Aim: Within the 6-month prospective timeframe: 1. to determine the extent to which clinicians referring to the cardiology advanced nurse practitioner (ANP) service have relied on automated electrocardiogram (ECG) machine software reports to confirm or rule out the diagnosis of atrial fibrillation; 2. to determine the false-positive or false-negative rates of AF made by the automated ECG software; 3. to identify factors that increase the likelihood of false-positive or false-negative rates of AF.
Design: This was a 6-month prospective review of formal cardiology consults provided to clinicians (emergency department [ED] and medical teams) for what they believed to be confirmed cases of AF. All patients have been referred or had self-presented to the ED. All ECGs were reviewed by a Registered cardiology ANP (RANP). Potential, contributory factors that may have compounded the software interpretation were considered by the RANP, including motion artefact, premature atrial/ventricular contractions (APCs/PVCs), visually low amplitude P waves and irregular QRS.
Results: A total of 21 consult requests, which were made to the Cardiology ANP service by clinicians in the ED or on-call medical teams as a result of what was believed to be a confirmed case of atrial fibrillation, were included. Overall, 19 automated ECG reports were deemed by the RANP to be false positives, which inaccurately reported AF. In 2 cases, the RANP was asked to provide consults for other reasons and identified false-negative reports where atrial flutter was not accurately reported.
Motion artefact was the biggest compounding factor affecting the accuracy of automated ECG reports seen in 62% (13/21) of false-positive reports followed by low amplitude P waves (48%, 10/21), irregular QRS (43%, 9/21) and presence of APC/PVCs in 38% (8/21) of reports. Interestingly, the 2 false-negative reports had no compounding factors to explain the inaccuracy.
Conclusions/recommendations: Significantly larger studies are aligned with ours and support the fact that clinicians are over-reliant on automated ECG reports, and that regardless of ECG machine manufacturer or cost, false-positive automated ECG software reports are worryingly high in general and seem inevitable when there are compounding factors, for example when: 1. there is motion artefact on the isoelectric line; 2. there is premature atrial or ventricular ectopy; 3. the QRS is irregular; 4. there are visually low amplitude P waves. These findings should serve as a warning to those involved in the management of AF, particularly when the cornerstone of AF management from the outset is predicated on accurate ECG diagnosis. The reality is that inexperienced clinical staff are more likely, in the absence of strict senior oversight and clinical governance with regards to manual ECG review, to over-rely on automated ECG machine software reports that have consistently proven to have high false-positive rates. This over-reliance has the potential, in the worst-case scenario, to put patients at risk of unnecessary lifelong anticoagulation and resulting serious bleed events. ❑
