Heart Health

Atrial fibrillation seen only on a device

How was it found? (the evidence differs) Device- or wearable-detected only

Recommendationaf-device-detectedNo one has reached this rule yet

Confirm on an ECG first; anticoagulation is a narrower call (NOAH-AFNET 6 / ARTESiA)

Clinical review

All reviewers

No clinician has signed off this rule yet. It reflects our reading of the guidelines cited below, which has not been independently checked.

What someone who reaches this rule is told

Word for word, as generated for the example person below. Figures in the text are theirs.

  • Worth raising with your doctor

    Your atrial fibrillation was picked up by a device rather than diagnosed clinically, and that genuinely changes the calculation. The first step is confirming it on a proper ECG.

    Brief high-rate episodes caught by a watch or an implanted monitor are not the same condition as clinically diagnosed AF, even though they share a name. The two trials that tested anticoagulation in exactly this situation — NOAH-AFNET 6 and ARTESiA — found the untreated stroke rate was only about 1% a year, and pooling them gives roughly a 32% reduction in stroke against roughly a 62% increase in major bleeding. That is a much narrower margin than in clinical AF, and it is why this is a genuine discussion rather than an automatic prescription. Episode length matters to that discussion, so bring the actual recordings.

    Outcome-provenTested in randomised trials measuring heart attacks, strokes, or death, not a stand-in for them.· Class 2b· Level A· checked 2026-09-12

    Meta-analysis of NOAH-AFNET 6 and ARTESiA — anticoagulation for device-detected atrial fibrillation (Circulation) · Class 2b — individualized decision

An example person who reaches it

“Atrial fibrillation from a smartwatch” — an illustrative profile, not a real user.

71-year-old manBP 124/78 on medicationLDL 120atrial fibrillation, seen on a device10-yr PREVENT-ASCVD 10.2%

Their path through atrial fibrillation

  1. Atrial fibrillation?Yes
  2. How was it found? (the evidence differs)Device- or wearable-detected only ← this rule
  3. CHA2DS2-VASc — untreated stroke risk above ~2% a year?Yes (score >=2 men / >=3 women)
  4. Applies whatever the scoreAlways
  5. Applies whatever the scoreAlways
  6. Applies whatever the scoreAlways

Where it sits

Asked after: Atrial fibrillation

Sources

Cited by the recommendations above.

What we are least sure of here

  • Q12Device-detected AF: stay at Class 2b, or collect episode duration and stratify as the guideline does?

Review this rule

For practising clinicians. Endorse it, propose a change, or dispute it. We verify every reviewer before a review is shown, and a review stays attached to the version of the rule you saw — if the rule changes, the page says so. You'll be asked to declare any relevant interests; they're shown with your review.

Log in or create an account to submit a review. An account is how we tie a review to a person we can verify.

Runs in lib/engine/afPathway.ts as af-device-detected.