Toria McMullin and Dr. Robert T Davidson, Nutrition, Dietetics and Food Science
Due to the current trend obesity rates and the increased public awareness of the connection between body composition and health, individuals are becoming more and more interested in assessing and changing their own body composition. The current methods used to assess body composition include the following: dual-energy x-ray absorptiometry (DEXA), computerized tomography (CT) scans, magnetic resonance imaging (MRI), bioelectric impedance analysis (BIA), Near-Infrared Interactance (NIR), BodPod measurements, skinfold and circumference anthropometric measurements, and hydrodensitometry. Multicomponent models most accurately predict body composition. However, such models—including hydrodensitometry and DEXA—are either too difficult or too costly for practical, widespread consideration. Other methods are also impractical due to their cost, large measurement error, or reliance upon skilled technicians. Consequently, more research needs to be done in order to accurately and inexpensively help individuals predict body composition and body composition changes associated with changes in weight.
Much of the research that attempts to predict body composition in an inexpensive and practical manner focuses on the formation of mathematical equations in which individuals can input easily obtained measurements (such as height and weight). Most of the equations developed use the data collected from expensive and accurate body composition measurements to predict total body fat percentages. While some studies have produced body fat percentage equations that are specific to various ethnicities – such as the Slaughter equations – other predictive equations are based on skin fold measurements or circumference ratios – such as those found by Garcia et. al. Although these studies incorporate circumferences when calculating total body fat, no reported research has used body composition data to predict regional body circumferences. Because researchers have focused only on unidirectional equations that move from anthropometric measurements to body composition, there are no reported equations that result in predicted circumferences. Consequently, there are no reported methods that predict the effects of body weight changes on regional body composition and circumferences.
Developing equations that predict regional circumferences will facilitate further research on the relationship between changes in weight and changes in regional circumferences and therefore provide information about body appearance and composition. The purpose of this research is to provide equations that will be useful in future research on body composition and weight changes. These equations use body composition data to predict regional circumferences.
For this study, 67 female and 32 male Caucasian college aged volunteers were recruited. They were informed that they would be participating in an energy balance study, where they would be compensated with a gift card and two DEXA (dual energy x-ray absorptiometry) scans. The duration of the study was eight months. Of these 99 volunteers, the data from 63 females and 30 males were used.
At the beginning of this observational study, the subjects came into the lab where they were scanned by a GE Medical Systems LUNAR DEXA machine, weighed, and had their anthropometric data measured. The DEXA machine divided the body into several different regions and calculated the Lean g, fat g, tissue g, % fat and % lean values for each region. The anthropometric measurements included body height along with the circumference measurements for the following 6 regions (which corresponded nicely with the regions designated by the DEXA machine): chest (across bust), waist (at navel), hips (at fullest part including buttocks), upper legs (upper-thigh), lower legs (at calf), and upper arms (at bicep). Each circumference measurement was taken at the fullest part of the region. An electric scale was used to measure weight and a standiometer was used to measure height. The same tension set tape measurer was used for each circumference measurement.
Using GraphPad InStat 3, multiple regression analysis was used to formulate mathematical equations that predict regional body circumferences based on the regional body composition data from the DEXA scans. Those equations that produced the highest R2 values were selected. Where possible, equations were also selected based on the variables involved and if those variables were common to other regional circumference equations for the population.
Six multiple regression equations, one for each of the anthropometric regions measure, that predict regional circumferences for males and females were developed. The significant variables used in the equation include the following: height, mass, regional lean mass, regional fat mass, body mass index (BMI), and lean body mass index (LBMI). All multiple regression p-values were <0.0001. Efforts were made to maintain consistency. The same tape measure, same DEXA machine, scale and standiometer were used. Because this is an observational study, the results only reflect the circumference of the group at hand. Another group of volunteers will be needed to validate the equations. However, the high r-squared values show an encouraging correlation between circumference and regional composition. Some limitations and possible opportunities for future research include the narrow population. Most of the participants were Caucasian, college-aged volunteers. Validation of the equations could involve groups of volunteers of different ages so as to asses which ages these equations are predictive of and which ages would need specific equations. Separate equations may also be needed for different ethnicities and perhaps for athletes and non-athletes. The equations presented in this study provide a foundation for future validation and population specific research.