Cardiologists Miss AFib on Single-Lead Strips About 1 in 4

This article covers how consistently and accurately cardiologists interpret single-lead ECG tracings from handheld and wearable devices, and how device algorithms compare in one study. It does not cover diagnosis or advice about any tracing.

Not reliably. Fifteen cardiologists read the same 200 single-lead ECGs from a handheld AliveCor KardiaMobile in patients aged 65 and over, against a same-day 12-lead ECG. Average sensitivity for AFib was 77.4% (range 50% to 90.6%) and specificity 73.0% (41.3% to 94.6%). In a separate screening study, two cardiologists agreed moderately (kappa 0.48 at participant level, 0.58 at ECG level). Neither study used a smartwatch.

How well do cardiologists read a single-lead ECG?

Modestly. Fifteen cardiologists each reviewed the same 200 handheld tracings from patients aged 65 or older in the VITAL-AF trial. The device was an AliveCor KardiaMobile. A same-day 12-lead ECG was the gold standard.

The cardiologists were blinded to the device algorithm and to the 12-lead result. Each classified the rhythm as sinus, AFib or unclassifiable and rated their confidence.

Average sensitivity for AFib was 77.4%, ranging from 50% to 90.6%. Average specificity was 73.0%, ranging from 41.3% to 94.6%. Variability between reviewers averaged 30.8% and reached 76.2%. Confidence averaged 3.6 out of 5.

Age, body mass index and artifact affected accuracy and confidence. The authors call the diagnostic accuracy modest.

Do two cardiologists agree with each other?

Moderately. A population-based AFib screening study had adults aged 65 or older record four single-lead ECGs a day for one to four weeks with a handheld recorder. Of 2,141 participants and 162,515 recordings, 1,843 ECGs from 185 participants were reviewed by both of two independent cardiologists.

Agreement was moderate, with a weighted kappa of 0.48 (95% CI 0.37 to 0.58) at participant level and 0.58 (0.53 to 0.62) at ECG level. Agreement was higher in people who recorded at least 67 adequate-quality ECGs. At ECG level it depended on quality and on whether the algorithm flagged possible AFib.

The authors suggest cardiologist training and getting enough ECGs per person could improve reliability.

Do the device algorithms do better?

In one clinic study, yes, on good-quality readings. A 2025 study of 122 participants compared KardiaMobile 6L, an Apple Watch and two smartphone apps (FibriCheck and Preventicus) against a 12-lead ECG. Thirty had AFib.

Sensitivity for AFib was 100% for all four devices. Specificity for sinus rhythm was 96.4% for KardiaMobile 6L, 97.8% for Apple Watch, 98.9% for FibriCheck and 97.8% for Preventicus. Recordings of insufficient quality ranged from 7.4% to 14.8%.

This is not a head-to-head with the cardiologist studies. The populations, devices and inclusion rules differ. Detection accuracy is covered in how accurately a wearable ECG detects AFib. Inconclusive rates are in wearable ECG comparison.

What can a single-lead strip miss?

Rhythms other than AFib. A case report describes a 70-year-old man with palpitations. His Apple Watch single-lead ECG classified sinus rhythm at 75 beats per minute. A same-day 12-lead ECG showed typical atrial flutter.

The authors note that AFib-validated algorithms may miss other supraventricular arrhythmias, especially with a regular ventricular response. They say persistent symptoms need a standard 12-lead ECG. It is one case.

What does this not settle?

Neither reader study used a smartwatch. Both used handheld recorders in adults aged 65 or older, one from a clinical trial and one from a screening program.

Neither studied whether reader disagreement changes patient care. The 12-lead ECG served as the reference in the first study and was not itself in dispute.

Common questions