337 JACC March 21, 2017 Volume 69, Issue 11
Arrhythmias and Clinical EP ACCURACY OF SMART WATCHES IN ARRHYTHMIAS: SMARTS STUDY Moderated Poster Contributions Arrhythmias and Clinical EP Moderated Poster Theater, Poster Hall, Hall C Sunday, March 19, 2017, 12:45 p.m.-12:55 p.m. Session Title: Novel Arrhythmia Diagnostics, Therapeutics, and “Wearables” Abstract Category: 5. Arrhythmias and Clinical EP: Devices Presentation Number: 1315M-05 Authors: Anoop Ninan Koshy, Jithin Sajeev, Mark Zureik, Maryann Street, Michael C. Wong, Louise Roberts, Andrew W. Teh, Eastern Health, Melbourne, Australia
Background: Smart watches that estimate heart rate (HR) are increasingly popular and show strong correlation with HR estimated electrocardiography (ECG) among healthy controls. Although not marketed for medical use, presentations to the emergency room due to device-detected heart rate abnormalities are increasing and a paucity of data exists for its utility in patients with arrhythmias.
Methods: Thirty-six inpatients were prospectively recruited from a single centre (64% male; aged 71 ±12.9 years). Patients were monitored with a 5 lead ECG, while simultaneously wearing a Fitbit Blaze, Apple Watch and finger plethysmograph for thirty minutes.
Results: Across all devices, 4,081 HR values were recorded. Arrhythmias were divided into three categories based on ventricular rate, Low (HR<60), Normal (HR 60-99), Fast (HR ≥100). They included atrial fibrillation (n= 12), atrial flutter (n=6), complete heart block (n=4), sinus bradycardia (n=7), sinus tachycardia (n=3), sinus rhythm (n=3) and ventricular tachycardia (n=1). Compared with the ECG, smart watch accuracy varied based on heart rate category, as assessed by the concordance correlation coefficient (rs value: Spearman’s rho) (Table 1).
Conclusions: The Apple Watch demonstrated strongest correlation to the ECG in HR estimation among all patients, specifically during normal HR. All three devices were robust in identifying HR during bradyarrhythmias, but unreliable during tachycardia.
All values N=4081 Low HR (<60 bpm) N=1725 Normal HR (60-99 bpm) N=805 High HR (≥100 bpm) N=1551
Mean HR(SD)
ECG
Fitbit Blaze
Apple Watch
Plethysmograph
81.7 (±36.1)
72.1 (±30.5) *
77.5 (±32.6) *
72.8 (±32.1)*
0.845*
0.900*
0.767*
47.2 (±5.6) *
48.2 (±6.8) *
49.3 (±16.8) *
0.815*
0.810*
0.865*
72.6 (±9.4) *
79.6 (±11.2) *
73.1 (±17.0) *
0.429*
0.864*
0.523*
99.5 (±30.6) *
108.9 (±27.2) *
98.8 (±31.1) *
0.218*
0.502*
0.241*
Correlation Coefficients Mean HR(SD)
45.8 (±4.9)
Correlation Coefficients Mean HR(SD)
81.1 (±10.6)
Correlation Coefficients Mean HR(SD)
122.0 (±16.8)
Correlation Coefficients HR = Heart Rate, SD=Standard deviation, * p<0.0001