What is the clinical evidence from this study?

Study design: Observational. Population: Electrocardiogram interpretation (n=294). Intervention: Smartphone-based 12-lead ECG machine computerized interpretation vs. Gold standard 12-lead ECG machine computerized interpretation. Primary outcome: Accuracy in correctly interpreting a normal ECG report compared to a cardiologist's diagnosis.

December 13, 2022Open Access

Evaluating the frequency of false-positive detection in electrocardiogram devices in comparison to cardiologist diagnosis: instrument validation study (Preprint)

Key Result

A smartphone-based 12-lead ECG machine demonstrated higher accuracy (86.2% vs 26.43%) in detecting a normal ECG compared to a gold standard ECG machine's computerized interpretation.

Study Design

Type

Observational (n=294)

Structured PICO

Does computerized interpretation by smartphone-based and gold standard 12-lead ECG machines accurately detect normal ECGs compared to cardiologist diagnosis in subjects undergoing electrocardiography?

Population

294 subjects whose 12-lead ECG reports from a gold standard machine and a smartphone-based machine were compared against a cardiologist's diagnosis.

Exposure

Computerized interpretation by a smartphone-based 12-lead ECG machine and a gold standard 12-lead ECG machine

Comparator

Cardiologist diagnosis based on observational changes in the morphology of the ECG traces

Outcome

Sensitivity, specificity, and accuracy in correctly interpreting a normal ECG reportsurrogate

Computerized ECG interpretation algorithms vary significantly in accuracy and should be used only as an assisting tool to a cardiologist's diagnosis rather than an independent diagnostic method.

Main Result

Absolute Event Rate: 86.2% vs 26.43%

Abstract

BACKGROUND Electrocardiography is the most convenient and cost-effective method at the primary level of healthcare infrastructure to localize and diagnose Myocardial infarction, ischemic heart diseases, and numerous arrhythmias. The manufacturers of ECG machines provide computer-generated Interpretations. However, these machines have differences in implementing the algorithms, which causes a change in diagnostic accuracy. Hence, the occurrence of False positives is the most commonly observed error that occurs during computer interpretation. OBJECTIVE This study aims to evaluate the differences observed in the computerized interpretation of ECG reports regarding the Cardiologist's diagnosis. METHODS The 12-lead ECG reports were collected from a 12-lead gold standard machine and a smartphone-based 12 lead ECG machine. The data of the 294 subjects out of 300 subjects were accessed from both ECG machines. The reports were evaluated by a Cardiologist based on the observational changes in the morphology of the ECG traces. RESULTS The gold standard ECG machine was 92% sensitive, 47.9% specific and 26.43% accurate in correctly interpreting a normal ECG report concerning the diagnosis provided by a Cardiologist. Whereas the Smartphone-based 12 lead ECG was found to be 95.9% sensitive, 88.9% specific, and 86.2% accurate in detecting a normal ECG concerning the diagnosis provided by a Cardiologist. CONCLUSIONS The ECG machine manufacturer's interpretation algorithm plays an important role in defining the accuracy of the ECG machine. Computerized interpretation is only an assisting tool for clinicians and not an independent tool to be relied on while treating a patient during health emergencies and check-ups.