Stroke and Thromboembolism in Warfarin-Treated Patients with Atrial Fibrillation: Comparing the CHA₂DS₂-VASc and GARFIELD-AF Risk Scores

BACKGROUND:  Evaluation of thromboembolic risk is essential in anticoagulated atrial fibrillation (AF) patients. The CHA2DS2-VASc score is largely validated and recommended by most guidelines. The GARFIELD-AF Stroke score has been proposed as an alternative risk score.

METHODS:  We analyzed warfarin-treated patients from SPORTIF III and V studies. Any thromboembolic event (TE) was an adjudicated study outcome. We compared the two scores' capacity in predicting any TE occurrence.

RESULTS:  A total of 3,665 patients (median [interquartile range] age: 72 [66-77] years; 30.5% female) were included in this analysis. After a mean (standard deviation) follow-up of 566.3 (142.5) days, 148 (4.03%) TEs were recorded. Both continuous CHA2DS2-VASc and GARFIELD-AF were associated with TE (hazard ratio [HR]: 1.37, 95% confidence interval [CI]: 1.22-1.53 and HR: 2.43, 95% CI: 1.72-3.42), with modest predictive ability (c-indexes: 0.63, 95% CI: 0.59-0.68 and 0.61, 95% CI: 0.56-0.66, respectively), with no differences. CHA2DS2-VASc quartiles showed an increasing cumulative risk, while in GARFIELD-AF only the highest quartile (Q4) demonstrated an increased TE risk. On multivariate Cox regression analysis, CHA2DS2-VASc quartiles were associated with increasing risk of TE, whereas for GARFIELD-AF only Q4 showed an association with TE. Discrimination analysis showed that GARFIELD-AF quartiles were associated with a 48.7% reduction in discriminatory ability. Using decision curve analysis, CHA2DS2-VASc was associated with improved clinical usefulness and net clinical benefit, compared with GARFIELD-AF.

CONCLUSION:  In a warfarin-treated trial cohort of AF patients, both CHA2DS2-VASc and GARFIELD-AF Stroke scores were associated with adjudicated TE events, with modest predictive capacity. The simpler CHA2DS2-VASc score improved discriminatory capacity compared with the more complex GARFIELD-AF score, demonstrating improved clinical usefulness and net clinical benefit.