Objective: Body composition measurements made using Hologic QDR-1000/W pencil-beam and QDR-4500W fan-beam scanners (Bedford, MA) were compared in a sample of 13- to 18-year-old white and black youth (n = 219).
Research methods and procedures: Total fat (FAT), fat-free soft tissue (FFST), bone mineral content (BMC), bone mineral density (BMD), and percent body fat (%BF) were compared between repeated measurements using the QDR-4500 and between the two scanners using mixed model ANOVA. Intraclass correlation coefficients and Bland-Altman limits of agreement were used to evaluate inter- and intrascanner reliability.
Results: Intraclass correlation coefficients for repeated measurements using the QDR-4500 ranged from 0.997 to 0.999 for FAT, %BF, FFST, and BMC and 0.987 for BMD. Mean measurements made using the two scanners differed significantly for FAT, %BF, BMC, and BMD (p < 0.0001), and scan by sex interactions were significant (all p < 0.0005). There were no significant differences in mean measurements between repeat scans using the QDR-4500 (all p > 0.19). Limits of agreement for measurements of FAT, FFST, and %BF made using the two scanners were approximately three times as wide as those for two measurements using the QDR-4500. For lower values of FAT and %BF, the QDR-4500 gave higher measurements than the QDR-1000, whereas at higher values, this relationship was reversed. The QDR-1000 tended to give higher BMC measurements, with larger differences for higher values.
Discussion: Using different models of DXA scanners within a study may reduce precision of body composition measurement. This issue needs to be considered in the design of longitudinal studies.