The Apple Watch Heart Study

Disclosure: I’m a huge Apple Fan.

Unless you’ve been living under a rock, you’ve heard about the Apple watch, many of you reading this are wearing one right now. On April 24th, 2015 it joined a long list of wearable wrist devices that can monitor things like steps, energy expenditure, and even heart rate (1). Last year Apple introduced the latest iteration of Apple Watch (series 4), and with it, a new ability. The Apple watch can now detect atrial fibrillation.  A huge accomplishment for sure, but how was this validated? How much can we trust a watch? How many overcalls will I have to deal with?

All great questions; The answers to which can be found within “The Apple Heart Study (AHS).”

First, Some background information. Atrial fibrillation and atrial flutter (AF) together represent the most common cardiac arrhythmia. Over 5 million Americans have or have had one or both of these rhythms (2) This is important because AF increases the risk of stroke by 5x and is estimated to be the cause of one-fourth of strokes in the united states (3). The use of anticoagulation to prevent stroke in patient’s with AF has been extremely effective, some studies showing relative risk reductions of up to 74% (4). Given that, AF is extremely prevalent, a precursor to a devastating disease, and somewhat treatable, it stands to reason that early detection IS key. Patients can go in and out of AF, making detection difficult. This is the problem that the AHS attempts to tackle. Since people are already wearing apple watches, can the watch be used to reliably detect atrial fibrillation that would otherwise have gone undetected?

The AHS was a prospective, single-arm study that enrolled 419,093 participants over nine months.  Inclusion criteria include having an iPhone, apple watch, age >22, US resident, fluent in English, valid phone number, and valid email address. Exclusion criteria were self-reported history of atrial fibrillation or atrial flutter, or patients already on anticoagulation.

Participants were asked to fill out medical screening evaluation, and download the AHS app on their phone. Once enrolled, the AHS app would use the Apple Watch’s plethsmography to make tachograms (pulse rate over time). These tachograms are taken at random intervals. If a tachogram is deemed abnormal, then 4 additional tachograms were taken to confirm the initial. If confirmed, the participant would receive a notification via Apple Watch and in the AHS app. Once alerted, if the participants had symptoms they were told to go to an urgent-care center or emergency department. If the had no symptoms, the participant was sent ambulatory ECG patches through BioTelemetry, which recorded the electrical rhythm of their hearts for up to a week, and asked to schedule a telemedicine conference.

The primary objective was to measure the fraction of participants with an irregular pulse detected by the Apple Watch, that then had AF on subsequent ambulatory ECG patch monitoring. The secondary objectives were two-fold: 1) characterize the concordance of pulse irregularity notification episodes from the Apple Watch with simultaneously recorded ambulatory ECGs; 2) estimate the rate of initial contact with a health care provider within 3 months after notification of pulse irregularity.

Notable findings summarized here:

• Overall, only 0.5 percent of participants received irregular pulse notifications.
• Comparisons between irregular pulse-detection on Apple Watch and simultaneous electrocardiography patch recordings showed the pulse detection algorithm (indicating a positive tachogram reading) has a 71 percent positive predictive value. Eighty-four percent of the time, participants who received irregular pulse notifications were found to be in atrial fibrillation at the time of the notification.
• One-third (34 percent) of the participants who received irregular pulse notifications and followed up by using an ECG patch over a week later were found to have atrial fibrillation.
• Fifty-seven percent of those who received irregular pulse notifications sought medical attention.

Limitations of this study are numerous, not in the least of which come from its source of funding…Apple. Secondly, patients who did not receive notifications were not mailed ECG patches, thus the study is unable to comment on NPV or LR. Thirdly, patients selected for this study were smartwatch users, and they may be different from the general population in a non-trivial way. Lastly, No comparison device was used.

The FDA has since greenlit the use of this technology, meaning like it or not, you can expect a patient in the near future saying they came in today because their watch told them to.

________________________________________________________________________________________

Work cited:

(1) Wallen MP, Gomersall SR, Keating SE, Wisløff U, Coombes JS. Accuracy of heart rate watches: implications for weight management. PLoS One. 2016;11(5):e0154420. doi: 10.1371/journal.pone.0154420.

(2) E.J. Benjamin, S.S. Virani, C.W. Callaway, et al.Heart Disease and Stroke Statistics-2018 Update: A Report From the American Heart Association. Circulation, 137 (12) (2018), pp. e67-492