Statistics from Altmetric.com
To understand the (high) prevalence of atrial fibrillation (AF) in the overall population and to identify subgroups at higher risk.
To understand the diagnostic procedures available to diagnose AF in patients deemed to be at higher risk, such as post-stroke.
To recognise and overcome barriers to anticoagulation.
To identify patients at higher risk of thromboembolic and haemorrhagic complications.
To understand indications for antithrombotic therapy in specific groups of patients with AF.
One in three of us will be diagnosed with atrial fibrillation (AF) at some point in our lives.1 Over 1.2 million individuals in the UK have been diagnosed with AF, and thousands still remain undiagnosed.2 3 Due to the ageing population, the number of individuals with AF in Europe will double in the next 50 years,3 contrasting with a fall in the incidence of myocardial infarction and stroke over recent decades.4 This shift in cardiovascular epidemiology is important as AF associates with a myriad of cardiovascular and non-cardiovascular complications (figure 1), and contributes to a 1.5-fold to 2.0-fold increase in all-cause mortality.5
The risk of ischaemic stroke in patients with AF is three to five times higher,6 with AF found in up to 30% of patients admitted with ischaemic stroke, and associated with a higher risk of fatal outcome.7 The risk of AF-related stroke can be mitigated through anticoagulation, with a 66% risk reduction with vitamin K antagonists (VKAs) and at least similar effectiveness with non-VKA.8 Stroke rates in patients with adequately anticoagulated AF are similar to those of patients without documented AF,9 with only 1 out of 10 patients with anticoagulated AF dying from stroke or bleeding.10 This article will expand on the topic of anticoagulation in patients with AF.
Comprehensive AF screening is key to prevent AF-related morbidity and mortality
Despite the bulk of evidence supporting anticoagulation use for stroke prevention, many patients with AF are not on this therapy. Two important reasons for this are (i) asymptomatic AF occurs more frequently than symptomatic events,11 and (2) screening for AF in patients at risk is often insufficient.
Detection of subclinical AF is possible through a variety of strategies and tools, including manual pulse palpation in those with suspicious symptoms,12 external surface monitoring with intermittent 12-lead ECG, Holter monitors, portable ECG recorders or handheld single-ECG devices with automated algorithms, adhesive patch electrocardiographic monitors and smartwatches. Intermittent short electrocardiographic recordings repeated over a longer-term period with mobile phones and smartwatches have higher sensitivity for AF detection compared with single-time point measurement.13 Implantable cardiac monitors are a reasonable option in patients at very high risk of AF but in whom all other options failed to provide conclusive results. Cardiac electronic devices (CIEDs) (pacemakers and defibrillators) can facilitate AF detection, but AF overdiagnosis can also occur, as CIED-detected atrial high-rate episodes can sometimes be caused by oversensing or other atrial arrhythmias. Close inspection of the stored electrogram is mandatory, particularly if anticoagulation prescription is expected.
AF has been recognised in 20%–30% of patients with a cryptogenic stroke when monitoring is long enough.14 In this case, AF is probably causal, but often there is no temporal relationship between the detection of AF and the occurrence of stroke. Still, even the unrelated detection of AF post-stroke predicts recurrent thromboembolism. Increased AF yield from monitoring is seen not only in patients with history of ischaemic stroke of unidentified cause but also elderly patients with additional risk factors for AF, such as hypertension, diabetes mellitus or heart failure, and in those with markers of atrial myopathy, such as frequent atrial ectopy, high natriuretic peptides, dilated left atrium or significant atrial scar. Presently, it is uncertain whether AF screening in the general population is cost-effective or improves health outcomes. Resources could be better invested in high-risk populations, and patients with stroke of undetermined source could be a potential candidate group. However, absence of definite data on the duration of AF requiring anticoagulation, the lack of proximate temporal relationship between AF and stroke, and the unclear benefit of anticoagulation in that scenario15 16 lead to questions about that strategy. American guidelines state that prolonged rhythm monitoring of approximately 30 days is reasonable within 6 months of a cryptogenic stroke,17 whereas European Society of Cardiology (ESC) guidelines recommend continuous ECG monitoring for ≥72 hours, and possibly non-invasive monitors or implantable loop recorders for longer periods.18
In summary, although the optimal AF monitoring device should ideally be inexpensive and simple to use, the ability to offer long-term monitoring and immediate feedback are of paramount importance. Particularly in patients with cryptogenic stroke, screening should be performed for as long as possible. In resource-limited regions where prolonged ECG monitoring may not be possible, repeat single-time point ECG is probably feasible and should be recommended. Figure 2 suggests a protocol for AF screening for patients with ischaemic stroke of unidentified cause.
Overcoming barriers to anticoagulation in patients with AF
Up to the early 2000s, less than 70% of high-risk patients with AF were receiving adequate oral anticoagulation with VKA in most parts of the world.19 The introduction of direct oral anticoagulants (DOACs) associated with improved rates of anticoagulation use, but up to 40%–50% of patients with AF may still not be on this medication (figure 3).20 There is a significant variability in DOAC prescription across different parts of the world, even when limiting this assessment to resource-rich countries,20 21 and the rate of anticoagulation prescription remains far from optimal, particularly for those at highest risk. An increase in the use of DOACs in younger patients with AF at lower risk of stroke has not been paralleled by a similar increase in more elderly patients at higher risk. Almost a third of patients with AF presenting with stroke were naïve to any antithrombotic therapy prior to the event.22
Persisting barriers to appropriate antithrombotic therapy include the difficulty in diagnosing AF, inadequate risk stratification, the advanced age of the patient and a high perceived risk of falls and haemorrhage. The requirement for regular monitoring, frequent dose adjustments and dietary restrictions with VKA is less of an issue nowadays as DOACs become preferred agents. When it comes to initiating anticoagulation in elderly patients, or those at presumably high risk of falls, many physicians still hesitate. While this is understandable to some extent, as falls may lead to subdural haematomas, intracerebral haemorrhage and hip fracture bleeding, which can be life-threatening, withholding anticoagulation in elderly patients leads to increased morbidity and mortality. Warfarin is superior to aspirin in reducing the risk of a composite endpoint of fatal or disabling stroke, intracranial haemorrhage or arterial embolism in elderly or very elderly patients,23 with antiplatelet treatment becoming progressively less effective in reducing stroke risk with advanced age.24 Apixaban reduces the risk of stroke by >50% compared with aspirin among patients with AF unsuitable for VKAs, while bleeding complications occur at a similar rate. The superiority of apixaban is most prominent among elderly patients.25 For every 1000 patients aged ≥75 treated with apixaban rather than aspirin, 41 strokes or systemic embolisms are prevented at the expense of four extra major bleeds. Aspirin is no longer recommended for cardioembolic stroke prevention.
The perceived risk of falls and bleeding and concerns about impaired cognition, limited compliance (such as in those with dementia) or advanced multisystem atrophy explain why anticoagulation is often withheld in the elderly population. While elderly patients with multiple risk factors for falling, such as sedative use, cognitive impairment and gait disturbance should have closer follow-up in clinic, there is evidence that (1) they are still at (very) low risk of anticoagulant-related major bleeding26 and (2) the benefits of anticoagulation outweigh its risks even in those who eventually fall.27 Furthermore, bleeding risk perceptions often do not correlate with the actual risk of bleeding, and accurately estimating bleeding risk is difficult as most risk factors overlap with those for stroke (figure 4). Although at least 2% of anticoagulated patients with AF experience bleeding events every year, ischaemic strokes are far more likely.24
To summarise, elderly patients in general should not be denied anticoagulation, or prescribed antiplatelet agents instead, on the assumption that they are at increased risk of falls and bleeding. Except for patients with active bleeding or severely impaired cognition and drug compliance, strong consideration should be given to a DOAC.
Anticoagulants for which patients with AF?
Even among non-valvular patients with AF at the highest risk category of the CHA2DS2-VASc score, nearly 80% will not sustain a stroke within a year in the absence of anticoagulation.28 In the four largest AF anticoagulation trials, the annual risk of bleeding-related mortality was 2%–3%.29 This suggests that some patients may not need anticoagulation and could be unnecessarily exposed to a low risk of bleeding-related fatality if put on this therapy. Major bleeding represents 6% of the causes of death in patients with AF on anticoagulation.10 This is a strong argument against the ‘one size fits all’ strategy of universal anticoagulation. The CHA2DS2VASc score performs well at identifying patients who should not receive anticoagulants, with a high negative predictive value for the lower-risk stratum. Conversely, the number of embolic events in non-anticoagulated patients outweighs that of severe bleeding episodes, and data confirm the unequivocal benefit of anticoagulation in stroke prevention. It is therefore essential that all patients are stratified for their thromboembolic and bleeding risks, as the combination of both will help the clinician and patient weight the potential benefits and risk of anticoagulation, and target correctable risk factors.
The 2014 National Institute for Health and Care Excellence (NICE) guidelines on AF recommend the use of the CHA2DS2-VASc score for assessing thromboembolic risk.12 This score was derived from data from the Euro Heart Survey on Atrial Fibrillation,30 and its simplicity and capability of providing a broad estimate of annual stroke risk have led to its wide usage. However, it has modest discriminative performance and may be overinclusive, leading to unnecessary anticoagulation of a few patients who may not require it at that time in their lives. NICE guidelines discourage using anticoagulants in men with a CHA2DS2-VASc score of 0 and women with a score of 1. Anticoagulation can be considered in men with a score of 1 and should be offered to individuals with a score of 2 or above. To avoid the cumbersome practice of selecting different CHA2DS2-VASc thresholds for men and women when recommending anticoagulation prescription, Australian clinical guidelines suggested the use of the sexless CHA2DS2-VA score.31 Patients with moderate to severe mitral stenosis, hypertrophic cardiomyopathy or mechanical heart valves represent subsets of high-risk patients requiring mandatory anticoagulation regardless of the CHA2DS2-VASc score.
Over the years, other risk scores like ATRIA,32 R-CHA2DS2VASc,33 ABC,34 QStroke35 and GARFIELD-AF36 (figure 4) were tested against CHA2DS2VASc but have yet failed to replace it. The potential improvement in discrimination is counterbalanced by the complexity of these tools, sometimes requiring calculators, and the need for education and training as well as validation studies.
Estimation of the risk of bleeding should also constitute a routine part of the management of patients with AF. NICE recommends using the HAS-BLED score for this purpose.37 This score allows not only an estimation of the bleeding risk but also raised awareness for correctable risk factors (eg, uncontrolled hypertension, labile international normalised ratio, concomitant use of antiplatelet agents and non-steroidal anti-inflammatory drugs and harmful alcohol consumption). As stroke events exceed bleeding episodes even in patients at high bleeding risk, anticoagulation should not be withheld in patients with high HAS-BLED score; rather, these patients should be reviewed regularly and modifiable bleeding risk factors addressed. As for thromboembolic risk stratification, multiple schemes have been developed (eg, HEMORR2HAGES, ATRIA, ABC, ORBIT and GARFIELD-AF)18 (figure 4), and a NICE appraisal on this matter and the role of patient involvement in shared-decision making should be made available soon.
Available anticoagulants: which one to choose?
While antiplatelet treatment is no longer recommended for this purpose, VKAs have been shown to prevent two of three embolic strokes and reduce all-cause mortality. However, their use is limited by its slow-onset of action, multiple drug and food interactions, unpredictable pharmacokinetic properties and need for regular and sometimes intensive monitoring. The DOACs have more rapid onset of action, do not usually require bridging with parenteral anticoagulants, have fewer drug or food interactions and do not require routine monitoring. Two classes of DOACs have been developed for stroke prevention in non-valvular AF: direct factor Xa inhibitors (rivaroxaban, apixaban and edoxaban) and direct thrombin inhibitors (dabigatran). A meta-analysis of all four agents showed reduced risk of stroke, intracranial bleeding and all-cause mortality, but increased risk of gastrointestinal bleeding compared with warfarin.8 There is no evidence to suggest that any of the DOACs is superior to others, although small differences in outcomes between different agents have been suggested by large observational studies.38
DOACs are not recommended for patients with prosthetic heart valves or valvular AF (eg, rheumatic mitral stenosis), in which case VKAs are recommended. The difficulty in achieving a time in therapeutic range (TTR) of >70% with VKAs associates with increased stroke and bleeding risk and these patients should be changed to a DOAC (provided there is no contraindication). For patients well controlled on a VKA, as defined by TTR of >70% and lack of thromboembolic or bleeding events, it is reasonable to continue VKA treatment. However, changing to a DOAC should be considered in case of complications or patient preference. For patients naïve to any antithrombotic therapy, the decision on which anticoagulant to choose must take several aspects into consideration, including patient weight (significantly underweight or overweight patients were less represented in DOAC studies), likelihood of adherence (crucial, particularly for DOAC patients given the shorter half-life), history of renal or hepatic impairment, limitations in gut absorption (eg, with previous gastric bypass), previous haemorrhagic events or recurrent stroke despite good drug adherence, intended pregnancy or breastfeeding (no data for DOACs), concerns with compliance and medication cost. Physicians should take an individualised approach taking the patient’s preferences into consideration, as well as underlying comorbidities and estimated risks of stroke and bleeding. Importantly, inappropriate underdosing of DOACs is common and physicians should be aware of the importance of correct dosing of these drugs to guarantee its safety and effectiveness. The off-label underdosing of DOACs associates with increased risk of stroke with no significant difference in major bleeding.39 An algorithm for helping physicians select the most appropriate anticoagulant according to the patient’s characteristics is suggested in figure 5.
Anticoagulation in specific groups of patients with AF
This chapter focus briefly on the management of anticoagulation in a few specific contexts, but for more detailed recommendations, interested readers should look elsewhere.40
The majority of cardiovascular interventions can be performed safely on continued anticoagulation, and heparin bridging is generally not recommended as it may result in increased risk of bleeding. For those undergoing endovascular procedures or catheter ablation, consideration should be given to uninterrupted anticoagulation,41 although temporary, short-term, discontinuation of a DOAC in patients at low stroke risk is also a reasonable strategy. AF ablation can be performed safely on uninterrupted or minimally-interrupted DOAC,42 and heparin bridging is not recommended for these patients as it leads to increased haemorrhagic risk.43 Continued warfarin treatment at the time of device implantation markedly reduces the incidence of clinically significant device-pocket haematoma compared with bridging therapy with heparin,44 while continued DOAC therapy is as safe as interrupted therapy with regard to clinically significant pocket haematoma.45 Patients with AF undergoing percutaneous coronary interventions should continue their anticoagulation during and after stenting, and a short period of triple therapy (anticoagulation, aspirin and clopidogrel) is recommended, followed by a period of dual therapy (preferably with anticoagulation and clopidogrel). The use of prasugrel or ticagrelor as part of triple therapy should be avoided whenever possible.
The decision to withhold anticoagulation should take into consideration not only the prior assessment of thromboembolic and bleeding risk but also the potential consequences for the patient in case of bleeding (eg, intracranial, cardiac or neuraxial surgical interventions are of particular concern). If a decision is made to discontinue the anticoagulant due to high bleeding risk, bridging should be limited to those at high or very high thromboembolic risk. Patients with mechanical heart valves, particularly those with a mitral prosthesis, should not have their anticoagulation withheld for procedures which are not considered high risk for bleeding events. However, when withholding anticoagulation is absolutely essential in these patients, heparin bridging is necessary. Patients undergoing low bleeding risk interventions should continue their anticoagulation (this is the case of most dental and cutaneous interventions). Elective surgery should be delayed if possible in patients who had ischaemic stroke the preceding month, and also patients with AF who were not adequately anticoagulated.
Asymptomatic AF detected by CIEDs
Approximately one-third of patients with CIED, CHA2DS2-VASc≥2 and no known history of AF are shown to develop subclinical AF episodes lasting >6 hours.46 This increases to almost 50% when considering episodes of >6 min.46 Although these patients are at increased risk of stroke,47 the risk associated with subclinical, short atrial high-rate episodes is less than what might be expected for clinically diagnosed paroxysmal AF,48 and a temporal relationship between these episodes and stroke is seen in only a minority of cases. There is currently no solid evidence from randomised studies that implanting a CIED to detect AF and initiating oral anticoagulation in those in whom AF is detected is beneficial. However, in the Asymptomatic Atrial Fibrillation and Stroke Evaluation in Pacemaker Patients and the Atrial Fibrillation Reduction Atrial Pacing trial, device-detected asymptomatic AF episodes of >24 hours associated with a stroke risk comparable to clinical AF,49 while in the TRENDS study, an atrial arrhythmia burden of ≥5.5 hours, but not less, appeared to double thromboembolic risk.48 It is reasonable to consider oral anticoagulation, preferably in the form of a DOAC, in patients with subclinical AF lasting longer than a few minutes and who would otherwise be considered for this therapy in the presence of clinical AF, provided the patient is not at high risk of bleeding. As half of all patients with subclinical AF transition to higher AF burden thresholds during follow-up,50 prompt detection of atrial high-rate episodes or subclinical AF by device diagnostics may allow closer follow-up and faster detection of progression to a longer AF burden, particularly if remote monitoring is used, in which case the benefit of anticoagulation should be less controversial.
Anticoagulation in patients with renal dysfunction or on dialysis
Renal dysfunction increases the risk of stroke and bleeding. While the benefit of oral anticoagulation in patients with mild to moderate renal dysfunction is clear,51 patients with end-stage renal disease on dialysis were systematically excluded from large cardiovascular-focused randomised trials. The superior bleeding risk profile of DOAC may seem attractive for the dialysis population, but anticoagulation may not be beneficial for these patients.52 Both dabigatran and rivaroxaban may be associated with a higher risk of bleeding than apixaban,52 although only dabigatran is mostly excreted via the renal route. Both agents are however safe and effective in patients with mild or moderate renal impairment. Apixaban is mainly excreted via the biliary route and has shown a favourable risk profile across all levels of renal function, including end-stage dysfunction on dialysis,52 with patients with impaired renal function experiencing the greatest reduction in major bleeding compared with warfarin.53 The recently presented Renal Hemodialysis Patients Allocated Apixaban versus Warfarin in Atrial Fibrillation trial suggested that apixaban is as safe as warfarin in dialysis patients, although the trial was interrupted prematurely due to loss of funding.54 In a very large cohort with end-stage renal disease and AF, apixaban use was associated with lower risk of major bleeding compared with warfarin, with a standard 5 mg two times per day dose also associating with reductions in thromboembolic and mortality risk.55 These findings suggest that apixaban may be a more favourable option in patients with renal impairment, including its severe stage. Still, the evidence for this is limited, given the lack of randomised data involving patients on dialysis. A left atrial appendage occlusion device is a reasonable option to reduce stroke risk in patients with AF who have a contraindication to long-term anticoagulation, but these devices have not been tested in patients on dialysis. Table 1 provides useful data on how to adjust DOAC dosage in patients with renal impairment.
Anticoagulation in pregnancy
Anticoagulation can cause miscarriage and haemorrhagic complications, including postpartum bleeding and retroplacental haemorrhage, but it is recommended for pregnant women with AF and a CHA2DS2-VASc score of ≥2 to prevent cardioembolic stroke.18 DOACs are not recommended during pregnancy as their safety has not been tested. Conversely, although VKAs can be used in the second and third trimesters, they can cross the placenta and increase the risk of foetal loss, as well as result in embryopathy in the first trimester in a dose-dependent manner. While low-dose VKA (eg, warfarin at less than 5 mg/day) is associated with low risk of teratogenicity, higher dosages can result in limb defects or nasal hypoplasia in up to 10% of cases. Thus, VKAs should ideally be replaced by continuous dose-adjusted unfractionated heparin (UFH) or subcutaneous low-molecular-weight heparin (LMWH) in the first trimester and in the 2–4 weeks prior to delivery to avoid foetal bleeding and maternal haemorrhage during labour. Heparin compounds are usually safe during pregnancy, although long-term use of UFH may be associated with maternal osteoporosis and thrombocytopenia. More frequent administration of UFH or LMWH is typically required due to the increase in plasma volume and glomerular filtration rate and placenta degradation heparinase. To reduce the risk of bleeding during labour, heparin compounds should be discontinued 12 hours before planned induction.
Managing bleeding in anticoagulated patients
Anticoagulation is associated with increased haemorrhagic risk. Although DOACs in general seem less likely to cause bleeding compared with warfarin, particularly intracranial, high-dose DOACs have been shown to be associated with increased risk of gastrointestinal bleeding.8 Fortunately, the majority of bleeding events are not life-threatening and can be managed conservatively. Figure 6 provides general advice on how to deal with bleeding in AF anticoagulated patients. This implies a prompt assessment of bleed severity and haemodynamic instability. Knowing the time of ingestion of last dose of anticoagulant and measuring anticoagulant activity are also crucial. Local measures such as manual compression whenever possible, fluid replacement and transfusion of blood products, direct procedural treatment (eg, endoscopy for gastrointestinal bleeding) and anticoagulation reversal are all possibly indicated, depending on the severity of the bleed.
Anticoagulation is an effective treatment for stroke prevention in patients with AF. However, the rate of anticoagulation prescription is still suboptimal even in the DOAC era, leading to increased morbidity and mortality. Reasons for this include the advanced age of the patient, an inaccurate perception of the risk of falls and bleeding, and the failure to diagnose AF when it is occurring in a paroxysmal fashion. AF screening for as long as possible is crucial in patients who sustained an ischaemic stroke of unidentified cause, as AF is eventually diagnosed in a significant percentage of these patients.
In patients with a history of ischaemic stroke of unidentified cause, atrial fibrillation (AF) screening should be performed for as long as possible. Twenty-four-hour ECG monitoring is insufficient for this purpose.
In elderly patients with AF, anticoagulation (preferably in the form of a direct oral anticoagulant) should not be withheld or replaced by an antiplatelet agent on the presumption of a high risk of falls and bleeding. Even in patients who fall, anticoagulation reduces stroke-related morbidity and mortality and the overall risk of death.
The risk of ischaemic stroke in patients with AF who are not on anticoagulation far exceeds the risk of severe haemorrhagic complications caused by this treatment.
Anticoagulation can be considered in men with a CHA2DS2VASc score of 1 and should be offered to individuals with a score ≥2, regardless of the HAS-BLED score.
Decisions on periprocedural management of antithrombotic therapy should be based on the patient’s clinical status and inherent risk of thromboembolism and bleeding, but also the procedure-related haemorrhagic risk. Uninterrupted oral anticoagulation for low-risk procedures appears to be safe and is gaining strong evidence support.
CME credits for Education in Heart
Education in Heart articles are accredited for CME by various providers. To answer the accompanying multiple choice questions (MCQs) and obtain your credits, click on the ‘Take the Test’ link on the online version of the article.The MCQs are hosted on BMJ Learning. All users must complete a one-time registration on BMJ Learning and subsequently log in on every visit using their username and password to access modules and their CME record. Accreditationis only valid for 2 years from the date of publication. Printable CME certificatesare available to users that achieve the minimum pass mark.
Contributors Both authors contributed to literature search and manuscript drafting.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, conduct, reporting or dissemination plans of this research.
Patient consent for publication Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement There are no data in this work
Author note References which include a * are considered to be key references
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.