Evaluation of Pediatric Electrocardiogram Diagnosis of Ventricular Hypertrophy by Computer Program Compared with Cardiologists

摘要

This study aimed to evaluate the diagnosis of pediatric left ventricular hypertrophy (LVH) and right ventricular hypertrophy (RVH) by the Glasgow electrocardiogram (ECG) interpretation program compared to interpretations provided by two pediatric cardiologists. ECGs had all identifying information removed and were sent to the cardiologists independently with the patient’s age and sex and the clinical indication for the ECG, if known. A total of 984 ECGs were included in the study, of which 664 were reported “with clinical indication” and 320 were reported “blind.” With respect to an averaged diagnosis of the two cardiologists, the sensitivity of the program for RVH was better when the cardiologists reported blind (73.3%) than with the clinical indication (53.5%), with the same trend for the program compared with individual cardiologists. The specificity of the program was at least 94.4% in all cases. For LVH, the program had high specificity (=95.8%) for “reported blind” and “with clinical indication” cases but low sensitivities throughout (the highest was 44.4% with respect to an averaged diagnosis of the two cardiologists reporting with the clinical indication). Subsequent discussion revealed that if the cardiologists had disagreed with one another initially, their consensus opinion was twice as likely to be in agreement with the program.