P values for centered main effects and interactions are from a regression model including body adiposity index (BAI) or body mass index (BMI), age, sex, race, and sex × BAI (sex × BMI) and race × BAI (race × BMI) interactions. For BAI centered effects, sex β = −5.2 (P < .001); race β = −0.7 (P < .001). For BMI centered effects, sex β = −12.2 (P < .001); race β = −1.6 (P < .001). The race × BAI interaction had P = .19; the 3 other interactions (sex × BAI, sex × BMI, and race × BMI) had P < .001. The lines indicate the linear relationships between the 2 variables.
Barreira TV, Harrington DM, Staiano AE, Heymsfield SB, Katzmarzyk PT. Body Adiposity Index, Body Mass Index, and Body Fat in White and Black Adults. JAMA. 2011;306(8):828-830. doi:10.1001/jama.2011.1189
Author Affiliations: Pennington Biomedical Research Center, Louisiana State University System, Baton Rouge (firstname.lastname@example.org).
To the Editor: Body mass index (BMI) is widely used as a proxy for body fat and has been shown to correlate with other measures of adiposity.1 However, its use is limited by differences in body fatness for a given BMI across age, sex, and race.2,3 To address this limitation, Bergman et al4 developed the body adiposity index (BAI) in samples of Mexican-American and black individuals. However, no sex-specific information was provided, and it is unknown how well BAI performs in white individuals. We investigated the sex-specific relationship between both BMI and BAI and body fat in white and black adults.
The participants were mainly healthy volunteers recruited from the greater Baton Rouge area for metabolic studies from 1992 to 2011.5 This analysis included only adults with dual-energy x-ray absorptiometry measures. Because a focus of the study was to understand racial differences, race was self-identified. The study was approved by the institutional review board of the Pennington Biomedical Research Center, and participants provided written informed consent.
Height and weight were measured using a stadiometer and digital scale, respectively. Hip circumference was measured at the level of the trochanters. The BMI was calculated as weight in kilograms divided by height in meters squared, and BAI was calculated as hip circumference in centimeters divided by height in meters to the 1.5 power minus 18. Fat mass and percentage of body fat were measured using a Hologic QDR4500 (n = 1549, 2001-2011) or QDR2000 scanner (n = 2302, 1992-2006); QDR2000 data were converted to QDR4500 output (Hologic, Bedford, Massachusetts).5
Pearson correlations were computed among BAI, BMI, fat percentage, and fat mass within each sex-by-race group. A linear regression model was used to assess the relationship between fat percentage and BAI (or BMI), age, sex, and race. Interaction terms were entered into the model for sex × BAI (or BMI) and race × BAI (or BMI). Body adiposity index and BMI were standardized to zero mean and unit variance prior to analysis using SAS version 9.2 (SAS Institute, Cary, North Carolina). Level of significance was P < .05, and statistical tests were 2-sided.
The sample included 1462 white women, 812 black women, 1262 white men, and 315 black men. The mean (SD) age, BMI, and fat percentage of the sample were 41.0 (13.4) years (range 18-69 years), 29.4 (6.1) (range, 17.2-57.7), and 32.0% (9.7%) (range, 7.8%-55.9%), respectively. The correlations with fat percentage across the 4 sex-by-race groups ranged from 0.75 to 0.82 for BAI and 0.80 to 0.83 for BMI, and the correlations with fat mass ranged from 0.77 to 0.86 for BAI and 0.90 to 0.96 for BMI (Table).
The regression model that included BAI explained 81.9% of the variance in fat percentage. The corresponding model for BMI explained 84.1%. Women had 5.2% and 12.2% more fat percentage than men (P < .001) at the mean BAI and BMI, respectively, while white individuals had 0.7% and 1.6% more fat percentage than black individuals (P < .001) at the mean BAI and BMI, respectively. The race interaction term was not significant for BAI (P = .19), but it was for BMI (P < .001); however, both sex interaction terms were significant, indicating that the associations between BAI and BMI with fat percentage differed by sex and by race for BMI (Figure).
Body mass index and BAI perform similarly in predicting body fat. In each sex-by-race group, the correlations with fat percentage and fat mass were similar for BMI and BAI. Moreover, the regression models including BAI or BMI explained a similar percentage of the variance in fat percentage. The sex and race differences in the relationship between both BAI and BMI with fat percentage make interpreting these measures in different population groups difficult. Our results are based on a sample of volunteers enrolled in clinical studies, and the representativeness of the results is not known. Neither BMI nor BAI measure obesity complications directly, and further research is required to determine the clinical significance of these measures.
Author Contributions: Dr Katzmarzyk had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Barreira, Harrington, Staiano, Heymsfield, Katzmarzyk.
Analysis and interpretation of data: Barreira, Harrington, Staiano, Heymsfield, Katzmarzyk.
Drafting of the manuscript: Barreira, Harrington, Staiano, Heymsfield, Katzmarzyk.
Critical revision of the manuscript for important intellectual content: Barreira, Staiano.
Statistical analysis: Barreira, Harrington, Staiano, Heymsfield, Katzmarzyk.
Obtained funding: Katzmarzyk.
Study supervision: Katzmarzyk.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.
Funding/Support: This research was supported by the Pennington Biomedical Research Center. Dr Katzmarzyk is supported, in part, by the Louisiana Public Facilities Authority Endowed Chair in Nutrition. This work was partially supported by a Nutrition Obesity Research Center grant 2P30-DK072476-06 entitled “Nutritional Programming: Environmental and Molecular Interactions” and sponsored by National Institute of Diabetes and Digestive and Kidney Diseases.
Role of the Sponsors: The sponsors had no role in the design and conduct of the study; in the collection, analysis, and interpretation of the data; or in the preparation, review, or approval of the manuscript.
Additional Contributions: We gratefully acknowledge the contributions of Emily Mire, MS, and Connie Murla, BS, for data management and the many clinical scientists and staff of the Pennington Biomedical Research Center who have contributed data to the Pennington Center Longitudinal Study, particularly Steven R. Smith, MD. Ms Mire, Ms Murla, and Dr Smith were all employees of the Pennington Biomedical Research Center and did not receive compensation beyond their normal salary.