Assessment of changes in dual-energy X-ray absorptiometry performance following a system upgrade

System upgrades are a constant feature of many highly technical specialities and are normally assumed to have a negligible effect on patient data. We report on the assessment of an upgrade made to a Norland XR-26 dual-energy X-ray absorptiometry (DXA) system. The upgraded version (2.3) consisted of new hip positioning aids, hip image analysis software and updates to the normal ranges. Measurements of femoral neck and trochanteric bone mineral density correlated well between versions (r = 0.969 and r = 0.971), those for femoral neck and trochanteric bone mineral content less well (r = 0.956 and r = 0.911), and Ward's triangle values and femoral neck area measurements were poorly correlated (r = 0.719-0.881). The upgrade was claimed to improve precision while reducing analysis and imaging time. Improved precision was not demonstrated, with good precision found both before and after the upgrade. Correlation between femoral neck Z-scores were good for the European normal range (r = 0.970), but unacceptable for the UK range (r = 0.769). These correlations were not close enough to allow patients scanned with the old version to be followed up with scans using the upgrade. The upgrade was, however, preferred for new attenders because of the improved speed and automation of hip analysis, without loss of precision. The most important implication in terms of patient management was the change made to the UK normal range, which had a very large effect on the number of patients deemed to be at risk from osteoporosis. It is essential that any upgrade introduced to an existing DXA system is carefully evaluated, since it may affect system performance and patient results.