Accuracy of systolic time intervals in detecting abnormal left ventricular performance in coronary artery disease

Sixty-six consecutive patients with a history of previous myocardial infarction and 48 patients with angina pectoris without evidence of previous myocardial infarction, all of whom had diagnostic coronary arteriography and left ventriculography, were studied in a prospective analysis of the accuracy of noninvasively determined systolic time intervals as a measure of global left ventricular performance. Forty-one patients who were evaluated for atypical chest pain and found to have normal coronary arteries and left ventricular performance served as control subjects. Six methods of statistical analysis were employed in assessing the accuracy of systolic time intervals in relation to the left ventricular ejection fraction: (1) analysis of variance, (2) cumulative distribution analysis, (3) correlation, (4) sensitivity and specificity, (5) percent agreement, and (6) logistic regression analysis. These tests permitted comparison between the systolic time intervals and the angiographic left ventricular ejection fraction. Analysis of variance revealed identical discriminating power for the ratio of the preejection period to left ventricular ejection time (PEP/ LVET) and left ventricular ejection fraction in separating the normal group and patients without previous myocardial infarction from the patients with previous myocardial infarction. The preejection period and left ventricular ejection time corrected for heart rate were less discriminating than left ventricular ejection fraction or PEP/LVET. The cumulative distribution plots for the left ventricular ejection fraction and PEP/LVET in the three groups of patients were remarkably similar. The correlation of PEP/LVET and left ventricular ejection fraction for all three groups of patients was 0.84. The sensitivity and specificity of the PEP/LVET in relation to the left ventricular ejection fraction were 88 and 96 percent, respectively. The overall agreement between the two measures in detecting the prevalence of abnormality in global left ventricular performance in subgroups of patients was 92 percent. By logistic regression analysis the two measures had equal capacity in discriminating the patients with previous myocardial infarction from the control group. The multiple strategies of comparison employed in this study document the close relation of measures of the timing of the left ventricular contraction cycle by systolic time intervals and estimates of the extent of left ventricular contraction by ejection fraction in patients with coronary artery disease. It is concluded that these measures afford independent and complementary methods of defining the presence of abnormal left ventricular performance in the resting supine patient with coronary artery disease.