Background: Dual-energy X-ray absorptiometry (DXA) requires phantoms for quality control and cross-calibration. No commercially available phantoms are designed specifically for whole-body scanning of infants.
Methods: We fabricated a phantom closely matching a 7-kg human infant in body habitus using polyvinyl chloride (PVC), nylon mix, and polyethylene for bone, lean tissue, and fat, respectively, for evaluating the comparability of instruments used in studies on infant body composition. We scanned the phantom multiple times for short- and long-term repeatability and then shipped it to six other sites for comparison scans. All instruments were Hologic Delphi or Discovery models. Scan analyses were in-house procedures (Hologic V12.1).
Results: Short- and long-term results were not significantly different. Nylon mix underrepresented expected lean mass values by 5%, PVC underrepresented bone by 12%, and polyethylene overrepresented fat by 30%. Precision values were as follows: lean mass ≈ 3%; bone ≈ 3.5%; and fat = 5.5-7.5%. Instruments differed significantly for bone mineral content and density results in most instances. Three instruments differed in fat and lean mass. The two Hologic models differed significantly in all compartments except bone density.
Conclusion: The phantom design came close to emulating bone, lean tissue, and fat and showed good reproducibility. Significant differences among various DXA instruments highlight the necessity of cross-calibration for any multicenter studies.