Diagnostic accuracy of computer-assisted electrocardiography in the diagnosis of left ventricular hypertrophy in left bundle branch block.

INTRODUCTION AND OBJECTIVES Left ventricular hypertrophy has important prognostic implications. Although electrocardiography is the technique most often recommended in the diagnosis of hypertrophy, its diagnostic accuracy is hampered in the presence of a left bundle branch block. METHODS In 1875 consecutive patients (56±16 years) undergoing studies to rule out heart disease and/or hypertension, 2-dimensional echocardiography and electrocardiography were performed simultaneously in an outpatient clinic. Digitized electrocardiograms were interpreted using an online computer-assisted platform (ELECTROPRES). Sensitivity, specificity, likelihood ratios, and predictive values of standard electrocardiographic criteria and of some diagnostic algorithms for left ventricular hypertrophy were determined and compared with the findings in patients with neither left bundle branch block nor myocardial infarction. RESULTS Left bundle branch block was present in 233 (12%) patients. Left ventricular hypertrophy was detected more frequently in patients with left bundle branch block (60% vs 31%). In patients with left bundle branch block, sensitivities were low but similar to those observed in patients without it, and ranged from 6.4% to 70.9%, whereas specificities were high, ranging from 57.6% to 100%. Positive likelihood ratios ranged from 1.33 to 4.94, and negative likelihood ratios from 0.50 to 0.98. Diagnostic algorithms, voltage-duration products, and certain compound criteria had the best sensitivities. CONCLUSIONS Left ventricular hypertrophy can be diagnosed in the presence of left bundle branch block with an accuracy at least similar to that observed in patients without this conduction defect. Computer-assisted interpretation of the electrocardiogram may be useful in the diagnosis of left ventricular hypertrophy as it enables the implementation of more accurate algorithms.