Clinical relevance of the single-lead ECG of the Apple Watch

Introduction

The Apple Watch (AW) is one of the first commercially available wearable with built-in electrocardiogram (ECG) electrodes to perform a single-lead ECG to detect atrial fibrillation (AF). The AW ECG application showed to have a high rate of unclassifiable notifications. The interpretation of the single-lead ECG by a physician can possibly reduce the total of unclassifiable notifications and contribute to the overall accuracy to detect AF.

Methods

This was a prospective, non-randomized, single center observational study to evaluate the accuracy and interrater agreement of the single-lead ECG of the AW. Patients scheduled for cardioversion for atrial fibrillation were asked for participation. The AW series 6 was used with WatchiOS 7.2 or 7.3 and ECG algorithm version 2.0. A single-lead ECG was obtained pre-conversion and, when cardioversion was successful, post-conversion. All single-lead ECGs and 12-lead ECGs were adjudicated by two physicians, agreement was reached by consensus. The AW single-lead ECG notification was compared to the physicians “gold-standard” interpretation of the 12-lead ECG. Sensitivity and specificity of the AW single-lead ECG and Kappa coefficient were calculated.

Results

In total, 74 patients were included. Mean age was 67.1±12.3 years and 20.3% were female. In total 65 AF and 64 sinus rhythm (SR) single-lead ECG measurements were obtained. Of these measurements, 27.9% (36/129) showed an unclassifiable notification. The AW ECG notification showed a sensitivity of 93.5% to detect AF and a specificity of 100% to detect SR (K= 0.94). After adjudication of the unclassifiable notifications by two independent physicians 1.6% (2/129) of the measurements remained unclassifiable. When including the by the physicians adjudicated unclassifiable recordings the sensitivity to diagnose AF was 89.2% and the specificity to detect SR was 93.8% (K=0.83). The adjudication of all single-lead ECGs by the two physicians showed a sensitivity of 90.6% to detect AF and specificity of 95.2% to detect SR (K= 0.83) (Table 1). All SR and AF notification by the AW were similarly adjudicated by the two physicians. The kappa coefficient for interrater agreement between both physicians was 0.69 for all single-lead ECGs and 0.58 for unclassifiable notifications specific.

Conclusion

Unclassifiable notifications of the AW can be reduced by physicians interpretation of the single-lead ECG, however the interrater agreement is only moderate. The physicians interpretation of the single-lead ECGs did not increase the accuracy to detect AF and the clinical relevance of the single-lead ECG, in addition to the AW notification, should therefore be questioned.

Funding Acknowledgement

Type of funding sources: None.

graphicOpen in new tabDownload slide