Assessment of left ventricular function from the electrocardiogram in left bundle branch block.

The QRS duration, maximum right precordial S amplitude, sum of amplitudes of the maximum right precordial S and T wave and T wave polarity in lead I have been analyzed in order to identify electrocardiographic predictors of left ventricular end-diastolic volume index and ejection fraction in 165 patients with complete left bundle branch block and various forms of heart disease. Multivariate analysis selected the duration, maximal amplitude of the S wave and polarity of the T wave in decreasing order of discriminatory power, which correctly identify 76.6% of the patients with a normal end-diastolic volume index less than or equal to 90 ml/m2 and a normal ejection fraction greater than or equal to 60% (n = 64) and 73.3% of those with an end-diastolic volume index greater than 90 ml/m2 or an ejection fraction less than 60% (n = 101). The comparisons of the QRS duration with the end-diastolic volume index and the ejection fraction give the best single correlations: r = 0.57 and -0.63, respectively. Multiple correlations lead to no substantial improvement of the r values: 0.06 and -0.65, respectively. A QRS duration less than 140 msec is almost always predictive of the presence of a normal end-diastolic volume index and a normal ejection fraction (sensitivity 100%, specificity 91.9%, positive predictive value 73.3%). A QRS duration greater than 170 msec is most accurate in predicting depressed left ventricular ejection fraction less than 55% (sensitivity 36.5%, specificity 98%, positive predictive value 92%). Thus, only the QRS duration provides a useful reference and guide for the evaluation of left ventricular function in the presence of left bundle branch block.