Wearable Technologies Show Promise and Present Challenges in Atrial Fibrillation Detection

Recent studies have explored the effectiveness of wearable technologies, including smartwatches and combined smartphone app/bed sensor systems, in detecting atrial fibrillation (AF) in high-risk patients. Research from Amsterdam UMC, known as the EQUAL trial, demonstrated that smartwatches significantly increased AF detection rates compared to standard care, identifying a notable number of asymptomatic cases. Conversely, the CARE-DETECT trial from Finland, evaluating a smartphone app and bed sensor system, also showed an increased AF detection rate but highlighted a significant challenge with a high incidence of false alarms.

Atrial Fibrillation Detection Efforts

Atrial fibrillation is a common heart arrhythmia associated with an elevated risk of stroke due to potential blood clot formation. Traditional AF monitoring methods, often involving ECG devices, can be inconvenient for patients and typically offer limited monitoring durations, sometimes up to two weeks.

The intermittent nature of AF often means it can be missed by short-term monitoring, prompting interest in continuous or prolonged monitoring solutions.

Smartwatch Monitoring in High-Risk Patients: The EQUAL Trial

The randomized EQUAL trial, conducted by researchers at Amsterdam UMC and published in the Journal of the American College of Cardiology (JACC), investigated the use of smartwatches for AF detection. The study involved 437 patients aged 65 and older (mean age 75) who were at elevated risk for stroke (CHA₂DS₂-VASc score of at least 2 in men and 3 in women).

Participants were randomized into two groups over six months:

• 219 patients wore an Apple Watch (Series 5 or 8, equipped with photoplethysmography (PPG) and single-lead ECG functions) for at least 12 hours daily.
• A control group of 218 patients received standard medical care.

Smartwatch users were instructed to record 30-second ECGs upon experiencing symptoms or receiving irregular pulse notifications. These recordings were transmitted to a telemonitoring app for review by an eHealth team and supervising cardiologist.

After six months, 21 patients (9.6%) in the smartwatch group were diagnosed with new AF episodes lasting at least 30 seconds, compared to 5 patients (2.3%) in the standard care group.

This represented a fourfold increase in detection, with a Hazard Ratio (HR) of 4.40. The absolute increase in AF detection was 7.3% over six months, with a number needed to screen (NNS) of 14.

A significant finding was that 57% (or 57.1%) of the AF episodes detected in the smartwatch group were in asymptomatic individuals. In contrast, all five diagnoses in the standard care group were among symptomatic patients.

Of the 37 initial smartwatch ECGs labeled as AF, 20 were confirmed as clinical AF, indicating a positive predictive value of 54%. All patients diagnosed with AF subsequently initiated oral anticoagulation. While the smartwatch arm observed a numerically lower rate of emergency department visits (5.9% vs 8.3%), this difference was not statistically significant, as the trial was not powered to detect differences in clinical outcomes.

Dr. Michiel Winter, a cardiologist at Amsterdam UMC and lead researcher, highlighted the study's achievement in demonstrating the feasibility of population-based screening using smartwatches in a high-risk AF population. Dr. T. Jared Bunch noted the evolving landscape of AF diagnosis, moving from traditional office-based methods to consumer-grade wearables.

Investigators suggested that embedding consumer wearables into telemonitoring workflows could offer a scalable model for integrating digital screening into routine cardiology practice.

Smartphone App and Bed Sensor System: The CARE-DETECT Trial

Another randomized clinical trial, the CARE-DETECT Part II study, published in Scientific Reports, investigated a different multi-device approach for AF detection. Conducted in Finland, this single-center study involved 150 patients hospitalized for coronary artery disease or valvular heart disease who were undergoing invasive cardiac procedures and had a high risk for AF and strokes (CHA2DS2-VASc score of ≥4 or ≥2 with enrichment criteria).

The intervention group (n=78) utilized a bed sensor (EMFIT QS) for overnight ballistocardiogram monitoring and performed twice-daily smartphone recordings using the CardioSignal app for three months post-discharge. Alerts from either device prompted a 12-lead ECG, followed by three-to-seven-day ECG Holter monitoring if the initial ECG was normal. The control group (n=72) received usual care, including in-hospital telemetry and standard ECG follow-up.

New AF within three months, the primary endpoint, was detected in 6 of 78 (7.7%) patients in the intervention group, compared to zero of 72 control patients (absolute risk difference 7.7%, p=0.029). Five of these six AF episodes occurred after discharge.

However, the study also found a high rate of false alarms, with 33 of 68 (48.5%) intervention patients experiencing device alarms that did not lead to ECG-confirmed AF.

In total, 47 long-term ECG Holter recordings were triggered for six true AF diagnoses, equating to 7.8 Holters per diagnosis. The patient-level positive predictive value among those with at least one alert was 15.4%. Most alerts were attributed to non-AF rhythms such as sinus arrhythmia or supraventricular or ventricular ectopic beats.

Recruitment for the CARE-DETECT trial was halted at 150 patients after an interim analysis indicated a high number of device alerts not resulting in confirmed AF diagnoses. Investigators noted that the increased AF detection might reflect greater surveillance rather than a true incidence difference, given the absence of continuous rhythm monitoring in the control group.

High exclusion and withdrawal rates in the intervention arm, partly due to device burden, also raised questions about the scalability of the approach.

The trial concluded that, while the intervention improved AF detection, the substantial volume of non-AF alerts and resulting diagnostic workload indicated that the current multi-device strategy was not suitable for routine clinical implementation.

Future Research Directions

Both studies underscore the potential of wearable technology in AF detection while also highlighting areas for further development. Investigators from the EQUAL trial suggested that the clinical impact of treating screen-detected or subclinical AF requires further confirmation from ongoing outcome trials.

Dr. T. Jared Bunch emphasized that the impact of diagnoses from this technology will be the next area of research, with studies like Heartline expected to provide further insights. Concerns regarding potential patient anxiety from inaccurate readings from smart devices were also noted. For the CARE-DETECT system, future research is needed to reduce false alerts, clarify the AF burden threshold for treatment, and evaluate overall cost-effectiveness before routine clinical integration.