Remote AF Screening by Patients Boosts Detection: REHEARSE-AF

Liam Davenport

August 31, 2017

BARCELONA, SPAIN — Atrial-fibrillation screening via submission of ECGs recorded using a handheld device substantially increases the detection rate of AF compared with routine care, a new study suggests[1]. The results, from a randomized, controlled trial, could pave the way for further use of such devices in the community.

Asking patients to submit readings twice weekly taken on a handheld ECG device (Kardia Mobile, AliveCor) that submits traces over the internet, the researchers were able to quadruple the selection rate of AF over routine primary care.

The research, was presented here at the European Society of Cardiology 2017 Congress and published simultaneously in Circulation, indicated that patients were extremely satisfied with using the devices and reported lower AF anxiety than other patients.

Prof Julian Halcox

Asked how easily this method of AF screening could be incorporated into routine clinical practice, Prof Julian Halcox (Swansea University, Wales) told | Medscape Cardiology, "At scale, it's something that's quite practical. I think managing patients individually on an ad hoc basis is a bit more of a challenge, but this is certainly something that's feasible."

"I think you need a dedicated 'hub-and-spoke'–type system, with someone who's specified to be collating ECGs as they come in and notifying the responsible cardiologist, but that's an individual cardiologist from a team of people with an appropriate [rotation]," he added during a press conference.

Carlos Manuel Tavares Aguiar (Hospital de Santa Cruz, Lisbon, Portugal), who was not involved in the study, explained that technologies such as these, alongside other systems for screening patients for AF currently under investigation, are addressing a common issue faced by clinicians.

"We see quite a lot of patients with stroke that looks like it could have been of embolic origin, and we simply cannot identify any rhythm disturbance in any means of monitoring that can help us decide on whether to put the patient on anticoagulation or not," he said.

He emphasized that this is "a heavy decision, since oral anticoagulation done at the therapeutic level does have a risk of bleeding, and so you feel much more comfortable as a physician to do this when you have an appropriate reason to do it."

Tavares Aguiar pointed out that, if everyone were screened at random for AF, "I doubt that you would pick up many atrial-fibrillation episodes, but when you select patients on the basis of these risk factors that have been well identified for stroke and for cardioembolic stroke, then you are much more likely to see it."

He continued: "So I think this is a very interesting area to find ways of picking up AF when we think there is strong reason to believe that it will be harmful or has already been harmful to the patient."

Not Benign

Presenting his study, Halcox observed that, although AF is often asymptomatic, even this less severe form of the condition is implicated in a significant proportion of ischemic stroke cases.

He emphasized that screen-detected AF is not a "benign" condition and, with previous studies showing that effective anticoagulation can reduce AF-related stroke risk by more than 60%, the combination of screen-detected AF and additional stroke risk factors is sufficient justification to start treatment.

To examine the potential of the handheld ECG device to provide verifiable traces for AF diagnosis, Halcox and colleagues conducted the Assessment of Remote Heart Rhythm Sampling Using the AliveCor Heart Monitor to Screen for Atrial Fibrillation (REHEARSE-AF) study.

This was a randomized, controlled trial in which they recruited patients aged over 65 years who had a CHA2DS2-VASc score of ≥2 and access to the internet via Wi-Fi.

They excluded all patients with a confirmed AF diagnosis, a known contraindication to anticoagulation, and a permanent cardiac pacing implant, as well as those who were unable to operate the ECG system after instruction.

After inviting 5846 potential participants, 1272 volunteered to take part in the study, and 1001 were randomized to either the intervention (iECG) group (n=501) or usual routine care with their local medical practitioner (n=500).

Patients in the iECG group were asked to record and submit an ECG trace using the device, which was connected to an iPod, twice weekly on Mondays and Wednesdays for 52 weeks, with the traces uploaded to a secure server for automated ECG analysis.

All ECGs were over read by an expert cardiac physiologist working for an outside company, and all abnormal ECGs were read by a cardiologist at the hospital. If AF or other significant arrhythmia was detected, prompt medical review of the patient was ordered.

Halcox told | Medscape Cardiology that the hospital cardiologists "were available during normal working hours, so the patients had to be aware of the fact that, if they were symptomatic and felt that they needed help, just recording the ECG and sending it to the server wouldn't trigger an immediate response from us."

"That seemed to work quite well, and we could respond reasonably promptly within 24 to 48 hours of patients uploading their ECG," he added.

All participants underwent telephone review at 12, 32, and 52 weeks. All adverse events were reported by patients and/or their care teams, and the researchers corresponded with the patients' primary-care physician at 52 weeks.

In addition, the patients' health records were reviewed for death and cardiovascular admissions, as well as all recorded adverse events. Patients also completed an experience survey at 1 year, and the team conducted a health economic evaluation of the cost per AF diagnosis.

There were no significant differences in baseline demographic and clinical and medication variables between the iECG and routine-care groups.

Overall, 60,440 iECGs were recorded and submitted during the study period, of which 76% were reported as normal by the automated algorithm. None of these were later found to be AF after read-over by a clinician.

The automated algorithm determined that 21% of the iECGs were "undetermined," of which only 6% were later found to have confirmed AF. Of the 1% of iECGs reported as AF by the algorithm, just 5% were finally confirmed as AF. In addition, 2.2% of the submitted iECGs were deemed uninterpretable.

Over the course of the study, 74% of participants did not miss a single week of iECG submissions, while approximately 80% submitted ≥1 weekly iECG reading during ≥90% of the study weeks and ≥2 iECGs during ≥75% of the study weeks.

Nineteen (3.79%) patients in the iECG group and five (1%) in the routine-care arm were diagnosed with AF over the 12-month follow-up, at a hazard ratio of 3.9 (95% CI 1.4–10.4, P=0.007). Twelve (63%) iECG patients were diagnosed with paroxysmal AF and seven (37%) with persistent AF vs none and five, respectively, of the routine-care patients.

Eight (42%) iECG patients were asymptomatic at the time of diagnosis, with four (21%) experiencing palpitations and seven (37%) aware of other symptoms. Among the routine-care patients, two (40%) were diagnosed with AF during palpitations while the other three (60%) were diagnosed while having other symptoms.

On univariate analysis, the predictors of AF were age ≥75 years (P=0.04), the presence of arterial disease (P=0.05), and a CHA2DS2-VASc score ≥4 (P=0.04). On multivariate analysis, the only significant independent predictor of AF was a CHA2DS2-VASc score ≥4, at an adjusted hazard ratio of 4.0 (95% CI 1.1–5.2, P=0.04).

Halcox and colleagues calculated that the cost per AF diagnosis in the iECG group was £8255 (~$10,700), the majority of which was accounted for by clinician read-over of the ECGs.

There were no significant differences in adverse events between the iECG and routine-care patients, including death, at three vs five events (P=0.51), stroke/transient ischemic attack, at six vs 10 events (P=0.34), and clinically significant bleeds, at two vs one event (P=0.56).

The researchers found small but significant differences between the two groups on a participant experience survey, with iECG patients reporting greater awareness of arrhythmia risk than those receiving routine care (P=0.001) and lower anxiety over arrhythmia risk (P=0.003).

iECG patients were also much less keen than routine-care patients to switch to the other study arm (P<0.0001).

These findings were backed up by iECG patients reporting that they were satisfied with the study, not anxious about their heart rhythm, and experiencing no lifestyle restrictions as a result of performing the ECGs. They also reported they were confident in using the device and comfortable with sharing the ECG data.

Halcox concluded that the screening approach they used was highly acceptable to their patients and "has the potential" to reduce stroke risk, although he noted that further study is required to determine its full clinical and cost impact.

Quadruple Detection Rate

Invited study discussant Dr Christophe Leclercq (Rennes University Hospital, France) commented that the study was well conducted, highlighting that the researchers were able to quadruple the AF detection rate in the iECG group, while reducing anxiety about AF risk.

However, he pointed out that close reading of the figures calls into question the willingness of the population for such screening methods, seeing as, from almost 6000 potential participants, only 17% took part in the study.

Leclercq also noted that only 20% of iECG patients completed the submissions exactly as set out in the protocol and that detecting 36 confirmed AF readings from 60,440 submitted iECGs represented a hit rate of just 0.06%.

Another important unanswered question was that of how the patients were treated after they were diagnosed with AF.

In his reply, Halcox said that all iECG patients diagnosed with AF were started on anticoagulants within 2 weeks of diagnosis, adding: "Of the five who were diagnosed with AF in the routine-care arm, four of them were anticoagulated; one of them, I'm sorry to say, got antiplatelet therapy."

As to the number of ECGs needed to detect a case of AF, he said that, despite the clinician read-over, "there was a large proportion of patients that had undetermined or uninterpretable ECGs; that's independent of those that we found with atrial fibrillation."

Halcox explained that the vast majority of these were due to the presence of bundle branch block or ectopy, which is "very, very common in this age group and with these risk factors," alongside which there were some patients with sinus arrhythmia.

"To have to undertake a human overread of all of these just to make sure you're not missing something clinically significant from among that large amount of chaff, you could call it, is quite expensive and quite challenging," he said.

Halcox noted that the device manufacturer is improving the diagnostic capabilities of its algorithm and that the research team is interested in reanalyzing its data using a newer algorithm.

"If we can have to analyze only a very, very tiny percentage of arrhythmias, which are really the ones that we would be clinically concerned about, it would be a far more practically feasible and cost-effective strategy," he said.

The study was funded primarily by the Welsh government via the Health and Technology and Telehealth Fund and in part by a grant from AliveCor. The authors report no relevant financial relationships

For more from | Medscape Cardiology, follow us on Twitter and Facebook.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.