Sonographic surveillance of abdominal aortic aneurysms: what is the smallest change in measured diameter that reliably reflects aneurysm growth?

OBJECTIVE: The decision to repair an abdominal aortic aneurysm may be based on an apparent increase in aneurysm diameter seen in successive sonographic examinations. However, true aneurysm growth can be diagnosed only if the measured increase in aneurysm diameter exceeds the variability inherent in the measurement. This study uses analysis of variance to determine the smallest change between 2 successive, independent sonographic measurements of aneurysm diameter that reliably indicates aneurysm growth. METHODS: Pairs of independent observers examined 63 patients with abdominal aortic aneurysms. Each observer obtained a sonographic measurement of the anteroposterior aneurysm diameter in the transaxial and longitudinal scan planes, and the transverse aneurysm diameter in the transaxial scan plane. Analysis of variance yielded the total variance associated with each type of measurement as well as the contributions made by variance among aneurysms, variance between observers and residual variance. These components were used to estimate interobserver standard error of measurement, interobserver reliability and the smallest measurement change needed to diagnose true aneurysm growth. RESULTS: Differences among aneurysms made the largest contribution to overall variance. Interobserver reliability was excellent, ranging from 0.89 to 0.94 (with perfect reliability being 1.00). The smallest difference between 2 successive, independent anteroposterior diameter measurements that indicated aneurysm growth at the 95% confidence level was 0.78 cm in the transaxial scan plane and 0.92 cm in the longitudinal scan plane. The smallest difference between 2 successive, independent transverse diameter measurements that indicated aneurysm growth at the same confidence level was 1.05 cm. CONCLUSION: Despite high interobserver reliability, only changes in measured aneurysm diameter greater than or equal to 0.78 cm indicate aneurysm growth at the 95% confidence level. Smaller changes in measured diameter may not be real, but due to variability in measurement.