CardiologyClinical Features

Subclinical Atrial Fibrillation – where are we now?

Introduction: Atrial Fibrillation (AF) is the most prevalent cardiac arrhythmia in the world with an estimated 50 million people worldwide living with the rhythm in 2020.1 It is expected that, by 2060 in the EU, over 14 million aged 65 years and over will have an AF diagnosis and that the lifetime of risk of someone a of EU descent aged over 55 years of developing AF in their lifetime is 37%, i.e. one in three.2 Whilst there are many aetiologies for atrial fibrillation, including valve disease, hypertension, heart failure with reduced or preserved ejection fraction (HFrEF or HFpEF), the rhythm is associated with a significant increased risk of mortality and morbidity. AF is associated with a 1.5-3.5fold increased risk of all-cause mortality, a 5-fold increased risk of stroke with AF accounting for 20-30% of all ischaemic strokes and an annual hospitalisation rate of 10-40%.2

The overall management of atrial fibrillation can be neatly summarised by either the SOS or ABC mnemonic quoted in the American or European Guidelines respectively.1,2 In the SOS nomenclature

  • S – assessment and treatment of Stroke risk
  • O – Optimisation of all modifiable risk factors
  • S – Symptom management with either a rhythm or rate control strategy

The European ABC scheme emphasises similar points

  • A – Avoid stroke
  • B – Better symptom control
  • C – Cardiovascular risk factor and concomitant disease optimisation

Prescribing of oral anticoagulation in patients who meet the threshold using the CHA2DS2-VASc score aims to reduce stroke risk with modifiable risk factors including obesity, smoking, alcohol consumption, increased exercise as well as good control of diabetes and hypertension.

Clinical AF vs Subclinical AF vs Device Detected AF

Clinical atrial fibrillation is that which is documented on 12-lead ECG, or an ECG rhythm strip of ≥30 seconds in either a symptomatic or asymptomatic patient. Often patients present for ECG due to symptoms, such as palpitations or dyspnoea, associated with their atrial fibrillation.

Subclinical atrial fibrillation can be defined as atrial fibrillation identified in a patient, with no previous history or ECG evidence of atrial fibrillation. The detection of this subclinical atrial fibrillation includes documentation by wearables and implanted cardiac devices. Because some wearables and consumer targeted single lead ECG devices, such as the Kardia device and app, can provide ECG evidence of atrial fibrillation with strips of 30s, the distinction between clinical and subclinical AF is becoming blurred.

Implanted devices such as pacemakers or implantable cardioverter defibrillators (ICDs) have advanced diagnostic features that can automatically record atrial high-rate episodes (AHREs) and such events are sometimes referred to as device detected atrial fibrillation but are also referred to as subclinical AF. For us to diagnose device detected AF from AHREs it is important that the stored device tracings are inspected to ensure they are true events and not artefactual.

Smartwatches and handheld ECG devices

There are number of handheld ECG devices available, typically marketed as patient owned devices, but they are also used by healthcare professionals in a clinical environment. These devices typically record a live single channel ECG trace, which then can be stored on a digital device, such as a phone or tablet, via an accompanying app. Some of these devices will give an interpretation of the rhythm, such as “possible arial fibrillation”, some with sensitivities of 98% and specificity of 97%.3

Many smartwatches or fitness trackers collect heart rate data and can flag up possible atrial fibrillation. These devices do not use ECG based technology but rather photoplethysmography (PPG). This is a clinically wellestablished technology, used in pulse oximeters. PPG detects pulse rate based on changes in the reflection of light, emitted from the back of the smartwatch, by the skin. The amount of light reflected changes in proportion to the about of blood under the skin, which changes with each pulse, thus the device can detect both heart rate and assess regularity. Some higher end smart watches such as the Apple or Samsung watch allow recording of single lead ECG by incorporating small electrodes in the back and side of the watch.

PPG based technology is easily integrated into devices and underpins apps, such as Fibricheck, which use a camera phone to check for possible atrial fibrillation. While this technology works well at rest there is some evidence that when subjects are ambulatory that motion artefact reduces the sensitivity and specificity.4 Further affecting the sensitivity for detection of AF, some older PPG based devices do not continuously assess rhythm regularity but sample for a minute every two hours, thus regularity is only checked for 12 minutes in a 24-hour window.

The incidence of irregular rhythm/ suspected AF alerts in smart watches is low. The Apple Heart Study monitored 419297 patients over 8 months and irregular rhythm notifications were identified in only 0.52% of subjects. 450 of these patients had 7-day ECG based monitoring with 34% demonstrating clinical AF.5 It should be noted that with handheld ECG asymptomatic paroxysmal atrial fibrillation can be missed and with wearable technology, pick up rates will be dependent on how long the device is worn.

European guidance6 on the use of digital devices recommends that if PPG based monitoring is indicative of AF, an ECG-based method should be used to confirm the diagnosis of AF. If ECG evidence of atrial fibrillation is obtained, then the patient should be managed using the SOS or ABC approach outlined above. Whilst the European guidance states that PPG-based or ECG-based devices are preferred to pulse palpation for AF screening, this is contradicted by NICE guidance from 2019 which states there is not enough evidence to recommend the routine use of single lead ECG devices to detect AF in patients with an irregular pulse.7

Patient owned devices have the potential to increase the detection of atrial fibrillation, although the reported incidence has been low in populations using them, and prompt them to seek medical advice. The true clinical consequence of these asymptomatic episodes, particularly if paroxysmal, are not fully clear. The main randomised trial which over 7000 participants, with no history of AF recorded twice daily handheld ECG over a 14-day period demonstrated a small 4% relative risk reduction in the primary combined endpoint of ischaemic or haemorrhagic stroke, systemic embolism, bleeding leading to hospitalisation, and all-cause death, over a median follow-up of 6.9 years, in those where AF was documented, and anticoagulation was commenced.18

Device detected atrial fibrillation

There is over 15 years of data demonstrating that implanted cardiac devices record AHREs or subclinical atrial fibrillation. However, many of these trials used different rate and duration cut-offs in their definitions. What we do know is that subclinical AF is common in patients ≥65 years, with the ASSERT II trial showing an incidence of an incidence rate of 34.4% per person-year for episodes >5mins.8 This study, using implanted loop recorders, also showed that the rate of detection for episodes ≥30 minutes was 21.8%/y, ≥6 hours was 7.1%/y and ≥24 hours, 2.7%/y.

With subclinical AF being common and of varying duration the question is what burden of device detected atrial fibrillation increases the risk of stroke. The TRENDS Study,9 which looked at AHREs in over 3000 pacemaker or ICD patients with and without a prior history of AF, demonstrated that a daily AF burden of ≥ 5.5hrs was associated with 2.2-fold increased risk of stroke (95%

CI 0.96-5.05, p=0.06), although this was not quite statistically significant. It is interesting to note that this increased risk of stroke is less that the 5-fold increased risk of clinical AF.

Data like this supported the original European Heart Rhythm consensus in 201710 recommending oral anticoagulation in patients with a CHA2DS2-VASc ≥2 and a daily AHRE duration of ≥5.5hrs. However, a later analysis of the original ASSERT Trial data,11 demonstrated that subclinical AF duration >24 h was associated with a significantly increased risk of subsequent stroke or systemic embolism. The 2023 ACC/AHA/ ACCP/HRS Guideline for the Diagnosis and Management of Atrial Fibrillation1 take the approach of combining AHRE daily burden and CHA2DS2-VASc in a decision spectrum. They recommend oral anticoagulation if the CHA2DS2-VASc ≥2 for men and ≥3 for women, the AHRE burden is ≥23.5 and AF is confirmed of the certainty is high. This left a large cohort with either shorter durations and high CHA2DS2-VASc scores or longer durations but lower scores. It was hoped, at the time of publication of these guidelines, that two, soon to be reported, randomised clinical trials, NOAH-AFNET 6 and ARTESIA, investigating the efficacy and safety of oral anticoagulation for device detected AF would inform this gap.

NOAH-AFNET 6 and ARTESIA

NOAH-AFNET 612 enrolled 2536 patient, aged over 65, with device detected AHREs >170bpm for >6 mins and at least one other risk factor for stroke. Patients were randomised to either edoxaban or placebo with patients in the placebo arm receiving aspirin if indicated. The primary outcome was a composite of CV death, stroke or systemic embolism with a safety outcome of death from any cause or major bleed.

The mean age was 78, with median AHRE durations of 2.8 hours and CHA2DS2-VASc scores of 4. There was no statistically significant difference in the primary efficacy outcome between the oral anticoagulation and placebo arm. Whilst there was a 35% relative risk reduction in a secondary end point of ischaemic stroke or systolic embolism in the group receiving anticoagulation, this did not reach statistical significance (CI 39%-107%). There was an increased risk the primary safety outcome being met with anticoagulation, HR 1.31 (95% CI 1.02-1.67) with 2.1-fold increased risk of major bleeding. This led to the trial being terminated early due to safety concerns and futility and led the authors to conclude, that despite the relatively high-risk population, that the rate of stroke was lower than similar populations and it was appropriate to withhold anticoagulation.

ARTESIA13 enrolled patients over 55 years of age with a CHA2DS2VASc scores ≥3 or patients with a history of stroke over 75 years and were randomised to apixaban or placebo. The placebo in this case was aspirin as this was commonly used at the time of the trial design. Patients had to have device detected AHREs ≥6 mins but unlike NOAH-AFNET 6 episodes >24 hours were removed from the trial medication and commenced on indicated anticoagulation.

The mean age of 76.8 across the 4012 patients enrolled was similar to NOAH-AFNET 6, with a similar stroke risk, mean CHA2DS2VASc score of 3.9. Apixaban significantly reduced the risk of stroke or systemic embolism with a HR of 0.63 (CI 0.45-0.88) as well as ischaemic stroke. There was no increase in haemorrhagic stroke, death from any cause or cardiovascular death. Consistent with the finding from NOAHAFNET 6 there was an increase in major bleeding with a HR of 1.36 (CI 1.01-1.82) but no increase in fatal bleeds. ARTESIA measured stroke severity using a modified Rankin scale with anticoagulation giving a 49% relative risk reduction in fatal or disabling stroke (CI 29-88%).

Both NOAH-AFNET 6 and ARTESIA found an increased risk of major bleeds in patients who are anticoagulated for device detected AHREs. However, they have divergent results in terms of benefit with ARTESIA demonstrating a 37% relative risk reduction in stroke or systolic embolism, but NOAH-AFNET 6 showed no reduction its primary end point. There were several differences in the trials, for example the inclusion of cardiovascular death as part of primary end point of NOAH-AFNET 6 potentially diluted a signal for stroke reduction as only 10% of CV deaths were likely to be due to stroke.14 Indeed, a significant limitation of NOAH-AFNET 6 was the low event rate, combined with the early termination of the trial resulted in the trial being underpowered to detect a small benefit in stroke reduction.

Another important finding across both studies is that 1 in 5 of patients with device detected AF went on to develop either episodes >24hr or clinical atrial fibrillation. Whilst a sub study of NOAH-AFNET 6 did investigate the benefit of oral anticoagulation in patients with AHREs >24hrs, there was no benefit, although the low event rates are a limitation with only 2 strokes per patient year in each group.15 Interestingly, 29.3% of patients with AHREs >24 hours developed clinical AF. So, at a minimum, device detected AF is a predictor of the development of clinical atrial fibrillation and episodes >24 hours are a stronger predictor.

A meta-analysis of the two trials,16 over 6,500 patients, which was published last year demonstrated that oral anticoagulation reduced the risk of ischaemic stroke, relative risk 0.68 (CI 0.5-0.92). It also reduced a composite of cardiovascular death, all cause stroke, peripheral arterial embolism, myocardial infarction, or pulmonary embolism with a relative risk of 0.85 (CI 0.73-1.0). Not surprisingly this benefit was at a cost of an increased risk of major bleeding, relative risk 1.62 (CI 1.05-2.5).

As healthcare professionals we may perceive the stroke reduction versus increased bleeding risk as an equipoise, almost like comparing apples and apples. However, the patient perspective is different and the literature suggests they compare them as apples and oranges. A metaanalysis of patient perspectives on oral anticoagulation showed that patients were willing to accept higher bleeding risks if a certain threshold in stroke risk reduction could be reached and that the preferences of patients may differ from the perspective of clinicians.17 The fact that, in ARTESIA, major bleeds were managed conservatively, including transfusion, in 90% of cases but there was a 47% relative risk reduction in disabling or fatal stroke is an important element to consider when using this new data to make decisions around anticoagulation of device detected AF.14

Conclusion

Prevalence of clinical atrial fibrillation is increasing and there is clear guideline on appropriate management. The increase in patient owned devices and implanted cardiac devices, such as pacemakers, has increased our detection of subclinical atrial fibrillation and there are remaining questions about how to best manage these.

The incidence of AF amongst patient with smart watches appear to be low and if AF is flagged through PPG technology, this must be confirmed with ECG based monitoring. If a patient owned device flags up an AF alert and can record a confirmatory ECG strip of 30s, then this meets the threshold for ECG evidence, particularly if reviewed by a clinician. There does remain a question about how much AF meets a threshold for anticoagulation.

For device detected AF, the two randomised trials tell us that the incidence of stroke is lower than for clinical AF at 1-1.2% per patient year, however in ARTESIA 43% of the stroke in the placebo arm were fatal or disabling.

Anticoagulation reduces the risk of ischaemic stroke across both trials and the risk of fatal or disabling stroke significantly. The bleeding risk was significantly increased but 90% were managed without need for surgical intervention.

For device detected AF a 1-1.2% per year stroke rate is the threshold for anticoagulation on the CHA2DS2-VASc score, although it should be noted that this was developed based on clinical AF. Some commentators state that the results tell us that there should be a clear discussion with patients regarding the potential benefits and risks of anticoagulation.14 If a decision is made not to commence anticoagulation, then given that one in five patients will develop episodes >24hrs or clinical AF, there should be careful monitoring for further episodes or the development of clinical atrial fibrillation. This is a task ideally suited to remote monitoring, where alerts can be set for various AHRE daily burdens.

References available on request

Read the April Edition

Read our Clinical Features

Leave a Reply

Your email address will not be published. Required fields are marked *

Please Confirm

This website is only for the eyes of medical professionals. Are you a medical professional?