Background
Reducing atrial fibrillation (AF)-related stroke requires timely AF diagnosis, but the optimal monitoring strategy is unknown.
Objective
We used insertable cardiac monitor (ICM) data from the REVEAL AF study to compare AF detection rates by various short-term continuous monitoring (STM) strategies.
Methods and Results
Patients without known AF, but with CHADS 2 scores ≥3 (or = 2 with ≥1 additional AF risk factor) received an ICM for AF detection. One-time STM strategies were assessed by computing AF incidence at 1, 2, 7, 14, and 30 days post-ICM insertion. Repeated STM strategies (quarterly 24-hour, 48-hour, 7-day, or monthly 24-hour monitoring) were modeled by randomly selecting day(s) within a 30-day window around each nominal evaluation date over a 1-year period (simulated 10,000 times). Endpoints included AF ≥6 minutes, AF ≥1 hour, and daily AF burden ≥1 and ≥ 5.5 hours. The impact of compliance on AF detection was evaluated using daily compliance rates of 85%, 75%, 65% and 55% during follow-up months 1-3, 4-6, 7-9, and 9-12, respectively. Based on data from 385 patients (71.5 ± 9.9 years; CHADS 2 score 3.0 ± 1.0) the incidence of AF ≥6 minutes via ICM at 12 months was 27.1% (95% CI, 22.5-31.5%). This exceeded the range of estimated rates from all modeled one-time and repeated STM strategies (0.8% for 24-hour Holter monitoring to 10.6% for quarterly 7-day monitoring). Findings were similar for all AF endpoints. Modeled non-compliance reduced AF detection by 4.5% to 22.9%.
Conclusions
Most AF episodes detected via ICMs would go undetected via conventional STM strategies, thus preventing optimal prophylaxis for adverse consequences.
Recognition of the relationship between atrial fibrillation (AF) and thromboembolism (TE), including ischemic stroke, and the value of prophylactic anticoagulation has progressively prompted efforts to identify AF in patients at risk before the occurrence of an embolic event. Ambulatory electrocardiographic monitoring approaches for AF range from short duration continuous monitoring (24 to 48-hour Holter) and intermittent single time point measures to more prolonged continuous ambulatory monitoring (up to 30 days) or extended non-continuous monitoring utilizing smartwatches and other related technologies. Diagnostic yields have varied due, in part, to differences in the populations screened, the type and duration of monitoring involved, and definitions of AF.
Studies utilizing continuous long-term monitoring with insertable cardiac monitors (ICM) in patients with no known AF but with risk markers for AF and stroke have reported AF incidences from 20.7% to 40% during 12 to 30 months of follow-up — much higher than the 1.1% to 7.9% noted with short-term and intermittent monitoring techniques. While it is intuitive that long-term continuous monitoring will detect more AF than shorter or less frequent monitoring, the difference has not been previously quantified in a population at high risk for AF and stroke. Direct comparison of AF incidence between studies is difficult due to differences in patient demographics, study design, device characteristics, and the minimum duration of AF recognized. Moreover, it is impractical to conduct a study in which patients simultaneously wear multiple external monitors or in which different monitoring strategies are assigned to different arms (due to high study cost). We therefore simulated various monitoring strategies from continuous monitoring data collected via ICM in the REVEAL AF study to quantify AF yields of different cardiac monitoring approaches in patients at high risk for AF and stroke. This approach allows for comparison of various monitoring approaches as if they were employed in an identical population.
Methods
Study design and patient follow-up
Results from the REVEAL AF study (Clinical Trial Registration: https://clinicaltrials.gov/ct2/show/NCT01727297; NCT01727297 ) have been previously published. In brief, REVEAL AF enrolled patients with no AF history but who were deemed to be at risk for AF based on demographics and/or non-specific but compatible symptoms with either a CHADS 2 score ≥ 3 or a score of 2 with coronary artery disease, renal impairment, sleep apnea, and/or chronic obstructive pulmonary disease. All patients underwent ≥24 hours of external monitoring within 90 days prior to enrollment or before ICM insertion; any AF was exclusionary. Enrolled patients who received an ICM were followed for ≥18 months; until either their 30-month visit or until the last patient completed their 18-month visit. Patients had in-office visits every 6 months (and unscheduled p.r.n.) and transmitted device data monthly via remote monitoring (CareLink, Medtronic). The study was sponsored and funded by Medtronic, Inc. and conducted in compliance with applicable local laws and regulations of each participating country. Ethics committee/institutional review board approval was obtained at each institution. A steering committee was responsible for the study design, conduct, and reporting. Data monitoring, collection, and analysis was performed by the sponsor and Steering Committee in partnership. The authors are solely responsible for the design and conduct of this study, all study analyses, the drafting and editing of the manuscript, and its final contents. All patients provided informed consent before participating in the study.
Simulated short-term continuous monitoring strategies
Using the continuous recordings from the 385 patients included in the primary endpoint analysis for REVEAL AF, we simulated one-time short-term continuous monitoring (STM) strategies by computing the AF incidence using 1, 2, 7, 14, and 30-day recording periods post-insertion from the ICM data. Repeated STM strategies (quarterly 24-hour, 48-hour, 7-day; or monthly 24-hour) were simulated by randomly selecting day(s) within a 30-day window around each nominal evaluation date over a 1-year period. Consecutive periods of 24 hours, 48 hours, and 7 days were randomly chosen for each quarter (3-month period) of each subject’s follow-up. Similarly, periods of 24 hours were simulated for each 30 days of each subject’s follow-up. This methodology has been described previously. Four analyses were performed. The first assessed occurrence of all AF episodes ≥6 minutes in duration, and the second assessed the occurrence of all AF episodes ≥1 hour in duration. We chose this second cut point because we believe most physicians would consider a 1-hour episode clinically significant even if they did not consider a 6-minute episode important. For these two analyses, only episodes adjudicated to be AF were evaluated. The third and fourth analyses assessed the occurrence of cumulative AF burden lasting ≥1 hour and ≥ 5.5 hours over the course of a day, respectively. We chose these thresholds as both have been independently associated with increased risk of stroke in patients with cardiac implantable electronic devices. Longer duration cut points including AF lasting 24 hours in duration could also have been tested but the number of patients meeting this threshold was too small for the analysis to be meaningful. Incidence rates were generated using the Kaplan-Meier method for each STM strategy. Repeated STM strategies were simulated 10,000 times. Twelve-month rates were averaged across the 10,000 simulations and minimum and maximum values and ranges were determined. As the same data set was used for all simulations, our modeling eliminated potential differences in results due to varying accuracy in AF detection algorithms between devices.
Compliance modeling
To evaluate the impact of compliance, AF detection was simulated for each one-time STM strategy assuming an 85% probability that a patient would complete each day of the monitoring period. All simulations were run 10,000 times. The 85% compliance rate was chosen based on the range of compliance rates previously reported in the literature (66%-97%) for continuous monitoring periods lasting 1 to 30 days.
For repeated STM strategies, compliance was assumed to be 85% during the first 3 months of follow-up, 75% for months 4 to 6, 65% for months 7 to 9, and 55% beyond month 9. These rates were based on previous studies reporting decreased compliance over time or overall low compliance with repeated cardiac monitoring protocols. Compliance was modeled on a day-by-day basis, assuming an 85%, 75%, 65%, or 55% chance that each day of monitoring was completed based on the period of follow-up during which the monitoring occurred. Repeated STM strategies were simulated 10,000 times, and for each time point post-insertion (1 day, 2 days, 7 days, etc) AF incidence rates were averaged across the simulated results; the corresponding minimum and maximum incidence rate among the 10,000 simulations was also determined for each time point. The endpoint for all compliance analyses was adjudicated AF ≥6 minutes occurring within each simulated period.
Results
Patients and follow-up
Of 446 patients screened in REVEAL AF, 394 underwent device insertion and 385 were included in the present analysis ( Figure 1 ). Subject characteristics are presented in Table I for the 385 patients analyzed.
| Characteristic | Patients (n = 385) |
|---|---|
| Age | 71.5 ± 9.9 |
| Sex (male) | 201 (52.2%) |
| Body mass index | 31.2 ± 6.5 |
| CHADS 2 score | 3.0 ± 1.0 |
| CHA 2 DS 2 -VASc score | 4.4 ± 1.3 |
| Renal dysfunction | 165 (42.9%) |
| Congestive heart failure | 81 (21.0%) |
| Coronary artery disease | 231 (60.0%) |
| Hypertension | 360 (93.5%) |
| COPD | 76 (19.7%) |
| Sleep apnea | 103 (26.8%) |
| Diabetes | 246 (63.9%) |
| Remote stroke | 79 (20.5%) |
| Transient ischemic attack | 75 (19.5%) |
Incidence of AF detection with simulated STM strategies
The incidence of AF lasting ≥6 minutes in duration with ICM monitoring was 27.1% (95% confidence interval [CI] 22.5%-31.5%) at 12 months. This exceeded the estimated rates from all forms of STM. Specifically, the incidence of AF ≥6 minutes ranged from 0.8% (95% CI, 0%-1.7%) with 24 hours of monitoring to 6.2% (95% CI, 3.8%-8.6%) with 30 days of continuous monitoring for the one-time STM strategies assessed. For simulated repeated STM strategies, AF detection at 12 months was lowest with quarterly 24-hour Holter monitoring (3.5%, range [min-max]: 1.3%-6.9%) and highest with quarterly 7-day monitoring (10.6%, range [min-max]: 7.3%-13.5%) ( Figure 2 A , Table II ).
| Cardiac monitoring strategy | 12-mo incidence rate of continuous AF ≥6 min | 12-mo incidence rate of continuous AF ≥1 h | 12-mo incidence rate of daily AF burden ≥1 h | 12-mo incidence rate of daily AF burden ≥5.5 h |
|---|---|---|---|---|
| ICM monitoring | 27.1% (22.5%, 31.5%) | 12.2% (8.8%, 15.4%) | 21.6% (17.3%, 25.7%) | 10.9% (7.7%, 14.1%) |
| One-time STM strategies | ||||
| 24-h | 0.8% (0%, 1.7%) | 0.5% (0%, 1.2%) | 4.4% (2.3%, 6.5%) | 1.8% (0.5%, 3.1%) |
| 48-h | 0.8% (0%, 1.7%) | 0.5% (0%, 1.2%) | 4.7% (2.5%, 6.8%) | 1.8% (0.5%, 3.1%) |
| 7-d | 2.3% (0.8%, 3.8%) | 1.6% (0.3%, 2.8%) | 6.8% (4.2%, 9.2%) | 2.6% (1.0%, 4.2%) |
| 14-d | 3.1% (1.4%, 4.8%) | 2.1% (0.6%, 3.5%) | 7.3% (4.6%, 9.8%) | 3.1% (1.4%, 4.8%) |
| 30-d | 6.2% (3.8%, 8.6%) | 3.1% (1.4%, 4.9%) | 9.1% (6.2%, 11.9%) | 4.4% (2.3%, 6.5%) |
| Repeated STM strategies | ||||
| Quarterly 24-h | 3.5% (1.3%, 6.9%) | 1.3% (0.3%, 3.0%) | 8.3% (5.8%, 10. 8%) | 4.3% (3.2%, 5.7%) |
| Quarterly 48-h | 5.1% (1.9%, 8.3%) | 1.9% (0.5%, 3.9%) | 9.5% (7.1%, 12.2%) | 4.7% (3.5%, 6.5%) |
| Quarterly 7-d | 10.6% (7.3%, 13.5%) | 4.6% (2.6%, 6.9%) | 12.9% (10.1%, 15.4%) | 6.4% (4.8%, 8.4%) |
| Monthly 24-h | 6.0% (3.0%, 9.8%) | 2.1% (0.5%, 4.6%) | 9.6% (7.4%, 12.7%) | 4.8% (3.2%, 6.6%) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
