Screening for Atrial Fibrillation using Economical and Accurate TechnologY – the SAFETY study

Talk Code: 
Mark Lown
Arthur M Yue2, Benoy N Shah2, Simon J Corbett2, George Lewith1ǂ, Beth Stuart1, James Garrard1, Paul Little1, Michael Moore1
Author institutions: 
Primary Care & Population Sciences, Faculty of Medicine, University of Southampton, Aldermoor Health Centre, Southampton SO16 5ST, Cardiology and Electrophysiology, Southampton General Hospital, Tremona Road, Southampton


The prevalence of Atrial Fibrillation (AF) is estimated at more than 3% in the adult population and there has been increased interest in screening for AF. Commercial, wearable devices that measure pulse or heart rate are now widely available that can store or transmit data to smartphones or tablets capable of running complex algorithms. In the SAFETY trial we chose to evaluate if inexpensive, wearable, consumer ECG sensing devices (Polar-H7 (PH7) and Firstbeat Bodyguard 2 (BG2)), could be used to detect AF accurately.


A case-control study of 413 participants aged > 65 (79 with AF at the study visit and 336 without) attending 3 general practice surgeries in Hampshire, UK for a single screening visit. The PH7 and BG2 devices were tested alongside two established AF detection devices (AliveCor and WatchBP) in random order and the diagnosis of AF was confirmed by 12-Lead ECG interpreted by a panel of cardiologists. We also evaluated usability metrics.


The sensitivity (95% CI range), specificity (95% CI range) and overall accuracy (95% CI range) of the four devices were: AliveCor: 91.1% (82.6-96.4%), 98.8% (97.0–99.7%), 97.4% (95.3-98.7%); WatchBP: 98.7% (93.1-100.0%), 93.5% (90.3-95.9%), 94.5% (91.8-96.5%): PH7: 98.7% (93.1-100.0%), 98.2% (96.2-99.3%), 98.3% (96.6-99.3%). BG2: 98.7% (93.1 – 100.0%), 98.5% (96.6-99.5%), 98.6% (96.9-99.5%). The PH7 & BG2 devices were highly reliable (the devices acquired sufficient data and obtained a diagnostic result in all but one participant on the first attempt.) Participant ratings (mean, SD) (0-10) of device comfort were for AliveCor 9.3 (1.2); WatchBP: 7.6 (2.2); PH7 8.6 (1.6); BG2 9.3 (1.2).


Inexpensive, consumer heart rate monitoring devices (PH7 and BG2) can be used to detect AF accurately with sensitivity and specificity > 95%. The consumer devices performed as well or better than WatchBP and AliveCor and have the capability to store or transmit ECG data which could be used to confirm AF. They could be used continuously for prolonged periods to detect paroxysmal AF.

Submitted by: 
Mark Lown
Funding acknowledgement: 
This study was funded by the National Institute of Health Research (NIHR) School for Primary Care Research (SPCR) FR 11: Project No:318