Can a smartwatch spot atrial fibrillation?

Why does atrial fibrillation often go unnoticed?

Atrial fibrillation is a heart rhythm problem: the atria, the upper chambers of the heart, beat in a chaotic way. It makes blood clots more likely and raises the risk of stroke.

The trouble is that it does not always show itself. It can come and go with no clear symptoms, and be gone on the day you have an electrocardiogram (ECG, a recording of the heart's electrical activity) at the doctor's.

Yet millions of people already wear a watch with an optical sensor, a small light that tracks the pulse through the skin. The question was simple: can that sensor flag episodes nobody was seeing?

How was such a large study run?

A team of US researchers ran the study, funded by Apple. Participants, who said they did not have atrial fibrillation, signed up through an app on their phone. No hospital visits were needed.

An algorithm checked the pulse measured by the watch. If it flagged an irregular rhythm several times, the person got a notification, a video consultation was arranged and an ECG patch was posted to them, to be worn for up to 7 days.

The team wanted to know how many notified people really had atrial fibrillation, and how often a new watch alert matched what the ECG showed at the same moment.

How many alerts, and how many confirmed cases?

Over eight months, 419,297 people joined. Over a median of 117 days of monitoring, 2,161 of them, or 0.52%, received an irregular pulse alert. Getting an alert was rare.

Of the 450 people who sent back an ECG patch with usable data, applied on average 13 days after the alert, 34% had atrial fibrillation, and 35% of those aged 65 or over. That may sound low, but atrial fibrillation comes and goes, so it can be absent during the days the patch is worn.

When the watch sent a new alert while the patch was on, the ECG showed atrial fibrillation in 84% of cases. And 57% of notified people who answered the 90-day survey had contacted a health professional outside the study.

What does this mean for people who wear a watch?

A watch alert is not a diagnosis. It is a prompt to talk to a doctor, who can arrange a proper ECG. The study shows that this prompt is rare and often matches real atrial fibrillation at the moment it goes off.

It also shows that a very large medical study can be run entirely at a distance, using devices people already own. The authors see it as a foundation for future studies that measure the effect of such alerts on health.

What are the limits of this study?

Only some of the notified people sent back a usable ECG patch, so we do not know the outcome for all of them. The patch went on an average of 13 days after the alert, which may have missed episodes.

The study does not tell us whether these alerts prevent strokes or save lives: that was not its aim. Nor does it measure how many cases the watch missed in people who got no alert. Finally, participants were volunteers with an iPhone and a compatible watch, not a cross-section of the whole population, and the study was funded by the manufacturer.