OBJECTIVE

We aimed to identify predictors of left atrial appendage (LAA) thrombus in patients with atrial fibrillation (AF) and to enhance the prognostic value of the CHADS-VASc score.

METHODS

Derivation cohort included 1033 consecutive AF patients referred for catheter ablation or direct current cardioversion, in whom transoesophageal echocardiography (TOE) was performed prior to the procedure. Logistic regression analysis was used to identify predictors of LAA thrombus on TOE. Receiver operating characteristic (ROC) curves were constructed to compare the newly developed score with the CHADS and CHADS-VASc scores in the derivation and the validation (n=320) cohort.

RESULTS

On TOE, LAA thrombus was present in 59 (5.7%) patients in the derivation cohort. Aside from variables encompassed by the CHADS-VASc score, LAA thrombus predictors included AF type (persistent/'permanent' vs paroxysmal) and renal dysfunction. These predictors were incorporated into the CHADS-VASc score. In ROC analysis, area under the curve (AUC) for the new score (CHADS-VASc-RAF score) was significantly higher (0.81) than those for the CHADS and CHADS-VASc scores (0.71 and 0.70, respectively). In the validation cohort, the CHADS-VASc-RAF score also performed significantly better (AUC of 0.88) than the CHADS and CHADS-VASc scores (AUC of 0.63 and 0.60, respectively).

CONCLUSION

In real-world AF patients with majority on oral anticoagulation, LAA thrombus was found in approximately 6%. Two variables not included in the CHADS-VASc score (AF type and renal dysfunction) proved strong, independent predictors of LAA thrombus and might improve thromboembolic risk stratification.