Assess bleeding risk with HAS-BLED and assess stroke risk with CHA₂DS₂-VASc in patients with atrial fibrillation

Toyoda et al.¹ used two stroke risk scores (that is, the CHADS₂ (Congestive heart failure, Hypertension, Age over 75 years, Diabetes mellitus and prior Stroke or TIA) and CHA₂DS₂-VASc (Congestive heart failure, Hypertension, Age over 75 years (doubled), Diabetes mellitus and prior Stroke or TIA (doubled), Vascular disease, Age 65–74 years and sex category)) as bleeding risk indices for Japanese patients with atrial fibrillation (AF). Stroke and bleeding risks are closely related. However, although there is an increased risk of bleeding with higher CHADS₂ and CHA₂DS₂-VASc scores,^{2, 3} this increase is not significant. In contrast, bleeding rates significantly increase with higher values on a specific bleeding risk score, the HAS-BLED (hypertension, abnormal renal/liver function, stroke, bleeding history or predisposition, labile international normalized ratio, elderly, drugs/alcohol concomitantly).²

The HAS-BLED score was developed and validated to predict bleeding events in anticoagulated AF patients; it adds one point for hypertension, abnormal renal/liver function (one point each), stroke, bleeding history or predisposition, labile International Normalized Ratio (INR), age older than 65 years and drugs/alcohol concomitantly (one point each).⁴ Furthermore, it has been validated in several external ‘real world’ cohorts.⁵ Importantly, the HAS-BLED outperforms the CHADS₂ and CHA₂DS₂-VASc when c-indices, net reclassification or discrimination indices are compared.^{2, 3}

Note that a high HAS-BLED score is not a reason to withhold anticoagulation, as such patients derive even greater net clinical benefits from anticoagulation (particularly with the novel oral anticoagulants) when balancing the reduced risk of ischemic stroke against the risk of serious bleeding.^{6, 7}

Note also that the risk of bleeding for patients administered vitamin K antagonists (VKAs) is largely dependent upon the quality of the anticoagulation control, which can be defined by the time in therapeutic range (TTR). A higher TTR (>70%) is associated with lower risks of thromboembolism and bleeding.^{8, 9} Recently, Apostolakis et al.¹⁰ have identified several clinical and demographic factors that are useful in predicting patients who will potentially have poor INR control. This new score, the SAMeTT₂R₂ score, aids clinical decision making by identifying those patients who would not perform well on warfarin and for whom an NOAC would be a better option.

For stroke risk, the CHA₂DS₂-VASc performs better than the CHAD₂ in predicting stroke and thromboembolism, and it is particularly good in identifying ‘low-risk’ patients.¹¹ Composite scores for a combined end point of thromboembolism and bleeding offer good predictive values but limited incremental benefits over the established individual scores.¹²

In summary, bleeding risk should be assessed with a well-validated specific bleeding risk score (HAS-BLED), and stroke risk should be assessed with well-validated stroke scores (for example, the CHA₂DS₂-VASc). The use of scoring systems to predict end points for which they were not designed nor adequately validated is strongly discouraged.