Agreement among Results of Bone Mineral Densitometry Using Dual-energy X-ray Absorptiometry Based on Race-Specific Normal Databases

To investigate the effect of the selection of different race-specific normal databases on the densitometric results and diagnosis. Eighty-two participants included in the study were selected consecutively from the archived scans. Analysis was conducted three times for each participant using different ethnicities (White, Asian and Hispanic) as the race-specific normal database for comparison with the patient’s data. The densitometric data, including bone mineral density (BMD), as well as T- and Z-scores, were extracted for three standard regions, including the spine, femoral neck, and total femur. The agreement among the diagnoses based on three different ethnicities (White Vs. Asian, White Vs. Hispanic, and Asian Vs. Hispanic) was tested by cross-tabulation and calculation of the coefficient of Kappa. The mean and standard deviation of the T-score of participants with the selected reference types were exactly the same in all three regions. The mean and standard deviation of the Z-score of all participants, in comparison with "Asian" and "white" references, was exactly the same, but with the cases where the "Latin" race was selected as the reference, in the region of the neck, thigh (p<0.001) and the whole thigh (p<0.001) was different. Z-score values of the spine were the same in all three racial groups. The Z-scores of the spine were the same in all three types of races. Bland-Altman analysis demonstrated a significant difference in the femoral neck but not in the spine and total femur. All the diagnoses based on WHO and ISCD criteria were the same (Agreement test: Kappa coefficient=1.00 and P-value<0.001). The study shows that the effect of the selection of different race-specific normal databases is insignificant on the densitometric results and diagnosis.