Atrial fibrillation (AF), a major cause of morbidity and mortality in the Western world, is considered to be a rising epidemic.1–3 In the US alone, approximately 2.7 million people are affected by AF4 and this figure is projected to increase to an estimated 16 million by 2050.5 The rising prevalence of AF is mainly attributable to the ageing population6 and survival from other cardiovascular disease. AF predominantly affects individuals over 80 years of age.4,7,8
AF is a strong independent risk factor for stroke and accounts for one in five strokes.9 Furthermore, stroke in AF has twice the mortality rate compared to other strokes,1 is a cause of severe disability compared to other strokes,10 more frequent hospital stays3 and is a major socioeconomic burden.11 Anticoagulation therapy with warfarin is recommended by evidence-based guidelines for stroke prevention in patients with AF and is widely regarded as the standard of care.12 However, warfarin therapy has severe limitations that make it difficult and inconvenient to use in clinical practice.13 Consequently, up to 50 % of AF patients at moderate to high risk of stroke do not receive appropriate anticoagulant treatment.14–19 Thus, there is an urgent need for improved antithrombotic therapy in AF management. This article aims to review the clinical consequences of stroke in AF and the recent improvements in risk stratification.12
Clinical Consequences of Atrial Fibrillation
AF is associated with palpitations, impaired exercise tolerance and symptoms from cardiac failure. Clinical deterioration may follow when AF complicates pre-existing cardiac disorders. In terms of cardiovascular events, women with a single electrocardiograph (ECG) recording showing AF have a five-fold increase in risk. The risk is increased two-fold in men with a single ECG of AF.2 Ten to 40 % of all elderly patients are asymptomatic.20 Additionally, in patients with either AF or congestive heart failure (HF) alone, development of the second condition as a concomitant disease carries a particularly poor prognosis. For example, in patients with AF, the subsequent development of congestive HF is associated with increased mortality (men: hazard ratio [HR] 2.7; 95 % confidence interval [CI], 1.9–3.7; women: HR 3.1; 95 % CI, 2.2–4.2).21
Stroke and thromboembolic disease are considered the most important complications of AF and their occurrence is increased in both paroxysmal and chronic AF.22 In an analysis of the Framingham study, the impact of cardiovascular disease in 5,070 participants after 34 years of follow-up found that patients with AF had a nearly five-fold increased risk of stroke compared with age-matched individuals with normal sinus rhythm.23 Indeed, AF is thought to cause as many as one in five of the 700,000 strokes that occur each year in the US;24 in individuals over the age of 80, AF is linked to nearly one in every three strokes.6
A retrospective study examined whether AF was an independent prognostic factor in 3,849 patients with ischaemic stroke.25 The study found that AF was associated with a poor prognosis in patients with ischaemic stroke, including longer periods of hospitalisation (median 15 versus nine days), an increased risk of in-hospital medical complications (adjusted relative risk = 1.48, 95 % CI, 1.23–1.79) and recurrent stroke (adjusted HR = 1.30, 95 % CI: 0.93–1.82) when compared with patients without AF.
A retrospective study of 1,061 patients with acute ischaemic stroke demonstrated that patients with AF had an increased stroke severity that was independent of advanced age and other stroke risk factors.26 The degree of disability from acute ischaemic stroke was significantly different in patients with AF versus those without AF in the 65–74 and 75–84 age groups.26 Patients with AF are also more likely to have both cerebral lesions and diminished functional status compared with patients without AF.26,27
The mortality rate of patients with AF is double that of patients in sinus rhythm and linked to the severity of underlying heart disease.1,28,29 In the Studies of left ventricular dysfunction (SOLVD), mortality was 34 % for those with AF compared with 23 % for patients in sinus rhythm (p<0.001). This difference was attributed to an increased risk of death due to HF rather than to thromboembolism.30 Indeed, AF has been found to be an independent risk factor for HF2 and is associated with poor outcomes in patients with chronic HF.31
Population-based data indicate that subjects with AF have markedly reduced survival compared with subjects without AF, with risk factor-adjusted odds ratios (OR) for death of 1.5 and 1.9 in men and women, respectively.1 The current treatments used for AF present additional challenges, which may further increase morbidity and mortality, including the potential to cause fatal proarrhythmia by the inappropriate use of antiarrhythmic drugs.28,32
Quality of Life
AF is associated with reduced quality of life (QoL).28 In the symptomatic patient, the spectrum of symptoms associated with AF range from palpitations and light-headedness to exacerbation of HF and chest pain, all of which negatively affect QoL.28,33 Furthermore, in patients with paroxysmal AF (PAF), QoL, as measured by a generic scale Medical Outcomes Study Short Form 36 (SF-36) has been found to be comparable to that of post-myocardial infarction patients and when compared with healthy subjects (n=47) patients with PAF (n=152) had significantly lower health-related QoL scores (p<0.001).34 Although impact on QoL can be mostly attributed to physical symptoms of AF, QoL is also influenced by patient factors.35 Even patients with an apparently clinically silent, asymptomatic AF do not have a typical QoL. Moreover, some patients initially deny that they have any symptoms but when the arrhythmia is ultimately treated, subsequently admit that their QoL was severely restricted. Others appear genuinely asymptomatic, with typical activity scores, although their QoL is evidently low.36 Consequently, relief of symptoms and improving QoL are often primary goals of therapeutic management of rhythm control interventions.33
A recent study has shown that control of symptomatic PAF patients has a positive beneficial impact on QoL.37 However, there is no gold standard for measuring QoL in AF patients and most questionnaires are cumbersome and time-consuming to complete. To address these limitations, a patient-guided AF-specific QoL questionnaire (Atrial Fibrillation Effect on Quality-of-life [AFEQT]) was recently developed and validated in a large prospective observational study. The 20-item self-administered AFEQT questionnaire assesses the impact of AF on patients QoL and evaluates patients’ perceptions of their symptoms, functional impairment, treatment concerns and satisfaction with treatment.
Thus, this tool may serve as an important QoL outcome measure in different clinical settings including clinical research, survey studies, or clinical practice.38 A simple, objective bedside measurement, such as the Canadian Cardiovascular Society Severity in AF scale or disease-specific AF QoL measure, may also be useful in determining the impact of symptoms on QoL.37,39,40
Economic and Healthcare System Costs of Atrial Fibrillation
AF represents a costly threat to the financial stability of healthcare systems. A prevalence-based analysis of the economic burden of AF in the UK estimated the direct cost of AF to be 0.62 % (￡244 million) of the total National Health Service (NHS) expenditure in 1995 and this was projected to account for 0.97 % (￡459 million) of the NHSbudget in 2000 − reflecting a predicted doubling in only a five-year time span.3 Yet, this study indicated that these costs were probably an underestimate. Factoring in the costs of stroke care and length of stay when AF complicates another illness would further increase the expenditure.
In the US, the total annual cost (2005 US dollars) for the treatment of AF was estimated at US$6.65 billion, including US$2.93 billion for hospitalisations.41 In the Cost of Care in AF (COCAF) survey of 671 AF patients from across France, the annual cost per patient was estimated at €3,209.42 Furthermore, the Euro Heart Survey on AF (2003−2004) reported that the total annual cost of AF amounted to €272 million in Greece, €3,286 million in Italy, €526 million in Poland, €1,545 million in Spain and €554 million in the Netherlands.43
The number of hospitalisations due to AF is also increasing.44,45 In the UK alone, AF accounts for 3 to 6 % of acute hospital admissions.46,47 The Euro Heart Survey on AF suggested that inpatient care and interventional procedures account for more than 70 % of total annual costs for AF in five European countries.43 A national costing report on AF by the National Institute for Health and Clinical Excellence (NICE) in 2006 estimated the unit cost of anticoagulation services to be an average of ￡565.8 per year per person and the annual cost per case of stroke was estimated to be ￡7,800. This report estimated the annual cost saving through stroke reduction to be ￡54.33 million, producing a net resource impact of ￡21.86 million.48 However, this may be an underestimation because it does not include the associated comorbidities that would require proactive management in AF.
Risk Stratification for Stroke
Risk stratification for stroke is critical in the management of patients with AF to ensure that patients receive appropriate anticoagulant therapy.12
Risk Stratification Schemes
The risk factors for stroke included in current stroke risk stratification schemes are mainly derived from analyses of clinical trials cohorts.21,49–51
The Framingham Stroke Risk Profile was one of the first risk stratification schemes for stroke to be developed. This scheme is gender-specific and includes AF, age, use of antihypertensive therapy, prior cardiovascular disease, cigarette smoking, diabetes mellitus, left ventricular hypertrophy by electrocardiogram and systolic blood pressure.51 Currently, the CHADS2 (one point each for congestive heart failure, hypertension, age ≥75 years and diabetes, and two points for a previous history of stroke, thromboembolism or transient ischemic attack) scheme is the most frequently used stroke risk stratification scheme (see Table 1).52,53 Risk factors included in the CHADS2 scheme were selected from those that independently predicted the risk of stroke in AF during clinical trials.52 However, many risk stratification models have been found to only have a modest predictive value in identifying patients with high thromboembolic risk.54,55 In fact, a large proportion of patients are classified into the moderate/intermediate-risk category, causing uncertainty amongst clinicians as to the appropriate treatment strategy.56,57
Recently, additional risk factors such as female sex, age 65−74 years and vascular disease have been demonstrated to influence the risk of stroke and thromboembolism in AF patients. Consequently, the CHA2DS2-VASc (congestive heart failure, hypertension, age ≥75 years, diabetes, previous stroke, thromboembolism or transient ischaemic attack, vascular disease, age 65–74 years, female sex category; age ≥75 years and previous stroke, thromboembolism or transient ischemic attack carry doubled risk weight) scheme was developed to complement the CHADS2 scoring systems (see Table 1).50
The performance of the CHADS2 scheme was compared to the CHA2DS2- VASc risk scheme by applying both scoring systems to a large nationwide registry of patients admitted to hospital with AF in Denmark.58 Although the c-statistics – a measure of the predictive value of a risk scoring scheme – were similar when the two risk schemes were tested as continuous point scales, CHA2DS2-VASc had unusually high c-statistics when applied using the traditional categories of low- (CHADS2 score of 0), intermediate- (CHADS2 score of 1–2) and high-risk groups (CHADS2 score of ≥3), which suggests that the new scheme provides improvements in predicting stroke risk over the CHADS2 scheme. The annual rate of thromboembolism including peripheral artery embolism, ischaemic stroke and pulmonary embolism in the CHA2DS2-VASc low risk group was 0.78 per 100 person years, compared with 1.67 per 100 person years in the CHADS2 low risk group.
Only 8.7 % of the study population met the low risk CHA2DS2-VASc criteria, compared with 22.3 % when using the CHADS2 criteria (see Figure 1).58 This finding is similar to other assessments, including a UK study in which only 8.6 % of patients with AF in general practice were considered low risk by CHA2DS2-VASc.59 In the Danish study, the CHA2DS2-VASc scheme performed better than CHADS2 in predicting patients at high risk; 80 % of the patients were considered high risk using CHA2DS2-VASc, however fewer than half would have met high-risk criteria if CHADS2 were used.58 The study also found that the rate for patients in the intermediate risk group was 4.75 with CHADS2 compared with 2.01 with CHA2DS2-VASc. Thus, the newer stratification methods such as CHA2DS2-VASc appear to offer improved risk stratification, reduce the number of people classified into the intermediate-risk group, which is the category that can cause ambiguity regarding whether to treat with warfarin or aspirin.50,55 Moreover, a recent study assessed the risk of stroke according to specific risk stratification schemes, in a cohort of 662 elderly AF patients treated with warfarin.
The results indicated that the CHADS2 and CHA2DS2-VASc schemes had the best c-statistics (0.717 and 0.724, respectively) for predicting the residual thromboembolic risk despite warfarin treatment and that other risk schemes had some limitations in this setting.55
Other Risk Factors, Risk Modifiers and the Role of Biomarkers
In addition to the risk factors included in current risk stratification schemes, several other risk factors have been recognised. These include a family history of stroke,60 the presence of chronic kidney disease, atrial high-rate episodes (AHRE) and atrial tachyarrhythmia (AT)/AF burden.61–65
Chronic kidney disease is a major cardiovascular risk factor, which is particularly common among elderly people.61 However, whether it does in fact independently raise the risk for ischaemic stroke is poorly understood. Among 13,535 adults with AF, chronic kidney disease raised the risk of thromboembolism in AF independently of other known risk factors.61 The presence of proteinuria has also been shown to reflect an increased thromboembolic risk in patients with normal or moderately impaired renal function.61
The duration of AF was previously not considered as a risk factor for stroke in AF patients due to the ambiguity of clinical symptoms.65 However, the diagnostic features of atrial-based pacemakers are now sufficiently advanced to supply information on the presence and duration of atrial arrhythmias.65 Three key studies have assessed the consequences related to these arrhythmias. In a study to evaluate the relationship between daily AT burden from implantable device diagnostics and stroke risk (The relationship between daily atrial tachyarrhythmia burden from implantable device diagnostics and stroke risk [TRENDS]), the overall rate of ischaemic events was low in this cohort of more than 2,400 patients (1.3 % per year). Patients with atrial rate >175 bpm, lasting >20 seconds, for <5.5 hours on a single day during a 30-day period experienced a clinical thromboembolism rate of 1.1 % per year, while the event rate among those with a high AT/AF burden was 2.4 % per year. After statistical adjustment for stroke risk factors and antithrombotic therapy, a high AT/AF burden <5.5 hours per day over 30 days was associated with a risk of thromboembolism similar to that of patients entirely free of AT/AF, while the risk doubled among patients with AT/AF during >5.5 hours per day. However, compared with zero burden, the difference in HR for the groups with low and high AT/AF burdens was not statistically significant.65
In preliminary results presented at the American Heart Association Scientific Sessions 2010 of the Asymptomatic atrial fibrillation and stroke evaluation in pacemaker patients and the atrial fibrillation reduction atrial pacing trial (ASSERT), episodes of device-detected atrial tachycardia greater than six minutes were found in 261 of 2,580 pacemaker patients, with hypertension but no history of AF over almost three years of follow-up. These arrhythmias were associated with a 2.5-fold increase in risk of ischaemic stroke and systemic embolism. Among the patients with a CHADS2 score of ≥2, device-detected ATs increased the absolute risk of stroke to 2.1 % per year. Among pacemaker patients without any prior history of atrial arrhythmias, 35 % of all strokes and systemic emboli were preceded by device-detected ATs.63,64
A study contributing further information comparing the duration of AF with the risk of stroke found that in patients without atrial arrhythmias, the risk of stroke was 1.2 % per year. If at least five minutes of atrial arrhythmias were detected, the event rate remained low at 1.7 % per year. If 24 hours or more of atrial arrhythmias were detected, the event rate was 4 % per year. However, by adding the patient’s CHADS2 score to the duration of atrial arrhythmia, groups at low risk (0.8 % per year) and at high risk (5 % per year) for thromboembolic events could be better identified.62 This is still an uncertain area, however there are several ongoing registries and trials that will report in the next several years and lead to a clarification.
Echocardiography is valuable to define the origin of AF and may add useful information in predicting the risk of stroke. Among high-risk patients, impaired left ventricular systolic function on transthoracic echocardiography, thrombus, dense spontaneous echo contrast or reduced velocity of blood flow in the left atrium appendage and complex atheromatous plaque in the thoracic aorta on transoesophageal echocardiography have been associated with increased thromboembolic risk.28
There has been increased interest in the role of biomarkers for improving stroke risk prediction models. C-reactive protein, D-dimer, fibrinogen, gamma glutamyl transferase, plasminogen-activator inhibitor type 1 and N-terminal pro-B-type natriuretic peptide (NT-proBNP) are some examples.66–71 Sub-study analyses of the 18,113- patient Randomisation evaluation of long-term anticoagulation therapy (RE-LY) trial,72 involving the new oral anticoagulant (OAC) dabigatran assessed the prognostic value of the biomarkers D-dimer and NT-proBNP for predicting cardiovascular events in patients with non-valvular AF.70,71
D-dimer concentration at initiation of treatment with warfarin or dabigatran was strongly associated with the risk of stroke, mortality and bleeding independently of CHADS2 risk factors and baseline anticoagulant treatment.70 In AF patients with risk factors for stroke, the NT-proBNP level was significantly raised compared with healthy subjects of similar age. A raised NT-proBNP level was associated with an elevated risk of stroke, major haemorrhage and death even after adjustment for the CHADS2 risk factors and study drug.71
The assessment of the contribution to improved risk prediction by considering of new risk factors and biomarkers has stimulated examination of optimal statistical techniques for these evaluations.73 However, the added value of these markers to current risk stratification schemes is yet to be determined.
Challenges and Unmet Needs in Antithrombotic Therapy for Stroke Prevention
Warfarin has long been established as the standard of care for prevention of stroke and thromboembolism in patients with AF. Several studies have evaluated the efficacy of antiplatelet therapy for the prevention of stroke in AF.74,75 In a meta-analysis of antithrombotic therapy to prevent stroke in patients with AF, antiplatelet therapy (eight trials, totalling 4,876 participants) reduced stroke by 22 % (CI, 6 to 35 %) compared to control. In contrast, adjusted-dose warfarin (six trials, totalling 2,900 participants) reduced stroke by 64 % (95 % CI, 49 to 74 %).74 Thus, the low efficacy of aspirin compared with warfarin is a severe limitation.
However, effective warfarin therapy depends on inter- and intra-individual variability in maintaining the international normalised ratio (INR), which is driven by a number of genetic and environmental factors.76,105–111 For effective thromboprophylaxis with warfarin, the target intensity of anticoagulation involves a balance between stroke prevention and avoidance of haemorrhage, particularly intracranial haemorrhage (see Figure 2).28,77–80 Maintaining the narrow therapeutic range of warfarin requires close and frequent monitoring of the patient’s INR, highlighting the benefit of monitoring in a designated anticoagulation clinic setting rather than community practice, although self-management can be effective in anticoagulation therapy.79,81
The greater the proportion of time spent in the therapeutic range of warfarin the more effective the anticoagulation. By contrast, increased time spent outside the therapeutic range is associated with an increased risk of mortality, ischaemic stroke or other thromboembolic event, myocardial infarction and major bleeding.10,74,82–89 Due to the risk of major bleeding, particularly intracranial haemorrhage, warfarin may not be recommended in patients at significant risk of bleeding.12,28 The risk of major bleeding is greater in elderly patients ≥80 years of age,90 who make up a large proportion of patients with AF.4,7
In antithrombotic therapy for stroke prevention in AF, the beneficial effect of antiplatelet therapy on ischaemic stroke appears to decrease with age compared with treatment with warfarin.12 In a meta-analysis, aspirin was associated with a lower rate of intracranial haemorrhage included in major bleeding event rates91 and extracranial haemorrhage compared with warfarin.74 These findings were not supported in a study comparing warfarin and aspirin treatment in an elderly population (mean age = 81.5 years), which observed similar rates of major bleeding, including intra- or extracranial haemorrhage, in the two treatment groups.92 Indeed, increasing age has a deleterious effect on the efficacy of aspirin, with little or no reduction in the risk of stroke in patients aged >75 years.93
The consequences of inadequate thromboprophylaxis were highlighted in a study examining the effect of pre-admission antithrombotic therapy on stroke severity in 1,938 patients with acute brain ischemia; 329 out of the 1,938 patients included had AF (17 %).94 Fewer severe strokes were seen in patients treated with warfarin to INR ≥2.0 (38 %) (p=0.01) compared to patients treated with warfarin to INR <2.0 (59 %), patients treated with antiplatelet therapy (55 %) or patients receiving no treatment (70 %). Moreover, adequate adjusted-dose warfarin therapy was associated with better rates of survival (p=0.004).94
Despite the demonstrated efficacy of warfarin for stroke prevention in AF, the perceived limitations of current therapeutic options have resulted in substantial underuse of antithrombotic therapy (see Table 2).95–98,107 This underuse has also been shown to result in increased risk for the combined endpoint of cardiovascular death, thromboembolism, or major bleeding.98 Underuse of vitamin K agonists (VKAs) may be due to a number of physician and patient related factors (see Table 3).14,17,18,99–101
Current Guidelines on Stroke Prevention
Risk factors can be incorporated either formally in a scoring system as has been suggested by the European Society of Cardiology (ESC) or in an informal fashion as recommended in the 2006 American College of Cardiology Foundation/American Heart Association/ESC guidelines.12,28,78 The current ESC guidelines have been framed with an emphasis on a risk factor-based approach and less emphasis on the use of the artificial low-/moderate-/high-risk strata due to the latter’s poor predictive value. The concept of stroke risk as a continuum has been adopted within the latest ESC guidelines, thus emphasising the cumulative effect of various risk factors.12
In relation to stroke prevention, the ESC recommendations for assessing antithrombotic therapy requirements utilise this risk factor-based approach. The CHADS2 scoring system is recommended for an initial rapid assessment of risk in non-valvular AF. For a more comprehensive assessment of risk, the CHA2DS2- VASc scheme, which considers ‘major’ and ‘clinically relevant non-major’ stroke risk factors, should be used.12 The antithrombotic strategy should be based on CHADS2/CHA2DS2-VASc scores (see Figure 3 and Table 4).12
The ESC guidelines also suggest the use of a new scoring system to evaluate the risk of bleeding, HAS-BLED (uncontrolled hypertension, abnormal renal/liver function, stroke, bleeding history or predisposition, labile INR, elderly e.g. >65 years, drugs/alcohol concomitantly) (see Table 5).12 Each factor scores one point, with a maximum score of nine. A score of three or more suggests high risk for bleeding. The predictive accuracy for one-year bleeding risk was consistent in different subgroups. This score is simple and provides a potentially practical tool in clinical decision-making.102 A regular review of those patients with a HAS-BLED score ≥3 upon initiation of antithrombotic therapy is highly recommended.12
Clinicians should consider whether risk stratification to identify patients at risk of stroke will be the best approach in the future. It may be more appropriate, as recommended in the ESC guidelines, to consider which patients should not be anticoagulated and to assume that all AF patients should receive treatment unless they are younger than 65 years with no heart disease. Certainly, if the CHA2DS2-VASc criteria were adopted, most people with AF would be offered anticoagulation.
It is not certain whether all women with AF should be deemed to be at least at intermediate risk and therefore be considered for anticoagulation. Gender has an inconsistent association with stroke, some observational studies have found higher rates of AF related stroke in women,103 while others have not. Although female gender may not prove powerful enough in itself to justify anticoagulation, when added to other risk factors such as hypertension or age 65 to 74 it raises the thromboembolic risk and substantially strengthens the case for anticoagulation.
CHA2DS2-VASc is certainly better than CHADS2 at identifying those individuals who are at very low risk. This is clinically important, particularly since many individuals with a CHADS2 score of zero or one are, in fact, at relatively high risk of thromboembolism (consider for example, a patient with widespread vascular disease aged 65 to 74 years).
Although the risk factors for stroke in AF are well recognised and there have been strong treatment recommendations for some time, neither physicians nor patients appear able to accept the guidance. This may be attributable to the problems associated with warfarin and has resulted in many patients at the highest risk for stroke, the elderly, not receiving anticoagulation therapy.
The decision-making of clinicians must consider the strong evidence base for antithrombotic prophylaxis, whilst at the same time considering the individual patient, comorbidities, patient values and preferences.
The development of new treatments for stroke prevention in AF might potentially address unmet needs in current therapeutic options and could persuade clinicians to lower the threshold at which they prescribe anticoagulants. Indeed, new OACs probably have demonstrated better safety profiles and may improve the ease of use compared with VKAs.104 Furthermore, OACs have also shown to have less drug and food interactions, and less haemorrhagic complications, particularly intracranial bleeding.104
The recent or imminent introduction into clinical practice of dabigatran, an example of a new OAC, has resulted in a shift of the guidelines towards identifying more patients at potential risk for stroke of less than 2 % per year, compared to the 3 % per year risk that was necessary with older anticoagulants. New OACs could affect the general status of anticoagulation and how clinicians decide who should be anticoagulated.
In this developing situation, where new therapeutic options for antithrombotic prophylaxis are being tested and validated as tools for prevention of stroke, the future challenge for physicians will be to make appropriate and timely clinical decisions based on the increasingly strong evidence base, after assessment of the individual patient’s clinical profile.