Differences in precision in bone mineral density measured by SXA and DXA: the NOREPOS study

Eur J Epidemiol. 2008;23(9):615-24. doi: 10.1007/s10654-008-9271-1. Epub 2008 Jul 2.

Abstract

The aims were to compare the precision (reliability) in single X-ray (SXA) and dual X-ray (DXA) absorptiometry, and to compare smallest detectable difference (SDD). An additional aim was to examine determinants for precision in bone mineral density (BMD). BMD was measured by SXA (DTX-100, Osteometer) in the forearm and by DXA (Lunar Expert) in the forearm and in the hip. Two measurements were performed at each site/method, and 195 of 207 participants had complete datasets. Participants were aged 47-49 and 71-74 years. The precision was estimated by Root Mean Square Standard Deviation (RMS SD) with 95% Confidence Interval (95%CI) and the corresponding coefficients of variation (CV%). Determinants (age, gender, BMD) were analysed by multiple linear regression with log (SD) and log (CV) as dependent variables. RMS SD tended to be largest in older women and in those with low BMD. RMS SD for SXA and DXA forearm was 4.6 (4.2-5.1) and 6.8 (6.1-7.4) and the corresponding CVs 1.0% and 1.4%. RMS SD for DXA hip was 11.0 (9.9-12.0) with CV 1.2%. To detect a 3% change in BMD one would need two repeated measurements by DXA in the distal forearm at each of two consultations, but only one measurement by SXA in the distal forearm and also only one measurement by DXA in the hip. Precision differed by type of densitometer affecting the number of repeated measurements needed to detect a given BMD difference.

Publication types

  • Comparative Study
  • Research Support, Non-U.S. Gov't

MeSH terms

  • Absorptiometry, Photon / methods*
  • Aged
  • Bone Density*
  • Female
  • Forearm / physiology
  • Hip / physiology
  • Humans
  • Linear Models
  • Middle Aged
  • Norway
  • Reproducibility of Results