Using Wearables to Manage Atrial Fibrillation: Pushing the Boundaries
with Consumer Devices
Marco Perez, MD1
1Stanford Center for Cardiovascular Research, Stanford
University, Stanford, CA
Word Count, excluding references: 1489
Corresponding author:
Marco Perez, MD
300 Pasteur Dr. M/C 5773
Stanford, CA 94305-5773
mvperez@stanford.edu
phone: 650-723-9363
The irregular pulse notification (IPN) algorithm on the Apple Watch
(Apple Inc., Cupertino, CA) was not designed for use by atrial
fibrillation (AF) patients. It is not FDA cleared for use in AF
patients. Before this study by Dr. Wasserlauf and colleagues, there were
no studies of its accuracy in AF patients. Yet, many AF patients could
not resist the temptation to use the feature. In the Apple Heart
Study1, even after making it clear that patients with
AF were not eligible for the study, 174 (18%) of the participants who
received an irregular pulse notification and connected with a study
visit doctor confessed that they knew they already had AF and were
excluded from the study. These participants were just too curious to
pass up the opportunity to see what the new technology was all about.
Patients pushing boundaries of what “should” be done with consumer
devices is not a novel concept. Patients with hypertension have been
using home blood pressure monitors to optimize their own
antihypertensive regimens for years. Clinical trials have shown that
empowering patients to self-measure blood pressure and to self-adjust
their medications may even outperform traditional management
strategies2. Similarly, diabetics who titrate their
own insulin based on frequent or continuous glucose monitoring can do so
safely and effectively3. More recently, as continuous
glucose monitoring devices have become less expensive and easier to use,
the number otherwise healthy people who are using them to optimize their
diet is growing. While the devices have been studied in obese
non-diabetic patients4, their utility in the healthy
population has not yet been validated.
It is natural that some AF patients will seek a similar sense of
empowerment. Until recently, the only way an asymptomatic AF patient
could confirm they were in AF was by going to a doctor’s office to get
an ECG. Savvy patients might measure their own pulse, or perhaps use a
home blood pressure monitoring device that could measure pulse
irregularity. However, most AF patients were in the dark until consumer
ECG devices became available5. With the ability to
measure ECGs at home and confirm AF, patients suddenly had opportunities
to explore novel management strategies.
The Irregular Pulse Notification Algorithm on the Apple Watch was the
first tool to use PPG on a smartwatch to detect AF. The technological
leap here was that monitoring for AF became something that could be done
passively and for prolonged periods. Shortly afterwards, other
smartwatches followed with similarly designed
algorithms6, 7. However, these tools were meant to
detect AF in consumers who had never been diagnosed with AF. The
algorithm was designed to be highly specific with very low false
positive rates. To do this, the algorithm relied on multiple passive
measurements of pulse irregularity, which was made practical by the
wearable nature of the smartwatch. Although the false positive rate of
the IPN algorithm was not measured in the Apple Heart Study, it had to
be lower than the total positive rate, which was 0.52% over a median
monitoring time of 117 days1.
Despite these caveats, patients with AF saw an opportunity to push the
boundaries of this screening tool. This could prove useful for patients
who had asymptomatic AF, or who wanted more data around onset of
paroxysms of AF. The potential to use AF notifications to manage rate
and rhythm controlling medications was obvious to some patients. One of
the challenges was that the IPN algorithm had never been validated in
patients with known AF. Even clinicians who might want to help patients
design personalized, albeit unvalidated, treatment strategies could not
tell their patients how accurate the algorithm might be.
In this study by Dr. Wasserlauf and colleagues, questions around the
accuracy of the IPN algorithm in AF patients are addressed [Wasserlauf
et al., JCE ]. They provided patients who already had insertable
cardiac monitors or cardiac implanted electronic devices and a history
of non-permanent AF with Apple Watches running the IPN application. Over
the course of 6 months, 11 patients had an episode of AF lasting at
least one hour while the watch was worn. Eight of these subjects had
successfully-detected AF episodes lasting > 1 hour with the
smartwatch, resulting in a sensitivity of 72% and specificity of 100%.
This was a relatively small study and therefore the caveats of small
sample size and number of events apply. In addition, the gold standards
in this study (insertable cardiac monitors and cardiac implanted
electronic devices) are also prone to some degree of
misclassification8. That said, the strength of this
study is in its novelty: while many others had looked at the sensitivity
and specificity of smartwatch and stand-alone ECG applications, no other
study had looked at the accuracy of a PPG-based irregular pulse
algorithm on a wearable device in AF patients.
To understand the limitations that the existing algorithm has for use in
AF patients, one needs to delve into details of the design of the IPN
algorithm9. The smartwatch makes one-minute-long pulse
measurements once every two hours, as long as the user is not moving.
However, if the device detects motion, a pulse check may not occur for
prolonged periods of time. The algorithm then classifies the pulse as
regular or irregular and if the latter is detected, the frequency of
pulse checks increase to once every 15 minutes. A total of 5 out of 6
pulse checks must then be classified as irregular before a notification
is delivered. Again, with motion detected by the smartwatch, this could
prolong the verification process and extend the time-to-notification
even further. This means that short episodes of AF are less likely to
result in a notification. For large-scale screening, the low sensitivity
of detecting short episodes may be acceptable, opting instead for high
levels of specificity. However, for an AF patient who might want to
consider waiting for episodes of AF before taking medications, the IPN
algorithm is not ideal. Finally, a wearable device can’t detect AF if it
is not being worn. In this study, about half of the AF episodes occurred
while the participant was not wearing the smartwatch. Currently,
limitations in battery life and charging habits limit wear times and
sampling frequency, though these may improve in future generation
devices.
Another limitation to the existing irregular pulse algorithms is that
they can be triggered by long periods of frequent premature atrial or
ventricular contractions10. In the Apple Heart Study,
in participants who received a notification and subsequently wore an ECG
patch, approximately 16% of the patients who received an irregular
pulse notification had a rhythm other than AF. This number in theory
could be higher in patients with a known history of AF, since high
burden of PACs are associated with AF, though this did not appear to be
an issue in this small sample.
Whether we as a clinical providers want our patients using wearables or
not, patients with AF are already using these devices to monitor their
AF. Apple released the AF History feature in 2022
(https://www.apple.com/legal/ifu/afhf/1-0/afhf-1-0-en_EN.pdf),
which is designed for AF patients to measure AF burden, defined as the
proportion of time that a patient has an irregular pulse. The History
algorithm is based on frequent sampling and artificial intelligence
distinguishing AF from other rhythms. The major limitation of this
feature is that the burden measurement is reported only on a weekly
basis. The algorithm was likely intentionally designed this way
precisely to prevent patients from using the feature to make immediately
reactive decisions about medication use.
Patients with AF who want to empower themselves to manage their disease
with more granular detail will need to look for algorithms that are more
sensitive to short periods of AF with more frequent sampling and lower
thresholds for AF classification. These ideal algorithms will also need
to provide passive and near-real-time monitoring of AF, and distinguish
AF from frequent ectopy. There is currently no such device with this
algorithm, much less one that has been validated, but the technologic
limitations can be overcome.
The onus will be on the medical community to help our patients figure
out the best ways to optimize their AF management in a safe and
effective manner. Dr. Passman, senior author of this study, is leading a
team to conduct REACT-AF, an NIH-sponsored, open-label, randomized
clinical trial testing the safety of using a modified irregular pulse
notification algorithm to guide anticoagulation use in patients with AF.
Patients randomized to the intervention arm will only take their
previously prescribed DOAC therapy for a limited period if their
smartwatch alerts them to an episode of AF. The hope is that this
strategy will not only be noninferior to standard management, but that
there will be fewer bleeds.
REACT-AF is pushing existing boundaries of wearable devices. However,
this is the direction the next generation of patients, who are
technologically savvy and who demand empowered self-care, want to go.
Our job as a clinical community is to help our patients push these
boundaries safely.
REFERENCES
1. Perez MV, Mahaffey KW, Hedlin H, Rumsfeld JS, Garcia A, Ferris T,
Balasubramanian V, Russo AM, Rajmane A, Cheung L, Hung G, Lee J, Kowey
P, Talati N, Nag D, Gummidipundi SE, Beatty A, Hills MT, Desai S,
Granger CB, Desai M, Turakhia MP and Apple Heart Study I. Large-Scale
Assessment of a Smartwatch to Identify Atrial Fibrillation. N Engl
J Med . 2019;381:1909-1917.
2. Zarnke KB, Feagan BG, Mahon JL and Feldman RD. A randomized study
comparing a patient-directed hypertension management strategy with usual
office-based care. Am J Hypertens . 1997;10:58-67.
3. Martens T, Beck RW, Bailey R, Ruedy KJ, Calhoun P, Peters AL,
Pop-Busui R, Philis-Tsimikas A, Bao S, Umpierrez G, Davis G, Kruger D,
Bhargava A, Young L, McGill JB, Aleppo G, Nguyen QT, Orozco I, Biggs W,
Lucas KJ, Polonsky WH, Buse JB, Price D, Bergenstal RM and Group MS.
Effect of Continuous Glucose Monitoring on Glycemic Control in Patients
With Type 2 Diabetes Treated With Basal Insulin: A Randomized Clinical
Trial. JAMA . 2021;325:2262-2272.
4. Fechner E, Op ’t Eyndt C, Mulder T and Mensink RP. Diet-induced
differences in estimated plasma glucose concentrations in healthy,
non-diabetic adults are detected by continuous glucose monitoring-a
randomized crossover trial. Nutr Res . 2020;80:36-43.
5. Ford C, Xie CX, Low A, Rajakariar K, Koshy AN, Sajeev JK, Roberts L,
Pathik B and Teh AW. Comparison of 2 Smart Watch Algorithms for
Detection of Atrial Fibrillation and the Benefit of Clinician
Interpretation: SMART WARS Study. JACC Clin Electrophysiol .
2022;8:782-791.
6. Lubitz SA, Faranesh AZ, Selvaggi C, Atlas SJ, McManus DD, Singer DE,
Pagoto S, McConnell MV, Pantelopoulos A and Foulkes AS. Detection of
Atrial Fibrillation in a Large Population Using Wearable Devices: The
Fitbit Heart Study. Circulation . 2022;146:1415-1424.
7. Guo Y, Wang H, Zhang H, Liu T, Liang Z, Xia Y, Yan L, Xing Y, Shi H,
Li S, Liu Y, Liu F, Feng M, Chen Y, Lip GYH and Investigators MI. Mobile
Photoplethysmographic Technology to Detect Atrial Fibrillation. J
Am Coll Cardiol . 2019;74:2365-2375.
8. Podd SJ, Sugihara C, Furniss SS and Sulke N. Are implantable cardiac
monitors the ’gold standard’ for atrial fibrillation detection? A
prospective randomized trial comparing atrial fibrillation monitoring
using implantable cardiac monitors and DDDRP permanent pacemakers in
post atrial fibrillation ablation patients. Europace .
2016;18:1000-5.
9. Turakhia MP, Desai M, Hedlin H, Rajmane A, Talati N, Ferris T, Desai
S, Nag D, Patel M, Kowey P, Rumsfeld JS, Russo AM, Hills MT, Granger CB,
Mahaffey KW and Perez MV. Rationale and design of a large-scale,
app-based study to identify cardiac arrhythmias using a smartwatch: The
Apple Heart Study. Am Heart J . 2019;207:66-75.
10. Perino AC, Gummidipundi SE, Lee J, Hedlin H, Garcia A, Ferris T,
Balasubramanian V, Gardner RM, Cheung L, Hung G, Granger CB, Kowey P,
Rumsfeld JS, Russo AM, True Hills M, Talati N, Nag D, Tsay D, Desai S,
Desai M, Mahaffey KW, Turakhia MP, Perez MV and Apple Heart Study I.
Arrhythmias Other Than Atrial Fibrillation in Those With an Irregular
Pulse Detected With a Smartwatch: Findings From the Apple Heart Study.Circ Arrhythm Electrophysiol . 2021;14:e010063.