Comparisons between three dual-energy X-ray absorptiometers used for measuring spine and femur

Lunar, Hologic and Norland dual-energy X-ray absorptiometry scanners have been compared for measurements of spine and femur bone density. Precision was not greatly different in realistic phantoms and volunteer subjects. Most clinical therapeutic trials are concerned with measuring changes of bone mineral, and interchangeability in this context was examined using phantoms of spine and femur in which changes of bone mineral density (BMD) were simulated. With each instrument the measured changes were closely linear. For the spine the biggest difference of slope between instruments was 15%. For the femur, in all areas of interest, the differences of slope were less than 10%. It is concluded that the three instruments can be satisfactorily used in multicentre clinical trials to investigate changes in bone mineral. 12 volunteers were measured with each scanner. There were significant mean differences between each pair of instruments, suggesting different calibration criteria. More importantly, those mean differences had appreciable standard deviations (SDs), in proportional terms from 3% to 10%. When the measurements were related to reference ranges and expressed in terms of age-matched normal values the mean biases disappeared, but the SDs did not improve. Results from different manufacturers' apparatus are not interchangeable for studying individual patients. Measurements from the phantoms were used to cross-calibrate the scanners. Those from the variable spine phantom predicted the in vivo ratio within 4%, but this was no better than measurements of the unmodified phantom alone. Results using the European Spine Phantom were less satisfactory. No phantom provided an adequate cross-calibration for femur measurements.