Automatic atrial fibrillation and flutter detection by a handheld ECG recorder, and utility of sequential finger and precordial recordings

Abstract

Background: Handheld ECG recorders may have algorithms which automatically inform the user of presence of arrhythmia. The main objectives of this study were to evaluate the accuracy of the arrhythmia diagnosis algorithm of Beurer ME90 recorder to diagnose atrial fibrillation (AF)/flutter, and to evaluate whether recording technique (finger versus precordial) affects diagnostic performance.

Methods: Consecutive patients admitted at the cardiology ward of a tertiary care hospital were enrolled. Handheld ECG recordings were performed by holding the device between index fingers (lead I), and by applying it to the chest (modified V4, mV4), with 12‑lead ECGs serving as the gold standard for presence of arrhythmia.

Results: A total of 127 patients were included. The automatic arrhythmia detection algorithm identified all 16 cases of AF, but specificity was poor (62-77%, with slightly better specificity of mV4 compared to lead I). Specificity improved to 84% (95% CI 76-91%) if both lead I and mV4 recordings had to be positive for diagnosis, with a positive predictive value of 48% (95% CI 30-67%). Interpretation of the tracings by an electrophysiologist was 100% specific. Atrial flutter with regular ventricular response was however missed by automatic and manual interpretation.

Conclusions: The automatic arrhythmia algorithm of the BeurerME90 device has excellent sensitivity for diagnosing AF, but with low specificity. Strategies such as first recording lead I (more practical to perform), and in case of arrhythmia detection, confirming with an mV4 recording, may be applied to reduce false positive readings requiring manual confirmation by a healthcare professional.

Keywords: Algorithm; Atrial fibrillation; Atrial flutter; Diagnosis; Electrocardiogram; Handheld recording.