those who are initially admitted. Future research is needed to better standardize admission/discharge decisions for AF patients in the ED.