Introduction: Mostly, it is impossible to establish the type of arrhythmias, based on signs and symptoms only. An ECG device is not always within reach. We presumed the heartscan, a handheld wireless device, to be of value to a GP in emergency situations. We therefore studied inter- and intra-observer variability and the accuracy of the screen readings.
Methods: All consecutive patients visiting the emergency department (ED) of the Gasthuisberg University Hospital in Leuven, Belgium, as well as patients hospitalised in three hospital wards on one day, were included. Immediately after the heartscan recording, a standard 12-lead ECG was recorded and read by an experienced hospital-based cardiologist. The recordings were read on the device screen by two general practitioners. All readers were blinded to the 12-lead ECG readings and vice versa, and for each other's ones. We compared both the heartscan reading of the first reader and the automatic reading of the device with the readings of the second GP and to the 12-lead ECG results, used as the gold standard. Intra- and inter-observer agreement was studied using total accuracy and kappa values with their 95% confidence interval (CI).
Results: Full data of 177 (73%) patients, 80 men (45%) and 97 women (55%), with a mean age of 55 years (range 18-94 y) were recorded. The specificity of the heartscan reading by a clinician was 88%, the sensitivity between 60 and 69%, PPV < 50% and NPV > 95%. Comparing codes of the heartscan with the ECG readings was difficult but sensitivity for atrial fibrillation was 92.3%. Inter- and intra-observer accuracy were high (> 0.86 and > or = 0.95, respectively), with low kappa values.
Conclusion: The detection of rhythm disorders by the device is incomplete. However, the heartscan can be a help for the GP. The performance of the heartscan could probably be improved by increasing screen resolution, but, in the future, more sophisticated heart monitors should become available. They should be small, light and affordable.