Standardization of spine and hip BMD measurements in different DXA devices

AimTo compare BMD values of lumbar and hip regions measured in two different DXA scanners in one laboratory, and to investigate the efficiencies of implemented and specifically derived standardization formulas.Materials and methodsPA lumbar (L2–L4) and right femoral neck BMD values were obtained in 100 women (aged 26–75), consecutively in GE-Lunar DPX-NT and Hologic QDR 4500 C DXA scanners. Standardization of BMD values obtained in two different DXA devices was done according to the method developed by International DXA Standardization Committee (IDSC), using the European Spine Phantom (ESP) to obtain the specific constant value. Mean corrected standardized BMD (sBMD) values in two scanners have been compared with each other and with the mean reported sBMD values, respectively.ResultsThe mean lumbar BMD values were 0.950 ± 0.117 g/cm2 for Hologic and 1.068 ± 0.135 g/cm2 for GE-Lunar (p < 0.05); mean corrected sBMD values were 1.035 ± 0.128 g/cm2 for Hologic and 1.035 ± 0.131 g/cm2 for GE-Lunar (p > 0.05). The mean femoral neck BMD values were 0.798 ± 0.114 g/cm2 for Hologic and 0.895 ± 0.111 g/cm2 for GE-Lunar (p < 0.05); mean corrected sBMD values were 0.869 ± 0.124 g/cm2 for Hologic and 0.867 ± 0.108 g/cm2 for GE-Lunar (p > 0.05). The difference between the mean values of BMD and sBMD, both corrected and reported, were statistically important in each scanner (p < 0.05). The mean values of corrected and reported sBMD were also statistically different in each scanner (p < 0.05; mean standard error in the spine was 1.3 for GE-Lunar and 1.8 for the Hologic device).ConclusionThe originally proposed standardization formulae may not optimally correct for manufacturer, model and device-specific differences. Therefore, use of sBMD is not recommended to compare results of individual patients obtained on scanners of different type and brand. The residual error of reported sBMD, however, is substantially smaller than for manufacturer-specific results, and therefore, reporting standardized results is useful for population studies.