Acta Cardiologica · Nov 6, 2026

Revisiting the role of sex in stroke risk stratification in atrial fibrillation: a systematic review and meta-analysis

By Yeh CY, Lee PH, Wu CS et al.

Background The 2024 European Society of Cardiology guidelines introduced the CHA₂DS₂-VA score, omitting female sex from stroke risk stratification in atrial fibrillation (AF). However, evidence on the relative predictive performance of CHA₂DS₂-VA versus CHA₂DS₂-VASc across diverse populations remains inconsistent. We therefore conducted a systematic evaluation of their discrimination and reclassification performance. Methods We performed a systematic review and meta-analysis of observational cohort studies comparing CHA₂DS₂-VA and CHA₂DS₂-VASc for prediction of ischaemic stroke or thromboembolic events in patients with AF. Discrimination was evaluated using the difference in C-statistics (ΔAUC) and integrated discrimination improvement (IDI), and reclassification using continuous net reclassification improvement (NRI). Pooled estimates were derived using random-effects models. Results Seven studies including 250,112 patients were analysed. In the overall population, discriminative performance was comparable between the CHA₂DS₂-VA and CHA₂DS₂-VASc scores (ΔAUC 0.00065, 95% CI −0.00288 to 0.00417; IDI 0.00010, 95% CI −0.00063 to 0.00083). Continuous NRI indicated a modest advantage for CHA₂DS₂-VASc (−0.05615; 95% CI: −0.10350 to −0.00880). Among patients not receiving oral anticoagulation, both discrimination and reclassification were similar between the two scores. Variability in reclassification performance was observed across regions and study periods. Conclusions The CHA₂DS₂-VA and CHA₂DS₂-VASc scores showed comparable discrimination for stroke risk prediction in patients with AF when evaluated as continuous scores, although reclassification modestly favoured the CHA₂DS₂-VASc score. The clinical contribution of female sex to risk prediction appears context-dependent and warrants further evaluation.

doi.org/10.1080/00015385.2026.2757891PubMed 42993272Biomarkers