Variability in Cobb Angle Measurements Using Reformatted Computerized Tomography Scans

Abstract

Survey of intraobserver and interobserver measurement variability. To assess the use of reformatted computerized tomography (CT) images for manual measurement of coronal Cobb angles in idiopathic scoliosis. Cobb angle measurements in idiopathic scoliosis are traditionally made from standing radiographs, whereas CT is often used for assessment of vertebral rotation. Correlating Cobb angles from standing radiographs with vertebral rotations from supine CT is problematic because the geometry of the spine changes significantly from standing to supine positions, and 2 different imaging methods are involved. We assessed the use of reformatted thoracolumbar CT images for Cobb angle measurement. Preoperative CT of 12 patients with idiopathic scoliosis were used to generate reformatted coronal images. Five observers measured coronal Cobb angles on 3 occasions from each of the images. Intraobserver and interobserver variability associated with Cobb measurement from reformatted CT scans was assessed and compared with previous studies of measurement variability using plain radiographs. For major curves, 95% confidence intervals for intraobserver and interobserver variability were +/-6.6 degrees and +/-7.7 degrees, respectively. For minor curves, the intervals were +/-7.5 degrees and +/-8.2 degrees, respectively. Intraobserver and interobserver technical error of measurement was 2.4 degrees and 2.7 degrees, with reliability coefficients of 88% and 84%, respectively. There was no correlation between measurement variability and curve severity. Reformatted CT images may be used for manual measurement of coronal Cobb angles in idiopathic scoliosis with similar variability to manual measurement of plain radiographs.