Interobserver variability in measuring the dimensions of the abdominal aorta: comparison of ultrasound and computed tomography

Abstract

Objectives: To examine the interobserver variability in measuring the abdominal aorta by ultrasound (US) and computed tomography (CT).

Design: A prospective clinical study in a university hospital.

Materials: Thirty-three patients whose abdominal aortas were scanned both with CT and US as a part of aneurysm investigation or for a variety of other reasons.

Methods: Three radiologists measured abdominal aortic diameters by US and CT. The interobserver differences (IOD) in US and CT and intraobserver differences for CT-US-pairs were analysed by various statistical methods. A new concept of "clinically acceptable difference" (CAD) was adopted denoting differences of less than 5mm.

Results: The IOD in US was 2mm or less in 65% of the anteroposterior and 61% of the transverse measurements and 5mm or more in 11% of the anteroposterior and 14% in the transverse measurements in 102 observer pairs for all aortas. The IODs were significantly larger in measuring the aneurysmal aortas compared with normal aortas (p < 0.001). The CAD-value for the aneurysmal aortas was 84% in the anteroposterior and 82% in the transverse directions. In CT the IODs were 2mm or less in 62% of the anteroposterior and 66% of the transverse measurements and 5mm or more in 12% of both anteroposterior and transverse measurements in 94 observer pairs for all aortas. The CAD-value in the aneurysmal aortas was 91% in the anteroposterior and 85% in the transverse directions. There was no significant difference between the US and CT CAD-levels. The absolute CT-US difference of an individual observer was 2mm or less in 54%, 5mm or more in 17% and 10mm or more in 2% of the anteroposterior measurements in the 95 CT-US pairs. In the transverse direction the corresponding figures were: 2mm or less in 63%, 5mm or more 13% and 10mm or more in 2% of the pairs. The diameters obtained by US were smaller in 84% of the cases compared with those of CT in measuring the maximum aortic diameter in anteroposterior direction, whereas the same figure for the transverse measurements was 59%.

Conclusions: Both US and CT measurements are subject to significant interobserver variability that must be taken into account in the clinical follow-up of small abdominal aortic aneurysms and in screening studies. Neither of these methods can be considered as a 'gold standard'.