Original Article Intrarater Reliability of Dual-Energy X-Ray AbsorptiometryeBased Measures of Vertebral Height in Postmenopausal Women Alison M. Bonnyman, 1 Colin E. Webber, 2 Paul W. Stratford, 1 and Norma J. MacIntyre * ,1 1 School of Rehabilitation Science, McMaster University, Hamilton, Ontario, Canada; and 2 Department of Nuclear Medicine, Hamilton Health Sciences, Hamilton, Ontario, Canada Abstract The primary purpose was to estimate intrarater reliability of vertebral body height (VH) measures in postmeno- pausal women based on duplicate analyses of vertebral fracture assessment (VFA) images. The secondary purpose was to determine the consistency in classification of vertebral deformity on duplicate analyses. Thirty-two VFA were randomly selected from a database of 464 scans acquired in postmenopausal women using dual-energy X-ray absorptiometry (Discovery A; Hologic Inc., Waltham, MA). Visible endplates were marked on each image on 2 occasions (4 wk apart) by a single rater; the semiautomated software derived measures of anterior, middle, and posterior VH and classified severity of vertebral deformity. Intrarater reliability was assessed using the intraclass correlation coefficient (with 95% confidence interval [CI]) when 22 VFA could be analyzed. Reliability of grading deformity of 267 vertebrae was assessed using Cohen’s unweighted kappa (with 95% CI). Reliability of anterior, middle, and posterior height measures from T8 to L4 was 0.85 and greater except for T8 anterior VH and T9 pos- terior VH (0.76 [0.43, 0.90] and 0.62 [0.15, 0.83], respectively). Chance-corrected agreement for 4 grades of ver- tebral deformity was 0.48 (0.30, 0.66) and for 2 categories (normal/mild and moderate/severe) was 0.70 (50, 0.90). Intrarater reliability was acceptable for VH measures from T10 to L4. Reliability in grading severity of de- formity was improved by classifying as !25% deformity (nonfracture) and as O25% deformity (fracture). Key Words: DXA; postmenopausal bone loss; reproducibility of results; spinal fractures; vertebrae. Introduction Dual-energy X-ray absorptiometry (DXA) vertebral frac- ture assessment (VFA) generates measures of vertebral body height (VH) in the thoracic and lumbar spine (T4eL4) based on operator placement of 3 markers on each endplate visible in the lateral scan. A novel rationale for measuring VH is to provide insight into the effect of exercise interventions target- ing the spine in postmenopausal women with osteoporosis (1). The relative risks and benefits of exercise on vertebrae in women with established osteoporosis of the spine are not clear. In postmenopausal women with lower risk for vertebral fracture, exercise appears to have a modest effect on bone structure and strength in the skeletal regions experiencing the increased forces (2). Methods of evaluating such site- specific adaptations to exercise in the spine in vivo are lim- ited. One report suggests that the rate of loss in VH over 12 mo was attenuated in a group of 21 women with osteopo- rosis and vertebral fracture who completed a home exercise program compared with a group of 25 women who did not ex- ercise (1). Measures of VH were available for only a subset of 74 women (all with spine T-score less than 2.5, at least 1 vertebral fracture and on antiresorptive therapy) randomized to exercise or control arms of a trial investigating the effect of a 12-mo home exercise program on bone mineral density, physical function, and quality of life (3). Apart from this 1 re- port, the use of VH measures as an outcome in trials of Received 10/19/11; Accepted 03/06/12. *Address correspondence to: Norma J. MacIntyre, PhD, School of Rehabilitation Science, McMaster University, IAHS-403, 1400 Main St W, Hamilton, Ontario, Canada. E-mail: macint@mcmaster.ca 405 Journal of Clinical Densitometry: Assessment of Skeletal Health, vol. 15, no. 4, 405e412, 2012 Ó Copyright 2012 by The International Society for Clinical Densitometry 1094-6950/15:405e412/$36.00 DOI: 10.1016/j.jocd.2012.03.005