@ NUTRITION/METABOLISM CLASSIC —-HHE-MAT BY COPYRIGHT LAW (IITLEPRIC 17, US. rane Onlythe man whois familiar with the art andscience ofthe past is competent to aid in its progress in the future. - Theodore Bulroth gege4 Chassic Editor: Harold Sandstead, MD Professor, Department of Preventive Medicine and Community Health, University of Texas Medical Branch, Galveston, Texas, USA VL SUTCTBT Body Composition from Fluid Spaces and Density: Analysis of Methods William E. Siri Reprinted with permission from Techniques of Measuring Body Composition. J. Brozek. ed. National Academy of Sciences National Research Council, Washington, DC, 1961:223-244 ... Prospective Overview Henry C. Lukaski, PhD From the U.S. Department of Agriculture, Agricultural Research Service, Grand Forks Human Nutrition Research Center. Grand Forks, North Dakota. USA Because methods for the assessment of human body composivon are indirect and based on assumptions regarding the chemical or physical characteristics of various components of the body, an important issue ts the variability to expect for estimates derived with anyof various methods available. Siri convincingly addressed this fundamental issue more than 37 vears ago when he evaluated the densitometric and isotope dilution methods, initially in a laboratory report! and later in a contribution in a conference proceedings.” The contributon serves as one of the landmarks in the field of bodycomposition methodology. In the early 1950s, two techniques, hydrodensitometry or underwater weighing and hydrogen isotope diluuon. were used routinely to assess body fatness in adults.* Each of these methods relies on assumptions regarding en aee BEST COPY AVAILABLE a unique chemical or physical property of the fat-free body ¢e.g.. constant hydration and protein-to-mineral raticy) and fat (e.g., densityoffat is less than that of bone. muscle. and protein) determined from chemical analy- ses. Although other investigators, e.g.. Behnke et al.tand Keys and Brozek. acknowledged the limitations of these assumptions, Siri? Challengedthe validity of these basic premises and formulated estimates of error in the prediction of body fatness based on the variability of the chemical composition of the fat-free body and adipose tissue. Siri was troubled bythe reliance on a “reference bady” for which the relative chemical composition ofthe fat- free body was assumed to be constantor at least constant within narrowlimits. Furthermore, it was assumed that. 480) NUTRITION VOL. 9 NO. 5, SEPTEMBER-OCTOBER 1993 when body weight changed, fat was ether added to or removed from the reference body without disruption of the basic assumption of constant composition of the fat- free body. Direct chemical analyses of animals indicated that these assumptions were valid.’ However, studies of the body composition of humans during weight change associated with dietary changes suggested that adipose tissue, or the Ussue that is gained or lost. is not onlyfat per se but also consists of water and cellular materials.*” These obser ations led Sir to assess the components of efror associited with the densitometric method. The error associated with any indirect method of hodyCOMposition assessment has two components. Measurement error and biological uncertainty Sint concluded that measurement error is minor relative to the uncertainty associated with the interindividual sariabilicy in the assurmptions of the chemical constancy of the fat-free body. He used a propagation-of-error model and determined that the error in estimating percentuge bodyfat determined with densitometry was ~4°o because of contributions from variability (standard des tation: in the water content (2.7) and protein-to-mineral ratio (2.190) of the fat-tree body and adipose ussue composition 1.) inthe general population, Siri indicated that this degree of uncertainty could be decreased if total-body water was used along with densitometry lo esumate body fatness. Siri emphasized the importance of treating measurements of componentsof the fat-free bodyas independent variables in the assessment of body fatness. This (Continued on Page 492)