Imagine a future, they’ve said, in which wearable devices track a variety of biomarkers, then automatically use this data to detect and help manage medical conditions based not just on your personal health history, but also on data from millions of other device users. While certain wearable fitness devices have been mainstream for several years now, any ability to predict serious health conditions in a scientifically verified way has remained mostly elusive — until now. Preliminary results from the Apple Heart Study, which has enrolled more than 400,000 participants, show that smart watches can identify heart rhythm irregularities like atrial fibrillation (afib) in a large majority of cases. But there are still a number of barriers to effectively using this technology as a screening tool.
Useful Predictive Tool, but Not Enough Follow-up
For this phase of the study, which was presented at the American College of Cardiology 2019 meeting in New Orleans on March 16, researchers at Stanford University in California examined how well the Apple Watch’s heart rate sensor could detect heart rhythm problems. Study participants all had an Apple Watch from series 1, 2, or 3 — before the current version, in which a built-in electrocardiogram (ECG) was added — as well as an iPhone that communicated with the Watch. The Apple Heart Study app periodically used the Watch’s built-in heart rate sensor to check participants for an irregular pulse. If an abnormal rhythm was detected, the participant received a notification and was asked to schedule a “telemedicine consultation” with a doctor involved with the study. After this consultation, participants with a possible heart rhythm problem were sent a wearable ECG — an approved medical device, not a consumer version. This device recorded their heart rhythm for up to a week, giving researchers an opportunity to compare the Apple Watch’s detection capabilities with an established medical tool. The data that emerged from this process was encouraging in many ways, but also showed that wearable devices have limitations. Possibly the most disappointing finding was that only 57 percent of participants who received irregular pulse notifications followed up with a telemedicine consultation — even though this was free and easy to do as part of the study. “It’s disappointing that number was so low,” says Matthew W. Martinez, MD, the associate chief of cardiology at Lehigh Valley Health Network in Allentown, Pennsylvania, who was not involved in the study. “It’s critical that they follow up. It’s an opportunity to educate your patient about what this means.” Among participants who did follow up after receiving an irregular pulse notification, the data hinted at the Apple Watch’s predictive value. Indeed, direct comparisons between the Watch and the wearable ECG showed that the Watch’s pulse detection algorithm was correct about abnormal heart rhythm 71 percent of the time. And at the exact time of an irregular-pulse notification, a Watch user was, in fact, experiencing afib 84 percent of the time, according to ECG readings. These numbers show, Dr. Martinez says, that “if someone comes in with an abnormal tachogram, it’s absolutely appropriate to pursue that further.” He adds that “confirming that there’s an abnormality is going to be critical.” So far, the study doesn’t indicate that this should be a serious concern. And the most reassuring number is that only 0.5 percent of participants ever received an irregular pulse notification. “The study didn’t show that there was a tsunami of extra phone calls,” says Martinez. “But clinically, I think we’re going to see more than the study showed. We’re going to have to figure out what to do with that extra volume,” as notifications from wearable devices become more widespread. It’s possible and even likely, of course, that some participants may have had an abnormal heart rhythm that wasn’t detected by the Watch, given its overall rate of prediction. This means it’s far too early to consider devices like these an invaluable, or sufficient, part of the screening process. All these considerations mean that the latest study “doesn’t show us that we’re going to identify afib by utilizing [wearable tech] for screening,” says Martinez. “But it is the first step to that [development], and I hope that we see more of it.” Ultimately, a device is only useful as a screening tool if it can be shown to improve health outcomes, Martinez stresses. “Just finding something doesn’t mean that you save anyone’s life,” he notes. “We have to implement the intervention arm. That’s a critical piece to improving outcomes.”
