Observer variability in the diagnosis of ST-Elevation Myocardial Infarction from Emergency Room Electrocardiograms of patients presenting with chest pain

ABSTRACT

Background@#The diagnosis of ST-elevation myocardial infarction on standard 12-lead electrocardiography is crucial to the timely institution of appropriate medical therapies and interventions. However, the electrocardiographic diagnosis may be prone to observer variability. @*Objectives@#We wanted to determine the 1) inter-observer variability among clinicians from a training hospital in the diagnosis of ST-elevation myocardial infarction (STEMI) from emergency room electrocardiograms (ECG) of patients presenting with chest pain, and 2) the accuracy of clinician readings compared to the final cardiac diagnosis in the official patient records @*Methods@#Forty electrocardiograms were independently interpreted by 40 clinicians which included 10 cardiologists (CC), 5 cardiology fellows (CF). 15 internal medicine (IM), and 10 internal medicine residents (MR), using uniform copies of the standard 12-lead electrocardiogram print-outs. All readers were blinded to the patient's clinical profile. The accuracy of clinician readings (% of tracings correctly identified as STEMI) were compared against a reference standard, which was the final cardiac diagnosis made based on historical, electrocardiographic, biochemical, and angiographic data. @*Results@#The overall level of agreement among all readers (intergroup variability) was only fair with a kappa of 0.24. The level of agreement within the groups (intra-group variability) was likewise "fair" for the IM (0.35), MR (0.32), and CC (0.30 groups. Agreement was much less (0.18) among the cardiology fellows (0.18) among the cardiology fellows (CF). Accuracy of ECG diagnosis varied among the groups IM (68%), MR (60%), CF (52%) and CC (50%). @*Conclusions@#There is substantial inter-group variability in the electrocardiographic diagnosis of STEM by clinicians. The accurate diagnosis of STEMI, based on pure electrocardiographic informa-tion, varied among readers with different training backgrounds. These findings suggest that the electrocardiographic diagnosis of STEM may be influenced by training background and experience. as well as presence or absence of supplemental medical information necessary to make a comprehensive and accurate cardiac diagnosis. More importantly, this underscores the need to harmonize electrocardiographic interpretations and recognize the value of reading ECGs in light of pertinent clinical data