← Back
NeuroTrials.ai
Neurology Clinical Trial Database

PER DIEM

Post-Embolic Rhythm Detection with Implantable vs External Monitoring

Year of Publication: 2021

Authors: Brian H. Buck, Michael D. Hill, F. Russell Quinn, ..., Derek V. Exner

Journal: JAMA

Citation: JAMA. 2021;325(21):2160-2168

Link: https://clinicaltrials.gov/ct2/show/NCT02428140


Clinical Question

In patients with a recent ischemic stroke, does 12 months of implantable loop recorder monitoring detect more atrial fibrillation compared with conventional external loop recorder monitoring for 30 days?

Bottom Line

Implantable loop recorder monitoring for 12 months detected significantly more AF than 30-day external loop recorder monitoring (15.3% vs 4.7%, RR 3.29). The difference emerged primarily after 30 days of monitoring. All patients with newly diagnosed AF were started on oral anticoagulation therapy.

Major Points

  • At 12 months, AF was detected in 15.3% of ILR patients vs 4.7% of ELR patients (RR 3.29, 95% CI 1.45-7.42, P=0.003)
  • At 30 days, AF detection was similar: 4.7% ILR vs 3.3% ELR (between-group difference 1.3%, P=0.77)
  • Between 30 days and 12 months, significantly more AF was detected: 10.7% ILR vs 1.3% ELR (RR 8.0, P=0.001)
  • 100% of patients with newly diagnosed definite AF were started on oral anticoagulation
  • 66.3% of patients had stroke of undetermined etiology; 16% had small vessel occlusion
  • Detection rate with ILR (15.3%) was comparable to CRYSTAL-AF (12.4%) despite including all stroke subtypes
  • ELR detection rate (3.3% at 30 days) was lower than EMBRACE trial (16.1%) due to more rigorous AF definition (≥2 min)
  • Older age was the strongest predictor of AF detection (RR 1.03 per year); patients >80 years had ~50% AF rate
  • NNT with ILR vs ELR to detect one AF case = 10 patients

Design

Study Type: Investigator-initiated, open-label, randomized clinical trial

Randomization: 1

Blinding: Open-label for patients and investigators; outcome adjudication by 2 cardiac electrophysiologists blinded to all clinical data

Enrollment Period: May 2015 to November 2017

Follow-up Duration: 12 months (final follow-up December 2018)

Centers: 3

Countries: Canada

Sample Size: 300

Analysis: As-randomized (intention-to-treat) primary analysis. Per-protocol secondary analysis. Unadjusted 2-sample comparison of proportions for primary outcome. Kaplan-Meier and Cox proportional hazards (adjusted post hoc for age and sex) for time-to-event. Fisher exact test for categorical variables. Stata version 16.1.


Inclusion Criteria

  • Age ≥18 years
  • Arterial ischemic stroke confirmed by neuroimaging
  • TIA patients eligible only if evidence of infarction on CT or diffusion-weighted MRI
  • Randomization within 6 months of ischemic stroke
  • No known atrial fibrillation
  • At least one 12-lead ECG after the stroke event

Exclusion Criteria

  • Previously documented atrial fibrillation
  • Pacemaker or implantable cardioverter-defibrillator that would allow AF detection
  • Prior extended (>7 days) external electrocardiographic monitoring after stroke

Arms

FieldControlImplantable Loop Recorder
InterventionBattery-operated SpiderFlash-t (Sorin Group) autotriggered external loop recorder worn as much as possible for 30 days. Patients provided diary to track hours worn and symptoms. Stored data uploaded to proprietary ECG viewing software at end of monitoring.Medtronic Reveal LINQ insertable cardiac monitoring system with wireless remote monitoring (MyCareLink). Programmed with standard AF detection algorithm plus tachycardia (>235/min minus age), bradycardia (<35/min), and pauses >4.5 seconds. Minimum AF detection duration 2 minutes.
Duration30 days monitoring, 12 months follow-up12 months monitoring and follow-up

Outcomes

OutcomeTypeControlInterventionHR / OR / RRP-value
Development of definite AF or highly probable AF (adjudicated new AF lasting ≥2 minutes within 12 months of randomization). Definite/highly probable AF defined as good quality recording showing continuously irregular QRS complexes for >2 minutes with no visible P waves or clearly identified fibrillatory atrial activity/flutter waves.Primary4.7% (7/150 patients)15.3% (23/150 patients)3.360.003
AF ≥2 min or death by 12 monthsSecondary6.7% (10 patients)17.3% (26 patients)2.640.009
AF detection at 30 days (post hoc)Secondary3.3% (5 patients)4.7% (7 patients)0.77
AF detection between 30 days and 12 months (post hoc)Secondary1.3% (2 patients)10.7% (16 patients)80.001
Recurrent ischemic strokeSecondary5.3% (8 patients)3.3% (5 patients)0.630.40
TIASecondary1.3% (2 patients)4.0% (6 patients)30.28
Intracerebral hemorrhageSecondary0.7% (1 patient)0.7% (1 patient)1
DeathSecondary2.0% (3 patients)2.0% (3 patients)1
Initiation of OAC in patients with definite AFSecondary100% (7/7 patients)100% (23/23 patients)
Possible AF (including lower probability events)Secondary5.3% (8 patients)20.7% (31 patients)3.880.001
Patient satisfaction - Very satisfiedSecondary20.2% (26/129 patients)42.9% (54/126 patients)<0.001
Device-related serious adverse eventsAdverse0% (0 patients)0.7% (1 patient - skin erosion requiring device removal)
Any serious adverse eventAdverse3.3% (4 patients)9.3% (14 patients)

Subgroup Analysis

In multivariable analysis adjusting for age, ILR resulted in significantly increased AF detection (RR 3.06, 95% CI 1.36-6.93, P=0.007). Older age was the strongest predictor of AF (RR 1.03 per year, P=0.002). AF detection rates by age decile ranged from 8.4% (ages 20-29) to >50% (ages >80). Distribution of TOAST classification was not significantly different between patients with or without AF (χ²=8.42, P=0.08). No significant site × randomized group interaction (P=0.26).


Criticisms

  • Open-label design for patients and investigators, though outcome adjudication was blinded
  • Median 2-month delay between stroke onset and enrollment may have decreased AF detection slightly
  • Variable pre-enrollment investigations; not all patients had echocardiography or 24-hour Holter before enrollment
  • AF definition required ≥2 minutes duration due to ILR technical limitations, which may miss shorter clinically relevant episodes
  • 8.6% of participants did not complete prescribed monitoring, potentially underestimating true AF rates
  • Phone assessments did not use validated questionnaire for stroke/TIA events
  • Study not powered to detect differences in clinical outcomes (stroke, death)
  • Included all stroke subtypes, not just cryptogenic, limiting direct comparison with CRYSTAL-AF
  • Cost-effectiveness analysis not performed
  • Single-country study (Canada) may limit generalizability

Funding

Alberta Innovates Health Solutions Collaborative Research and Innovations Opportunities; Partnership for Research and Innovation in the Health System (government of Alberta); unrestricted in-kind support from Medtronic Canada (ILR devices and ECG core laboratory)

Based on: PER DIEM (JAMA, 2021)

Authors: Brian H. Buck, Michael D. Hill, F. Russell Quinn, ..., Derek V. Exner

Citation: JAMA. 2021;325(21):2160-2168

Content summarized and formatted by NeuroTrials.ai.