Anomalies in the measurement of changes in bone mineral density of the spine by dual-energy X-ray absorptiometry.

Dual-energy X-ray absorptiometry (DXA) is frequently used for longitudinal studies of bone mineral status because of the high precision obtained, but evidence is emerging that the accuracy of measurements of changes may be a limitation because of artefacts of the analysis procedure, in particular, a dependence of the measured bone area (BA) on the bone mineral content (BMC). Results of spine bone mineral measurements taken at intervals with two DXA scanners, a Hologic QDR 1000 W, and a Norland XR 26 HS, were examined. There was a consistent correlation between changes in BA and in BMC, with a slope of approximately 0.25 when expressed as percentages. A real change of BA of the magnitude observed is not feasible. There were no differences among the correlations for different instruments, genders, ages, or weight changes. There would appear to be an underestimation of changes in bone mineral density (BMD), but there is a possibility that some of the anomaly is manifested as an overestimation of a change in BMC. Phantom measurements were undertaken with the DXA scanners mentioned above and with a Lunar DPX. The phantoms consisted of simulations of the spine cut from aluminium sheet, so that the effective BMD could be varied. The dependence of the measured BA on BMC varied with the phantom outline, particularly the thickness of the transverse processes. Evidence was obtained of both an underestimate of BMD changes and an overestimate of BMC changes. There are errors in measuring spine changes, but these do not seem to be as serious as a previous report suggests for the Hologic scanner and are not likely to lead to misinterpretation of results.