Body composition by three-compartment model and relative validity of some methods to assess percentage body fat in Mexican healthy elderly subjects

Background: In Mexico, there is scarce information about the body composition in the elderly. Some researchers have measured body fat by anthropometry and bioelectrical impedance. These techniques are practical and can be used in the field; however, proper validation is required. This implies the use of accurate and precise methods. In the elderly, Siri's three-compartment (3C) model has been proposed as feasible alternative to the gold standard four-compartment model. Objective: To evaluate the body composition by the 3C model and the relative validity of densitometry and hydrometry methods to assess the percentage body fat (%BF) in 37 healthy Mexican elderly subjects. Methods: The body density was measured by air displacement plethysmography, total body water was evaluated by means of an isotopic dilution technique, and two different hydration factors were used to determine the fat-free mass. These measurements were used to calculate %BF by the 3C model. Accuracy and precision of the methods were tested by a two-way analysis of variance and regression procedures and bias by Bland and Altman analysis. Results: In men and women as a whole group, the %BF by the 3C model was 34.4 ± 8.01. Women had higher values of %BF as compared with men (41.7 ± 3.7 vs. 29.4 ± 6.4; p < 0.0001). The regression procedures showed that air displacement plethysmography was accurate and precise only in women and in the whole group. Regression analyses of %BF by total body water (using both hydration factors) and 3C model showed that the intercepts were not different from zero and that the slopes were not different from 1.0 in men and women separately and as group. Conclusion: In this study, the total-body water method revealed a good relative validity (accuracy, precision, and freedom from bias) as compared with the 3C model in men and women both separately and as a group and may be acceptable for the estimation of %BF in individuals or groups of healthy elderly subjects.