Conference Coverage

Apple Watch algorithm showed 84% positive predictive value for Afib

"The next 10 years or so will see the creation of a new field within cardiovascular medicine, whose skilled practitioners might be called ‘digitalists’ – experts in collecting, moving, and safeguarding massive quantities of digital health data."



– Detection of an irregular pulse rhythm by an algorithm installed on a smartwatch had a positive predictive value of 84% for simultaneous ECG-confirmed atrial fibrillation in the landmark Apple Heart Study, investigators reported at the annual meeting of the American College of Cardiology.

This was a single-arm, prospective, open-label, observational study of unprecedented size and speediness of completion. It included nearly 420,000 self-enrolled adults living in the U.S., with 8 months of monitoring. But despite the study’s flashy size and trendy digital health theme, the researchers were careful not to oversell the findings.

Bruce Jancin/MDedge News

Dr. Marco Perez (left) and Dr. Mintu Turakhia

“This study was just meant to be a very, very first step in trying to learn if this kind of technology can help us to prevent stroke. It was not a randomized trial of a public health intervention for screening. This is the first half of the first inning. Rigorous investigation of this technology and its potential use in clinical settings will need to happen. But we do think from a trial and operational standpoint the Apple Heart Study provides a solid foundation upon which further research in digital health can be conducted,” according to Mintu Turakhia, MD, co-principal investigator and an electrophysiologist as well as executive director of the Center for Digital Health at Stanford (Calif.) University, which conducted the study.

The study was conducted virtually. Screening, consent, and data gathering were performed electronically by smartphone. Participants had to have an Apple Watch Series 1, 2, or 3, and an Apple iPhone 5 or more recent model in order to join. The majority of subjects were under age 40, and just 6% were age 65 or older, when the risks of atrial fibrillation (AFib) and stroke are higher. All participants self-reported having no history of AFib nor currently being on anticoagulation.

The study algorithm utilized the Apple Watch’s built-in light sensor technology to opportunistically sample the time interval between pulses when the wearer was still. An irregular time interval triggered a cascade of more frequent sampling. If 5 of 6 samples were irregular within a 48-hour period, the wearer received an irregular rhythm notification along with a prompt to contact a participating physician via telemedicine. The physician could then arrange for an ECG patch to be mailed to the participant, who wore it for up to 7 days before mailing it back for analysis.

Among the key findings in the Apple Heart Study: the irregular pulse notification rate was low overall, at 0.5%, ranging from 0.16% in the under-40 group to 3.2% in subjects age 65 or older. As a result, the study population of particular interest nosedived from an initial 419,297 to the less than 2,100 who received an irregular pulse notification. Of the 658 participants who were subsequently sent an ECG patch, 450 returned it for analysis.

An average of 13 days went by between an irregular pulse notification and ECG patch receipt and activation, so it wasn’t particularly surprising that only 34% of the patches were positive for AFib, since early-stage paroxysmal AFib comes and goes. However, of the 86 subjects who received a new notification of an irregular rhythm while they were wearing a patch, 72 simultaneously showed AFib on their patch. That translates to an 84% positive predictive value for an irregular rhythm notification as an indicator of AFib.

Of the 153 subjects with evidence of AFib on their ECG patch, 20% proved to be in AFib for the full week they wore it. Of those with AFib, 89% had a longest episode of at least 1 hour in duration.

Several discussants expressed reservations about this approach to finding individuals with previously undetected AFib. Jeanne E. Poole, MD, an electrophysiologist and professor of medicine at the University of Washington in Seattle, observed that the question of whether patients with asymptomatic AFib should receive oral anticoagulation therapy is as-yet unanswered and is the focus of ongoing randomized trials. The Apple Heart Study approach, she said, “might lead a lot of patients into being treated unnecessarily or prematurely, or flooding doctors’ offices with a lot of young people.”

Co-principal investigator Marco Perez, MD, an electrophysiologist at Stanford, replied, “Stroke is important, and we all worry about it. But it’s also important that there are other things atrial fibrillation is associated with, like cardiomyopathy and heart failure. So finding atrial fibrillation in a young population might be important. Maybe they don’t need anticoagulation, but maybe there’s something else going on.”

Patrick T. O’Gara, MD, professor of medicine at Harvard University, Boston, was concerned about what he called “the signal-to-noise ratio – the noise that will come in when there is an irregularity detected on the watch that could range from anything from ventricular premature beats to atrial fibrillation.” He is also leery of what he considers to be at this point the excessive hype surrounding direct-to-consumer wearable digital health technology.

“I applaud your circumspection,” he told Dr. Turakhia and Dr. Perez. “I understand very directly from you that these are limited observations. But it’s a good step forward.”

Dr. Perez reported receiving research funding from and serving as a consultant to Apple. Dr. Turakhia reported serving as a consultant to AliveCor and Cardiva Medical.

Their presentation was immediately followed by a related panel discussion titled, “Digital Disruption at Our Doorstep – Implications for Clinicians and Patients.” Session moderator John Rumsfeld, MD, chief innovation officer at the ACC and professor of medicine at the University of Colorado, Denver, kicked things off by observing, “Digital health technology certainly exists. There’ve been billions of dollars invested in digital health. Outside of health care there’s been successful digital transformation of almost every other sector of the economy except for health care. But we deliver care pretty much the same as we have for the past 50 or more years.”

Paul Stoeffels, MD, chief scientific officer at Johnson & Johnson, said physicians and payers want to see evidence of benefit before adopting change. Towards that end, he announced that Johnson & Johnson and Apple are collaborating on a randomized controlled trial called the HEARTLINE study. The active intervention arm in the study involves utilization of the Apple Watch’s irregular pulse notification algorithm, with confirmation of AFib to be achieved using the ECG app incorporated in the latest version of the watch, coupled with a medication adherence app developed by Johnson & Johnson. Enrollment of 150,000 U.S. adults age 65 and older is planned to begin this summer. The study, conducted on a digital platform akin to the Apple Heart Study, will look at the intervention’s impact on rates of stroke, MI, and death as well as AF detection.

Maulik Majmudar, MD, a cardiologist and chief health officer for health and wellness at Amazon, declared, “There’s no doubt in my mind that digital solutions will become a mainstay in our care delivery going forward. The question is really not if, but when.”

He predicted that just as the past two decades have seen the birth of new medical specialties, including hospitalists and cardiovascular intensivists, the next 10 years or so will see the creation of a new field within cardiovascular medicine, whose skilled practitioners might be called ‘digitalists’ – experts in collecting, moving, and safeguarding massive quantities of digital health data.

Robert Califf, MD, vice chancellor for health data science at Duke Health in Durham, N.C., addressed the issue of how society is going to pay for a shift to digital health: “It’s very simple. I don’t see any way that fee-for-service medicine can deal with this. It’s just not possible. If you think we’re going to add on more cost to the system by doing virtual visits, I just do not see that happening. Our solution to the payment system is to get rid of fee-for-service medicine and go to pay-for-value. The minute you’re in pay-for-value, virtual visits and digital information will become the way to do it – to move the treatment and the interaction more to home and less of having people wait in doctors’ offices and spending time in hospitals.”


Recommended Reading

What cardiologists need to know about ARVC
MDedge Cardiology
ICYMI: Andexanet alfa reduces anti–factor Xa activity from apixaban, rivaroxaban
MDedge Cardiology
Not all AF maze operations are aMAZE-ing
MDedge Cardiology
Infective endocarditis isn’t what it used to be
MDedge Cardiology
Apple Watch algorithm brings wearables closer to clinical practice
MDedge Cardiology