Risk Stratification for Stroke in Atrial Fibrillation: A Critique

Ammar M. Killu; Christopher B. Granger; Bernard J. Gersh


Eur Heart J. 2019;40(16):1294-1302. 

In This Article

Limitations of Current Risk Scores

Generally, stroke risk scores use a 'one size fits all' approach. This poses a problem given that the AF population is highly heterogeneous. As such, the scores are only as good in the population from which they were derived and performance varies according to the population's baseline risk. Furthermore, incorporation of other, non-stroke embolic episodes into determinants of stroke risk may be misleading. A major limitation is that current risk schemas are based primarily on clinical risk factors and fail to account for the differential weight of individual risk factors. As shown in Figure 3, the HR estimates for the individual risk factors are not equal, which greatest importance for age. Thus, older individuals with age as their only risk are at greater stroke risk than their younger counterparts. This may be related to increasing atrial fibrosis with age, though remains to be fully elucidated.[59] Finally, given that vascular pathology may largely underpin AF, with stroke and bleeding risk closely related, it is not surprising that the CHA2DS2-VASc score has an equivalent C-statistic for predicting bleeding rates compared with dedicated bleeding scores such as ORBIT, ATRIA, and HAS-BLED.[60] The recent European guidelines recommend against using bleeding scores to decide who should receive OAC, since this can lead to failure to treat those at higher risk of stroke.[61] This is in distinction to the ACCP guidelines which, by emphasizing that attention to modifiable bleeding risk factors should be made at each patient contact, recommend use of the HAS-BLED score to gauge the frequency of patient follow-up.[25] Again, for patients at high risk of bleeding, an individualized approach is needed, with focus on treatment to prevent stroke while also treating modifiable factors to reduce the risk of bleeding.