ARS | Publication request: Field method to measure changes in percent body fat of young women: The TIGER Study

Submitted to: Obesity
Publication Type: Abstract Only
Publication Acceptance Date: August 1, 2007
Publication Date: September 20, 2007
Citation: Jackson, A., Ellis, K., Sailors, M., McFarlin, B., Turpin, I., Miller, F., Bray, M. 2007. Field method to measure changes in percent body fat of young women: The TIGER Study[abstract]. Obesity. 15(Suppl):A73.

Technical Abstract: Body mass index (BMI), waist (W) and hip (H) circumference (C) are commonly used to assess changes in body composition for field research. We developed a model to estimate changes in dual energy X-ray absorption (DXA) percent fat (% fat) from these variables with a diverse sample of young women from the Training Interventions and Genetics of Exercise Response (TIGER) Study. The sample was 652 women, age 17 to 35 y who had their body composition assessed 1 to 3 times over 9 months for a total of 994 observations. The sample race/ethnic (R/E) composition was: White (W), 28.2%; Hispanic (H), 25.0%; African-American (AA), 33.9%; Asian-Indian (AI), 4.3%; and Asian (A), 8.6%. Random-intercept, linear mixed model (LMM) regression was used to develop the model. The dependent variable was DXA %fat and the fixed effect variables were log transformed BMI (L-BMI), sum of WC and HC (WHC) and R/E group. Subjects were the random component of the LMM. A maximum likelihood solution fit the solution. The Wald statistic for the model was statistically significant (p <0.0001)and all fixed effects regression coefficients were significantly greater than 0 (p <0,001). Using the White women as the referent group, the fixed effect coefficients (95% CI) were: L-BMI, 22.38 (19.29, 25.48); WHC, 0.05 (0.03, 0.08); H, 1.717 (0.92, 2.52); AA, -1.64(-2.38, -0.90); AI, 5.37 (3.91, 6.84); A, 2.71 (1.56, 3.87); and the intercept, -49.90 (-55.98, -43.81). The standard errors (SE) for the fixed and random models were 3.77 and 1.10 DXA %fat. The within subject correlation, controlling for the fixed effect model, was 0.94. The fixed effect model provides an accurate field method to estimate DXA %fat (SE = 3.57% fat) of diverse young women. The obtained LMM provides a valid and feasible field method to measure change in DXA %fat.