Screening for Atrial Fibrillation: Closing the "LOOP"

Naga Venkata K. Pothineni, MD; Rajat Deo, MD, MTR

Disclosures

Circulation. 2020;141(19):1523-1526. 

Atrial fibrillation (AF) is a public health epidemic that is associated with significant morbidity. The known association between AF and stroke has resulted in clinical trials that have provided overwhelming evidence for the benefits of anticoagulation to prevent thromboembolic complications. This treatment paradigm has also led to longer continuous cardiac rhythm monitoring among cryptogenic stroke patients, with the goal of detecting AF so that anticoagulation can be initiated. Large randomized trials such as CRYSTAL AF (Cryptogenic Stroke and Underlying Atrial Fibrillation) have established the potential to diagnose AF with insertable cardiac monitoring (ICM) for up to 3 years after stroke.[1] Given the potential benefit of anticoagulation in this high-risk population, long-term, continuous cardiac rhythm monitoring is recommended in the poststroke patient population.[2,3]

Broader population-wide screening for AF has also been the subject of recent studies and guideline documents. The STROKESTOP study (Systematic ECG Screening for Atrial Fibrillation Among 75 Year Old Subjects in the Region of Stockholm and Halland, Sweden) evaluated intermittent ECG screening during a 2-week period in the elderly and demonstrated an AF prevalence of 12.3%.[4] The European Society of Cardiology subsequently recommended "opportunistic" screening for AF either by pulse taking or rhythm strip in patients older than 65 years.[5] These guidelines also suggest that more systematic ECG screening can be considered to detect AF in individuals >75 years of age or those at high stroke risk. However, before deciding on potential strategies for AF monitoring, it is critical to have a sense of the prevalence of the underlying arrhythmia. Some cohort studies have used continuous rhythm monitoring to evaluate estimates of AF prevalence and burden in older individuals. In ASSERT-II (Prevalence of Sub-Clinical Atrial Fibrillation Using an Implantable Cardiac Monitor), ICMs were implanted in 256 patients ≥65 years of age, CHA2DS2-VASc ≥2, and evidence of left atrial enlargement. After 1 year, 34% of the study population had subclinical AF defined as an episode of >5 minutes of AF.[6] The median AF burden was reported to be only 3 minutes, suggesting short and infrequent episodes. Similarly, in REVEAL AF (REVEAL AF: Incidence of AF in High Risk Patients), 446 patients with a CHADS2 score ≥3 and nonspecific cardiac symptoms had an ICM and were observed for 23 months. AF detection rate at 12 months was 27%.[7] These studies in elderly individuals with cardiovascular risk factors suggest that AF is common and that the overall burden is low. Future studies will clarify the significance of these subclinical or screening episodes. In particular, the ongoing ARTESIA (Apixaban for the Reduction of Thrombo-Embolism in Patients With Device-Detected Sub-Clinical Atrial Fibrillation), NOAH (Non-vitamin K Antagonist Oral Anticoagulants in Patients With Atrial High Rate Episodes), and LOOP (Atrial Fibrillation Detected by Continuous ECG Monitoring Using Implantable Loop Recorder to Prevent Stroke in High-risk Individuals) trials are evaluating the benefits of anticoagulation for stroke prevention in this population. Of equal importance is recognizing the differences in the diagnostic yield for AF detection across various devices and monitoring strategies.

In this issue of Circulation, Diederichsen et al[8] compare the diagnostic yield for detecting AF across different theoretical monitoring strategies. They use more than 3 years of continuous rhythm monitoring data from the LOOP study to evaluate AF detection using either an ECG or ambulatory rhythm monitor. Specifically, after a rigorous protocol for adjudicating AF, the investigators designed a simulation study that compared the sensitivity and negative predictive value for AF detection using 10-second ECGs, bidaily 30-second ECG strips, and external monitoring at 24 hours, 48 hours, 7 days, and 30 days. Similar to other population-based screening studies, the participants in LOOP are at high risk of developing AF (mean CHA2DS2-VASc 4.0, mean age 76 years), and >20% had a previous stroke/transient ischemic attack. After 1 year, 131 participants (22% of the study population) had AF; after the complete monitoring period of 3+ years, 205 (35% of the study population) had AF. In addition, the mean AF burden in this population was 2.7%, and two-thirds of individuals had an AF burden <0.5% (median AF burden was 0.12%). The simulated screenings demonstrate how AF burden impacts the likelihood of detection using the various monitoring strategies. A strategy of screening with a single 30-day monitor during a 1-year period resulted in a sensitivity <25%. Additional 30-day monitoring periods each year enhanced the sensitivity for AF detection to 55% after 3 years. This sensitivity still corresponds to a miss rate of 45%—a near coin toss probability for AF detection after serial screenings during the 3-year period. This finding is important because most clinicians would have considered annual 30-day monitoring to be a rigorous approach for AF screening in the elderly. In contrast, the investigators demonstrate markedly enhanced sensitivity for AF detection with a higher AF burden across all screening strategies. Among individuals with paroxysmal AF who have at least a 24-hour episode of AF, the sensitivity of AF detection using a 30-day monitor in a 1-year period was 56%. This rose to 85% after serial monitoring with annual 30-day monitors for 3 years. The findings from these analyses remind us of our clinical decision making when selecting the duration of ambulatory monitoring according to the frequency of a patient's symptoms such as palpitations or syncope. Another important insight from these screening simulations was that AF detection was higher when the same monitoring duration was spread over separate periods compared with being compiled into a single period.

The study also provides greater precision around estimates of AF in the elderly and associated burdens that are consistent with those found in ASSERT-II and REVEAL AF. Although ongoing clinical trials are evaluating the risks and benefits of anticoagulation therapy, it is hard to imagine that AF detection would not be clinically relevant in the elderly population with comorbidities. Unfortunately, short-term, intermittent monitoring—even when conducted at serial timepoints—seems like an ineffective strategy for AF screening in the general population. The results are also relevant for other clinical situations that require monitoring such as post-AF ablation when the burden is expected to be low, but the presence of AF could signal the need for additional therapies or interventions. Diederichsen et al provide compelling data for the use of long-term continuous monitoring when the goal is AF screening. At this time, ICMs are not a practical option for every high-risk individual in the population. However, alternative technologies, including wearable devices, might one day play an important role in identifying AF and its burden in the population.[9] These devices are clearly not medical grade at this point;[10] however, with ongoing advances in engineering, it is likely that our population will have access to options for long-term AF screening.

In addition to the partnerships required for developing wearable technologies and novel treatment algorithms, stakeholders need to assess care models that leverage the benefits of early AF detection. ICMs are assumed to be the gold standard for rhythm monitoring. Although they have a high sensitivity for AF detection, ICMs are known to have a significant number of false positives—up to 46% for AF detection.[11] As such, healthcare providers need to review tracing level data before making management decisions. In the current study, all recordings were adjudicated by at least 2 senior cardiologists. This extent of detailed adjudication can only be possible in a clinical trial and will be much more challenging to replicate in a clinical setting. Alternatively, methods that use machine learning algorithms, such as deep neural networks for accurate rhythm identification, have already proven effective with shorter-term ambulatory monitors[12] and may be applied to the abundant data that will be generated from wearable devices. In the current analysis and as part of the study protocol, >80% of AF patients were started on anticoagulation. In real-world practices, however, a clinic visit and subsequent shared decision making are necessary before initiating anticoagulation and discussing other AF-specific therapies. Immediate access to specialized providers may present a significant barrier, and future studies will need to assess whether telemedicine may play a role in streamlining widespread screening programs to interventions (Figure). In addition, selective screening of higher risk subgroups such as those with advanced age, large left atrial size, or elevated NT-proBNP (N-terminal pro-B-type natriuretic peptide) may help identify those who are more likely to have AF, limit the number of individuals who require care, and potentially maximize the overall treatment effect. Efforts to streamline these clinical issues should also take into account appropriate reimbursements for the time and effort involved in data adjudication and follow-up clinical work, all of which will be crucial to derive the benefits of early AF detection and potential stroke prevention. As we await results of outcome trials regarding the management of subclinical AF, it is important for healthcare systems to prepare for the surge in AF that will be detected through reliable and widely available continuous cardiac rhythm monitors.

Figure.

Schematic illustrating opportunities and challenges associated with atrial fibrillation screening using continuous rhythm monitoring.

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....