Abstract
AIMS: The high costs associated with treatment for atrial fibrillation (AF) are primarily due to hospital care, but there are limited data to understand the reasons for and predictors of hospitalization in patients with AF. METHODS AND RESULTS: The ROCKET AF trial compared rivaroxaban with warfarin for stroke prophylaxis in AF. We described the frequency of and reasons for hospitalization during study follow-up and utilized Cox proportional hazards models to assess for baseline characteristics associated with all-cause hospitalization. Of 14 171 patients, 14% were hospitalized at least once. Of 2614 total hospitalizations, 41% were cardiovascular including 4% for AF; of the remaining, 12% were for bleeding. Compared with patients not hospitalized, hospitalized patients were older (74 vs. 72 years), and more frequently had diabetes (46 vs. 39%), prior MI (23 vs. 16%), and paroxysmal AF (19 vs. 17%), but less frequently had prior transient ischaemic attack/stroke (49 vs. 56%). After multivariable adjustment, lung disease [hazard ratio (HR) 1.46, 95% confidence interval (CI) 1.29-1.66], diabetes [1.22, (1.11-1.34)], prior MI [1.27, (1.13-1.42)], and renal dysfunction [HR 1.07 per 5 unit GFR < 65 mL/min, (1.04-1.10)] were associated with increased hospitalization risk. Treatment assignment was not associated with differential rates of hospitalization. CONCLUSION: Nearly 1 in 7 of the moderate-to-high-risk patients with AF enrolled in this trial was hospitalized within 2 years, and both AF and bleeding were rare causes of hospitalization. Further research is needed to determine whether care pathways directed at comorbid conditions among AF patients could reduce the need for and costs associated with hospitalization.