In order to most effectively treat people with atrial fibrillation (AF), clinicians need to understand key issues, including the indications, and evidence for rhythm control versus ventricular rate control and how to reduce the risk of ischaemic stroke. This review aims to outline the current challenges posed by AF in older adults and to provide a practical guide to the current evidence regarding its management.

The prevalence of AF in the developed world is estimated at 1.5-2% of the general population.  Prevalence increases with age and it is more common in males. Average age of patients with the condition is steadily increasing and mean age is now between 75 and 85 years.1 Prevalence from age 50 years on doubles every 10 years.2 The ATRIA study found the prevalence was 0.1% in females below 55 years of age, while in those above 85 years old, it was 9.1%; for males, figures were 0.2% and 11.0%, respectively.3

In addition to intrinsic cardiac causes such as valve disease and congestive heart failure, risk factors for cardiovascular disease also predispose to AF.4

AF is by no means a benign condition. It is associated with significant morbidity and mortality, including a fivefold increase in stroke and a threefold increase in cardiac failure.1 Hospitalisation of patients with AF is also common. Follow up data from two large trials has shown it to be an independent predictor for mortality.5, 6 It therefore carries a significant social and economic burden. Potential complications of AF are outlined below.

Thromboembolism in AF

Mechanism of thromboembolism

Thromboembolism is a major cause of morbidity in AF. There are a number of suggested reasons for the increased incidence of thromboembolism in AF including: stasis of blood caused by dilated left atrium/left atrial appendage and a hypercoagulable state in AF (elevated fibrin D-dimers and vWf).


Risk stratification

There are a number of tools for assessing the risk of thromboembolism in AF. The most commonly used are CHADS2 and CHA2DS2 VASc models. The main advantage of the CHADS2 score is its simplicity.7 Patients with CHADS2 score of ≥2 should be recommended for anticoagulation. However, a low CHADS2 score (eg. <2) does not reliably identify AF patients who are "truly low-risk."8,9

Therefore, patients with a low CHADS2 score can be evaluated using the CHA2DS2-VASc score; the accumulated evidence suggests that CHA2DS2-VASc is better at identifying "truly low-risk" patients with AF10,11 and is as good as, and possibly better than CHADS2 in identifying patients who develop stroke and thromboembolism.12 Use of the CHA2DS2-VASc model will lead to recommendation for antithrombotic therapy in a higher percentage of at risk individuals.13 Maximum score is 9 since age may contribute 0, 1, or 2 points.


Prediction of stroke with CHADS2 and CHA2DS2-VASc

These scoring systems are useful tools in predicting the risk of stroke in a population. They are therefore useful in helping the clinician to choose the appropriate treatment for the patient.

Anti-thrombotic therapy in AF

There have been a large number of trials on the prevention of thromboembolism in AF, the major findings relevant to clinical practice are outlined below:


Anticoagulation with Vitamin K antagonists (VKA)
ie. warfarin versus control

VKA treatment should be considered for all patients with AF with high CHADS2 or CHA2DS2-VASc score as mentioned above, provided there are no contraindications. The risk-benefit ratio and an appreciation of the patient's preferences should be taken into account. In a meta-analysis, the relative risk reduction (RRR) with VKA was highly significant and amounted to 64% (absolute annual risk reduction in all strokes of 2.7%).14 This reduction was similar for both primary and secondary prevention and for both disabling and non-disabling strokes. All-cause mortality was significantly reduced (26%) by adjusted-dose VKA versus control. The risk of intracranial haemorrhage was small. Cohort studies suggest a 2-fold increase in stroke risk at INR 1.5-2.0, suggesting time in therapeutic range is very important. Therefore INR should be maintained between 2 and 3.  Even in clinical trials, patients stay within the intended INR range of 2.0-3.0 for only 60-65% of the time on average.


Antiplatelets versus control

In seven large trials aspirin was compared with placebo or no treatment. Treatment with aspirin was associated with a non-significant 19% reduction in the incidence of stroke.13 There was an absolute risk reduction of 0.8% per year for primary prevention trials and 2.5% per year for secondary prevention. In a meta-analysis, RR reduction with aspirin versus placebo or control was 22% (95% CI 6-35). However, it is broadly similar to that seen when aspirin is given to vascular disease subjects not in AF. The modest benefit seen for aspirin in AF is likely to be related to its effects on vascular disease rather than being AF specific.

Aspirin versus Vitamin K antagonists

VKAs are significantly superior to aspirin in preventing stroke in AF. Studies showed a RR reduction of 39% with no difference in the risk of major haemorrhage between warfarin and aspirin.13 It is therefore important that warfarin is strongly considered in all patients with AF and a CHA2DS2-VASc score of ≥2.


Combination of aspirin and clopidogrel versus Vitamin K antagonist

In the ACTIVE W trial, anticoagulation therapy was superior to the combination of clopidogrel plus aspirin (RR reduction 40%; 95% CI 18-56), with no difference in bleeding events between treatment arms.15 Dual antiplatelets should therefore not be routinely used in those with AF. However, dual therapy with aspirin and clopidogrel may be reasonable for patients who require it for other reasons, such as those with coronary stent placement.


Combinations of VKA (INR 2.0-3.0) with antiplatelet

Combinations of VKA (INR 2.0-3.0) with antiplatelet therapy have been studied, but no beneficial effect on ischaemic stroke or vascular events were seen, while more bleeding was evident.16 Thus, in patients with AF who sustain an ischaemic stroke despite adjusted dose VKA (INR 2.0-3.0), raising the intensity of anticoagulation to a higher INR range of 3.0-3.5 may be considered, rather than adding an antiplatelet agent, given that an appreciable risk in major bleeding starts at INRs >3.5.


Novel oral anticoagulants (NOACs)

When anticoagulation is indicated, warfarin offers excellent and cost effective stroke prevention with surprisingly low rates of significant bleeding. However, fears regarding bleeding, the need for regular blood tests and dose adjustments, and potential for multiple drug interactions lead to high rates of discontinuation. It is estimated that only half of warfarin eligible patients receive anticoagulation.17 In recent years a number of alternative drugs have become available with potential advantages over warfarin. Dabigatran (a direct thrombin inhibitor), rivaroxiban and apixaban (both factor Xa inhibitors) have all shown non- inferiority to warfarin in the prevention of stroke and systemic emboli. Their advantages are convenience of use, predictable anticoagulant effects, low propensity for food and drug interactions,18 and lower rates of intracranial bleeding than with warfarin. They do not require monitoring of blood clotting, therefore negating the need for frequent blood tests. A current concern of the NOACs is the lack of a specific reversal agent in the event of major bleeding although they all have, relative to warfarin, a short half-life.


Direct thrombin inhibitors versus VKA

In the RE-LY study, dabigatran 110mg b.i.d. was non-inferior to VKA for the prevention of stroke and systemic embolism with lower rates of major bleeding, whilst dabigatran 150mg b.i.d. was associated with lower rates of stroke and systemic embolism with similar rates of major haemorrhage, compared with VKA.19 If a patient has an increased risk of bleeding, dabigatran 110mg bd may be considered, in view of a similar efficacy in the prevention of stroke and systemic embolism (but lower rates of intracranial haemorrhage and of major bleeding compared with VKA).

Factor Xa inhibitors versus VKA

In the ROCKET AF study, rivaroxaban was non-inferior to warfarin for the prevention of stroke or systemic embolism.20 There was no significant difference in the risk of major bleeding, although intracranial and fatal bleeding occurred less frequently in the rivaroxaban group.

In the ARISTOTLE study, apixaban was superior to warfarin in preventing stroke or systemic embolism, caused less bleeding, and resulted in lower mortality.21

Percutaneous left atrial appendage closure

The PROTECT AF study showed that percutaneous left atrial appendage (LAA) closure with a filter device (Watchman) was non-inferior to warfarin for stroke prevention in patients with AF. Although there was a higher rate of adverse safety events in the intervention group than in the control group, events were mainly a result of periprocedural complications.22 Closure of the LAA might provide an alternative strategy to chronic warfarin therapy for stroke prophylaxis in patients with non-valvular atrial fibrillation. Current ESC guidance recommends that it be considered in those with a high stroke risk and contraindications for long-term anticoagulation.

Bleeding risk with anticoagulation

Major bleeding risk with aspirin is similar to that with VKA, especially in elderly individuals.23 The fear of falls may be overstated, as many falls are not associated with intracranial haemorrhage to outweigh the benefit of OAC in stroke prevention. Given the above, and the clear superiority of OACs versus aspirin in stroke prevention, OACs should be strongly considered in all patients with AF.

The EuroHeart Survey helped to develop a simple bleeding risk score, HAS-BLED, which offers useful predictive capacity for identifying higher risk patients.24 A score of ≥3 indicates "high risk". The HAS-BLED score can be used as a measure of bleeding risk, to identify modifiable bleeding risk factors and to indicate patients in whom regular review is needed after OACs/antiplatelets are initiated. It should not be used on its own to exclude patients from OAC therapy.

Anticoagulation following a stroke or TIA

It is likely that clinicians will see more and more patients on oral anticoagulation for AF. It is therefore important to be aware of current guidance regarding when to commence anticoagulation post TIA/stroke, and what to do if a patient has an event whilst on oral anticoagulants. The following section outlines key practical points.

Up to 6% of ischaemic stroke patients will sustain a symptomatic haemorrhagic transformation and more will have asymptomatic bleeding. The risk of haemorrhagic transformation is greater with larger infarcts and the risks of starting or continuing anticoagulation immediately after a significant infarct are high. Recent RCP stroke guidance recommends the following:

•     TIA: Oral anticoagulation can be commenced immediately, as long as brain imaging has ruled out haemorrhage

•     Disabling Ischaemic Stroke: Anticoagulation should be delayed/stopped for at least 14 days from stroke onset.  Aspirin 300mg should be used until this time 

•     Minor/Non Disabling Ischaemic Stroke:  It may be appropriate to start anticoagulation earlier as the risk of haemorrhagic transformation is low.  This is at the discretion of the physician. It should be started within 14 days and aspirin used to bridge the gap

•     Recurrent TIA/Ishaemic stroke: Antithrombotic treatment should be the same as for those who have had a single event

•     Disabling stroke in those already on anticoagulation with prosthetic valves: In people who are at significant risk of haemorrhagic transformation, anticoagulation treatment should be stopped for one week and aspirin 300mg substituted.25


Anticoagulants and haemorrhagic stroke

Clotting levels in people with a primary intracerebral haemorrhage (ICH) who were receiving anticoagulation with a vitamin K antagonist (eg. warfarin) before their stroke, should be returned to a normal international normalised ratio (INR) as soon as possible using a combination of prothrombin complex concentrate and intravenous vitamin K.25

In those with haemorrhagic stroke on newer oral anticoagulants, local guidelines should be adhered to and haematology advice sought. Specific antidotes are not available. Evidence supporting non-specific therapies (eg. prothrombin complex concentrate, activated prothrombin complex concentrate) is limited and clinical outcome data are lacking.26

Anticoagulation post haemorrhagic stroke/ICH in patients with AF

Patients with haemorrhagic stroke and AF can present a therapeutic dilemma. In the acute post-haemorrhagic stroke situation, the risks associated with bleeding outweigh considerations  of thromboembolism and anticoagulation should be stopped or-if indicated-reversed. Long-term management for such patients is less clear and should be based on weighing the risks of recurrent bleeding with the risk of a further thromboembolic event. There is little evidence to guide clinicians with no randomised studies and few observational studies. One systematic review of observational data did not suggest a large increase in haemorrhagic stroke risk with antithrombotic therapy use, though numbers were small.27 Decision analysis tools have reported that for patients with prior lobar ICH, withholding anticoagulation therapy is preferred, improving quality-adjusted life expectancy by 1.9 QALYs.28

Overall risk of recurrent haemorrhagic stroke is 1% at three months, then 2-4% per patient year. Anticoagulation is likely to increase the risk, though there are no population based studies to quantify this.29 Risk of recurrent haemorrhage appears to relate to underlying vasculopathy. Most intracranial haemorrhages in deep hemispheric (basal ganglia, thalamus) or brainstem territories are likely caused by hypertensive vasculopathy and have a recurrence rate of 2% of cases per year, whereas lobar ICH is often associated with cerebral amyloid angiopathy, and has a recurrence rate of 4% of cases per year.29-31

No guideline can cover all eventualities and there is a need for more observational data. Assessment of risk of recurrent haemorrhage (taking location and likely aetiology into account) versus risk of thromboembolic event (using CHADS2/ CHA2DS2-VASc) should be made.  Decisions should be made on an individual basis after open discussion with the patient. In the long term, many patients with lobar hemorrhage, cerebral amyloid angiopathy, or other risk factors may remain at higher risk of recurrence of haemorrhage than of fatal or disabling thromboembolic events and would therefore be best managed without anticoagulants. Conversely, those with deep hemispheric ICH, hypertension that can be well controlled, and a high risk of disabling thromboembolism may receive a net benefit from anticoagulation.


AF is the commonest cardiac arrhythmia, its prevalence increasing with age. It is by no means a benign condition and is associated with significant morbidity and mortality. AF significantly increases the risk of stroke and heart failure and carries significant medical, social and economic burden. Opportunistic screening is effective in detecting AF and should carried out for all those over 65 years.  Current evidence suggests that in the majority of patients with persistent or permanent AF, who are haemodynamically stable, a rate control (rather than rhythm control) strategy is preferable. Scoring systems (eg. CHAD2-VASc) are useful in determining thromboembolic risk. Warfarin (VKA) is far superior to aspirin or placebo in preventing stroke and should be considered in all those with AF, except those at very low thromboembolic risk. NOACs have shown non-inferiority with similar or reduced bleeding risks and are being used in increasing frequency. Clinicians should be mindful that bleeding risk with aspirin is similar to warfarin. New opportunities for increasing OAC use in patients with AF and hence preventing stroke are offered by the NOACs.


Conflict of interest: none declared

References available online at