Can Radiologists Detect Osteopenia on Plain Radiographs

M J Garton, Elizabeth Robertson, Fiona Jane Gilbert, L Gomersall, David M Reid

Research output: Contribution to journalArticlepeer-review

24 Citations (Scopus)

Abstract

The main purpose of this study was to compare subjective estimates of vertebral osteopenia with bone mineral density (BMD) measurements of the same region in 200 (107 male, 93 female) subjects, aged 52-90 years, invited randomly from the community. Standardized plain thoracolumbar radiographs were examined by three senior radiologists, and given a semiquantitative osteopenia score using the method of Saville. The BMD of the anteroposterior (AP) lumbar spine and femoral neck was measured by dual energy X-ray absorptiometry. In addition BMD measurements were compared in subjects without significant vertebral deformity, and those with mild (20-25%) or definite (> 25%) reductions of vertebral height. For the Saville score, intraobserver agreement was moderate to good (kappa 0.46-0.57), and interobserver agreement was fair to moderate kappa 0.25-0.41). Although the overlap between gradings was considerable, BMD was significantly related to visually estimated osteopenia. Subjects without apparent radiographic osteopenia (Saville grade 0) had a low risk (9-15%), compared to those with definite osteopenia (64-86%), of falling below the lowest quartile of BMD at either the femoral neck or the AP spine. In women (but not men), BMD measured at the hip and spine was related to vertebral deformity.

Original languageEnglish
Pages (from-to)118-122
Number of pages5
JournalClinical Radiology
Volume49
Issue number2
DOIs
Publication statusPublished - Feb 1994

Keywords

  • BONE DENSITOMETRY
  • VERTEBRAL FRACTURES
  • OSTEOPOROSIS
  • DEFORMITY
  • WOMEN
  • SPINE
  • MASS
  • Bone Densitometry
  • Vertebral Fractures
  • Osteoperosis
  • Deformity
  • Women
  • Spine
  • Mass

Fingerprint

Dive into the research topics of 'Can Radiologists Detect Osteopenia on Plain Radiographs'. Together they form a unique fingerprint.

Cite this