Does osteoporosis classification using heel BMD agree across manufacturers?

Osteoporos Int. 2002 Aug;13(8):613-7. doi: 10.1007/s001980200082.

Abstract

The lack of standardization in bone mineral density (BMD) measurements is known. Several studies have been carried out to cross-calibrate the axial dual X-ray absorptiometry (DXA) devices. Recently, a number of peripheral DXA (pDXA) densitometers have been introduced. In this study we evaluated the agreement between two heel DXA devices on BMD and T-scores. A total of 99 females aged 21-78 years (ca. 16 per decade) had their non-dominant heel BMD measured using the PIXI (Lunar Inc.) and the Apollo (Norland Medical) pDXA scanners. The mean BMD values were 0.492 and 0.607 g/cm(2) and the mean T-scores using manufacturers' specified reference data were -0.07 and -0.25 for the PIXI and Apollo, respectively. Both the BMD and T-score intermachine relationships were highly correlated but showed significant nonidentity slopes and non-zero offsets. The diagnostic comparison on T-scores resulted in 86% agreement between the instruments (weighted kappa score of 0.550). Normalizing the reference peaks and SDs using this study's young adult population BMD results removed the systematic T-score disagreement. We found that PIXI and Apollo are highly correlated. Differences in BMD values are mainly due to different region of interest (ROI) definitions and additional T-score disagreement reflects the difference in normative databases.

MeSH terms

  • Absorptiometry, Photon / instrumentation
  • Absorptiometry, Photon / methods
  • Absorptiometry, Photon / standards*
  • Adult
  • Aged
  • Bone Density / physiology
  • Calcaneus
  • Female
  • Humans
  • Linear Models
  • Middle Aged
  • Osteoporosis / classification
  • Osteoporosis / diagnostic imaging*
  • Phantoms, Imaging / statistics & numerical data
  • Reference Standards