External validation of atrial fibrillation stroke risk scores for predicting ischemic risk in patients after left atrial appendage occlusion

AIM To externally validate the atrial fibrillation (AF) stroke risk scores for predicting ischemic risk in patients following left appendage occlusion (LAAO).

METHODS From April 1, 2019, to October 31, 2020, a total of 3,082 AF patients who underwent successful LAAO at 39 hospitals were prospectively included. The discriminatory and calibration abilities of seven AF stroke risk scores — CHADS₂, Modified-CHADS₂, CHA₂DS₂-VASc, ATRIA, mCHA₂DS₂-VASc, GARFIELD-AF, and GARFIELD-AF II — were evaluated for predicting 1-year postoperative ischemic risk using the C-statistic, receiver operating characteristic (ROC) curve, Hosmer-Lemeshow test, and calibration curves.

RESULTS  At 1-year follow-up, 74 patients (2.4%) experienced stroke, of whom 52 (1.7%) had ischemic stroke, 4 (0.1%) had systemic embolism (SE), and the composite endpoint of death/stroke/SE occurred in 133 (4.3%) patients. Among the seven scores, the GARFIELD-AF II score demonstrated the highest discriminative ability in predicting stroke, ischemic stroke, stroke/SE, and death/stroke/SE (C-statistics: 0.685, 0.675, 0.685, and 0.692, respectively). Furthermore, the Hosmer-Lemeshow test and calibration curves indicated that the GARFIELD-AF II score was well-calibrated for predicting ischemic stroke in patients after LAAO (χ² = 3.150).

CONCLUSION The GARFIELD-AF II score exhibits strong discriminatory and calibration properties, making it a reliable tool for predicting stroke risk in AF patients following LAAO.