Improving stroke and thromboembolism risk stratification in chronic atrial fibrillation

Atrial fibrillation (AF) is the commonest sustained cardiac rhythm disorder, and is encountered in everyday clinical practice.  Irrespective of whether we use a rate-control or rhythm-control strategy, stroke prevention with appropriate thromboprophylaxis still remains central to the management of this common arrhythmia.  When strokes occur in AF patients, the risk of mortality and disability, as well as recurrent stroke is substantially much higher.

Stroke risk in AF is not homogeneous [1].  Various risk factors have been identified largely from non-warfarin arms of clinical trial cohorts, and a few epidemiological/cohort studies [1,2]. Reliance on clinical trial cohort data is problematic, especially since <10% of those screened in the historical trials (those with a placebo or no antithrombotic therapy comparator arm) were ultimately randomized into the intervention.  These risk factors have been used to inform the development of stroke risk stratification schema, which (perhaps artificially) classify AF patients into high, moderate and low risk strata [2], despite stroke risk being a continuous variable.  Various management guidelines have essentially recommended oral anticoagulation (essentially warfarin or other Vitamin K antagonist) is recommended for high risk patients, ‘warfarin or aspirin’ for moderate risk, and aspirin for low risk patients.

Over the last 15 years of so, the various published stroke risk schema only have modest predictive value for thromboembolism, with no improvement in predictive ability over the years [2-4]. Many stroke risk assessment schema classify a large proportion of subjects into the ‘moderate risk’ category where treatment guidelines recommend either warfarin or aspirin, and risk stratification schema that result in classification of a large proportion of AF subjects into the ‘moderate risk’ category could potentially be less useful in everyday clinical practice, since current treatment guidelines recommend the use of either warfarin or aspirin in such patients, causing confusion over which therapy should really be prescribed. Alternatively, classification as ‘moderate risk’ is often used as an excuse not to give anticoagulation, since the guidelines ‘allow’ aspirin.

Given the modest predictive ability for identifying ‘high risk’ subjects and the availability of new oral anticoagulant drugs that overcome the shortcomings of warfarin, stroke risk stratification schema perhaps need to focus more on identifying the ‘truly low risk’ category of patients where no antithrombotic therapy may even be an option, given the increasing debate over the effectiveness of aspirin and potential for harm [5].  This concept was first proposed by van Walraven et al [6] and more recently revisited by Lip and Halperin [4].

Thus, rather than focus on artificial categories of high/moderate/low risk strata, a risk factor based approach is perhaps the best [4]. The CHADS2 schema [Table 1], whilst easy to remember, does not include all potential stroke risk factors, and indeed, ‘clinical heart failure’ (the ‘C’ in CHADS2) is regarded as an inconsistent stroke risk [1].  ‘Major risk factors’ are those with previous stroke or thromboembolism and those age ≥75, whilst ‘clinically relevant non-major risk factors’ are diabetes, hypertension, age 65-74, systolic heart failure (or moderate-severe left ventricular dysfunction), female gender and vascular disease [4,7]. 

These risk factors can be expressed with an acronym, CHA2D2-VASc [Table].  Those patients with one major risk factor or ≥2 ‘clinically relevant non-major’ stroke risk factors (essentially CHA2D2-VASc score of ≥2) should be treated with oral anticoagulation. Those with one ‘clinically relevant non-major’ stroke (ie. CHA2D2-VASc score of 1) can be treated with oral anticoagulation or aspirin, although oral anticoagulation is suggested rather than aspirin, given recent data in such patients [8,9].  Low risk patients (CHA2D2-VASc score=0) are those with no risk factors, and given that such patients are ‘truly low risk’ treatment with aspirin or (preferably) no antithrombotic therapy is appropriate.  



Table 1  - Stroke risk stratification with the CHADS2 and CHA2DS2-VASc scores