Abstract
Background
Insertable cardiac monitors (ICMs) are often implanted after cryptogenic stroke (CS) to detect atrial fibrillation (AF) and guide anticoagulation. However, the impact of this practice on stroke recurrence remains unclear.
Objectives
This study sought to compare stroke recurrence between CS patients with and without ICM-detected AF and identify predictors of recurrent stroke.
Methods
A retrospective analysis was conducted of consecutive patients admitted to a stroke center with no history of AF who received an ICM for CS or transient ischemic attack.
Results
Among 840 patients (median follow-up 990 days), AF was detected in 235. Recurrent strokes occurred in 112 patients, of which 70.5% occurred in patients without AF ever being detected. Patients with AF were observed to have increased stroke recurrence when analyzing AF as a time-varying exposure (adjusted HR: 2.73; 95% CI: 1.33-5.59), despite most patients with AF being started on anticoagulation (97.9%). Other covariates associated with increased risk of recurrent stroke in multivariable analyses included an elevated CHA2DS2-VASc (Congestive heart failure, Hypertension, Age ≥75 years, Diabetes, history of Stroke or transient ischemic attack, VAscular disease, Age 65-74 years, and Sex category [female]) score ≥4 and male sex. In exploratory subgroup analysis of patients with a CHA2DS2-VASc score <4 and age of <65 years (n = 107), only 1 patient with a recurrent stroke had AF detected (<1%).
Conclusions
Despite the high rate of detection of AF following CS by ICMs, most strokes occurring in this population may not be due to AF. Further randomized trials are needed to verify these observations and clarify the benefit of routine ICM implantation for secondary CS prevention.
Key words: arrhythmia, atrial fibrillation, cryptogenic stroke, implantable loop recorder, insertable cardiac monitor
Central Illustration
Ischemic stroke is among the leading causes of death and disability in the United States.1 Approximately 30% of strokes have no identifiable cause after a standard initial diagnostic work-up and are labeled cryptogenic.2,3 Atrial fibrillation (AF) is a potential embolic source for cryptogenic stroke (CS) which can easily evade initial detection due to its paroxysmal and often asymptomatic nature.4 Nevertheless, patients who have had a stroke due to AF face a high risk of stroke recurrence.5 Thus, strategies to improve the detection and appropriate treatment of AF may potentially reduce the risk of stroke recurrence in this population.
Among stroke patients without evidence of AF on initial evaluation, prolonged cardiac monitoring is recommended to increase the detection of AF.6,7 Insertable cardiac monitors (ICMs) have shown the highest yield for detecting AF following stroke, with up to 30% of patients with CS having ICM-detected AF at 3 years.8,9 The increased rate of AF detection with ICM monitoring also leads to a corresponding increase in anticoagulation (AC) initiation as compared to conventional monitoring.9 However, the impact of ICM monitoring on mitigating subsequent stroke risk, and the contribution of device-detected AF to recurrent strokes, has not been well established. Prior studies have not shown a clear reduction in stroke recurrence with the use of ICMs as compared to conventional cardiac monitoring, despite the increased detection of AF and initiation of AC with ICMs.9,10 To address this gap in knowledge, we conducted a retrospective analysis of CS patients admitted to a stroke center who received an ICM following CS to identify characteristics and potential predictors of recurrent stroke.
Methods
Study population
We performed a retrospective analysis of all patients who received an ICM (Medtronic Reveal LINQ or Abbott Confirm RX) after admission for CS or transient ischemic attack (TIA) between January 2015 and December 2022 at the Buffalo General Medical Center/Gates Vascular Institute, a large comprehensive stroke center. Evaluation by the stroke-neurology service for all included patients was required to establish the diagnosis of stroke or TIA, which was defined as new neurological deficits concordant with findings on neurovascular imaging (brain magnetic resonance imaging or computed tomography). Stroke or TIA was determined to be cryptogenic after initial evaluation with vascular imaging, transthoracic or transesophageal echocardiography, and inpatient telemetry monitoring was unrevealing for any clear etiology. It was standard practice at our institution that patients without AF detected during their hospitalization or any other clear etiology for their stroke would undergo ICM implantation prior to discharge. Patients with a hemorrhagic stroke, prior diagnosis of AF, or those who were already on AC for alternative indications, such as pulmonary embolism or deep vein thrombosis, were excluded from the analysis. Patients without follow-up data were also excluded. The study received proper ethical oversight and was approved by the University at Buffalo Institutional Review Board. The study was conducted in accordance with the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) and RECORD (REporting of studies Conducted using Observational Routinely collected health Data) guidelines for observational studies.11,12
Data collection
Patient characteristics, including demographics and comorbidities, were obtained by accessing the medical center’s “Get With The Guidelines” Stroke Database (American Heart Association). Information was collected on age, sex, and race in addition to type of stroke and medical comorbidities (Table 1). The patients’ demographic information subsequently underwent an independent, 2-physician verification process utilizing the medical center’s electronic health record. The patients’ data were deidentified after the demographic information was recorded. The CHA2DS2-VASc (Congestive heart failure, Hypertension, Age ≥75 years, Diabetes, history of Stroke or transient ischemic attack, VAscular disease, Age 65-74 years, and Sex category [female]) score was calculated, and the medication list was reconciled using the electronic prescription fill history.
Table 1.
Baseline Characteristics
| All (N = 840) | AF Group (n = 235) | Non-AF Group (n = 605) | P Value | |
|---|---|---|---|---|
| Age (y) | 67.9 ± 12.5 | 72.4 ± 11.7 | 66.1 ± 12.4 | <0.001 |
| Female | 405 (48.2) | 102 (43.4) | 303 (50.1) | 0.082 |
| Index event | ||||
| Stroke | 768 (91.4%) | 209 (88.9%) | 559 (92.4%) | 0.11 |
| TIA | 72 (8.6%) | 26 (11.1%) | 46 (7.6%) | 0.11 |
| CHA2DS2-VASc | 5.2 ± 1.6 | 5.7 ± 1.5 | 5.1 ± 1.5 | <0.001 |
| Hypertension | 720 (85.7) | 215 (91.5) | 505 (83.5) | 0.003 |
| Diabetes mellitus | 325 (38.7) | 89 (37.9) | 236 (39.0) | 0.76 |
| Heart failure | 201 (23.9) | 84 (35.7) | 117 (19.3) | <0.001 |
| Hyperlipidemia | 637 (75.8) | 182 (77.4) | 455 (75.2) | 0.50 |
| Peripheral arterial disease | 103 (12.3) | 39 (16.6) | 64 (10.6) | 0.017 |
| Obstructive sleep apnea | 102 (12.1) | 27 (11.5) | 75 (12.4) | 0.72 |
| Coronary artery disease | 250 (29.8) | 87 (37.0) | 163 (26.9) | 0.004 |
| Deep vein thrombosis/pulmonary embolism | 8 (1.0) | 0 (0) | 8 (1.3) | 0.077 |
| Hormone replacement therapy | 20 (2.4) | 3 (1.3) | 17 (2.8) | 0.19 |
| Migraine | 42 (5.0) | 7 (3.0) | 35 (5.8) | 0.094 |
| Previous TIA/stroke/SAH | 388 (46.2) | 112 (47.7) | 276 (45.6) | 0.60 |
| Chronic kidney disease | 137 (16.3) | 42 (17.9) | 95 (15.7) | 0.45 |
| Dementia | 56 (6.7) | 22 (9.4) | 34 (5.6) | 0.064 |
| Obesity | 434 (51.7) | 120 (51.1) | 314 (51.9) | 0.83 |
| Sudden cardiac arrest | 6 (0.7) | 1 (0.4) | 5 (0.8) | 0.54 |
| Prosthetic heart valve/repair | 30 (3.6) | 11 (4.7) | 19 (3.1) | 0.28 |
| COVID-19 | 39 (4.6) | 11 (4.7) | 28 (4.6) | 0.97 |
| LAVI (mL/m2) | 29.9 ± 17.3 (n = 538) | 33.8 ± 19.2 (n = 166) | 28.1 ± 16.0 (n = 372) | <0.001 |
| LVEF | 0.002 | |||
| LVEF >50% | 784 (93.3) | 210 (89.4) | 574 (94.9) | |
| LVEF 35%-50% | 46 (5.5) | 18 (7.7) | 28 (4.6) | |
| LVEF <35% | 10 (1.2) | 7 (3.0) | 3 (0.5) | |
Values are mean ± SD or n (%). P values compare AF to non-AF groups. P values <0.05 (ie, statistically significant differences) are in boldface.
AF = atrial fibrillation; CHA2DS2-VASc = Congestive heart failure, Hypertension, Age ≥75 years, Diabetes, history of Stroke or transient ischemic attack, VAscular disease, Age 65-74 years, and Sex category (female); LAVI = left atrial volume index; LVEF = left ventricular ejection fraction; SAH = subarachnoid hemorrhage; TIA = transient ischemic attack.
After hospital discharge, ICM data were obtained at regular intervals, either remotely or in person, for arrhythmia detection. The ICM analyzed electrocardiogram (ECG) rhythms in discrete 2-minute intervals. If the ICM algorithm determined that AF was present during a 2-minute interval, the ECG rhythm strip was classified as AF. While there are many nuances to the algorithm, the odds of detection of AF are high if at least 60 seconds of AF is detected in any 2-minute window. Subsequent 2-minute intervals were analyzed until an interval did not show AF, in which case the AF episode was determined to be terminated. All rhythm strips initially labeled as AF by the ICM detection algorithm were adjudicated by experienced electrophysiologists. For those patients who had AF detected, AC was initiated at the discretion of the treating cardiologist. In most cases, AC was initiated when AF was detected regardless of the duration of AF. It was standard practice to use antiplatelet agents, such as aspirin and/or clopidogrel, in most patients following their CS, switching to oral AC only if AF was detected. While the majority of patients had in-office follow-up appointments, all patients were contacted via phone, at a minimum, when a diagnosis of AF was made.
Following ICM implant, recurrent strokes/TIAs were identified through medical chart review of subsequent hospitalizations. All recurrent stroke/TIA events were defined as per the inclusion criteria and required evaluation by a neurologist to establish the diagnosis. All patients were followed up until their last clinical encounter recorded in the electronic medical records or 3 years (the standard battery life of ICMs at the time of the study), whichever occurred first.
Outcomes
The primary outcome of interest was the rate of stroke/TIA recurrence among patients with ICM-detected AF as compared to those without AF detected. Secondary outcomes of interest included rates of AF detection, AC initiation following ICM implantation, and longest AF episode duration.
Statistical analysis
Baseline demographics and characteristics were summarized with means and SDs for continuous variables and with counts of patients in each group for categorical variables. Continuous variables with a skewed distribution, such as follow-up time, were summarized using medians and 25th-75th percentiles (Q1-Q3). These baseline variables were compared between the 2 cohorts using chi-square tests for categorical variables and 2-sample t-tests for continuous variables. Since AF can occur at any time during follow-up, we reframed it to be a time-varying exposure to mitigate potential immortal time bias. We formulated the AF exposure as a time-varying indicator to identify whether the patient experienced AF by a certain time point when there was no recurrence of stroke. To assess the association of AF with the risk of having the first recurrent stroke/TIA, we utilized a Cox regression model with time-varying AF status and estimated the HR for recurrent stroke/TIA with 95% CIs. Furthermore, we explored the association of baseline demographics and characteristics with the risk of the first stroke/TIA by conducting univariable Cox regression models. The baseline variables included in the model were age, sex, AF duration, index event (ischemic stroke vs TIA), left atrial volume index (LAVI), chronic kidney disease, and CHA2DS2-VASc score (≥4 vs < 4). These covariates were specified a priori based on their clinical relevance and known or proposed association with the outcome (stroke). The duration of the longest AF episode was categorized into the following groups: <6 minutes, 6 minutes to 24 hours, and >24 hours. We also performed multivariable Cox regression using all prespecified covariates in the univariable model to estimate the adjusted HR. The proportional hazards assumption for the Cox regression models was assessed with the Schoenfeld residual test. The linearity assumption was visually assessed by plotting residuals vs each continuous variable. No significant violations were detected. Missing data were handled using multiple imputations. All P values <0.05 were considered statistically significant. All statistical analyses were completed using RStudio.
Results
The cohort consisted of 840 patients with an ICM inserted after an initial CS or TIA. Baseline patient characteristics are shown in Table 1. The index event was an ischemic stroke in 768 patients (91.4%) and TIA in 72 (8.6%). The median follow-up time was 990 days (Q1-Q3: 480-1,095).
There were 235 patients who had confirmed AF episodes during follow-up, with a median time to AF diagnosis of 268 days (Q1-Q3: 59-682). Patients diagnosed with AF tended to be older, with a higher CHA2DS2-VASc score, larger LAVI, and lower left ventricular ejection fraction (Table 1). The cumulative incidence for AF detection at 3 years was 32.8% (95% CI: 29.0%-36.4%) (Figure 1). When analyzed by the duration of the longest AF episode, 40 patients had AF <6 minutes, 158 had AF ranging 6 minutes to 24 hours, and 26 had AF longer than 24 hours. In 11 patients, the longest AF duration was not available.
Figure 1.
Cumulative Incidence of AF Detection
Cumulative incidence estimate for AF detection was 32% at 3 years (95% CI: 29.0%-36.4%). AF = atrial fibrillation.
Recurrent strokes occurred in 112 patients, of which 83.9% were ischemic, 14.3% were TIAs, and 1.8% were hemorrhagic strokes. Forty-one patients had multiple recurrent strokes or TIAs. Of those patients with ICM-detected AF, 230 patients (97.9%) were started on AC, regardless of duration of AF. In patients without AF, 40 patients (6.6%) were also started on AC during follow-up. Among those patients who had AF and received AC, 86% received a direct oral anticoagulant, 7.8% received warfarin, and 5.7% received a left atrial appendage occlusion device.
Characteristics of recurrent stroke/TIA in patients with and without AF
Out of all 112 recurrent strokes, 79 (70.5%) of them occurred in patients without AF being detected at any time during follow-up. The occurrence of recurrent stroke according to the presence of AF is shown in Table 2. The overall crude rate of recurrent stroke was 14% (33/235) in patients with AF as compared to 13% (79/605) of patients without AF (HR: 1.09; 95% CI: 0.69-1.67).
Table 2.
Recurrent Stroke According to AF Diagnosis
| Recurrent Stroke | No Recurrent Stroke | Total | |
|---|---|---|---|
| AF | 33 (29.5%) | 202 (27.7%) | 235 |
| No AF | 79 (70.5%) | 526 (72.3%) | 605 |
| Total | 112 | 728 | 840 |
Abbreviation as in Table 1.
In patients with AF who suffered a recurrent stroke (n = 33), 13 of these recurrent events occurred while on AC and 20 occurred while off AC, either due to AC not being prescribed, AC being temporarily held for a procedure, or patient noncompliance. With regard to the timing of AF and subsequent strokes, 20 patients had AF detected before their recurrent event while 13 had AF detected only after the recurrent stroke.
To mitigate immortal time bias for AF, we fit a Cox regression model with AF status as a time-varying covariate. The likelihood of recurrent stroke was higher in patients with AF detected as compared to no AF (HR: 1.76; 95% CI: 1.07-2.89; P = 0.026). Figure 2 shows the cumulative incidence of the first recurrent stroke occurring following ICM implantation, stratified by the timing of AF detection. The cumulative incidence of recurrent stroke at 3 years was 13.6% (95% CI: 10.8%-16.3%) and 22.6% (95% CI: 18.3%-26.8%) for patients with no AF vs AF diagnosed on day 1, respectively.
Figure 2.
Cumulative Incidence of Stroke Recurrence by AF Status
Recurrent stroke risk at 3 years when treating AF as a time-varying exposure. The AF group is further stratified based on the timing of AF detection following the index ischemic event. The cumulative incidence of recurrent stroke was higher among patients with detected AF as compared to those without AF (HR: 1.76; 95% CI: 1.07-2.89; P = 0.026). Abbreviation as in Figure 1.
HR estimates from univariable and multivariable Cox regression models for the risk of recurrent stroke with variables of interest are shown in Table 3. AF status was significantly associated with an increased risk of stroke (HR: 1.76; 95% CI: 1.07-2.89). Age, sex, duration of longest AF episode, type of index event, chronic kidney disease, and LAVI were not significantly associated with recurrent strokes, while CHA2DS2-VASc score was significantly associated with recurrent events in the univariable Cox regression model (HR: 1.32; 95% CI: 1.16-1.49). The association between high CHA2DS2-VASc score and the risk of recurrent stroke remained significant in the multivariable model (adjusted HR: 1.49; 95% CI: 1.27-1.75) and AF also remained significantly associated with the risk of stroke (adjusted HR: 2.73; 95% CI: 1.33-5.59). Additionally, we identified that sex is a significant risk factor associated with stroke recurrence in the multivariate regression (adjusted HR: 0.57; 95% CI: 0.38-0.85). Sensitivity analysis repeating the multivariable regression analysis excluding patients with incomplete data (LAVI missing in 302 patients) yielded similar results as analysis using multiple imputations. In further exploratory subgroup analysis of patients with a CHA2DS2-VASc score <4 and age of <65 years (n = 107), 8 recurrent strokes occurred in total and only 1 patient with a recurrent stroke was found to have AF detected (0.9%).
Table 3.
Univariable and Multivariable Cox Regression Results for Risk of Recurrent Stroke
| Covariate | Univariable Model |
Multivariable Model |
||||
|---|---|---|---|---|---|---|
| HR | 95% CI | P Value | Adjusted HR | 95% CI | P Value | |
| Age | 1.01 | 0.99–1.03 | 0.12 | 0.99 | 0.97–1.01 | 0.24 |
| Female | 0.80 | 0.55–1.17 | 0.25 | 0.57 | 0.38–0.85 | 0.005 |
| AF | 1.76 | 1.07–2.89 | 0.026 | 2.73 | 1.33–5.59 | 0.006 |
| AF duration | ||||||
| No AF/Unknown | - | - | - | - | - | - |
| <6 min | 1.04 | 0.45–2.39 | 0.92 | 0.56 | 0.22–1.41 | 0.22 |
| 6 min–24 h | 1.13 | 0.72–1.78 | 0.60 | 0.62 | 0.32–1.17 | 0.14 |
| >24 h | 0.30 | 0.04–2.15 | 0.23 | 0.14 | 0.02–1.06 | 0.057 |
| Index event | ||||||
| Stroke | - | - | - | - | - | - |
| TIA | 0.79 | 0.39–1.62 | 0.52 | 0.70 | 0.34–1.45 | 0.34 |
| LAVI | 0.99 | 0.97–1.01 | 0.38 | 0.98 | 0.97–1.00 | 0.081 |
| CHA2DS2-VASc ≥4 | 1.32 | 1.16–1.49 | <0.001 | 1.49 | 1.27–1.75 | <0.001 |
| CKD | 0.65 | 0.41–1.01 | 0.057 | 0.87 | 0.54–1.39 | 0.55 |
P values <0.05 (ie, statistically significant differences) are in boldface.
CKD = chronic kidney disease; other abbreviations as in Table 1.
Discussion
To our knowledge, the present study is the largest to date evaluating stroke recurrence in patients undergoing ICM implantation following CS or TIA. In our cohort, approximately 33% of patients were ultimately found to have AF, which is similar to previously reported rates of ICM-detected AF in this patient population.9,13,14 However, despite the common occurrence of device-detected AF in our cohort, we found that the majority of recurrent strokes (70%) occurred in patients who had no AF ever detected (Central Illustration). Additionally, patients with AF were seen to have a higher risk of recurrent stroke in time-varying analysis by AF status, despite virtually all patients with AF being initiated on AC (97.9%).
Central Illustration.
ICM-Detected AF Following CS and Risk of Recurrent Ischemic Events
In this retrospective study of 840 consecutive patients undergoing ICM implantation after CS, most recurrent strokes in this population occurred in patients with no AF ever detected, challenging the notion that most recurrent strokes in this population may be due to undetected AF. The likelihood of recurrent stroke was increased in patients with AF detected, despite most patients with AF being started on AC, regardless of duration. In exploratory subgroup analysis, patients <65 years of age and with a CHA2DS2-VASc score <4 may represent a low-risk group in whom deferral of ICM implantation may be reasonable. CHA2DS2-VASc = Congestive heart failure, Hypertension, Age ≥75 years, Diabetes, history of Stroke or transient ischemic attack, VAscular disease, Age 65-74 years, and Sex category (female); CS = cryptogenic stroke; ICM = insertable cardiac monitor; other abbreviation as in Figure 1.
Altogether, these observations may suggest that the majority of CS are not due to AF. Similar findings have been seen in recent clinical trials, which have demonstrated no difference in recurrent strokes with empiric AC vs aspirin in patients with embolic stroke of undetermined source.15, 16, 17 While it is possible that AC initiated after AF detection may have partially mitigated the risk of recurrent stroke in our cohort, it is notable that patients with AF still had a significantly elevated risk of recurrent stroke as compared to those with no AF detected. Previous studies have demonstrated a much lower rate of residual stroke in anticoagulated AF patients, ranging from approximately 1.5% to 5% per year.18,19 In patients with device-detected AF, whether these episodes represent a direct thromboembolic source or are an indirect marker of increased adverse structural, cellular, and pro-thrombotic processes that predispose to stroke is unclear. Most AF detected in our cohort was of short duration (<24 hours) and found remotely after the index event, which may argue against the causal relevance of such episodes. Similar findings were observed in the ASSERT (Asymptomatic Atrial Fibrillation and Stroke Evaluation in Pacemaker Patients and the Atrial Fibrillation Reduction Atrial Pacing Trial) trial, in which very few patients had subclinical AF (SCAF) in the month before experiencing a stroke.20 On the other hand, when stratifying patients by the timing of AF detection, we observed an increase in the risk of recurrent stroke following a diagnosis of AF, regardless of the timing of AF.
Recent trials of AC in SCAF have suggested a potential benefit with AC, although the risk of stroke conferred by SCAF is small and the absolute risk reduction is much less as compared to clinically detected AF.21, 22, 23 Likewise, our study suggests that low burden AF found remotely after an index stroke only through prolonged intensive monitoring may not confer the same risk of stroke as clinically overt AF, and the benefit of routine AC is unclear given the potential competing sources for stroke in this population. Notably, out of the 33 patients who had AF detected and a recurrent stroke, we observed that 13 patients (39%) still had a repeat stroke while on AC. Nevertheless, these observations should be viewed as hypothesis-generating only, as the low number of patients not initiated on AC after a diagnosis of AF, no matter how brief in duration, and the small number of long duration AF episodes limit any firm conclusions on the true impact of AC and cannot exclude a causal relationship of higher burden AF.
While AF may not be responsible for most CS, whether an ICM-based strategy may reduce recurrent strokes over conventional monitoring cannot be determined from the present study given the lack of a comparator group. Randomized trials evaluating the use of ICM following ischemic stroke for this endpoint have been equivocal, although in pooled analyses, there may be a potential benefit based on study design.24 Whether these effects can be fully extrapolated to the CS population is uncertain. It is possible that the early detection of SCAF may lead to a reduction in stroke by motivating more aggressive risk factor modification and optimization of comorbidities rather than the direct effects of AC.
In multivariable analyses, sex and CHA2DS2-VASc score were the only other significant predictors of recurrent stroke in our cohort. Notably, age and left atrial size, which have been shown to predict AF on ICM monitoring following CS,25,26 were not significantly associated with increased stroke recurrence. Similarly, the recent ARCADIA (Atrial Cardiopathy and Antithrombotic Drugs in Prevention After Cryptogenic Stroke) trial showed no difference in recurrent stroke risk with apixaban vs aspirin in patients with CS and with evidence of atrial cardiopathy without AF.27
The latest American College of Cardiology/American Heart Association/American College of Chest Physicians/Heart Rhythm Society AF guidelines give a class 2a recommendation for cardiac monitoring in patients with CS.28 However, there is little guidance for the proper patient selection for upfront ICM monitoring due to a paucity of data. In exploratory subgroup analysis of patients with a CHA2DS2-VASc score <4 and age <65 years (n = 107), we observed that only 1 patient with a recurrent stroke had AF ever detected. This observation suggests that it may be reasonable to defer ICM implantation in younger patients with a low CHA2DS2-VASc score, as this group comprises a very low-risk group in whom recurrent stroke due to clinically relevant AF is unlikely (<1%). Subgroup analysis of the ARTESIA (Apixaban for the Reduction of Thrombo-Embolism in Patients With Device-Detected Sub-Clinical Atrial Fibrillation) trial based on CHA2DS2-VASc score also demonstrated that in patients with a score of <4, the risks of AC treatment for SCAF outweigh the benefits.29 In these low-risk patients, external ambulatory ECG monitoring after CS may be sufficient to provide the highest yield for detecting clinically relevant AF for secondary stroke prevention, with diminishing incremental benefits with ICM implantation. The ongoing Find-AF 2 trial is anticipated to provide further clarity on the benefit of extended cardiac monitoring for secondary stroke prevention.30
Strengths and limitations
A strength of our study is that ICMs were inserted in all patients with CS at our institution, creating a large patient cohort. However, our study has several important limitations that warrant consideration. First, this study is a single-center retrospective study without a control group of CS patients who did not undergo ICM implantation. Furthermore, due to a lack of complete mortality data in our cohort, we were unable to perform a competing risks analysis with death as a competing event, and thus the current analysis treated patients who died during the follow-up as censored. We acknowledge that not accounting for competing events may yield bias in studying the relationship between AF and subsequent stroke. Second, the very small number of long-duration AF episodes limits any conclusive analyses on the graded impact of AF burden and stroke risk by reducing the statistical analyses to a largely binary comparison between no AF and any AF detected. However, this may not necessarily limit the generalizability of our findings to clinical practice, where a binary approach of any AF vs no AF is often utilized to determine AC initiation, particularly in patients with a prior history of CS where the threshold to initiate AC may be lower. Third, it is possible that some patients who had repeat strokes were not documented if they were treated at a facility outside of our hospital network. Fourth, the workup for CS was not standardized, and transesophageal echocardiography was not performed in all patients before proceeding with ICM implantation.
Conclusions
Despite the high detection of AF following CS by ICM monitoring, most strokes occurring in this population may not be due to AF. Younger patients with a low CHA2DS2-VASc score may constitute a low-risk group in whom ICM implantation is unlikely to detect clinically relevant AF. Further large, randomized studies are needed to confirm these findings and clarify the role of routine ICM-based strategies for secondary CS prevention.
Perspectives.
COMPETENCY IN MEDICAL KNOWLEDGE: Although ICM-detected AF is common in CS patients, most recurrent strokes in this population may not be attributable to AF.
COMPETENCY IN PATIENT CARE: Deferral of ICM implantation for the work-up of CS may be reasonable in younger patients with a low CHA2DS2-VASc score, who constitute a low-risk subgroup.
TRANSLATIONAL OUTLOOK: Further adequately powered randomized trials comparing ICM vs extended external monitoring following CS are needed to clearly delineate the role of ICMs in secondary stroke prevention in this population.
Funding support and author disclosures
Dr Curtis has served on medical advisory boards for Janssen Pharmaceuticals, Medtronic, and Abbott; and has received honoraria for speaking from Sanofi Aventis and Medtronic. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
Footnotes
The 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 Author Center.
References
- 1.Martin S.S., Aday A.W., Almarzooq Z.I., et al. 2024 heart disease and stroke statistics: a report of US and global data from the American Heart Association. Circulation. 2024;149(8):e347–e913. doi: 10.1161/cir.0000000000001209. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Kolominsky-Rabas P.L., Weber M., Gefeller O., Neundoerfer B., Heuschmann P.U. Epidemiology of ischemic stroke subtypes according to TOAST criteria: incidence, recurrence, and long-term survival in ischemic stroke subtypes: a population-based study. Stroke. 2001;32(12):2735–2740. doi: 10.1161/hs1201.100209. [DOI] [PubMed] [Google Scholar]
- 3.Li L., Yiin G.S., Geraghty O.C., et al. Incidence, outcome, risk factors, and long-term prognosis of cryptogenic transient ischaemic attack and ischaemic stroke: a population-based study. Lancet Neurol. 2015;14(9):903–913. doi: 10.1016/s1474-4422(15)00132-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Ntaios G. Embolic stroke of undetermined source: JACC review topic of the week. J Am Coll Cardiol. 2020;75(3):333–340. doi: 10.1016/j.jacc.2019.11.024. [DOI] [PubMed] [Google Scholar]
- 5.Independent predictors of stroke in patients with atrial fibrillation: a systematic review. Neurology. 2007;69(6):546–554. doi: 10.1212/01.wnl.0000267275.68538.8d. [DOI] [PubMed] [Google Scholar]
- 6.Kleindorfer D.O., Towfighi A., Chaturvedi S., et al. 2021 guideline for the prevention of stroke in patients with stroke and transient ischemic attack: a guideline from the American Heart Association/American Stroke Association. Stroke. 2021;52(7):e364–e467. doi: 10.1161/str.0000000000000375. [DOI] [PubMed] [Google Scholar]
- 7.Spooner M.T., Messé S.R., Chaturvedi S., et al. 2024 ACC expert consensus decision pathway on practical approaches for arrhythmia monitoring after stroke: a report of the American college of cardiology solution set oversight committee. J Am Coll Cardiol. 2025;85(6):657–681. doi: 10.1016/j.jacc.2024.10.100. [DOI] [PubMed] [Google Scholar]
- 8.Buck B.H., Hill M.D., Quinn F.R., et al. Effect of implantable vs prolonged external electrocardiographic monitoring on atrial fibrillation detection in patients with ischemic stroke: the PER DIEM randomized clinical trial. JAMA. 2021;325(21):2160–2168. doi: 10.1001/jama.2021.6128. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Sanna T., Diener H.C., Passman R.S., et al. Cryptogenic stroke and underlying atrial fibrillation. N Engl J Med. 2014;370(26):2478–2486. doi: 10.1056/NEJMoa1313600. [DOI] [PubMed] [Google Scholar]
- 10.Tsivgoulis G., Triantafyllou S., Palaiodimou L., et al. Prolonged Cardiac monitoring and stroke recurrence: a meta-analysis. Neurology. 2022;98(19):e1942–e1952. doi: 10.1212/wnl.0000000000200227. [DOI] [PubMed] [Google Scholar]
- 11.von Elm E., Altman D.G., Egger M., Pocock S.J., Gøtzsche P.C., Vandenbroucke J.P. The strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. J Clin Epidemiol. 2008;61(4):344–349. doi: 10.1016/j.jclinepi.2007.11.008. [DOI] [PubMed] [Google Scholar]
- 12.Benchimol E.I., Smeeth L., Guttmann A., et al. The REporting of studies conducted using observational routinely-collected health data (RECORD) statement. PLoS Med. 2015;12(10) doi: 10.1371/journal.pmed.1001885. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Ratajczak-Tretel B., Lambert A.T., Al-Ani R., et al. Underlying causes of cryptogenic stroke and TIA in the nordic atrial fibrillation and stroke (NOR-FIB) study - the importance of comprehensive clinical evaluation. BMC Neurol. 2023;23(1):115. doi: 10.1186/s12883-023-03155-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Todo K., Okazaki S., Doijiri R., et al. Atrial fibrillation detection and ischemic stroke recurrence in cryptogenic stroke: a retrospective, multicenter, observational study. J Am Heart Assoc. 2024;13(3) doi: 10.1161/jaha.123.031508. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Hart R.G., Sharma M., Mundl H., et al. Rivaroxaban for stroke prevention after embolic stroke of undetermined source. N Engl J Med. 2018;378(23):2191–2201. doi: 10.1056/NEJMoa1802686. [DOI] [PubMed] [Google Scholar]
- 16.Diener H.C., Sacco R.L., Easton J.D., et al. Dabigatran for prevention of stroke after embolic stroke of undetermined source. N Engl J Med. 2019;380(20):1906–1917. doi: 10.1056/NEJMoa1813959. [DOI] [PubMed] [Google Scholar]
- 17.Geisler T., Keller T., Martus P., et al. Apixaban versus aspirin for embolic stroke of undetermined source. NEJM Evid. 2024;3(1) doi: 10.1056/EVIDoa2300235. [DOI] [PubMed] [Google Scholar]
- 18.Johnson L.S., Benz A.P., Shoamanesh A., et al. Residual stroke risk among patients with atrial fibrillation prescribed oral anticoagulants: a patient-level meta-analysis from COMBINE AF. J Am Heart Assoc. 2024;13(17) doi: 10.1161/jaha.123.034758. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.McCabe J.J., Cheung Y., Foley M., et al. Residual risk of recurrent stroke despite anticoagulation in patients with atrial fibrillation: a systematic review and meta-analysis. JAMA Neurol. 2025;82(7):696–705. doi: 10.1001/jamaneurol.2025.1337. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Brambatti M., Connolly S.J., Gold M.R., et al. Temporal relationship between subclinical atrial fibrillation and embolic events. Circulation. 2014;129(21):2094–2099. doi: 10.1161/circulationaha.113.007825. [DOI] [PubMed] [Google Scholar]
- 21.McIntyre W.F., Benz A.P., Becher N., et al. Direct oral anticoagulants for stroke prevention in patients with device-detected atrial fibrillation: a study-level meta-analysis of the NOAH-AFNET 6 and ARTESiA trials. Circulation. 2024;149(13):981–988. doi: 10.1161/circulationaha.123.067512. [DOI] [PubMed] [Google Scholar]
- 22.Kirchhof P., Toennis T., Goette A., et al. Anticoagulation with edoxaban in patients with atrial high-rate episodes. N Engl J Med. 2023;389(13):1167–1179. doi: 10.1056/NEJMoa2303062. [DOI] [PubMed] [Google Scholar]
- 23.Healey J.S., Lopes R.D., Granger C.B., et al. Apixaban for stroke prevention in subclinical atrial fibrillation. N Engl J Med. 2024;390(2):107–117. doi: 10.1056/NEJMoa2310234. [DOI] [PubMed] [Google Scholar]
- 24.Tsivgoulis G., Palaiodimou L., Triantafyllou S., et al. Prolonged cardiac monitoring for stroke prevention: a systematic review and meta-analysis of randomized-controlled clinical trials. Eur Stroke J. 2023;8(1):106–116. doi: 10.1177/23969873221139410. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Riordan M., Opaskar A., Yoruk A., et al. Predictors of atrial fibrillation during long-term implantable cardiac monitoring following cryptogenic stroke. J Am Heart Assoc. 2020;9(15) doi: 10.1161/jaha.120.016040. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Apple S.J., Parker M., Flomenbaum D., et al. Predictors of paroxysmal atrial fibrillation in patients with a cryptogenic stroke: selecting patients for long-term rhythm monitoring. Heart Rhythm. 2025;22(1):13–20. doi: 10.1016/j.hrthm.2024.07.004. [DOI] [PubMed] [Google Scholar]
- 27.Kamel H., Longstreth W.T., Jr., Tirschwell D.L., et al. Apixaban to prevent recurrence after cryptogenic stroke in patients with atrial cardiopathy: the ARCADIA randomized clinical trial. JAMA. 2024;331(7):573–581. doi: 10.1001/jama.2023.27188. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Joglar J.A., Chung M.K., Armbruster A.L., et al. 2023 ACC/AHA/ACCP/HRS guideline for the diagnosis and management of atrial fibrillation: a report of the American college of cardiology/American Heart Association joint committee on clinical practice guidelines. Circulation. 2024;149(1):e1–e156. doi: 10.1161/cir.0000000000001193. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Lopes R.D., Granger C.B., Wojdyla D.M., et al. Apixaban vs aspirin according to CHA(2)DS(2)-VASc score in subclinical atrial fibrillation: insights from ARTESiA. J Am Coll Cardiol. 2024;84(4):354–364. doi: 10.1016/j.jacc.2024.05.002. [DOI] [PubMed] [Google Scholar]
- 30.Uhe T., Wasser K., Weber-Krüger M., et al. Intensive heart rhythm monitoring to decrease ischemic stroke and systemic embolism-the find-AF 2 study-rationale and design. Am Heart J. 2023;265:66–76. doi: 10.1016/j.ahj.2023.06.016. [DOI] [PubMed] [Google Scholar]