Why the study?
Do existing stroke risk scores accurately identify truly low-risk patients with atrial fibrillation in an Australian population?
Population
224,451 oral anticoagulant-naïve patients with a first hospital admission for atrial fibrillation between…
Design
Cohort
Follow-up
≥12 months
Key result
The ATRIA score demonstrated the best predictive accuracy for ischemic stroke (c-statistic 0.66; 95% CI 0.66-0.67), while the guideline-endorsed CHA2DS2-VA score performed the weakest (c-statistic 0.61).
Authors
Loading...
ATRIA may better identify low-risk AF patients than CHA2DS2-VA; leaves open prospective validation before changing practice.
Cohort (n=224,451)
Yes
Do existing stroke risk scores accurately identify truly low-risk patients with atrial fibrillation in an Australian population?
Effect estimate: c-statistic 0.66 (ATRIA) and 0.61 (CHA2DS2-VA) (95% CI 0.66-0.67 (ATRIA) and 0.60-0.61 (CHA2DS2-VA))
The guideline-endorsed CHA2DS2-VA score showed the lowest discrimination for identifying low-risk AF patients, highlighting the need for locally validated risk tools.
Zubrzycki et al. (2026) conducted a cohort in Atrial fibrillation (n=224,451). Stroke risk scores (CHA2DS2-VA, CHADS2, ATRIA, Modified-CHADS2) was evaluated on Predictive accuracy for ischaemic stroke within 12 months (concordance-statistic) (c-statistic 0.66 (ATRIA) and 0.61 (CHA2DS2-VA), 95% CI 0.66-0.67 (ATRIA) and 0.60-0.61 (CHA2DS2-VA)). The ATRIA score demonstrated the best predictive accuracy for ischemic stroke (c-statistic 0.66; 95% CI 0.66-0.67), while the guideline-endorsed CHA2DS2-VA score performed the weakest (c-statistic 0.61).