Stroke risk scores to predict hospitalization for acute decompensated heart failure in atrial fibrillation patients.

Atrial fibrillation (AF) is the most frequent hospitalized arrhythmia. It associates increased risk of death, stroke and heart failure (HF). Stroke risk scores, especially CHA2DS2-VASc, have been applied also for populations with different diseases. There is, however, limited data focusing on the ability of these scores to predict HF decompensation.
We conducted a retrospective observational study on a cohort of 204 patients admitted for cardiovascular pathology to the Cardiology Ward of our tertiary University Hospital. We aimed to determine whether the stroke risk scores could predict hospitalisations for acute decompensated HF in AF patients.
C-statistics for CHADS2 and R2CHADS2 showed a modest predictive ability for hospitalisation with decompensated HF (CHADS2: AUC 0.631 p=0.003; 95%CI 0.560-0.697. R2CHADS2: AUC 0.619; 95%CI 0.548-0.686; p=0.004), a marginal correlation for CHA2DS2-VASc (AUC 0.572 95%CI 0.501-0.641 with a p value of only 0.09, while the other scores failed to show a correlation. A CHADS2≥2 showed a RR=2.96, p<0.0001 for decompensated HF compared to a score <2. For R2CHADS2 ≥2, RR= 2.41, p=0.001 compared to a score <2. For CHA2DS2-VASc≥2 RR=2.18 p=0.1, compared to CHA2DS2-VASc <2. The correlation coefficients showed a weak correlation for CHADS2 (r=0.216; p=0.001) and even weaker for R2CHADS2 (r=0.197; p=0.0047 and CHA2DS2-VASc (r=0.14; p=0.035).
Among AF patients, CHADS2, CHA2DS2-VASc and R2CHADS2 were associated with the risk of hospitalisation for decompensated HF while ABC and ATRIA failed to show an association. However, predictive accuracy was modest and the clinical utility for this outcome remains to be determined.