Noninvasive tests to evaluate the severity of aortic stenosis. Limitations and reliability.

Fifty patients were examined with phonocardiograms, carotid pulse tracings, and M-mode echocardiograms to evaluate the ability of noninvasive tests to identify the severity of aortic valvular disease as determined at cardiac catheterization. Linear and multivariate analysis showed these noninvasive approaches to have only fair correlation with the severity of the disease. A binary division of the population under study into severe and nonsevere subgroups based on aortic valvular area (less than or equal to 0.8 sq cm in severe group [n = 25]; greater than 0.8 sq cm in nonsevere group [n = 25]) allowed sensitivity, specificity, and likelihood ratios to be determined. The likelihood ratio increased fourfold as the interval from the ECG Q wave to the murmur's peak (Q-MP) prolonged to 320 msec and increased sevenfold when the rate-corrected left ventricular ejection time (delta LVET) was more than 40 msec beyond values predicted from standard regression equations. Echocardiographic measurements were less helpful. Prolonged values of Q-MP and delta LVET proved to be the best discriminators of severe aortic valvular disease in this population where the prevalence of severe and nonsevere disease was equal.