Simple scores to predict 1-year mortality in atrial fibrillation

Traditional scores as CHADS2 and CHA2DS2-Vasc are suitable for predicting stroke and systemic embolism in patients with atrial fibrillation (AF) and have shown to be also associated with mortality. Other more complex scores have been recommended for survival prediction. The purpose of our analysis was to test the performance of different clinical scores in predicting 1-year mortality in AF patients.

CHADS2 and CHA2DS2-Vasc scores were calculated for AF patients of the BLITZ-AF register and compared to R2-CHADS2, R2-CHA2DS2-Vasc and CHA2DS2VASc-RAF scores in predicting 1-year survival. Scores including renal function were calculated both with glomerular filtration rate (GFR) and creatinine clearance.

One-year vital status (1960 alive and 199 dead) was available in 2159 patients. Receiver-operating characteristic curves displayed an association of each score to all-cause mortality, with R2(ClCrea)-CHADS2 being the best [area under the curve (AUC) 0.734]. Differences among the AUCs of the eight scores were not so evident, and a significant difference was found only between R2(ClCrea)-CHADS2 and CHADS2, CHA2DS2VASc, (ClCrea)-CHA2DS2-VASC-RAF.All the scores showed a similar performance for cardiovascular (CV) mortality, with CHA2DS2VASc-RAF being the best (AUC 0.757), with a significant difference with respect to CHADS2, CHA2DS2VASc, and (ClCrea)CHA2DS2Vasc-RAF.

More complex scores, even if with better statistical performance, do not show a clinically relevant higher capability to discriminate alive or dead patients at 12 months. The classical and well known CHA2DS2VASc score, which is routinely used all around the world, has a high sensitivity in predicting all-cause mortality (AUC 0.695; Sensit. 80.4%) and CV mortality (AUC 0.691; Sensit. 80.0%).