ABSTRACT

1. Introduction

Atrial fibrillation (AF) is a supraventricular tachyarrhythmia characterized by uncoordinated atrial activation with consequent deterioration of atrial mechanical function [1]. As one of the most studied cardiovascular diseases, its understanding has evolved through decades. Currently affecting about 2.7–6.1 million people in the USA [1] and 33 million people worldwide [2], with an annual risk of ischemic stroke estimated to be about 5–7%[3], the cornerstone of management remains interventions for rate control, rhythm control and thromboprophylaxis. More recently, the focus has been to streamline patient management pathways, to approach AF management in a holistic or integrated manner, as follows (the ABC or Atrial fibrillation Better Care Pathway- Figure 1): ‘A’ Avoid stroke with Anticoagulation; ‘B’ Better symptom care, with patient-centred symptom directed decisions on rate or rhythm control; and ‘C’ Cardiovascular and comorbidity risk management, including attention to risk factors and lifestyle changes [4]. The ABC approach has shown that compliance with such optimised care is associated with a significant reduction in mortality and hospitalisations [5], and a reduction in healthcare cost associated with cardiovascular events [6]. This review aims to provide a historic perspective on the evolution of stroke risk classifications in AF with particular emphasis on the most widely used system – the CHA₂DS₂VASc score, its individual components and their contribution to the overall risk of stroke.

2. Individual risk factors for stroke in atrial fibrillation

Among many other factors that increase the risk of stroke, AF is an independent risk factor for ischemic stroke and studies showed that this risk is increased by a factor of five in patients with AF [7,8]. AF is associated with major systemic thromboembolism and about a third of patients with ischemic stroke have been found to have either clinical or subclinical AF [9] due to high prevalence of left atrial thrombosis [10]. Since oral anticoagulation (OAC) is known to reduce the incidence of stroke in AF, timely diagnosis and use of anticoagulant is imperative for stroke prevention [11].

Over the years several stroke risk stratification systems have been developed using common and validated stroke risk factors in patients with AF, to aid decision-making for thromboprophylaxis. Most of these systems utilise acronyms to represent various individual risk factors and are scored accordingly. Although stroke risk is a continuum, patients have been artificially classified as low, intermediate and high risk for stroke and decision to commence thromboprophylaxis are made to avoid unnecessary use of OAC [12]. A major shift in the stroke risk stratification is the identification of low risk group to minimize needless anticoagulation.

One of the earliest stroke risk stratification system was the Framingham score [13], a points-based system assessed various clinical factors and assigned scores as follows: Age (0–10), female sex (6), systolic hypertension (0–4), diabetes mellitus (5), and prior stroke or transient ischemic attack (TIA) (6). Scores >8 were considered increased risk and required thromboprophylaxis [14]. The Atrial Fibrillation Investigators (AFI) sought to identify patient features predictive of stroke risk and found that patients with AF less than 65 years without hypertension, previous stroke or transient ischemic attack or diabetes were at very low risk for stroke and were not anticoagulated [15,16]. The stroke prevention in atrial fibrillation (SPAF) trial also identified recent congestive heart failure (within 3 months), history of hypertension (systolic blood pressure >160 mmHg) and previous arterial thromboembolism as independent risk factors for thromboembolism in atrial fibrillation [17]. In 2001, Gage et al devised the CHADS₂ score by merging the AFI and SPAF classification schemes, testing this in a cohort of hospitalised AF admissions [18]. The CHADS₂ score assesses the individual risk factors and scores 1 point for congestive cardiac failure, hypertension, age 75 years and above and diabetes mellitus. Prior stroke or transient ischemic attack is assigned 2 points [18].

In 2010, a revised clinical risk stratification tool for predicting stroke and thromboembolism in AF [19] provided some improvement in the predictive value of the existing CHADS₂ schema by including age 65–74 years, vascular disease and female gender to form the CHA₂DS₂VASc score (Table 1). This is currently a widely used scoring system in guidelines for stroke in AF as it usefully identifies low risk patients with AF. Piccini et al in 2012 noted that renal dysfunction was an independent risk factor for stroke in patients with AF and thus the R₂CHADS₂ scoring system was introduced with 2 points assigned to creatinine clearance less than 60ml/min [20]. This is perhaps unsurprising as renal impairment often coexists with components of the CHADS₂ and CHA₂DS₂-VASc scores. Latter studies disputed the relevance of renal impairment as an independent risk factor for stroke in AF as it failed to improve the predictive ability of CHADS₂ and CHA₂DS₂-VASc score [21]. Despite the use of the various risk stratification systems, a significant number of strokes are still noted in patients with AF classified as low risk and not requiring thromboprophylaxis. The need to further identify other risk factors for thromboembolic event led to the introduction of the ATRIA (anticoagulant and risk factors in atrial fibrillation) score [22] which in addition to existing risk factors, also included age categories, presence of proteinuria and estimated glomerular filtration rate less than 45 or end stage renal disease. While a study by Van Den Ham et al noted that the ATRIA score better identified low risk patients for stroke than the CHA₂DS₂-VASc score [23], various other cohort studies concluded that CHA₂DS₂-VASc score had better correlation with thromboembolism in non-anticoagulated patients with AF than the newer R₂CHA₂DS, ABC-stroke score and ATRIA scoring systems [24,25] The Canadian Cardiovascular Society adopted the CHADS65 risk score, which incorporated the CHADS₂ and CHA₂DS₂-VASc risk factors proposing a guideline for thromboprophylaxis in patients with AF. This risk score considered age 65 years and above as a cut-off and recommend the use of OAC in AF patients with CHADS₂ ≥ 1 and age ≥ 65 years. Aspirin was recommended for patients with stable coronary artery disease (CAD)/peripheral vascular disease (PVD), or those age< 65 years, and no thromboprophylaxis in patients < 65 with CHADS₂ of 0, in the absence of CAD/PVD [26]. In 2013, Hippisley-Cox et al developed and published the Qstroke algorithm, which is an electronic model that calculates the risk of stroke or TIA without prior stroke or TIA based on several risk factors such as age, ethnicity, sex, smoking status, clinical diagnosis of diabetes, treated hypertension, kidney disease, rheumatoid arthritis, angina, coronary heart disease, congestive cardiac failure, valvular heart disease, and clinical values including ratio of total serum cholesterol to high density lipoprotein cholesterol concentrations, body mass index as well as family history of coronary heart disease in first degree relative under 60 years. Its performance was compared to the existing scoring systems mainly the CHADS₂ and CHA₂DS₂VASc scores. Qstroke was found to improve the performance of both CHADS₂ and CHA₂DS₂VASc in assessing high risk patients who might benefit from anticoagulation [27]. The quest to find an ideal stroke risk stratification tool continued as a significant population of non-anticoagulated low risk patient was seen to have thromboembolic events. The Global Anticoagulant Registry in the FIELD–Atrial Fibrillation (GARFIELD study) [28] used a computer-generated risk model that predicted all-cause mortality, ischaemic stroke/systemic embolism (SE) and haemorrhagic stroke/major bleeding in low-risk atrial fibrillation (AF) patients – and proposed the GARFIELD-AF score which was superior to CHA₂DS₂-VASc in predicting risk of ischemic stroke in AF patients with low risk of stroke [29]. In the Southern hemisphere, the National Heart Foundation of Australia (NHF) and the Cardiac Society of Australia and New Zealand (CSANZ) published a national guideline for stroke prevention in patients with AF. They recommended the use of the ‘sexless CHA₂DS₂-VASc score’ ie. the CHA₂DS₂-VA risk score, and preference was given to non-vitamin K antagonist oral anticoagulants (NOACs) over vitamin K antagonists (VKA) in patients with scores ≥ 1 [30].

Currently, given its relative simplicity the CHA₂DS_2-VASc score is still the most widely used scoring system and the 2018 American College of Chest Physicians (ACCP) CHEST guidelines recommend the use of oral anticoagulants, preferably NOACs for patients with a single non-sex CHA₂DS₂-VASc risk factor (score of ≥ 1 in males, ≥ 2 in females) rather than no anticoagulation or use of single or dual anti-platelet therapy [31].

3. Congestive heart failure

Congestive heart failure (CHF) which represents the ‘C’ in the CHA₂DS₂VASc scoring system, is considered a cause and effect of AF and is associated with increased morbidity and mortality [38]. In older scoring systems such as CHADS₂, heart failure only referred to recent decompensated heart failure and there was some uncertainty as to whether or not heart failure with preserved ejection fraction (HFpEF; left ventricular ejection fraction >50%) or heart failure with reduced ejection fraction (HFrEF; left ventricular ejection fraction <40%) or asymptomatic systolic or diastolic dysfunction was being included.

Banerjee et al sought to establish the role of ejection fraction in risk prediction in patients with non-valvular AF and heart failure. They found no difference in rates of stroke, systemic thromboembolism or death between the different heart failure categories [39]. Sandhu et al also assessed this relationship among participants of ACTIVE (Atrial Fibrillation Clopidogrel Trial with Irbesartan for Prevention of Vascular Events) trial in which, among patients with heart failure, neither the presence of left ventricular systolic dysfunction nor the degree of symptom severity influenced the risk for embolic events [40]. Therefore, the decision about risks and benefits of anticoagulation for heart failure patients with AF should not be influenced by ejection fraction or symptom status.

4. Hypertension

Hypertension is the most prevalent risk factor for new-onset atrial fibrillation. Controlled or uncontrolled hypertension and history of hypertension scores a point in the CHA₂DS₂VASc score. In one study, the observed absolute stroke risk from hypertension alone as a risk factor in non-anticoagulated patients with atrial fibrillation was about 1.5–3.0% per year [41].

A cross-sectional longitudinal analysis using data from Stroke prevention using an Oral Thrombin inhibitor in atrial Fibrillation (SPORTIFF) III and V trials showed an increase in stroke rate with increasing systolic blood pressure in patients with AF. Hypertension contributes to increased stroke with events rates increased at systolic blood pressure (SBP) levels of 140 mmHg and above. However, controlled hypertension with mean SBP <140 mmHg, is associated with a lower risk of stroke compared with patients with poorly controlled hypertension. Notably, in the aspirin arm of SPAF III trial, even subjects with a history of hypertension had an increased risk of stroke in AF [42].

5. Age

The Framingham study and more recent studies established age as a significant and independent risk factor for AF associated stroke [7,43–45]. Studies also showed that in patients with lone AF (without other risk factors), the risk of stroke significantly increased in patients >60 years [46]. Studies have also shown that the highest risk of stroke in AF was found in patients aged 75 years and above. For this reason, the CHA₂DS₂-VASc score assigned a score of 2 for age ≥ 75 years and 1 for ages 65–74 years. In Asia, the age threshold may even be lower, with ischemic stroke rates approximately 1.5%/year in patients 50–64 years [47] leading to a suggestion of a modified CHA₂DS₂VASc score for Asians, with 1 point given for age 50–74 years.

Physicians are often reluctant to use OAC in elderly patients due to the potential life-threatening complications associated with its use. Studies have shown that the benefits of OAC also increases with advancing age as stroke is known to increase with age. A large cohort study in Asia demonstrated that among patients >90 years, warfarin use was associated with a lower risk of ischemic stroke (HR, 0.69 [0.49–0.96]) and a positive net clinical benefit [47]. NOACs were associated with lower risk of ICH (HR, 0.32 [0.10–0.97]) without difference in ischemic stroke risk compared to warfarin.

Although age is a component of several bleeding risk scores, it should not be used solely to negate the use of OACs in elderly patients [45]. Anticoagulation in elderly often requires benefit-risk conversation with patients and relatives. It is also important not to over-emphasize the bleeding risk associated with elderly falls as studies have shown that the absolute risk of bleed in those on anticoagulants is small [48].

6. Diabetes mellitus

Diabetes mellitus (DM) is known to increase both the risk of AF and development of stroke in AF [49–51]. Arrhythmogenicity may be associated with the accelerated atherosclerosis in diabetes with structural, electrical and autonomic remodelling [52]. DM independently increases the risk of stroke in patients with AF by 1.7 fold with other studies noting an absolute stroke rate of 2.0–3.5% per year in the absence of other risk factors in non-anticoagulated AF patient [46].

Among participants of the ATRIA (Anticoagulation and Risk Factors in Atrial Fibrillation) study, duration of diabetes (>3 years) was strongly associated with increased risk of stroke compared to having diabetes for less than 3 years (HR: 1.74, 95% CI 1.10–2.76). They also found that neither poor glycemic control (HA1c > 9.0%, HR 1.04, 95% CI: 0.57–1.92) nor moderately increased HA1c (7.0–8.9%, adjusted HR: 1.21, 95% CI: 0.77–1.91) were significantly associated with an increased rate of ischemic stroke compared with patients who had HA1c <7.0%. Therefore, duration of diabetes is a more important predictor of ischemic stroke than glycemic control in diabetic patients with AF [53].

A study by Fangel et al using data from Danish nationwide registry also found that among patients <65 years, there was higher risk of thromboembolism among those with type 2 diabetes compared to those with type 1 diabetes. However, overall no independent association was found between type of diabetes (type 1 versus type 2) and the risk of thromboembolism in non-anticoagulated patients with AF [54].

7. Stroke/transient ischemic attack

Previous stroke or transient ischemic attack (TIA) is perhaps the strongest independent risk factor for stroke in patients with AF. Studies have shown that this risk is increased by 2.5-fold following a previous stroke of TIA [55]. Interestingly, a significant association was found between prior intracranial hemorrhage (ICH) and ischemic stroke in patients with AF in a study by Friberg et al. whereby AF patients with prior ICH are at increased risk of having an ischemic stroke [56].

8. Vascular disease

This component is unique to the CHA₂DS₂-VASc scoring system and not present in other stroke risk scores. This refers to atherosclerotic vascular diseases such as prior myocardial infarction, complex aortic plaques (>4mm thick or presence of mobile debris [57]) and peripheral arterial disease [58]. Peripheral arterial disease (PAD) is also observed to increase the risk of incident AF in elderly patients with risk increasing with progressively lower ankle-brachial pressure index (ABPI).

Recent evidence showed that vascular diseases including acute coronary syndrome (ACS) and peripheral artery disease (PAD) are a predictor of AF related stroke and thromboembolism and improves the predictive ability of CHADS₂ score [59]. Studies showed that peripheral artery disease alone increases the risk of stroke with poor outcome in patients with AF whereas an uncomplicated ACS is less a risk factor than myocardial infarction and angina pectoris [60]. In patients with atrial fibrillation and vascular disease with no additional non-sex risk factors, the risk of ischemic stroke was increased by 1.68 in males and 2.15 in females [61].

In a large cohort study using the Danish National Patient Registry, coronary artery disease (defined as obstructive (≥50%) coronary stenosis in ≥1 coronary vessel or non-obstructive coronary stenosis in ≥2 coronary vessels) conferred a 29% increased risk of ischemic stroke, TIA or thromboembolism among patients with AF suggesting that coronary artery disease (CAD) was an independent risk factor for ischemic stroke hence the recommendation to include this component in stroke risk scores among AF patients [62].

9. Sex category

Even in the absence of atrial fibrillation, women have higher lifetime risk of stroke than men. The CHA₂DS₂-VASc scoring system introduced sex as a factor for stroke in AF. Various studies showed that female sex is a strong independent risk factor [63–65]. This may be related to post-menopausal vascular changes related to the reduction in estrogen. This affects lipid metabolism, increases the risk of left ventricular remodelling and hypertension, and leads to increase in inflammatory and procoagulant markers all contributing to thromboembolism [64]. In a large cohort study using data from Atrial Fibrillation Follow-up Investigation of Rhythm Management (AFFIRM) trial, hormone replacement therapy (HRT) did not independently predict mortality, thromboembolism or bleeding among women with atrial fibrillation [66].

A score of 1 point is attributed to the female gender in the CHA₂DS₂-VASc score. In a cohort study using the Danish national register involving 87,202 patients with atrial fibrillation, the risk of stroke and thromboembolism in females less than 74 years was not increased compared to men; however, females aged ≥ 75 years had increased stroke rates compared to males in the same group with HR 1.20 (95% CI 1.12–1.28) recommending that female sex should not be an independent score in the absence of other risk factors [63].

In a meta-analysis of 17 studies comparing males and females with atrial fibrillation for the outcome of stroke and thromboembolism, 10 demonstrated increased risk of stroke in women revealing 1.31 fold (95% CI 1.18–1.46) elevated risk of stroke in women with atrial fibrillation, more so in women aged 75 years or more [67]. The annual stroke rate for males with only one risk factor and a CHA₂DS₂-VASc score of 1 is similar to that of females with 1 additional risk factor and CHA₂DS₂-VASc a score of 2 (1.96–3.50% versus 1.91–3.34%).

Thus, the female gender should be considered a risk modifier rather than a risk factor on its own in the absence of other risk indices in patients with AF and the decision to start OAC should be guided by the CHA₂DS₂-VASc score without the sex category component because the sex category risk component accentuates the stroke risk in women already eligible for OAC based on the presence of 2 or more risk factors [61,68].

10. Non CHA₂DS₂-VASc risk factors

Apart from the well-established CHA₂DS₂-VASc risk factors, other factors may independently contribute to increased stroke risk in AF patients, one of such is obstructive sleep apnea (OSA). Several studies noted an association between OSA and increased risk of ischemic stroke [69,70]. It remains unclear whether OSA should be factored independently into the risk stratification as patients with OSA may have co-existing CHA₂DS₂-VASc factors. Kabra et al proposed a modification of the CHA₂DS₂-VASc score to the CHA₂DS₂-VASc-R score based on the findings of increased risk of stroke in African-American patients with AF [71]. However, the conclusion that race presented an additional risk in patients with AF was met with scepticism [72,73]. Studies have linked inflammation to the risk of AF and AF related thromboembolism. Higher levels of pro-Inflammatory cytokines have been documented in patients with AF, although it remains unclear whether the raised levels are directly due to AF or an underlying cardiovascular disease [74]. Irrespective of this fact, inflammation confers a prothrombotic state and increases stroke risk in AF patients [74]. The role of inflammatory makers in AF management and risk stratification is still undetermined.

11. Conclusion

Atrial fibrillation is a major risk factor for stroke and the incidence of stroke remains high in patients with AF with or without valvular heart disease. As the prevalence of AF approaches epidemic proportions, stroke prevention remains one of the cornerstones of management. To minimize the risk of stroke, careful evaluation of stroke and bleeding risk must be undertaken with prompt initiation of thromboprophylaxis where necessary. The development of various scoring systems over the years has simplified the evaluation, with the CHA₂DS₂VASc scoring system being the most widely used. Use of the CHA₂DS₂VASc scoring system reliably excludes patients at low risk for thromboembolism. Current guidelines recommend prompt initiation of thromboprophylaxis, preferably NOACs for AF patients with one or more non-sex CHA₂DS₂VASc risk factor (≥1 in males, ≥2 in females).

References