Vertebral fracture assessment using the lateral scoutview of computed tomography in comparison with radiographs.

Semiquantitative vertebral fracture assessment was compared between lateral computed tomography (CT) scoutviews and conventional thoracolumbar spinal radiographs. Vertebral levels T4-L4 were assessed by both techniques in a group of 56 women (mean age 60 +/- 13 years). In order to compare inter- and intra-observer variabilities for the two techniques, the images were analyzed twice by two independent observers, and percentage agreement and kappa statistics were measured both between readings and between observers. Percentage agreement and kappa statistics were also used to quantify differences between techniques. In the CT scoutviews, noise and artifacts from overlying tissues in the thoracic spinal levels rendered 3.4% of the vertebrae unreadable for the first observer and 8.3% for the second observer. For the CT scoutviews the agreement between readings was 98.1%, 97.3% and 100% (kappa = 0.87, 0.83 and 1.0) on T4-L12, T4-12 and L1-4, respectively for the first observer, and 97.8%, 97.1% and 99.5% (kappa = 0.79, 0.73 and 0.92) for the second observer. For the lateral radiographs, the agreement between readings was 97.7%, 96.9% and 100% (kappa = 0.87, 0.85 and 1.0) on T4-L12, T4-12 and L1-4, respectively for the first observer, and 98.4%, 97.7% and 99.5% (kappa = 0.86, 0.82 and 0.95) for the second observer. The agreement between observers was 96.1%, 94.4% and 100% (kappa = 0.68, 0.58 and 1.0) on T4-L12, T4-12 and L1-4, respectively for the CT scoutviews and 96.8%, 95.9% and 99.0% (kappa = 0.79, 0.76 and 0.91) for the lateral radiographs. The inter-technique was 95.8%, 94.2% and 99.5% (kappa = 0.73, 0.68 and 0.95) on T4-L12, T4-12 and L1-4, respectively for the first observer and 95.6%, 94.2% and 99.0% (kappa = 0.64, 0.55 and 0.90) for the second observer, with the scoutview technique detecting, on average, 23% fewer fractures than the lateral radiographs. Although the vertebral fracture detection in lumbar spine is quite comparable to that of conventional radiographs, given its reduced sensitivity for vertebral fracture detection in thoracic spine, the lateral CT scoutview technique should not be substituted for conventional radiographs where diagnosis of all vertebral fractures is of primary importance.