This study assessed agreement between the GE Lunar iDXA and Prodigy densitometers for bone measurements and used 3 statistical methods to derive cross-calibration equations: linear regression, the Deming method, and multivariate analysis. Compatibility of machines for the measurements of bone mineral density, bone mineral content, and bone area also was explored. Eighty-five adults, age: 45.5 (SD 12.8) years; body mass index: 25.6 (SD 3.7) kg.m(-2) were measured once at the lumbar spine: L1-L4 and total hip on each densitometer, within 24 hours. Both linear regression and Deming analysis indicated that cross-calibration equations were required at the lumbar spine and total hip but not at the femoral neck. Multivariate analysis identified femur thickness and femur percent fat as predictive variables at the femoral neck and total hip. Bland Altman analysis (Prodigy-iDXA) indicated significant positive bone mineral density bias at the lumbar spine and femoral neck. Significant bone mineral content biases were observed at all 3 sites and bone area biases at both hip sites. These initial results suggest there are small significant differences in the bone parameters and that all 3 bone parameters should be evaluated when comparing densitometers, especially when there are differences in pixel size between the densitometers.
Keywords: Agreement; DXA; analysis; bone; cross calibration.
Copyright © 2015 The International Society for Clinical Densitometry. Published by Elsevier Inc. All rights reserved.