Radiographic measures of thoracic kyphosis in osteoporosis: Cobb and vertebral centroid angles

Abstract

Several measures can quantify thoracic kyphosis from radiographs, yet their suitability for people with osteoporosis remains uncertain. The aim of this study was to examine the validity and reliability of the vertebral centroid and Cobb angles in people with osteoporosis. Lateral radiographs of the thoracic spine were captured in 31 elderly women with osteoporosis. Thoracic kyphosis was measured globally (T1-T12) and regionally (T4-T9) using Cobb and vertebral centroid angles. Multisegmental curvature was also measured by fitting polynomial functions to the thoracic curvature profile. Canonical and Pearson correlations were used to examine correspondence; agreement between measures was examined with linear regression. Moderate to high intra- and inter-rater reliability was achieved (SEM = 0.9-4.0 degrees ). Concurrent validity of the simple measures was established against multisegmental curvature (r = 0.88-0.98). Strong association was observed between the Cobb and centroid angles globally (r = 0.84) and regionally (r = 0.83). Correspondence between measures was moderate for the Cobb method r = 0.72), yet stronger for the centroid method (r = 0.80). The Cobb angle was 20% greater for regional measures due to the influence of endplate tilt. Regional Cobb and centroid angles are valid and reliable measures of thoracic kyphosis in people with osteoporosis. However, the Cobb angle is biased by endplate tilt, suggesting that the centroid angle is more appropriate for this population.