Stroke Risk Scores and Treatment Decisions in Atrial Fibrillation

Study Questions:

What are the predictive abilities of the ATRIA, CHADS2, and CHA2DS2-VASc risk scores in patients with atrial fibrillation (AF), and their implication for anticoagulant treatment decisions?


AF patients not using warfarin from the United Kingdom Clinical Practice Research Database were identified between 1998 and 2012, with a mean follow-up of 2.1 years. Patients were followed from the time of AF diagnosis until the occurrence of ischemic stroke, prescription of warfarin, death, or the study’s end date. Predictors of stroke were identified, along with calculation of the C statistic and net reclassification improvement (NRI).


A total of 60,594 AF patients (mean age 74.4 years) were identified, with an annualized stroke rate of 2.99%. Event rates were lower in moderate- and high-risk categories based on the CHA2DS2-VASc score than for the ATRIA and CHADS2 scores. Age and prior stroke were the strongest predictors of ischemic stroke. C statistics were 0.70 (95% confidence interval [CI], 0.69-0.71) for the ATRIA risk score, and 0.68 (95% CI, 0.67-0.69) for both the CHADS2 and CHA2DS2-VASc scores. NRI was 0.23 (95% CI, 0.22-0.25) for the ATRIA score compared to the CHADS2 and CHA2DS2-VASc scores, and 0.14 (95% CI, 0.12-0.15) for the ATRIA score compared to the CHADS2 score.


The authors concluded that the ATRIA score performed better in the United Kingdom Clinical Practice Research Database AF population. The authors also concluded that reclassification of stroke risk could prevent overuse of anticoagulants in low stroke risk patients with AF.


The authors compare the newer ATRIA stroke risk score (comprised of age, gender, diabetes, heart failure, hypertension, proteinuria, and renal failure) to the more established CHADS2 and CHA2DS2-VASc scores for predicting stroke risk in AF patients. While the ATRIA score was very mildly more discriminatory for stroke risk (C statistic 0.02 points higher with overlapping 95% CI), the NRI did show improvement over the CHADS2 and CHA2DS2-VASc scores. Perhaps most strikingly was the higher percentage of AF patients categorized as ‘low risk’ when using the ATRIA score. While the authors argue that the ATRIA score may lead to less ‘overutilization’ of anticoagulation, this claim has not been tested in a prospective fashion. Similarly, the ATRIA score is not as intuitive to calculate, which may limit its adoption. While overutilization of anticoagulation in AF may lead to bleeding complications, the problem of underutilization of anticoagulation for AF is currently the larger issue. Clinicians should assess stroke risk factors for all AF patients using any available risk score (ATRIA, CHADS2, or CHA2DS2-VASc) and offer anticoagulant therapy to all moderate- and high-risk patients.

< Back to Listings