Amid the explosive popularity of the Apple Watch and other wearable fitness devices, along with new health care apps designed to track heart rate and other health measures, consumers have more opportunities than ever to monitor their own health data and make decisions about what to do with the information. The Apple Heart Study was designed to evaluate how well the Apple Watch can identify and prompt subsequent clinical evaluation for a common, yet often undiagnosed, heart rhythm disorder known as atrial fibrillation, or AFib, which is a leading cause of major stroke and hospitalizations. According to preliminary data from the study presented at the American College of Cardiology’s 68th Annual Scientific Session, the Apple Watch was able to detect AFib in a small group of people who had been alerted by the app as having an irregular heartbeat.
AFib is estimated to affect up to 6 million Americans, yet many people don’t know they have it. AFib can lead to blood clots, disabling stroke, heart failure and other heart-related complications.
For this study, researchers at Stanford University School of Medicine specifically evaluated a mobile app that uses the watch’s existing light sensor technology (called photoplethysmography) to intermittently measure blood flow activity and detect subtle changes that might indicate an irregular contraction or heartbeat. The Apple Watch generates a tachogram, which is a plot of time between heart beats. If an irregular tachogram is seen, then more frequent tachogram collection occurs. The tachograms are analyzed by an algorithm to determine if an irregular rhythm may be present and, if so, a notification is sent to the watch.
The main study goals were to determine: the proportion of participants with an irregular pulse watch notification who have AFib on a subsequent electrocardiogram (ECG) patch worn at home, how well the algorithm on the watch matched the ECG findings and the percentage of patients who received a notification who went on to seek medical help through the app. Of the 419,297 people who self-enrolled in this prospective study between November 2017 and July 2018, 0.5 percent (nearly 2,100 participants) received an irregular pulse notification, which was triggered if the sensor detected 5 out of 6 repeat episodes (or tachograms) of an irregular pulse within a 48-hour period.
“The app continuously gathers data in the background without the wearer of the device doing anything, so it’s very opportunistic in this way,” said Mintu Turakhia, MD, associate professor of cardiovascular medicine, Stanford School of Medicine, and the study’s co-principal investigator, adding that the entire user experience, end-to-end was directly through the study application. “[Overall], this study improves our understanding of how this wearable technology and app works in the real-world setting and how well the technology can detect long periods of AFib. Notifications of heart rhythm irregularities were low, an important finding given concerns about over notification, and we were able to see what happened downstream after participants received a notification.”
Participants who got a notification were also prompted to contact the study doctor through the app. Based on a video consultation on their device, it was decided whether the person should wear an electrocardiogram (ECG) patch, sent by mail, to measure their heart’s activity and potentially identify AFib. Participants wore it for up to seven days and then sent it back to determine whether AFib was evident. In total, 658 participants were sent a patch and 450 were returned and included in the analyses. One-third (34 percent) of the participants who had received an irregular pulse notification on their Watch and wore the ECG patch over a week later were then found to have AFib based on the ECG reading, which researchers said is not entirely unexpected.
“AFib can come and go, particularly early on in the course of the disease. It’s not surprising for it to go undetected in subsequent ECG patch monitoring. So while only 34 percent of people who were still having [signs of] AFib on the ambulatory ECG, that doesn’t mean that 66 percent didn’t have AFib. It just means that AFib may not have been there at the time,” Turakhia said. “These parameters help us understand how we, as clinicians, should think about these notifications.”
When comparing pulse detections on the watch with simultaneous ECG patch recordings among 450 participants who wore both at the same time, researchers found that the positive predictive value for the tachogram was 71 percent, and the positive predictive value for the notification was 84 percent. This means if a notification occurred, 84 percent of the time, the ambulatory ECG monitor also showed AFib.
Of the participants who received an irregular pulse notification via the app to confer with a study doctor to determine next steps, about half ended up initiating contact with the study doctor, but researchers said the others may have sought care elsewhere. In subsequent surveys, 57 percent of people who got an alert said they sought medical attention outside of the study regardless of whether they had been seen virtually by a study doctor.
Participants were required to have an Apple Watch (series 1, 2 or 3) and a compatible iPhone (iPhone 5S or later), be age ? 22 and a U.S. resident. The newer Apple Watches that feature a built-in ECG app were not part of this study. Participants downloaded the application to their device and were subsequently given information about the study, questions to help verify their eligibility and prompts to provide signed consent. They were not included in the study if they said they were taking an anticoagulant for any reason or had current or prior AFib. Surveys were used to collect information about patient reported outcomes.
Notification rates in people under 40 years of age were very low (0.16 percent) and were more frequent in those over age 65 (just over 3 percent).
“This is encouraging because it tracks with our understanding of AFib as being more common as you get older,” Turakhia said, adding that the overall study population represented a striking cross section of cardiovascular risk.
Five percent had diabetes, 21 percent had high blood pressure, 1 percent had a prior stroke and 38 percent were obese based on body mass index.
About one-third of people who received a notification had a CHADS-VASc score of two or greater, which is used to predict the risk of stroke in people with AFib and is the threshold to start anti-coagulation. People with AFib have more than five times the risk of having a stroke than someone without it.
There were several study limitations, including researchers’ reliance on self-reported data from participants. Researchers said they also did not reach their target enrollment of 500,000 participants, of which they hoped 75,000 would be age 65 and older. But they explained this virtual study is step one in creating the groundwork for future studies that look to leverage wearable technologies as a way to support heart health outside of the clinic. They said an obvious next step is to study how this technology can be used in complement with other technologies such as real-time ECG and other validated tools to manage heart rhythm problems.
“It really represents a paradigm shift for how clinical studies can be conducted,” Turakhia said. “We don’t have to bring people into a brick and mortar clinic and give the study intervention.”