Signal-averaged electrocardiogram. Improved identification of patients with ventricular tachycardia using a 28-lead optimal array.

BACKGROUND Although the signal-averaged ECG (SAECG) is currently the best noninvasive test to identify patients with ventricular tachycardia (VT) following myocardial infarction (MI), it is still a relatively insensitive test. Body surface mapping has improved the sensitivity of ECG in detecting various cardiac diseases. This study applied body surface mapping to the SAECG in the form of a clinically practical, 28-lead optimal array and compared its sensitivity and specificity with those of an orthogonal array. METHODS AND RESULTS Two hundred twenty-three patients with previous MI (82 with inducible VT) underwent SAECG using 28 surface electrodes from which were obtained a three-lead orthogonal array and a 28-lead optimal array (optimal). From the orthogonal array, two QRS durations (QRSd) were obtained using the combined vector magnitude method (CVM) and the earliest onset to latest offset of the three individually filtered leads (individual). From the optimal array, 28 QRSd were obtained, each defined as the duration from the earliest onset of any of the 28 leads to the offset of each individually filtered lead. QRSd > 120 msec in > or = 3 leads was considered abnormal. For CVM and individual, QRSd of > 120 msec were considered abnormal. While the specificity of each method was comparable (84%, 86%, and 84% for CVM, individual, and optimal, respectively), the sensitivity of optimal (70%) was significantly greater than the sensitivity of CVM (54%) (p = 0.001) or individual (59%) (p = 0.004). The magnitude of improvement in sensitivity, 16% and 15%, respectively, was equal for anterior (n = 120) and inferior (n = 103) infarctions. CONCLUSIONS Body surface mapping using the 28-lead optimal array significantly improved the sensitivity of the SAECG without loss of specificity. The increased sensitivity was of equal magnitude for inferior and anterior infarctions. The superiority and practicality of the 28-lead optimal array make it worth pursuing as an option for further refinement in SAECG: