Precision of Dual-Energy X-Ray Absorptiometry Reflection Scans in Muscular Athletes

Background: Offset scanning procedures, such as reflection scanning, allow for dual-energy x-ray absorptiometry (DXA) body composition assessment of individuals who are too broad for standard scanning dimensions. However, limited information is available concerning the precision of this procedure, particularly in athletes. Methodology: Twenty-seven muscular athletes (n = 17 males, BMI: 28.8 +/- 2.0 kg/m(2), DXA body fat: 12.5 +/- 2.7%; n = 10 females, BMI: 22.8 +/- 1.6 kg/m(2), DXA body fat: 19.2 +/- 3.4%) underwent consecutive DXA scans on a GE Lunar Prodigy scanner using the reflection scanning technique. The fully automated output was obtained for each scan, and an additional version of each scan's output was saved after manual adjustment of regions of interest (ROI). Metrics of reliability and precision were calculated for total and regional body mass (BM), lean mass (LM), fat mass (FM) and bone mineral content (BMC). These metrics included the precision error (PE), least significant change, Delta Mean, technical error of measurement, intraclass correlation coefficient, smallest worthwhile effect and minimum difference considered real. Results: Reflection scanning produced small errors for BM (PE: similar to 0.5%), LM (PE < 1%) and BMC (PE: similar to 1.2%), with larger errors observed for total FM (PE: similar to 3%). Manual ROI adjustment produced lower errors for total BM, LM, FM, and BMC, as well as lower errors for most regional estimates. The utilization of automated ROI revealed concerns unique to reflection scanning, including unnecessary estimation of trunk composition, which appreciably increased error in this region. Regional estimates produced higher errors for all variables as compared to whole-body estimates, although which regions produced the highest errors differed between BM, LM, FM, and BMC. Conclusions: Reflection scanning allows DXA body composition assessment in individuals exceeding traditional scanning dimensions, including broad athletes. Although this procedure introduces error, it may be minimized through manual adjustment of ROIs and consistency of analysis methods.