Pilot Randomized Trial of Outpatient Cardiac Monitoring After Cryptogenic Stroke
Background and Purpose—Observational studies indicate that outpatient cardiac monitoring detects previously undiagnosed atrial fibrillation (AF) in 5% to 20% of patients with recent stroke. However, it remains unknown whether the yield of monitoring exceeds that of routine clinical follow-up.
Methods—In a pilot trial, we randomly assigned 40 patients with cryptogenic ischemic stroke or high-risk transient ischemic attack to wear a Cardionet mobile cardiac outpatient telemetry monitor for 21 days or to receive routine follow-up alone. After thorough investigation, we excluded patients with documented AF or other apparent stroke pathogenesis. We contacted patients and their physicians at 3 months and at 1 year to ascertain any diagnoses of AF or recurrent stroke or transient ischemic attack.
Results—The baseline characteristics of our cohort broadly matched those of previous observational studies of monitoring after stroke. In the monitoring group, patients wore monitors for 64% of the assigned days, and 25% of patients were not compliant at all with monitoring. No patient in either study arm received a diagnosis of AF. Cardiac monitoring revealed AF in zero patients (0%; 95% confidence interval, 0%–17%), brief episodes of atrial tachycardia in 2 patients (10%; 95% confidence interval, 1%–32%), and nonsustained ventricular tachycardia in 2 patients (10%; 95% confidence interval, 1%–32%).
Conclusions—In the first reported randomized trial of cardiac monitoring after cryptogenic stroke, the rate of AF detection was lower than expected, incidental arrhythmias were frequent, and compliance with monitoring was suboptimal. Our findings highlight the challenges of prospectively identifying stroke patients at risk for harboring paroxysmal AF and ensuring adequate compliance with cardiac monitoring.
Failure to diagnose atrial fibrillation (AF) may result in suboptimal therapy for secondary stroke prevention.1 However, paroxysmal AF is not always apparent on presentation.2 In studies without control groups, cardiac monitoring after stroke has been shown to detect previously undiagnosed AF in at least 5% of patients.3 However, AF eventually may have been diagnosed even without monitoring. Therefore, we performed a pilot randomized trial of outpatient cardiac monitoring in patients with ischemic stroke or transient ischemic attack.
We enrolled adult patients with ischemic stroke or high-risk transient ischemic attack (ABCD2 score ≥4). Hospitalized patients underwent ≥24 hours of cardiac telemetry, and those with AF before discharge were excluded. We excluded patients with lacunar infarcts, ≥50% stenosis of relevant arteries, likely cardioembolism, or other apparent cause. We excluded patients ineligible to receive anticoagulation or with onset of symptoms >60 days previously. This trial was approved by the University of California, San Francisco, Committee on Human Research. All patients provided written informed consent.
Patients were assigned to cardiac monitoring or routine follow-up using random permuted blocks of varying sizes. For practical reasons, blinding was not performed. Patients were discharged with antiplatelet therapy, with a plan to begin anticoagulation if AF had been diagnosed. All patients were scheduled to see their primary care physician within 1 month and our stroke clinic within 3 months, and they were educated to report symptoms of AF at these visits. The monitoring group was additionally assigned 21 days of Cardionet Mobile Cardiac Outpatient Telemetry, which has >99% sensitivity for AF lasting >30 seconds.4 To ensure specificity, all device-labeled AF episodes were manually reviewed by a cardiologist (G.F.).
Our primary feasibility outcomes were enrollment of 40 patients in 2 years, completion of assigned monitoring in ≥70% of patients, and full follow-up for ≥90% of patients. Our primary safety outcome was any adverse event resulting directly from use of the cardiac monitoring device. Secondary outcomes included new diagnoses of AF within 3 months and 1 year.
We used the t test or rank sum test for comparisons of continuous measures and Fisher exact test for categorical variables. Feasibility and efficacy analyses were performed using the intention-to-treat principle, whereas safety analyses involved the per-protocol population. Post hoc, we examined the relationship between stroke severity and monitoring compliance using multiple regression. Statistical analysis was performed using Stata (version 11; StataCorp). Further details about study design are available in the online-only Data Supplement.
We enrolled 40 patients between October 29, 2009 and May 24, 2011. Baseline characteristics were well-balanced between the 2 groups (Table). Patients underwent a mean 49 hours (±32 hours) of inpatient cardiac telemetry before discharge. Outpatient monitoring began 22 days (±12 days) after symptom onset.
All feasibility criteria were met, with full follow-up for 38 patients (95%) and completion of assigned monitoring in 15 of 20 patients (75%). However, 4 of these 15 patients were not fully compliant, resulting in overall compliance of 64%. After controlling for age, we found a nonsignificant trend toward less compliance in patients with higher baseline National Institutes of Health Stroke scale scores (β coefficient, −0.24; P=0.35) and modified Rankin scale scores (β coefficient, −0.26; P=0.32).
No patient received a diagnosis of AF. Cardiac monitoring revealed AF in zero patients (0%; 95% confidence interval, 0%–17%), but did reveal brief episodes (<10 seconds) of atrial tachycardia in 2 patients (10%; 95% confidence interval, 1%–32%). One of these episodes was incorrectly labeled as AF by the automated telemetry software (online-only Data Supplement). Nonsustained ventricular tachycardia was detected by cardiac monitoring in 2 patients (10%; 95% confidence interval, 1%–32%).
No serious adverse event occurred that was attributable to the monitoring intervention. One of the 15 monitored patients (7%; 95% confidence interval, 2%–32%) had development of contact dermatitis, which resolved within a few days of removing the leads.
In the first reported randomized trial of cardiac monitoring after cryptogenic stroke, we have established the safety and feasibility of randomizing patients to outpatient cardiac monitoring or usual follow-up. However, patients wore monitors for only 64% of the assigned time. This may be because of the inconvenience of devices with multiple leads, particularly for patients with functional limitations. Our results indicate that the tolerability of outpatient monitoring must be carefully considered when devising strategies to rule out AF as a cause of stroke.
The low rate of AF detected in our monitoring group differs from most published studies of similar monitoring techniques after stroke.4–6 This may be because of chance, because our pilot study aimed to establish feasibility and lacked power to examine clinical outcomes, which is apparent in the wide confidence interval around our AF detection rate (0%–17%) and was exacerbated by imperfect compliance with monitoring. The low AF rate also may reflect differences in populations, although comparison of our cohort with other published cohorts does not reveal obvious differences (online-only Data Supplement). Last, our results may indicate publication bias discouraging the reporting of negative studies. Regardless, the failure of monitoring to detect any AF in our study reinforces the importance of randomized trials for establishing the role of cardiac monitoring after stroke.
Despite not detecting arrhythmias meeting the standard definition of AF, cardiac monitoring did reveal 2 brief episodes of atrial tachycardia. Nonspecific supraventricular tachyarrhythmias have been reported to occur in >23% of patients with cryptogenic stroke undergoing cardiac monitoring.4 Recent evidence indicates that atrial tachycardias increase AF risk and may increase stroke risk, even in the absence of clinically apparent AF.7 The significance and optimal management of these nonspecific arrhythmias await clarification by future studies.
Given the remaining uncertainties in this field and the different monitoring strategies available, further comparative studies such as ours and the ongoing Event Monitor Belt for Recording Atrial Fibrillation After a Cerebral Ischemic Event (EMBRACE) and CRYptogenic STroke And underLying Atrial Fibrillation (CRYSTAL-AF) trials will be needed to identify optimal strategies for diagnosing AF as a cause of cerebral ischemia, thereby reducing the incidence of recurrent stroke.
Sources of Funding
This work was funded by Cahill Family Foundation.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.112.679100/-/DC1.
- Received October 2, 2012.
- Revision received October 2, 2012.
- Accepted October 4, 2012.
- © 2013 American Heart Association, Inc.
- Connolly S,
- Pogue J,
- Hart R,
- Pfeffer M,
- Hohnloser S,
- Chrolavicius S,
- et al
- Ziegler PD,
- Glotzer TV,
- Daoud EG,
- Wyse DG,
- Singer DE,
- Ezekowitz MD,
- et al
- Liao J,
- Khalid Z,
- Scallan C,
- Morillo C,
- O’Donnell M
- Tayal AH,
- Tian M,
- Kelly KM,
- Jones SC,
- Wright DG,
- Singh D,
- et al
- Flint AC,
- Banki NM,
- Ren X,
- Rao VA,
- Go AS