Intertest variability of echocardiographic and chest X-ray measurements: implications for decision making in patients...

Echocardiograms and chest X-ray examinations are commonly employed for serial measurements of left ventricular size and function in patients with chronic aortic insufficiency and often support or even determine therapeutic decisions. This study was undertaken to assess the intertest variability of these measurements made from M-mode echocardiograms and X-ray films performed 3 months apart without intervening clinical or therapeutic changes in 22 patients with significant but asymptomatic aortic insufficiency. End-diastolic and end-systolic dimensions, fractional shortening and cardiothoracic ratios were measured by the same reader, with the initial and 3 month tests being read both independently and together for comparison. The mean values for the initial and 3 month studies were similar, but the intertest variability was substantial, especially when the two tests were read independently. The 95% prediction limits are approximately 50% smaller when the serial studies are read together for comparison. The coefficient of variation for end-diastolic and end-systolic dimensions was 6.1 and 10.1%, respectively, and that for fractional shortening was 17.1%. These findings translate into 95% level prediction limits exceeding +/- 8 mm for left ventricular dimensions and 0.12 for fractional shortening; changes on serial evaluations would have to exceed these values to be considered with a high degree of certainty to represent more than random variability. Although this variability may reflect a number of biologic and technical factors, it emphasizes the need to be cautious in making decisions based solely on changes between two tests, particularly if they are not evaluated together.