The Accuracy Of Four Component Models Using Body Volume Equations And Field-Based Water Measurements

Several recent investigations have been altering the traditional 4 component (4C) model for determining percent fat (%Fat) by using body volume (BV) from dual-energy x-ray absorptiometry (DXA) and total body water (TBW) from bioelectrical impedance analysis (BIA) to avoid using underwater weighing (UWW) and TBW from deuterium oxide dilution. PURPOSE: To determine the accuracy of using either Db from DXA, TBW from BIA or both in a 4C model to estimate %Fat. METHODS: A 4C model to determine %Fat was measured using body density (Db) from UWW, bone mineral from DXA and TBW from deuterium oxide dilution in 45 males and females. The Lohman (1986) 4C model equation was used to determine %Fat. BV was calculated from DXA scans using an equation from Wilson (2013) which was used to determine Db. TBW was estimated from BIA using an equation from Evans (2001). An alternate 4C model was created using Db from DXA (4C-DXA Db) and TBW from BIA (4C-BIA TBW) to determine the accuracy of either or both (4C-DXA/BIA) of these compared to the 4C model (4C-Criterion). UWW with residual volume measured simultaneously (by nitrogen dilution) was performed and converted to %Fat (Siri, 1961). A Hologic QDR-1000 W was used for DXA bone measurements and BIA (RJL Systems) was performed following standard procedures. Each person was measured in a fasted, rested condition during a single 3-4 hour session. RESULTS: The sample was 21.0 ± 2.9 years old and weighed 61.3 ± 9.4 kg. %Fat from 4C-Criterion was 18.0 ± 7.2%. Strong (p < 0.05) associations existed in %Fat between 4C-Criterion versus from UWW and all alternate 4C models (Adj R2s = 0.829 to 0.919). SEEs from these analyses were 2.05% (UWW) to 2.98% (4C-DXA/BIA). No significant mean differences in %Fat existed (range: -0.033% (UWW) to -0.687% (4C-BIA TBW). Standard deviations of the differences were 2.04% (UWW), 2.39% (4C-DXA Db), 2.46% (4C-BIA TBW), and 3.10% (4C-DXA/BIA). Bland Altman analyses (differences vs 4C-Criterion %Fat) showed nonsignificant slopes except for UWW (r = 0.395) and 4C-BIA TBW (r = 0.299). CONCLUSIONS: %Fat from 4C-DXA Db did not provide a substantial improvement over using UWW. The highest errors and large individual differences resulted from 4C-DXA/BIA, therefore this practice of substituting BIA TBW and DXA Db should be discouraged.