P0381 - Comparison of Electrocardiogram Interpretations by Computer Algorithms, Pediatric Emergency Department Physicians, and Pediatric Cardiologists

Background
Electrocardiograms (EKGs) are a fundamental tool for detecting cardiac abnormalities. Computer interpretation of EKGs are commonly used to increase efficiency and facilitate interpretation. However, computer interpretation of EKGs is based on pattern recognition, and measured intervals are not always accurate. Therefore, reliance on computer interpretations alone may lead to errors.

Further, concerns have been raised about the consistency and accuracy of EKG interpretations performed by non-cardiologists compared to interpretations by cardiologists. We evaluated the accuracy of EKG interpretations at a single pediatric academic institution, comparing the interpretations of a computer (C), Emergency Department physicians (EDP), and Pediatric Cardiologists (PDC). We evaluated our data based on the incidence of discrepant interpretations between the three groups, as well as the significance of the discrepancies.

Methodology
284 random EKGs from the pediatric Emergency Department were randomly selected over a 4-year period, and interpreted independently by EDP and PDC. Interpretations were considered to be identical if the differences between studies was unlikely to be of clinical significance (eg. small differences in the measurement of intervals, EKG findings which were within normal limits).

Discussion
EDP and C interpretations were the same in 96.4% of EKGs (270 of 284). EDP and PDC as well as EDP and C interpretations were the same in 90.8% of EKGs (258 of 284 for both comparisons). In all cases, where the EDP and PDC interpretations were discrepant, the EDP matched the C interpretation.

For the 26 EKGs with discrepant interpretations by the EDP and PDC, 24 (92.3%) were related to incorrect measurement of the corrected QT interval (QTc). In 75.0% (18 of 24) of EKG's with errors in QTc, a prolonged or mildly prolonged QTc was missed, in 25.0% (6 of 24) a normal QTc was diagnosed as prolonged. The other 2 discrepant interpretations were related to a misdiagnosis by EDPs of an infarct (3.8%), or of an arrhythmia (3.8%).

Cohen’s Kappa coefficient (K) and 95% confidence intervals were calculated to determine the inter-rater reliability of EKG interpretation by C, EDPs, and PDCs. Statistically, there was no significant difference in EKG interpretation between EDP v C, PDC v C, and EDP v PDC. Although differences in EKG interpretation were not statistically significant. However, individual errors in EKG interpretation can still be clinically relevant.

Conclusion
The most common difference between pediatric EDPs and PDCs that could impact patient management was in measuring QTc intervals. Abnormally long or short QT intervals are associated with an increased risk for life‐threatening ventricular arrhythmias and sudden cardiac death. Findings from this study suggest that a focused intervention directed towards the measurement of the QTc is likely to be successful in improving the accuracy of EKG interpretation in the pediatric Emergency Department at our institution.