Evaluation of Methods for Determining Various Components of Body Composition

Thumbnail Image
Journal Title
Journal ISSN
Volume Title
University of Alabama Libraries

Doubly indirect methods of assessing body composition are commonly used in laboratory and practical settings. The purpose of this dissertation was to expand upon the methodological discrepancies associated with various techniques, and to provide improved equations to overcome these limitations. A series of three studies was conducted to 1) improve the estimation of underwater residual lung volume (RLV), 2) systematically review and quantify the error associated with single-frequency bioimpedance analysis (SFBIA) for the determination of total body water (TBW), and 3) develop a novel equation for predicting percent body fat (%BF) from skinfolds using a criterion multi-compartment model. The first study developed an equation for the prediction of underwater RLV in healthy adults using age and height as predictor variables. The new equation produced superior validity statistics upon cross-validation compared to four existing equations, indicating that it may be used by practitioners to accurately estimate underwater RLV during hydrostatic weighing. The second study systematically reviewed and meta-analyzed 264 effects from 51 original studies designed to compare SFBIA to criterion dilution methods for TBW estimation. Although a non-significant overall effect was identified, there was significant variability associated with SFBIA methodology (i.e., frequency and resistivity index) and sample sex (% female). Moderator analyses indicated that SFBIA procedures utilizing Ht2/R at 100 kHz produced the most accurate estimate of TBW when compared to isotope dilution techniques. The third study developed a skinfold-based equation for the prediction of five-compartment model %BF in a sample of healthy adults. The new equation outperformed selected existing equations when cross-validated, indicating its potential utility for practitioners concerned with obtaining accurate estimates of %BF in the general population.

Electronic Thesis or Dissertation