Apple Heart Study: App Detects Some Changes Suggestive of AFib

An app for the Apple Watch was able to detect atrial fibrillation (AFib) in a small group of people and alert them of an irregular heartbeat, according to preliminary data from a new study presented at the 2019 American College of Cardiology Annual Scientific Session in New Orleans.

The researchers for the prospective, single-arm study, working out of Stanford University School of Medicine, sought to fund the proportion of people with an irregular pulse watch notification who then went on to have diagnosed AFib via electrocardiogram (ECG). The app functions by using the Apple Watch’s existing light sensor to generate a tachogram (time point between heat beats). The generated tachograms are then analyzed via algorithm for irregularities, triggering notifications sent to the watch if irregular rhythms are detected. The notification recommended that the wearer contact a study doctor through the app.

Some Interesting Data Points

The overall study cohort consisted of 419,297 participants, 2,161 of whom were sent a notification via Apple Watch and 450 of whom ended up wearing (and returning for analysis) ECG patches based on a clinical visit following the notification. The 2,161 patients receiving the notifications completed a 90-day survey following the notification. About 21% of the study cohort had hypertension and about 5% had diabetes.

According to the presenter presentation, the positive predictive value of the tachograms was 0.71 (97.5% CI, 0.69 to 0.74) and 0.84 (97.5% CI, 0.76 to 0.92) for the irregular pulse notifications. About half of participants who received a notification went on to seek medical attention outside of the study regardless of whether they had been assessed by a virtual doctor working for the study. The also reported that notifications in younger participants (under 40) were low (1.6%), and higher (3% or more) in older participants (over 65). Overall irregular pulse notifications were low at just 0.52% (0.49 to 0.54).

Limitations, and a Foundation

While the virtual study design was “large, quick, and pragmatic,” according to the researchers, the study limitations included self-reporting of outcomes from participants, which can be unreliable, as well as missing the target study enrollment of 500,000 participants. They also reported higher-than-anticipated drop-offs following notification, and fewer than anticipated telehealth study visits. Exposure to the app was safe.

“The positive predictive value does support the ability to identify AFib, and we think that these can provide context in the overall clinical context of everything we do,” author Mintu Turakhia, MD, of Stanford University School of Medicine, said in a press conference. “We think from a trial and operational standpoint this provides us with a foundation to do more digital health research.”

The study was funded by Apple, Inc.