Concerns And Limitations Of Dual-Energy X-Ray Absorptiometry (Dxa) For The Evaluation Of Fat And Fat-Free Mass In Older Men And Women.

Abstract

The purpose of the present study was to evaluate Dual-energy X-ray absorptiometry (DXA) for estimating fat mass (FM), percent fat (%FAT), and fat-free mass (FFM) in older men and women compared to a four-compartment model (4C). Forty men and women (65-84yr, 20 men and 20 women) participated in the study. Body fat calculations included a criterion 4C model and DXA-derived FFM, %FAT, and FFM. The criterion 4C model included body volume from air-displacement plethysmography (ADP), total body water from deuterium dilution, and bone mineral content from DXA. Constant error (CE), standard error of estimate (SEE), total error (TE), and the 95% limits of agreement were used to compare DXA values to the 4C model. All DXA mean values (CE) were significantly different than the 4C model. DXA %fat values were the least accurate with the largest TE and a y-intercept significantly different than zero (p = 0.035). Bland and Altman analysis indicated that no DXA-derived value produced a significant trend. Compared to DXA, the Siri two-compartment (2C) model using ADP produced non-significant (p > 0.106) CE values, lower TE values ( < 1.66), lower SEE values ( < 1.65), and smaller individual errors (95% limits of agreement < 3.17). DXA appears to be biased in older men and women when comparing mean values of FM, FFM, and %FAT compared to a 4C model. %FAT produced the largest individual errors with limits of agreement ± 5.23%FAT. However, DXA produced acceptable TE values and can be considered a valid method for predicting body composition. Nonetheless, regardless of the significant CE values for DXA, using ADP alone in a 2C model produced more accurate, non-biased results compared to the 4C model for all values, as indicated by a reduced SEE and smaller individual errors. FM and FFM DXA values were more accurate than %FAT values compared to a 4C model. However, the Siri 2C model was more accurate than all DXA derived values. Additionally, DXA produced significantly different FM, FFM, and %FAT values compared to a 4C model. Therefore, DXA is not suggested for use in older men and women when attempting to predict FM, FFM, or %fat. Furthermore, choosing ADP for use in the Siri 2C model to predict FM, FFM, and %FAT appears to be more accurate than DXA in this population.