Long-Term Versus Short-Term Precision of Dual-Energy X-Ray Absorptiometry Scans and the Impact on Interpreting Change in Bone Mineral Density at Follow-up

Abstract

We hypothesized that increased time between repeat scans would be a critical factor resulting in long-term precision (LTP) being greater than short-term precision (STP) because time erases the short-term mind’s eye memory of positioning that makes STP easy to perform. In 2 prospective studies, STP, in which repeat scans were performed within minutes, was compared with LTP in the same patients in whom repeat scans were separated by 3–10 d (n = 35) or 2–12 mo (n = 30). The first baseline scan and the follow-up scan were used to calculate LTP. Rigorous exclusion criteria were applied to patients returning 2–12 mo after baseline to avoid the effects of biological change that could affect bone mineral density (e.g., bisphosphonate treatment). Results showed that LTP was significantly greater than STP irrespective of whether the interval between baseline was short (3–10 d) or averaged 188 d. For example, LTP for left femoral neck (LFN) was 0.017 g/cm 2 with 3–10 d separating follow-up from baseline, with STP being 0.007 g/cm 2 (duplicate scans at baseline, Day 1). For the second prospective cohort, LTP for LFN when the mean follow-up interval was 188 d was 0.024 g/cm 2; STP in the same patients was 0.009 g/cm 2 ( p < 0.05 by F test). A sampling of patients returning for follow-up to an osteoporosis clinic was used to evaluate the impact of using LTP vs STP to interpret change. The use of a least significant change value based on LTP (rather than STP) resulted in significantly fewer patients having “significant” change at follow-up and eliminated discrepancies in which BMD apparently increased at 1 skeletal site but decreased at another. We conclude that (1) time is the primary variable accounting for the difference between STP and LTP, and (2) STP underestimates true dual-energy X-ray absorptiometry variability, resulting in overdiagnosis of change at follow-up.