Author + information
- Received April 5, 2017
- Revision received June 8, 2017
- Accepted June 15, 2017
- Published online December 18, 2017.
- Tine J. Philippsen, MDa,b,∗ (, )
- Lene S. Christensen, MD, PhDa,
- Michael G. Hansen, MDa,
- Jordi S. Dahl, MD, PhDb and
- Axel Brandes, MD, DMScb,c
- aDepartment of Cardiology, Hospital of Southern Jutland, Aabenraa, Denmark
- bDepartment of Cardiology, Odense University Hospital, Odense, Denmark
- cDepartment of Clinical Research, University of Southern Denmark, Odense, Denmark
- ↵∗Address for correspondence:
Dr. Tine J. Philippsen, Department of Cardiology, Hospital of Southern Jutland, Kresten Philipsens Vej 15, 6200 Aabenraa, Denmark.
Objectives The study sought to determine the incidence of subclinical atrial fibrillation (AF) in high-risk patients and to compare the effect of continuous versus intermittent monitoring.
Background AF often occurs in a subclinical form, which makes it difficult to detect. The authors do not know the incidence of subclinical AF among patients ≥65 years of age with hypertension and diabetes mellitus. This group of patients has increased risk of developing AF and in addition a high thromboembolic risk, if AF is present.
Methods A total of 82 outpatients ≥65 years of age (median age 71.3 years [interquartile range [IQR]: 67.4 to 75.1 years]) with hypertension and diabetes mellitus, and no history of AF or any other cardiovascular disease, were consecutively included. All patients received an insertable cardiac monitor (ICM) and were followed for a median of 588 days (IQR: 453 to 712 days). We compared continuous monitoring with 72-h Holter monitoring 1 month after ICM insertion. The primary endpoint was AF ≥2 min for the ICM and AF ≥30 s for the Holter monitoring.
Results During follow-up 17 (20.7%) patients were found to have subclinical AF detected by ICM with a median time to first detected episode of 91 days (IQR: 41 to 251 days) from inclusion. Only 2 (2.4%) patients also had AF episodes on the 72-h Holter monitoring. All detected episodes were completely asymptomatic.
Conclusions The incidence of subclinical AF in this group of patients was surprisingly high. Continuous monitoring with ICM detected significantly more AF episodes than 72-h Holter monitoring. (Detection of Subclinical Atrial Fibrillation in High Risk Patients Using Implantable Loop Recorder; NCT02041832)
Atrial fibrillation (AF) is the most common cardiac arrhythmia, affecting approximately 3% of the general population with greater prevalence in older persons and in patients with underlying conditions such as hypertension and diabetes mellitus (1–3). In general the prevalence of AF is estimated to increase substantially over the next decades (4,5) particularly in patients 65 years of age or older (4). Furthermore AF is the most frequent cardiac source of thromboembolism and a major cause of stroke (6,7).
Both diabetes mellitus and hypertension are known risk factors for AF, where hypertension accounts for at least 20% of AF cases (8). Patients with abnormal glucose metabolism are shown to have certain electrophysiological changes, which are consistent with an increased risk of AF (9,10). In addition to a high risk of developing AF, patients with hypertension and diabetes mellitus also have a high stroke risk, if AF is present. Treatment with oral anticoagulants (OACs) substantially reduces this stroke risk in patients with documented AF (11,12). AF often occurs in a subclinical form meaning that patients have either no or only mild and frequently unspecific symptoms (13,14). Moreover, it is often paroxysmal in nature, making it difficult to detect by intermittent monitoring. It is currently being discussed whether patients should actively be screened for AF. Moreover, the screening methods used and possible target populations are still a matter of debate (3). The recently published European guidelines for the management of AF recommend opportunistic screening for AF by pulse taking or electrocardiogram (ECG) rhythm strip in patients >65 years of age (3). Current European guidelines on diabetes, pre-diabetes, and cardiovascular diseases have a less clear recommendation for AF screening among patients with diabetes mellitus (15). There are no guidelines regarding screening for subclinical AF in these high-risk patients. Screening of a population at increased risk of AF and thromboembolism, if AF is present, but without any known cardiovascular diseases or other risk factors than hypertension and diabetes mellitus, using continuous monitoring with an insertable cardiac monitor (ICM) has never been performed. The aim of this study was to determine the rate of subclinical AF in high-risk patients ≥65 years of age with hypertension and diabetes mellitus using continuous ECG monitoring with an ICM. Furthermore we wanted to investigate the efficacy of continuous monitoring compared with conventional 72-h Holter monitoring in this study population.
The study was designed as a prospective single-center observational study.
Patients, ≥65 years of age without known or suspected AF attending the diabetes and cardiology outpatient clinics at the Hospital of Southern Jutland, were eligible for inclusion if they met the following criteria for hypertension and diabetes mellitus: treatment with a minimum of 2 antihypertensive drugs and treatment with any antidiabetics or insulin. All medications had been stable for at least 1 month prior to inclusion. Patients were excluded if they had any other risk factors for AF or met 1 or more of the following exclusion criteria: known AF; ongoing OAC treatment; left ventricular ejection fraction <45%; significant valve disease needing intervention; implanted pacemaker or implantable cardioverter-defibrillator; known ischemic heart disease, stroke, or transient ischemic attack (TIA) or peripheral artery disease; thyrotoxicosis; end-stage renal failure; or severe obesity expected substantially to compromise echocardiography and the signal from the ICM. A total of 248 patients met the inclusion criteria, and 151 of these patients declined participation. At the day of inclusion, all patients underwent advanced transthoracic echocardiography as well as standard blood tests. In particular, thyroid-stimulating hormone, creatinine, C-reactive protein, and leukocytes were analyzed. We excluded 15 patients because of new onset AF on 12-lead ECG (n = 2); severe obesity, expected substantially to compromise echocardiography and the signal from the ICM (n = 3); severe aortic stenosis on the echocardiography (n = 1); no longer meeting the inclusion criteria because of change in antihypertensive or antidiabetic treatment (n = 4); and declining ICM implantation after inclusion in the study (n = 5). Thus, the final population comprised a total of 82 consecutive patients. Figure 1 shows a flowchart of the enrollment process.
Device characteristics, implantation procedure, and interpretation of AF episodes
All patients received an ICM (Reveal XT, Medtronic, Minneapolis, Minnesota), which was implanted subcutaneously in the left parasternal region after careful mapping for the best ECG signal with clearly visible P waves during sinus rhythm. In 1 patient, a Reveal LINQ (Medtronic) was reimplanted after a minor pocket infection of the initially implanted Reveal XT.
The Reveal XT detects AF by analyzing the irregularity and incoherence of successive R-R intervals to identify and classify patterns in the ventricular conduction. Detection of AF requires a minimum episode duration of 2 min, wherein the R-R intervals are analyzed, and the difference in duration between consecutive R-R intervals (ΔR-R) is calculated. A Lorenz plot is constructed from the calculated variability of the ΔR-R intervals. The rhythm is classified as AF, when the variability of the ΔR-R intervals shows a certain pattern of uncorrelated irregularity (16,17). The ICM was at baseline programmed with default settings (Table 1). All patients were followed using the secure online database, CareLink Network (Medtronic). This system allows the transmission of full device data and diagnostic information. Patients were instructed to transmit data once every month. We found visible P waves on the ECG strips from the first transmission 1 month after implantation and onward in all patients. The episode log and all recorded ECGs were adjudicated by 2 experienced electrophysiologists, who were blinded for other study data. All patients were asked about any cardiopulmonary symptoms at the time of their initial detected AF episode and after 1 year of follow-up. One month after ICM implantation all patients underwent a 2-channel 72-h Holter monitoring (Lifecard CF, SpaceLabs Healthcare, Snoqualmie, Washington). All recordings were analyzed by trained staff using dedicated software (Pathfinder SL, SpaceLabs Healthcare) and adjudicated by 2 experienced electrophysiologists.
AF definition and treatment initiation
For the ICM the presence of AF was defined as at least 1 episode of absolutely irregular rhythm without P waves of ≥2 min duration, due to the detection algorithm of the device. Because we did not expect a high frequency or long duration of AF episodes in our study population, we chose the 2-min cutoff defining an AF episode. On Holter monitoring the presence of AF was defined as ≥1 episode of absolutely irregular rhythm without P waves lasting 30 s or more in accordance with common recommendations (3). The inclusion criteria correspond to a CHA2DS2-VASc (congestive heart failure, hypertension, age ≥75 years, diabetes mellitus, prior stroke or TIA or thromboembolism, vascular disease, age 65–74 years, sex category) score of 3 to 5. OAC treatment was initiated upon an AF episode lasting ≥6 min. If necessary, either rate or rhythm control was initiated as appropriate.
The study was approved by the Regional Scientific Ethical Committees for Southern Denmark (Project-ID S20130062) and the Danish Data Protection Agency (case no. 13/12874). The study is registered on ClinicalTrials.gov (NCT02041832) and complies with the Declaration of Helsinki. Prior to enrollment, all patients were carefully informed both orally and in writing, that the indication for ICM was solely study related, just as they were informed in the same way of the study procedure and potential consequences. All patients gave written informed consent before enrollment.
Data acquisition and statistical analysis
Data were obtained from the patient’s medical records. All data were entered into a database (Excel 2010, Microsoft Corporation, Redmond, Washington) and analyzed using STATA 14 (StataCorp LP, College Station, Texas). Categorical variables are presented as percentages and were analyzed using Fisher’s exact or chi-square test where appropriate. Continuous variables are presented as mean ± SD, if data follow a normal distribution, and Student’s t test was used to test for differences between groups. Non–Gaussian-distributed continuous variables are reported as median with interquartile range (IQR), and Wilcoxon rank-sum test was used. A 2-sided p-value <0.05 was considered statistically significant. Because of the similarity in the baseline characteristics within the 2 groups, no adjustment was made. The Kaplan-Meier function was used to construct a cumulative incidence curve for subclinical AF.
A total of 82 patients ≥65 years (median age 71.3 years [IQR: 67.4 to 75.1 years]) with diabetes mellitus and arterial hypertension, and no history of AF or any other cardiovascular disease, were consecutively included between December 1, 2013, and November 1, 2015. Table 2 shows the baseline characteristics for the entire cohort and for patients with and without subclinical AF, respectively. There were no significant differences between patients with and without subclinical AF regarding age, sex, body mass index, CHA2DS2-VASc score, and medical treatment. Patients with subclinical AF had significantly higher creatinine, 99 μmol/l (IQR: 91 to 109 μmol/l) versus 85 μmol/l (IQR: 71 to 117 μmol/l) (p = 0.03), than did those without AF.
New onset AF
All patients received an ICM after a median of 16 days (IQR: 12 to 22 days) from inclusion. Patients were followed for a median of 588 days (IQR: 453 to 712 days). During follow-up, 17 patients (20.7%) had subclinical AF detected by the ICM with a median time to first-detected episode of 91 days (IQR: 41 to 251 days) from inclusion. Fourteen (17%) patients had 1 or more AF episodes of ≥6 min duration and were therefore offered OAC treatment, which was initiated in all of them. Among patients with detected subclinical AF, rate control, either beta-blocker or digoxin, was initiated in 4 patients, whereas 5 patients already received beta-blocker at the time of inclusion. The duration of the longest AF episode in each patient is shown in Figure 2. All patients denied any cardiopulmonary symptoms at the time of their initial AF episodes. At the annual control they were again asked for any cardiopulmonary symptoms, which again they all refused to have. Compared with continuous ICM monitoring, only 2 (2.4%) patients had AF episodes on 72-h Holter monitoring. These episodes were also detected by the ICM and thus lasted longer than 2 min. Almost all patients had their first AF episode within the first year of follow-up (Figure 3). During follow-up 1 patient presented with advanced second-degree atrioventricular block 2:1 conduction for 34 s during daytime. The heart rate was 30 beats/min, and the patients did not have any symptoms during the episode. According to guidelines a pacemaker was implanted. There were no other significant arrhythmias requiring treatment.
Incidence of subclinical AF
Seventeen of 82 (20.7%) older patients had at least 1 episode of paroxysmal subclinical AF on continuous rhythm monitoring with the ICM. All patients also underwent 72-h Holter monitoring, which only detected subclinical AF in 2 (2.4%) patients. Thus continuous monitoring with the ICM was superior to intermittent Holter monitoring in this group of patients.
Almost all cases of subclinical AF were found within the first year after ICM implantation, suggesting only modest benefit of further monitoring beyond 1 year of follow-up (Figure 3). Patients with subclinical AF did not have a higher CHA2DS2-VASc score than did patients with no subclinical AF (p = 0.44). They had slightly higher creatinine levels than patients without subclinical AF, but the levels between the 2 groups overlap, and the medians are still within the normal range for serum creatinine. The incidence of paroxysmal subclinical AF in this population is surprisingly high, taken into consideration that they had no other risk factors for AF apart from age, diabetes mellitus and hypertension. The STROKE-STOP (Systemic ECG Screening for Atrial Fibrillation Among 75 Year Old Subjects in the Region of Stockholm and Halland, Sweden) study investigated more than 7,000 patients 75 to 76 years of age with no history of AF. Beyond older age these patients had several other risk factors for AF and underlying cardiovascular conditions, including stroke or TIA, compared with our study population. They underwent intermittent monitoring using handheld single-lead ECG recorders. A 30-s ECG strip was recorded and automatically transmitted to a database twice daily for 2 weeks. The overall AF incidence rate in this population was 12.3% (18), which would probably have been even higher if continuous monitoring had been performed.
The ASSERT (Asymptomatic Atrial Fibrillation and Stroke Evaluation in Pacemaker Patients and the Atrial Fibrillation Reduction Atrial Pacing) trial found subclinical atrial tachyarrhythmias in 34.6% of 2,580 patients with hypertension, and indication for pacemaker or implantable cardioverter-defibrillator, and no previous history of AF during a mean follow-up of 2.5 years using the monitor function of the devices (2). For several reasons, this trial is also different from our study. Patients included in this trial were older than in our study and were also allowed to have underlying cardiovascular diseases, whereas only about one-third of the patients had diabetes. As all patients had a device indication, the condition affecting their conducting system might have made them also more prone to AF with a higher baseline risk compared with the patients in our study. Moreover, the definition of an atrial tachyarrhythmia was different with an episode lasting at least 6 min at an atrial rate of >190 beats/min.
The randomized CRYSTAL-AF (Continuous Cardiac Monitoring to Assess Atrial Fibrillation After Cryptogenic Stroke) study included patients with a previous stroke or TIA and randomized them to monitoring with an ICM similar to our study or usual care. In the usual care group ECG monitoring was performed at the discretion of the investigator. After 1 year of follow-up 12.4% of the patients in the ICM group had AF compared with 2% in the conventional follow-up group, showing that continuous monitoring detects more episodes of AF than conventional monitoring (19). We also found that continuous monitoring was superior to conventional Holter monitoring in our study population, even though patients were different in both studies. Because of the lack of symptoms and no cardiovascular comorbidity, the patients in the present study are not likely to attend any health services regarding their cardiac status. They are typically offered annual 12-lead ECG at their diabetes control, which most likely will not detect paroxysmal AF. Nevertheless these patients are at substantial risk of being undiagnosed with AF and not receiving timely OAC treatment. We pre-specified the criterion for initiation of OAC treatment to be ≥1 episode of AF lasting at least 6 min regardless of the detection method, as the AF burden, whether of symptomatic or subclinical character, related to a substantially increased thromboembolic risk, is still unknown. This has also been pinpointed in a recent review by Camm et al. (20). The ongoing ARTESiA (Apixaban for the Reduction of Thrombo-Embolism in Patients With Device-Detected Sub-Clinical Atrial Fibrillation) trial aims to determine if treatment with Apixaban, compared with acetylsalicylic acid, will reduce the risk of ischemic stroke and systemic embolism in patients with device-detected subclinical AF and additional risk factors for stroke (NCT01938248). The also ongoing NOAH (Non-vitamin K Antagonist Oral Anticoagulants in Patients With Atrial High Rate Episodes) trial investigates whether OAC treatment with Edoxaban is superior to acetylsalicylic acid treatment in patients with atrial high-rate episodes and at least 2 stroke risk factors but without AF regarding prevention of stroke, systemic embolism, or cardiovascular death (NCT02618577). These studies may answer the question of whether subclinical AF should be treated with OACs as well as symptomatic AF and find a more accurate risk-benefit ratio of OAC treatment in these patients.
Due to aging of the population and current lifestyle, the prevalence of AF is expected to grow in future decades as well as the burden from its associated complications (21). Beyond primary prevention of the risk factors for AF, it is highly relevant to consider screening for AF in high-risk patients, which might prevent stroke. The ASSERT-II (The Prevalence of Sub-Clinical Atrial Fibrillation Using an Implantable Cardiac Monitor in Patients with Cardiovascular Risk Factors) trial aimed to determine the detection rate of subclinical AF (≥5 min) within an average of 12 months following implantation of an ICM in patients with known cardiovascular risk factors and left atrial enlargement, but without prior AF (NCT01694394). These patients were allowed to have heart failure, prior stroke, and valvular heart disease and are, thus, not comparable to the population of this present study. Our study, therefore, differs from these prior and ongoing studies by investigating patients with no other underlying cardiovascular conditions apart from the risk factors hypertension and diabetes mellitus. Although these conditions were not present, we still found a substantially high incidence rate of subclinical AF affecting 1 of 5 patients.
Due to the previously mentioned detection algorithm, the ICM has limitations regarding estimation of the AF burden. The ICM, Reveal XT, can store up to 49.5 min of recorded ECG, which is allocated to 27 min of automatically activated events and 22.5 min of patient-activated events. In addition, the ICM has an episode log that can store 30 automatically detected AF episodes and up to 10 patient-activated episodes. When the memory is full, an additional episode will overwrite the oldest stored episode (16). High numbers of premature beats often lead to false positive AF episodes because of the alterations in R-R intervals. These false positive AF episodes can be eliminated during the manual analysis of the stored ECG examples, as long as the ECG memory is not filled up. In patients with a high number of false positive AF episodes, the diagnostic accuracy of the ICM may decrease due to the limited ECG storage capacity of the device. When the memory is filled up with false positive episodes, additional AF episodes are only listed in the episode log of the ICM without corresponding ECGs available for verification. The XPECT (Reveal® XT Performance Trial) trial also pointed to this limitation of the ICM and found false positive AF episodes in 15% of the study population. The study was performed in a specific patient population with prior documented AF, and the results may not be generalized to other patient populations (16). During the study the ICM needed to be reprogrammed in 6 patients (7%) due to numerous premature beats, which caused many false positive AF episodes filling up the ECG memory. We reprogrammed the ICM from balanced sensitivity regarding AF detection to least sensitivity with ectopy rejection on. In spite of reprogramming there were still cases of full memory in these patients within the follow-up time, and the incidence of subclinical AF in this study population might therefore be slightly underestimated. In patients with both new onset subclinical AF and numerous premature beats it is not possible to manually verify all actual AF episodes, when there are more episodes than the memory can store with ECG examples making the calculation of AF burden and the true number of AF episodes impossible.
Holter versus ICM
We also compared ICM to conventional Holter monitoring to estimate any difference in the detection rate of subclinical AF by the 2 devices in the present study population. We chose 72-h Holter monitoring to simulate an everyday clinical setup regarding screening of AF in patients with symptoms according to national practice. We found that ICM was superior to Holter monitoring in terms of detection of subclinical paroxysmal AF in this study population. This finding is consistent with those in the XPECT and CRYSTAL-AF studies (16,19). Because of the costs related to the ICM, this monitoring method is less suitable for screening for subclinical AF in large unselected populations. Screening with an ICM should probably be reserved for patients with even high-risk or with rare symptoms (22). In addition the ICM only detects episode of AF ≥2 min, where the definition of AF is ≥30 s (3). It is still a matter of ongoing debate when AF, in terms of pattern, episode duration, and AF burden, becomes clinically relevant with respect to comorbidity, in particular stroke.
Because of the detection algorithm and storage capacity of the ICM as mentioned previously, the incidence rate of subclinical AF at 20.7% in this study population might potentially be underestimated. The results of this study refer only to patients ≥65 years of age with hypertension and diabetes mellitus and cannot be extrapolated to other patient populations. Another important limitation is the absence of a control group, but implantation of an ICM in an age- and gender-matched group of healthy individuals serving as control group would be ethically untenable and not be approved by the ethics committee.
Our study differs from prior and ongoing studies by investigating patients with no cardiovascular risk factors apart from ≥65 years of age, hypertension, and diabetes mellitus. We found a considerably high incidence rate of subclinical AF in these patients, suggesting that this population with a substantial risk of thromboembolism should be taken into consideration when discussing relevant patient populations for AF screening. None of the patients presented any symptoms of AF, potentially putting them at risk of stroke or other thromboembolic events, if OAC treatment is not timely initiated, because the patients remain undiagnosed. Furthermore, continuous monitoring with an ICM detected significantly more AF episodes than 72-h Holter monitoring in this group of patients, which is in line with previous studies in different study populations. Whether the thromboembolic risk is the same as for symptomatic AF patients is still unknown. Future studies are necessary to elucidate this.
COMPETENCY IN MEDICAL KNOWLEDGE: Current guidelines do not recommend monitoring with ICMs of patients without symptoms. Nevertheless, the same guidelines recommend OAC treatment in case of documented AF regardless of symptoms, when CHA2DS2-VASc score is ≥2. We demonstrate that in patients ≥65 years of age without any known cardiovascular disease, other than hypertension and diabetes mellitus, subclinical AF was common. As these patients have a high thromboembolic risk with a CHA2DS2-VASc score ≥3, they could possibly benefit from OAC treatment. However, it is unknown if the thromboembolic risk is the same as for symptomatic AF.
TRANSLATIONAL OUTLOOK: Future randomized controlled trials are warranted to elucidate the risk-benefit ratio of OAC treatment in these patients with subclinical AF. Also we do not know the rate of subclinical AF detected by an ICM in people ≥65 years of age without hypertension, diabetes mellitus, or any known cardiovascular disease. In these patients OAC treatment is still indicated due to their high age, and it would therefore be interesting to reveal the rate of subclinical AF in this population.
The authors thank the patients and support staff that participated in the study, and the financial support that made the study possible. There were no sponsorships by industry.
The study was funded by the University of Southern Denmark; Department of Cardiology, Hospital of Southern Jutland; Department of Cardiology, Odense University Hospital; A.P. Møller Foundation for the Advancement of Medical Science, Copenhagen, Denmark; Knud and Edith Eriksen’s Memorial Foundation, Sønderborg, Denmark; and Brødrene Hartmann’s Foundation, Copenhagen, Denmark. The authors have reported that they have no relationships relevant to the contents of this paper to disclose.
All authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’ institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information, visit the JACC: Clinical Electrophysiology author instructions page.
- Abbreviations and Acronyms
- atrial fibrillation
- insertable cardiac monitor
- interquartile range
- oral anticoagulant
- transient ischemic attack
- Received April 5, 2017.
- Revision received June 8, 2017.
- Accepted June 15, 2017.
- 2017 American College of Cardiology Foundation
- Lloyd-Jones D.M.,
- Wang T.J.,
- Leip E.P.,
- et al.
- Wolf P.A.,
- Abbott R.D.,
- Kannel W.B.
- Huxley R.R.,
- Lopez F.L.,
- Folsom A.R.,
- et al.
- Rienstra M.,
- Lubitz S.A.,
- Mahida S.,
- et al.
- Freeman J.V.,
- Simon D.N.,
- Go A.S.,
- et al.
- Ryden L.,
- Grant P.J.,
- Anker S.D.,
- et al.
- Hindricks G.,
- Pokushalov E.,
- Urban L.,
- et al.
- Bjorkenheim A.,
- Brandes A.,
- Chemnitz A.,
- Magnuson A.,
- Edvardsson N.,
- Poci D.
- Svennberg E.,
- Engdahl J.,
- Al-Khalili F.,
- Friberg L.,
- Frykman V.,
- Rosenqvist M.
- Camm A.J.,
- Simantirakis E.,
- Goette A.,
- et al.
- Alonso A.,
- Bengtson L.G.
- Giada F.,
- Gulizia M.,
- Francese M.,
- et al.