Variability of maximal aortic aneurysm diameter measurements on CT scan: significance and methods to minimize.

OBJECTIVES

We noted substantial differences when measuring repeatedly the same abdominal aortic aneurysm (AAA) on the same computed tomography (CT) scan. This study quantitated this variability, and methods to minimize it were developed.

METHODS

The CT maximal diameter of 25 AAAs was measured by eight experienced observers, including six vascular surgeons and two radiologists, using two methods: an unstandardized protocol, and a standardized protocol using fine calipers to carefully measure the largest diameter perpendicular to the estimated aneurysm centerline, from outer aneurysm wall to outer wall. The average measurement difference between observers was calculated for each method. The average difference between each observer's measurement and the official radiology report value was also calculated. Agreement between the two measurement methods was assessed with Bland-Altman plots.

RESULTS

The difference in maximal diameter measurements between each observer averaged 4.0 +/- 5.1 mm (range, 0.0-35.0 mm) with the unstandardized method. The mean measurement difference with the standardized protocol was significantly lower, and averaged 2.8 +/- 4.4 mm (range, 0.0-26.0 mm; P<.05). Measurements taken from the official radiology report differed from each of the observer's standardized measurement by an average of 5.0 +/- 6.3 mm (range, 0.0-28.0 mm). This difference was similar for both the unstandardized and standardized methods. Bland-Altman plots confirmed the wide variation of the maximal diameter measurements when the unstandardized method was compared with the standardized method (95% confidence interval, -9-9 mm).

CONCLUSIONS

Routine CT maximal diameter measurement of AAAs can have substantial interobserver variability. Standardized measurement protocols can decrease, but not eliminate, this measurement variability. Thus apparent size changes based on CT measurements may represent measurement artifact rather than actual aneurysm growth or shrinkage, particularly when a standardized system is not used.