Inaccuracies in using aortic valve gradients alone to grade severity of aortic stenosis.

The severity of aortic stenosis is an important determinant of prognosis in patients with symptoms who do not undergo valve replacement. To assess the pitfalls of using valve gradients alone 636 patients with aortic stenosis in whom the aortic valve area had been calculated by the Gorlin formula were studied. The correlation between valve area and aortic gradients was poor. No gradient was found that was both sensitive and specific for aortic stenosis. The maximum predictive accuracy was 81% for a mean gradient of 30 mm Hg and 80% for a peak gradient of 30 mm Hg. A mean gradient of 50 mm Hg or a peak gradient of 60 mm Hg were specific with a 90% or more positive predictive value. It proved difficult, however, to find a lower limit with a 90% negative predictive value. Patients with severe aortic stenosis and low gradients (peak or mean gradient of less than 30 mm Hg) had small ventricles (on both angiographic and echocardiographic data) with good ejection fractions and so were unlikely to be detected subjectively. In comparison patients with mild aortic stenosis and low gradients tended to have more aortic regurgitation but have similar degrees of left ventricular hypertrophy on echocardiographic or electrocardiographic criteria. The aortic valve area should be measured in all patients with the suspicion of severe aortic stenosis with a mean gradient of less than 50 mm Hg (50% of patients in this study) or a peak gradient of less than 60 mm Hg (47% of patients in this study).