ESPE Abstracts (2015) 84 P-2-236

Body Composition Measures on Different DEXA Scanners are not the Same

Udo Meinhardt, Fabienne Witassek, Christiane Fritz & Urs Eiholzer


PEZZ Center for Pediatric Endocrinology Zurich, Zuerich, Switzerland


Background: Body composition measures differ between DEXA scanners. If an old DEXA is replaced a transition period for double measurements on the old and the new scanner is needed.

Objective and hypotheses: To evaluate differences between the old (Hologic QDR 2000) and new (Hologic Discovery Wi) scanner and to calculate formula transforming measurements.

Method: 51 double measurements were performed on a group of 41 children and adults (mean (SD) age 18.57 (10.06) y, BMI x (y) kg/m2). Results for fat, lean, BMD and BMC were compared using Bland-Altmann plots. Linear regression analysis was used for transformation formula. Least significant change (LSC) was calculated using triple measurements of a separate group of 15 healthy adults.

Results: LSC on the Discovery Wi for fat, lean, BMD and BMC was 0.76 kg, 0.83 kg, 0.02 g/cm2 and 0.065 kg. The individual’s mean of QDR 2000 and Discovery Wi measurements for fat mass was 4.6–34.9 kg; the mean (±2 SD) difference being 1.9 (4.3) kg, this difference was increasing as fat mass increased. Mean lean mass was 19.2–72.7 kg; independent of lean mass, the mean (SD) difference between QDR2000 and Wi was −1.5 (4.2) kg. Mean BMD was 0.7–1.5 g/cm2; the mean (SD) difference between QDR2000 and Wi was −0.02 (0.06) g/cm2; this difference was increasing as BMD increased. Mean BMC was 0.78–3.8 kg; independent of BMC the mean (SD) difference between QDR2000 and Wi was 0.04 (0.14) kg. Using transformation formula, maximum differences between true and calculated measurements for fat, lean, BMC and BMD are 11.8, 12.8, 11.9 and 22.5% respectively.

Conclusion: Except for BMC and BMD, mean differences between the DEXA scanners exceed LSC two-fold. Exchanging DEXA scanners, an overlapping period with double measurements is mandatory. Transforming formulas induce significant variability.

Article tools

My recent searches

No recent searches.