Single-lead ECGs with wearable technology: diagnostic accuracy in patients with cardiovascular disease

FUNDING ACKNOWLEDGEMENTS: Type of funding sources: None. INTRODUCTION: The 2020 ESC AF management guideline states that a 30-second single-lead electrocardiogram (sl-ECG) can be used to diagnose atrial fibrillation, but advises caution as most validation studies used small, carefully selected cohorts and are prone to bias. Patients with frequent ectopy, other atrial arrhythmias, and cardiac implantable electronic devices, which are important sources of false-positive detection, are noticeably absent from many studies. PURPOSE: We sought to compare the diagnostic accuracy of the Apple Watch and CART-I ring automated AF detection to physician interpretation of a sl-ECG in a broad cohort of patients with pre-existing cardiovascular disease. METHODS: This prospective single-centre randomised study recruited patients with known cardiovascular disease in a large UK tertiary centre. All patients underwent simultaneous 12-lead ECG recordings with a 30-second sl-ECG from an Apple Watch Series 6 and a CART-I Ring. The order of devices was randomly assigned. Two cardiologists independently adjudicated all ECGs; any disagreements were resolved by a third cardiologist. The wearable devices’ algorithm labelled ECGs as "AF", "Not AF" or "Unclassified". Unclassified ECGs were categorised as false-positive (no AF on the 12-lead ECG) or false-negative (AF on the 12-lead ECG). Diagnostic accuracy was determined by calculating AF sensitivity and specificity for each device and comparing it to 12-lead ECGs. RESULTS: A total of 1600 ECGs from 400 consecutive patients were recorded. Seven patients were excluded due to uninterpretable 12-lead ECGs. The median age was 63 (58-75) years; 24% were female; 75.1% had coronary artery disease; 64% had a history of AF; and 10 % had undergone previous cardiac surgery. Interobserver reproducibility for sl-ECGs was high, with levels of agreement for the Apple Watch and the CART-I ring being 89.8% and 91.1%, respectively (p=0.54). The CART-I ring’s AF sensitivity from automated detection was significantly higher than the Apple Watch’s (mean difference 18.6%; 95%CI 17.0%-20.6%, p=0.03). Physician interpretation of the Apple Watch sl-ECGs led to a significant increase in AF sensitivity and specificity, 95.8% (95% CI: 91.6 %-97.9%) and 90.8 % (95% CI: 86.2 %-93.9%), respectively (Figure 1B). In contrast, physician interpretation of the CART-I ring’s sl-ECG resulted in higher AF sensitivity (94.6% [95% CI: 90.0%-97.1%]) but a marginal decrease in specificity (89.4% [95% CI: 84.8%-92.8%]). There was no statistically significant difference in AF detection between physician-interpreted sl-ECG from the wearable devices: 4.74% difference in AF sensitivity (p=0.12) and 1.32% difference in AF specificity (p=0.75). CONCLUSION: In patients with cardiovascular disease, automatic AF detection algorithms from wearable devices have a higher rate of false-positive detections than previously reported. Physician-interpreted sl-ECGs have excellent diagnostic accuracy for AF. [Figure: see text] [Figure: see text]