During World War II, radar operators in coastal England faced the challenge of distinguishing enemy airplanes from flocks of birds. If the radar was too sensitive, every colony would be mistaken for Luftwaffe. If specificity was prioritized, some enemy aircraft could pass undetected along with the flocks. Similarly, in medicine, technological advances have allowed us to find subclinical disease states that would otherwise go unrecognized in the absence of signs or symptoms. With these advances, however, comes the realization that distinguishing between true disease and insignificant findings can be extremely challenging and can expose the patient to the risks of unnecessary treatment. There are few better examples than in the field of atrial fibrillation (AF) monitoring, where our ability to detect brief, infrequent, asymptomatic AF episodes has outpaced our grasp of their significance. The fundamental questions we must ask ourselves is whether these AF episodes are more akin to innocuous birds or enemy aircraft and where we should set our threshold to distinguish the two.
It is in this context that 2 important contributions to the field, both published in this month’s American Heart Journal , should be viewed. First, in the REVEAL AF study, 385 patients with vascular risk factors underwent placement of an implantable cardiac monitor (ICM). Approximately 90% had nonspecific symptoms compatible with AF, such as fatigue, dyspnea, and/or palpitation prior to device placement. The detection rate of AF episodes lasting > 6 minutes was 27.1% at 12 months, 29.3% at 18 months, and 40% at 30 months. Of those with any AF detected, 41.4% had at least 1 episode lasting 6 or more hours, and 10.2% (or 3.3% of the entire cohort) had at least 1 episode > 24 hours. In the second study from the ongoing LOOP trial, 597 participants age 70 years or older who also had stroke risk factors were recruited from the community and implanted with an ICM. Over the 40-month follow-up period, the cumulative rate of AF > 6 minutes was 33.8%, and the rates of AF > 5.5 hours and > 24 hours were 16.1% and 5.7%, respectively. Combined with the findings of the similarly designed ASSERT II and PREDATE AF trials, these studies demonstrate unequivocally that AF is extremely common in older individuals with additional vascular risk factors, particularly when we screen with continuous monitoring over long time horizons. However, given the invasive nature, cost, and infrastructure required for ICM use for this purpose, it is important to understand how much AF we would be missing if we relied on other noninvasive and less expensive screening options and whether we can prospectively define a high-risk subgroup who would benefit most from screening.
The first question was addressed in this current analysis by the REVEAL AF authors, who simulated the rate of AF detection using the various forms of external monitors used in routine clinical care. The authors varied the duration, frequency, and compliance rates of monitoring lasting 24 hours to 30 days and calculated the AF detection rates compared to the “gold standard” ICM data. The results show that for every scenario evaluated, the rate of AF detection by the ICM was higher than any and all forms of external monitoring. The differences in detection rates were staggering in many cases as exemplified by the finding that a single 24-hour Holter monitor showed an AF detection rate that was 34-fold lower compared to that using the ICM. Although never approaching the detection rates of the implanted device, detection using external monitors increased when longer AF durations and burdens were used as the target end point, when simulated patient compliance was improved, and when more prolonged and repeated external monitoring was performed. Even so, a single 30-day monitor would fail to detect 75% of those with AF > 1 hour, and even quarterly 7-day monitors would miss almost two thirds of cases. These data replicate the findings observed in the CRYSTAL-AF trial where the ICM was consistently superior to various forms of external monitoring for AF detection in a cryptogenic stroke population.
The findings from REVEAL AF highlight the inadequacies of relative short-term AF monitoring but should hardly come as a surprise. Mathematical models of AF density constructed from episodes obtained from dual-chamber devices show that the chances of finding a single, brief episode of AF with intermittent, short-term monitoring are slim. But before we can understand the best way to screen for AF, we must first determine whom to screen; how much AF is necessary for a stroke; and whether device-detected, subclinical AF of short duration (ie, <24 hours) is as responsive to anticoagulation as clinically detected AF. At face value, the first issue appears the easiest to address. It makes sense to screen only those with both an elevated risk of AF and an indication for anticoagulation should AF be found. As confirmed by the LOOP investigators and others, the vascular risk factors used in the CHA 2 DS 2 -VASc score alone appear to define a population at high risk of incident AF. The addition of resting sinus rate, N-terminal prohormone of brain natriuretic peptide, and troponin T further delineates a particularly high-risk group of individuals, although only for AF episodes > 24 hours and at an added logistical and financial cost. Thus, both REVEAL AF and LOOP support the assertion that an effective AF screening strategy will rely, in large part, on how much AF constitutes “disease.”
Prior studies of implantable and wearable devices demonstrate that various durations and burdens of AF are associated with increased risk of stroke, but the risk is consistently lower than that of clinically detected AF even with similar clinical risk scores until the duration exceeds 24 hours. Although a recent meta-analysis suggests that more intensive cardiac monitoring, including monitoring with ICMs, can lead to a reduction in recurrent events in those with cryptogenic stroke, the results of pivotal trials such as ARTESIA, NOAH, and the main LOOP trial that are specifically designed to address the efficacy of anticoagulation in these individuals will be needed to better understand how best to manage short-duration, subclinical AF detected on implanted devices. Regardless of the results of these studies, the progressive nature of AF will require repeated measures of duration and/or burden over the individual’s lifetime to assess if and when a treatment threshold is crossed.
Current guidelines recommend opportunistic screening for AF using pulse palpation not because this is the best way of finding AF but more because we have yet to prove that treating AF found through screening does more good than harm. Still, there appears to be a wide range of reasonable options between the gross undersampling that results from pulse palpation at the time of a routine examination and the continuous monitoring afforded by ICMs. The former will find only those with very high AF burdens, whereas the latter may result in potential overtreatment of those with rare, brief episodes whose clinical significance and response to anticoagulation remain untested. When one adds to that the costs and resources needed for device implantation and monitoring of ICM data, it becomes clear that this option is not scalable as a screening tool for the tens of millions of people worldwide who are at high risk of developing AF and its consequences. A reasonable compromise may lie somewhere between. With 3.3 billion smartphones currently in use worldwide and 4 out of 5 Americans owning such a device, AF detection could best be accomplished if based on consumer-grade digital technologies. Smartphone-based ECG monitors have successfully been used for this purpose, and smartwatches and other wearables that can provide near-continuous long-term monitoring with excellent sensitivity are already being used in mass screening efforts including the Huawei Heart Study and the upcoming Heartline study.
If ICMs are comparable to radar, the REVEAL AF and LOOP investigators have shown us that the device can see both the flock of birds and airplane correlates of AF. Before we can decide how best to target our search for AF to prevent its consequences, however, it is incumbent upon us to first determine how much AF we need to be on the lookout for.
Disclosures
Advisory boards: Medtronic, Abbott, Johnson and Johnson.
Royalties: UpToDate.
Research support: Medtronic, Pfizer.
References
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
