Clinical Decision Guidelines
Background: The HEART Score, a clinical decision aid for emergency department (ED) patients with chest pain, incorporates subjective variables that clinicians must interpret, like electrocardiograms (ECG’s). However, instructions to correctly score ECG’s are limited. Further, a single point separates low and high-risk patients, making correct ECG scoring vital. Our study assessed the inter-rater agreement of HEART score ECG classification and identified common ECG patterns difficult to score.
Methods: A prospective observational study was performed at a single academic center comparing ED providers’ clinical HEART scores with scores derived through structured data collection. A convenience sample of adult patients presenting with symptoms concerning for acute coronary syndrome were included. ED providers using the HEART score were asked to complete a survey detailing their score, including their ECG classification. ECG’s were then independently reviewed and scored by two board-certified emergency physicians engaged in HEART score research. First, the researchers categorized each ECG according to published predefined criteria. Researchers then discussed common challenges to ECG scoring and independently re-categorized ECG’s. Finally, researchers identified ECG’s with conflicting scores and discussed the patterns that led to differences. Agreement and weighted kappa (WK) were used to compare scores.
Results: During the 24-month study, 288 provider patient pairs were enrolled. Using predefined criteria, researchers independently agreed on 75.4% of ECG’s (WK 0.64). After discussing challenges and re-categorizing, agreement improved to 85%(WK 0.72). The most common reason for scoring differences in the 42 remaining ECG’s included: diffuse flattening of T-waves (13), lateral T wave inversions (11), ECG artifact (3), and biphasic V2 or V3 (3). The final research scores agreed with ED provider’s ECG classification in 71.2% of cases (WK 0.48), including 3 cases in which scores diverged by 2 points.
Conclusion: Even when scored by ED physician experts, inter-rater agreement on HEART score ECG classification remained only moderate to good. Common ECG patterns in conflicting scores included T wave flattening and lateral T wave inversions, both of which are not defined in published criteria.