How was it found? (the evidence differs) Device- or wearable-detected only
af-device-detectedNo one has reached this rule yetConfirm on an ECG first; anticoagulation is a narrower call (NOAH-AFNET 6 / ARTESiA)
No clinician has signed off this rule yet. It reflects our reading of the guidelines cited below, which has not been independently checked.
Word for word, as generated for the example person below. Figures in the text are theirs.
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.
Proven in trialsTrials counted real events, like heart attacks, strokes, or deaths, and this changed them.
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.
ACC/AHA (US)
The guideline grades device-detected episodes by how long they lasted. 2b is for episodes of 5 minutes to 24 hours with a score of 3 or higher. The rationale quotes two trials, which are graded as randomized evidence on their own; the guideline's class and level for the decision are shown here.
2023 ACC/AHA/ACCP/HRS Guideline for the Diagnosis and Management of… · Recommendation behind this rule · class and level read from the guideline tables on 2026-09-30
An episode of 24 hours or more with a CHA2DS2-VASc score of 2 or higher: anticoagulation is reasonable
2023 ACC/AHA/ACCP/HRS Guideline for the Diagnosis and Management of… · Graded separately within the same recommendation · class and level read from the guideline tables on 2026-09-30
An episode under 5 minutes with no other reason for anticoagulation: do not anticoagulate
2023 ACC/AHA/ACCP/HRS Guideline for the Diagnosis and Management of… · Graded separately within the same recommendation · class and level read from the guideline tables on 2026-09-30
Meta-analysis of NOAH-AFNET 6 and ARTESiA — anticoagulation for device-detected atrial fibrillation (Circulation) · Class 2b · Level B-NR
“Atrial fibrillation from a smartwatch” — an illustrative profile, not a real user.
Asked after: Atrial fibrillation
Cited by the recommendations above.
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.