Many patients with AF are asymptomatic, and AF may only be first detected during presentation with complications related to the arrhythmia, such as heart failure, stroke or thromboembolism. Indeed, only 1 in 12 paroxysms of AF are actually symptomatic, amongst patients with paroxysmal AF . Other studies with pacemakers confirm that AF burden varies widely, and many paroxysms are asymptomatic [14,15].
Assessing stroke risk in AF
Whilst AF increases the risk of stroke by 5-fold, this risk is not homogeneous as the risk of stroke is altered by the presence of other stroke risk factors .
Two systematic reviews have investigated the impact of various risk factors on stroke, based on epidemiological cohorts and non-warfarin arms of clinical trial cohorts [21,22]. However, data from trial cohorts are limited depending on whether particular risk factors were systematically examined and/or recorded. An additional complication is that there have been inconsistencies in the definitions of certain risk factors between different trials.
The Stroke in AF Working Group  identified the following risk factors for stroke in AF: previous stroke/TIA [adjusted relative risk (RR) 2.5], age (RR 1.5/decade), hypertension (RR 2.0), diabetes (RR 1.8) and female gender (RR 1.6). History of heart failure was not significant (despite being the ‘C’ in the CHADS2 score that is commonly used for stroke risk stratification – see later), although the presence of moderate-systolic left ventricular dysfunction was still an independent predictor of thromboembolism upon multivariate analysis. The systematic review of stroke risk factors as part of the UK National Institute for Health and Clinical Evidence (NICE) guidelines identified history of stroke or TIA, increasing age, hypertension and structural heart disease (left-ventricular dysfunction or hypertrophy) to be good predictors of stroke risk in AF patients, whilst the evidence regarding diabetes mellitus, gender and other patient characteristics was less consistent .
These stroke risk factors have been used to formulate various stroke risk stratification schema . Despite stroke risk in AF being a continuum, these schema have been used to ‘artificially’ categorise patients into low, moderate and high risk stroke strata, so that the patients at highest risk can be identified for warfarin therapy, given the disutility and limitations of the latter therapy that requires regular anticoagulation intensity monitoring and interactions with diet, drugs and alcohol .
Clearly, things have moved on with new information on stroke risk factors, and the availability of new OAC that are alternatives to warfarin. Thus, we need to be more inclusive (rather than exclusive) of common stroke risk factors, to get better at identifying the ‘truly low risk patients’ with AF [2,23]. This is relevant given that the latter category of patients can even be managed with no antithrombotic therapy, whilst those AF patients with ≥ 1 stroke risk factor could be treated with OAC, whether with well-controlled warfarin or the new agents, such as dabigatran [2,24].
Stroke risk stratification schema
The most common stroke risk stratification schema, given its simplicity, is the CHADS2 score [Congestive heart failure, Hypertension, Age ≥ 75, Diabetes mellitus, and prior Stroke or transient ischemic attack]  (see Table 1). The CHADS2 score was derived by amalgamation of the AF Investigators and SPAF-1 risk schema (both trial-based risk stratification schema) and validated in a hospitalised cohort of AF patients, the Non-Rheumatic AF cohort . The limitations of the CHADS2 score have been discussed and debated , particularly since it does not include many common potential stroke risk factors. Based on its original validation study , it categorises a score of 0 as ‘low risk’, 1–2 as ‘moderate/intermediate risk’ and ≥ 3 as ‘high risk’. Thus, a patient with previous stroke or TIA alone as a risk factor would have a score of 2 and have been categorised as ‘moderate risk’ using the original categorisation, despite this sort of patient having the highest risk for subsequent stroke or thromboembolism. Nonetheless, current guidelines have partly remedied this, by defining a CHADS2 score ≥ 2 as ‘high risk’.
Table 1. Assessment of stroke [CHA2DS2-VASc] and bleeding risk [HAS-BLED] in atrial fibrillation patients
|Congestive heart failure||1||Hypertension (systolic blood pressure > 160 mmHg)||1|
|Hypertension||1||Abnormal renal and liver function (1 point each)||1 or 2|
|Aged ≥ 75 years||2||Stroke||1|
|Diabetes mellitus||1||Bleeding tendency/predisposition||1|
|Stroke/TIA/TE||2||Labile INRs (if on warfarin)||1|
|Vascular disease (prior MI, PAD, or aortic plaque)||1||Elderly (e.g. age > 65)||1|
|Aged 65–74 years||1||Drugs (e.g. NSAIDs, aspirin) or alcohol abuse (1 point for each category)||1 or 2|
|Sex category (i.e. female gender)||1|| || |
|Maximum score||9||Maximum score||9|
Furthermore, subsequent validation studies have shown a poor predictive value for the CHADS2 (and other) schema (c-statistics approximately 0.6). Furthermore, the CHADS2 score based on its original validation would categorise nearly 60%–65% of various AF populations into the ‘moderate/intermediate risk’ category, where older management guidelines would recommend ‘warfarin or aspirin’ as suitable treatments. This would give some uncertainty on what should one prescribe (warfarin or aspirin?) or some clinicians prescribing aspirin rather than warfarin, as ‘the guidelines allow it’. However, current guidelines would redefine ‘moderate/intermediate risk’ as those with a CHADS2 score = 1, and ‘warfarin or aspirin’ is recommended.
To complement the CHADS2 schema, the new European Society of Cardiology guidelines  de-emphasises the low, moderate and high risk categorisation and instead recommends a risk factor based approach with a new schema, the ‘CHA2DS2-VASc’ score to complement the CHADS2 scheme (see Table 1). The CHA2DS2-VASc schema places greater emphasis on what it terms ‘major risk factors’, that is, age ≥ 75 years and previous stroke/TIA, by allocating two points to each, with one point for the presence of each of the other ‘clinically relevant non-major’ risk factors (systolic heart failure, hypertension, diabetes, age 65–74, vascular disease and female gender), with total scores ranging from 0 to 9.
Of note, female gender, age 65–74, and vascular disease are not ‘new’ risk factors per se in AF, but rather, have considered as potential risk factors for stroke in AF in some guidelines. The ‘CHA2DS2-VASc’ scheme tries to formalise these risk factors in an attempt to become more inclusive of common stroke risk factors in AF. After all, any stroke risk factor confers ‘clinical risk’per se when AF is present, and the result may ultimately be a devastating stroke. Indeed, the CHA2DS2-VASc score is more inclusive of common stroke risk factors seen in most patients with non-valvular AF, so that it would at least be applicable most of the time, and easily implemented (informally, guidelines should be applicable for > 80% of the time, and in > 80% of patients). Thus, some less common risk factors associated with thromboembolism in AF (such as end-stage renal failure, amyloid heart disease, etc…) are not part of the CHA2DS2-VASc score as such risk factors have not been adequately studied in clinical trials, nor the balance between mortality, stroke and bleeding clearly defined.
The CHA2DS2-VASc score was first validated in a European cohort from the EuroHeart survey on AF . This study concluded that the CHA2DS2-VASc was good at identifying ‘truly low risk’ patients with AF (≤ 1%/year, with a CHA2DS2-VASc score = 0), categorised the lowest proportion into the ‘moderate/intermediate risk’ strata and that the point estimate using the c-statistic was marginally better than the CHADS2 schema. However, the EuroHeart survey cohort had various limitations (including a proportion of patients lost to follow-up), and validation in other independent cohorts was necessary.
A further large validation was performed in a cohort of 79884 AF patients aged ≥ 18 years in the UK General Practice Research Database, who were followed for an average of 4 years (average of 2.4 years up to the start of warfarin therapy) . This analysis found that all 15 published AF stroke risk stratification schemes had modest discriminatory ability in AF patients, with c-statistics for predicting thromboembolism that ranged from 0.55 to 0.69 for strokes recorded by the general practitioner or in hospital, from 0.56 to 0.69 for stroke hospitalisations, and from 0.56 to 0.78 for death resulting from stroke as reported on death certificates. The proportion of patients assigned to individual risk categories also varied widely across the schemes, with the proportion categorised as moderate risk ranging from 12.7% (CHA2DS2-VASc) to 61.5% (modified CHADS2). Low-risk subjects were truly low risk (with annual stroke events < 0.5%) with the CHA2DS2-VASc schemes. Thus, the CHA2DS2-VASc schema was able to discriminate those at ‘truly low risk’ and minimised classification of subjects into the moderate/intermediate risk category.
A further cohort study used nationwide data on 73,538 hospitalised patients with AF who were not treated with VKAs in Denmark in the period 1997–2006 . In ‘low risk’ subjects (score = 0), the rate of thromboembolism per 100 person-years was 1.67 (95% confidence interval 1.47–1.89) with CHADS2 and 0.78 (0.58–1.04) with CHA2DS2-VASc, at 1 year follow-up. In ‘moderate/intermediate risk’ subjects (score = 1), this rate was 4.75 (4.45–5.07) with CHADS2 and 2.01 (1.70–2.36) with CHA2DS2-VASc. When patients were categorised into low, intermediate, and high-risk strata, the c-statistics at 10 years follow-up were 0.812 (0.796–0.827) with CHADS2 and 0.888 (0.875–0.900) with CHA2DS2-VASc, respectively. In this huge cohort, therefore, the CHA2DS2-VASc scheme performed better than CHADS2 in predicting those at ‘high risk’, and those categorised as ‘low risk’ using CHA2DS2-VASc were ‘truly low risk’ for thromboembolism.
In a trial-based anticoagulated AF cohort (n = 7329 subjects) , c-statistics for stroke and thromboembolism were broadly similar among the contemporary risk stratification schema tested and varied between 0.575 (NICE 2006) and 0.647 (CHA2DS2-VASc). CHA2DS2-VASc classified 94.2% as being at high risk (after all, the patients needed to be at some risk of stroke to enter the trial), whereas most other schemes only categorised two-thirds as being at high risk. Of the 184 thromboembolic events, 181 (98.4%) occurred in patients identified as being at high risk by the CHA2DS2-VASc schema, which had the highest hazard ratio (3.75) among the tested schemes. The negative predictive value (i.e. the percent categorised as ‘not high risk’ actually being free from thromboembolism) for CHA2DS2-VASc was 99.5%. Of the contemporary stroke risk stratification schemes, the CHA2DS2-VASc scheme correctly identified the greatest proportion of AF patients at high risk, despite a similar predictive ability of most schemes as evidenced by the c-statistic.
In a ‘real world’ of 662 consecutive elderly anticoagulated AF patients , all stroke risk schema had modest discriminating ability, with c-statistics ranging from 0.54 (AF Investigators) to 0.72 (CHA2DS2-VASc). The CHADS2 and CHA2DS2-VASc schemes had the best c-statistics (0.717 and 0.724, respectively) with significant discriminating value between risk strata (both P < 0.001). The proportion of patients assigned to individual risk categories varied widely across the schema, with those categorised as ‘moderate-risk’ ranging from 5.3% (CHA2DS2-VASc) to 49.2% (CHADS2 classical).
A comparison of thromboembolism event rates with the CHADS2 and CHA2DS2-VASc score is shown in Table 2, using data from Olesen et al. .
Table 2. Event rates (95% CI) of hospital admission and death due to thromboembolism per 100 person years, based on the CHADS2 and CHA2DS2-VASc scores
|Score/risk category||1 year’s follow-up|
| 0||1.67 (1.47–1.89)|
| 1||4.75 (4.45–5.07)|
| 2||7.34 (6.88–7.82)|
| 3||15.47 (14.62–16.36)|
| 4||21.55 (20.03–23.18)|
| 5||19.71 (16.93–22.93)|
| 6||22.36 (14.58–34.30)|
| Low risk (0)||1.67 (1.47–1.89)|
| Intermediate risk (1)||4.75 (4.45–5.07)|
| High risk (2–6)||12.27 (11.84–12.71)|
| 0||0.78 (0.58–1.04)|
| 1||2.01 (1.70–2.36)|
| 2||3.71 (3.36–4.09)|
| 3||5.92 (5.53–6.34)|
| 4||9.27 (8.71–9.86)|
| 5||15.26 (14.35–16.24)|
| 6||19.74 (18.21–21.41)|
| 7||21.50 (18.75–24.64)|
| 8||22.38 (16.29–30.76)|
| 9||23.64 (10.62–52.61)|
| Low risk (0)||0.78 (0.58–1.04)|
| Intermediate risk (1)||2.01 (1.70–2.36)|
| High risk (2–9)||8.82 (8.55–9.09)|