As of 2023, Hispanic Americans comprise approximately 19.5 % of the U.S. population, larger than any other minority ethnic group.1 Recent data from the All of Us Program, highlighted that the prevalence of obesity (body mass index, BMI > 30.0 kg/m2) among Hispanic women is approximately 49 % compared to 36 % in non-Hispanic white women.2 While these estimates are based on BMI and not measured percent body fat (%BF), it is clear that obesity is prevalent in the largest minority population in the United States. Additionally, there is a paucity of literature evaluating methods of body composition estimates for validity and accuracy for this population.

Densitometric-based methods of body composition determination including under-water weighing (hydrostatic weighing, [HW]) and air displacement plethysmography [ADP], have long been utilized as gold standard methods for fat mass determination in human research and clinical applications. These methods are based on fundamental, empirical relationships between body components (e.g. fat and fat-free mass) and the physical properties of tissue densities (i.e. grams/volume). Two commonly used approaches for deriving body composition from body density measurements are the Siri3 and Brozek et al.4 two-component (2-C) model equations. However, the validity of these equations to accurately estimate %BF is limited to 1) the population from which the equations were derived, and 2) how well the “tested” population compares to a reference population. A 2-C model divides the body into fat mass (FM) and fat-free mass (FFM) and therefore assumes that the properties of the individual components of the FFM compartment (e.g., total body water, mineral, protein) are the same among individuals. However, these assumptions do not hold true for different ethnic groups or populations with differing physical fitness, such as athletes. As an example, Schutte et al. determined that the density of lean body mass was greater in black than white men (1.113 g/cc vs 1.100 g/cc, respectively), and this difference was influenced by greater bone mineral content and bone mineral density for black men.5 Use of the Brozek equation in a group of black female athletes found %BF was underestimated by approximately 3 % when compared to a 4-C model, and the authors attributed this difference to increased density of the FFM of these women, calculated as ∼1.109 g/cc, significantly greater than the 1.100 g/cc standard used by the Brozek equation6. Furthermore, in a sample of Hispanic women from New Mexico, %BF was significantly underestimated compared to a 4-C reference model7 by the Siri 2-C and Lohman 3-C models8,9. From this data, Heyward et al. (1995) estimated the density of the FFM component in Hispanic women to be 1.105 g/cc.10 Recent data comparing cadaver reference values to measured FFM values in Hispanic adults further highlighted the importance for the use of ethnic/racial-specific equations for body composition estimates, as all FFM characteristics (total body water, bone mineral content, body volume) were different from reference values.11

Dual-energy x-ray absorptiometry (DXA) is now a commonly used method for estimation of body fat over HW for several reasons: the measurement is relatively quick, technical error is minimized, and it requires little effort from the participant. Additionally, DXA machines are prevalent in laboratories with research or clinical facilities, oftentimes being more frequently utilized and/or replacing HW equipment. However, similar to the aforementioned issues for HW, the accuracy of DXA for estimated %BF is based on several assumptions. The principle of DXA analysis is based on the attenuation of two different energy levels of x-rays which is influenced by the density and composition of a tissue.12 These attenuation energies are assumed to be constant for all individuals; however, as stated above, the density of tissues is not constant across race or ethnic groups, athletes, or age. The assumed coefficients of FM and FFM by DXA software are also influenced by the hydration level of the tissue, and the assumption that the amount of fat in a pixel that includes bone is the same as the amount of fat in a pixel that is calculated from bone-free tissue.13 In other words, a pixel containing bone is seen as “bone + not bone” while a bone-free pixel is calculated as “fat mass + lean mass.” Given that approximately 40–45 % of the pixels from a DXA scan contain bone, the total estimated fat and lean mass is calculated from only ∼60 % of the body.12,13

Laboratory reference methods such as HW and DXA are convenient, cost-efficient options utilized for an accurate estimation of %BF. However, in minority populations, these estimations of %BF may induce additional error due to the aforementioned assumptions made that are based on ethnic-specific and demographic-specific information from limited sample size cadaver analyses. The purpose of this study to compare body composition components (%BF, FM, and FFM) estimated by HW and DXA for a sample of pre-menopausal Hispanic women using data collected from a previous study.14