Figures
Abstract
Objective
The worldwide prevalence of obesity mandates a widely accessible tool to categorize adiposity that can best predict associated health risks. The body adiposity index (BAI) was designed as a single equation to predict body adiposity in pooled analysis of both genders. We compared body adiposity index (BAI), body mass index (BMI), and other anthropometric measures, including percent body fat (PBF), in their correlations with cardiometabolic risk factors. We also compared BAI with BMI to determine which index is a better predictor of PBF.
Methods
The cohort consisted of 698 Mexican Americans. We calculated correlations of BAI, BMI, and other anthropometric measurements (PBF measured by dual energy X-ray absorptiometry, waist and hip circumference, height, weight) with glucose homeostasis indices (including insulin sensitivity and insulin clearance from euglycemic clamp), lipid parameters, cardiovascular traits (including carotid intima-media thickness), and biomarkers (C-reactive protein, plasminogen activator inhibitor-1 and adiponectin). Correlations between each anthropometric measure and cardiometabolic trait were compared in both sex-pooled and sex-stratified groups.
Results
BMI was associated with all but two measured traits (carotid intima-media thickness and fasting glucose in men), while BAI lacked association with several variables. BAI did not outperform BMI in its associations with any cardiometabolic trait. BAI was correlated more strongly than BMI with PBF in sex-pooled analyses (r = 0.78 versus r = 0.51), but not in sex-stratified analyses (men, r = 0.63 versus r = 0.79; women, r = 0.69 versus r = 0.77). Additionally, PBF showed fewer correlations with cardiometabolic risk factors than BMI. Weight was more strongly correlated than hip with many of the cardiometabolic risk factors examined.
Citation: Lichtash CT, Cui J, Guo X, Chen Y-DI, Hsueh WA, Rotter JI, et al. (2013) Body Adiposity Index versus Body Mass Index and Other Anthropometric Traits as Correlates of Cardiometabolic Risk Factors. PLoS ONE 8(6): e65954. https://doi.org/10.1371/journal.pone.0065954
Editor: Massimo Federici, University of Tor Vergata, Italy
Received: January 29, 2013; Accepted: May 1, 2013; Published: June 11, 2013
Copyright: © 2013 Lichtash et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This study was supported in part by National Institutes of Health grants R01-HL088457, R01-DK079888, P30-DK063491, M01-RR00425 (General Clinical Research Center Grant from the National Center for Research Resources), and UL1TR000124 (University of California Los Angeles Clinical & Translational Research Workshop), and the Cedars-Sinai Winnick Clinical Scholars Award (to M.O.G) and the Cedars-Sinai Board of Governor’s Chair in Medical Genetics (J.I.R.). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
The prevalence of obesity worldwide and across age groups has made it a focus of much investigative research. In the United States, the prevalence of obesity has been estimated at approximately one-third of the population, with the combined proportion of overweight and obese individuals encompassing approximately two thirds of the country [1]. Obesity is associated with an increased incidence of poor health outcomes, including cardiovascular disease, hypertension, diabetes mellitus, dyslipidemia, osteoarthritis, and certain cancers [2].
The most widely used method to categorize overweight and obese individuals is the body mass index (BMI, (weight in kilograms)/(height in meters)2), first named the Quetelet index and described by Adolphe Quetelet in 1832 [3]. In 1998, the National Heart, Lung, and Blood Institute of the National Institutes of Health (NIH) published clinical guidelines in which BMI was used in the systematic classification of overweight and obese individuals [4].
However, BMI is an imperfect measure of body adiposity. The weight term in BMI does not distinguish between muscle mass and fat mass. Furthermore, BMI has been shown to be age-, sex-, and in some cases, race-, dependent [5], [6]. At equivalent BMIs, women have significantly greater amounts of total body fat than men, and older individuals have significantly greater amounts of total body fat than those who are younger [5], [6]. Finally, it does not describe the depot of total fat in a person. Despite these disparities, however, the same BMI cutpoints for overweight and obesity are often applied across sex and age groups.
In May 2011, Bergman et al. introduced a new parameter, the body adiposity index (BAI, (hip circumference in centimeters)/(height in meters)1.5–18), derived via analyses of Mexican-American subjects, and validated in African-Americans [7]. Among its proposed advantages, the BAI presents a method of estimating body adiposity without requiring assessment of body weight, offering a simple-to-use tool that can be accessed globally [7]. Additionally, the demonstration of similar linear relationships between BAI and percent body fat (PBF) in men and women suggests that sex-specific adjustments of BAI as an estimate of PBF may not be necessary [7].
Several subsequent studies evaluated BAI as a predictor of body fat composition, yielding inconsistent results. While BAI was more strongly correlated than BMI with dual-energy X-ray absorptiometry (DXA)-derived PBF in sex-pooled analyses [8], [9], results have varied in sex-stratified analyses, with BAI being similarly correlated [8], [10] or less correlated [9] than BMI with PBF. Conflicting results have also been observed in states of extreme adiposity; BAI was more correlated than BMI with PBF in women with familial partial lipodystrophy [11]. In contrast, in a study of severely obese women, BMI, but not BAI, was significantly correlated with DXA-derived PBF [12].
To date, several published studies have examined BAI’s association with health risks or outcomes [9], [13], [14], [15], [16], [17], [18]. Findings of the studies reported thus far illustrate the need for further clarification on the clinical utility of BAI as a measure of body adiposity and correlate of disease [13], [19]. The goal of our study was to define the correlations of BAI, BMI, and other measured anthropometric variables with (a) glucose homeostasis traits, (b) lipid parameters, (c) cardiovascular traits, and (d) biomarkers. In cases wherein both BMI and BAI were associated with a trait, we statistically assessed whether one was more strongly associated than the other, to elucidate which may be more associated with cardiometabolic risk. A secondary aim of our study was to examine the associations of the various anthropometric variables with PBF, and to determine the correlation of PBF itself with cardiometabolic risk factors.
Materials and Methods
Study Subjects
Metabolic and cardiovascular phenotypes were assessed in participants of the UCLA/Cedars-Sinai Mexican-American Coronary Artery Disease (MACAD) project, a study of Mexican-American families from Los Angeles [20], [21]. To be classified as Mexican and qualify for the study, subjects had to report at least three grandparents of Mexican origin. In the present report, 698 subjects from 193 families (299 male and 399 female) with BAI values were studied, comprising adult offspring (age 18 or older) of probands with coronary artery disease, and the spouses of those offspring (if available) [20], [21]. By design, offspring were free of overt cardiovascular and metabolic disease, thus avoiding secondary changes in phenotype caused by overt disease.
Ethics Statement
All studies were approved by Human Subjects Protection Institutional Review Boards at UCLA and Cedars-Sinai Medical Center. All subjects gave written informed consent prior to participation.
Phenotyping Procedures
Subjects underwent a phenotyping protocol that included testing of glucose homeostasis indices, lipid parameters, cardiovascular traits, and biomarkers. In the original MACAD study design, a three-day phenotyping protocol was to take place within one week for each subject in the offspring generation. As executed, the time window for completion of all phenotyping ranged from one week to several years. The median time to completion of studies was 25.5 days; 81% of subjects completed phenotyping within 6 months. Only subjects completing studies within one year were included in the current study. On one day, fasting blood was obtained, followed by a 75 g oral glucose tolerance test (OGTT). On a separate day, B-mode ultrasound was performed for measurement of common carotid artery intima-media thickness (IMT) and dual-energy X-ray absorptiometry (DXA) scan was performed to assess body fat distribution. On a further day, a euglycemic-hyperinsulinemic clamp was performed.
During the clamp, a priming dose of human insulin (Novolin, Clayton, NC) was given and followed by infusion for 120 minutes at a constant rate (60 mU·m−2·min−1) with the goal of achieving a steady state plasma insulin concentration of 100 µIU/ml or greater [20], [22]. Blood was sampled every 5 minutes, and the rate of 20% dextrose coinfused was adjusted to maintain plasma glucose concentrations at 95 to 100 mg/dl. The glucose infusion rate (M value, mg·m−2·min−1) over the last 30 minutes of steady-state insulin and glucose concentrations reflects glucose uptake by all tissues of the body (primarily insulin-mediated glucose uptake in muscle) and is therefore directly correlated with tissue insulin sensitivity [22]. The insulin sensitivity index (M/I, mg·m−2·min−1·µIU−1·mL) was calculated as M divided by the steady state plasma insulin level (I). The metabolic clearance rate of insulin (MCRI, mL·m−2·min−1) was calculated as the insulin infusion rate divided by the steady state insulin level of the euglycemic clamp, as previously described [22], [23].
Fasting lipid parameters including low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), and triglycerides (TG), were examined in this study.
Cardiovascular traits included systolic and diastolic blood pressure (SBP and DBP) and carotid intima-media thickness (IMT) [24], [25]. Carotid artery images were obtained by high-resolution B-mode ultrasound using the Toshiba SSH-140A ultrasound system with a 7.5-MHz probe, at the University of Southern California Atherosclerosis Research Unit [26]. The IMT measure represents the distance between the blood-intima and media-adventitia echoes taken at the right distal common carotid artery [26]. Values are reported as the average of 80 to 100 individual IMT measurements made over 1 cm of the right distal common carotid artery [26].
Fasting biomarkers included C-reactive protein (CRP), adiponectin, and plasminogen activator inhibitor-1 (PAI-1) levels.
Also tested were percent body fat (PBF) measured by DXA, as well as anthropometric indices (BMI, BAI, height, waist circumference, weight, and hip circumference) [27].
Data Analysis
Log-transformed (BAI, BMI, weight, hip, HDL-C, TG, carotid IMT, adiponectin, CRP, 2-hour glucose, fasting insulin) or square-root transformed (M/I, MCRI, PAI-I) trait values were used to normalize the distribution for statistical analyses. We computed correlation coefficients (r) between the cardiometabolic phenotypes mentioned above, and the following measures: BMI, BAI, waist, hip, height, weight, and PBF and the P values indicating the difference of the correlation coefficients from zero. Correlation coefficients were also used to compare anthropometric measures in their degree of association with PBF. Hotelling’s t-test was used to determine whether correlation coefficients were significantly different. P values of <0.05 were considered significant.
Because the MACAD cohort consists of families, we also computed correlation coefficients using generalized estimating equations (GEE), adjusting for familial relationships. The weighted GEE1 [28] was computed assuming an exchangeable correlation structure and using the sandwich estimator of the variance to account for familial correlation present in family data.
Results
General Characteristics
The clinical characteristics of the 698 subjects are shown in Table 1. BMI did not differ significantly by gender, while BAI was higher in women than in men. Of the anthropometric measurements, waist circumference, weight, and height were significantly higher in men, while hip circumference and PBF were higher in women. With the exception of MCRI, fasting insulin, and PAI-1, all of the cardiovascular and metabolic phenotypes differed significantly between men and women. Men had more adverse lipid and cardiovascular profiles, while no clear pattern was observed within the glucose homeostasis and biomarker categories.
Correlations of Anthropometric Measurements with Cardiometabolic Risk Factors
We found that waist circumference, weight, and BMI were significantly correlated with all lipid parameters, glucose homeostasis traits, cardiovascular traits, and biomarkers, with the exception of carotid IMT, LDL-C, and fasting glucose in men (Table 2). Hip circumference was correlated with fewer of the cardiometabolic variables; lack of correlation was found with DBP and LDL-C in both sex-pooled and sex-stratified analyses, and with 2-hour glucose, fasting glucose, adiponectin levels, and carotid IMT in men. Comparatively, height showed the least number of significant relationships with the cardiometabolic variables. BAI and PBF shared a comparable pattern of lack of correlation, with neither having significant associations with LDL-C, TG, fasting glucose, carotid IMT, adiponectin, and SBP in sex-pooled analyses, as well as adiponectin in women and carotid IMT in men. We also conducted correlation analyses taking family relationships into account, and found that the correlation coefficients were essentially the same (Table S1).
Comparison of the Correlations of BAI and BMI with Cardiometabolic Risk Factors
Comparison of BAI and BMI in the strength of their correlations with cardiometabolic traits (Table 3) revealed the following. BMI was more strongly correlated than BAI with TG, M/I, fasting insulin and SBP in all analyses (both sex-pooled and sex-stratified). BMI was also more strongly correlated than BAI with LDL-C, HDL-C, MCRI, fasting glucose, carotid IMT, DBP, adiponectin, and PAI-1 in sex-pooled analyses; with HDL-C, 2-hour glucose, DBP, CRP, and PAI-1 in men; and with MCRI, carotid IMT, CRP, and adiponectin in women. BMI and BAI were similar in the strength of their correlations with 2-hour glucose and CRP in sex-pooled data; with MCRI, fasting glucose, LDL-C, adiponectin, carotid IMT in men; and with LDL-C, HDL-C, 2-hour glucose, fasting glucose, DBP, and PAI-1 in women.
Comparison of the Correlations of Hip Circumference and Weight with Cardiometabolic Risk Factors
We compared the two variables that distinguish the calculations of BAI and BMI, namely hip circumference and weight, respectively (Table S2). Weight was more strongly correlated than hip circumference with TG and SBP in all analyses (both sex-pooled and sex-stratified); in sex-pooled data with LDL-C, HDL-C, carotid IMT, DBP, and adiponectin; in men with 2-hour glucose, DBP, and PAI-1; and in women with M/I, MCRI, fasting insulin, CRP and adiponectin. Hip circumference outperformed weight in the strength of its association with only CRP in pooled data. Hip circumference and weight were similarly associated with all other cardiometabolic traits.
Comparison of the Correlations of Anthropometric Variables with PBF
Comparison of the various anthropometric variables in their ability to predict PBF is displayed in Table 2. BAI outperformed BMI in the strength of its correlation with DXA-derived PBF in sex-pooled analysis; however, when data was sex-stratified, BAI was weaker than BMI in predicting PBF in men and women. Figure 1 illustrates these findings. The larger percentage of overlapping data points between men and women when PBF is plotted against BAI leads to greater correlation between BAI and PBF than between BMI and PBF (r = 0.78 versus r = 0.51; P<0.0001) in sex-pooled analyses. However, in sex-stratified data, correlations with PBF were greater for BMI than for BAI (in men, r = 0.79 versus r = 0.63, P<0.0001; in women, r = 0.77 versus r = 0.69, P<0.0001) (Figure 1). Similarly, when DXA-derived total fat mass was examined, both BMI and BAI correlated with this measure. However, BMI was stronger than BAI in its correlation with total fat mass in both sex-pooled (r = 0.84 versus r = 0.71; P<0.0001) and sex-stratified analyses (in men, r = 0.88 versus r = 0.61, P<0.0001; in women, r = 0.91 versus r = 0.71, P<0.0001).
A. BAI versus PBF. B. BMI versus PBF. The graphs are generated on untransformed data for the BAI and BMI variables, while the main analyses in our study are based on log transformations of these variables. Therefore, the correlation coefficients (r) here are slightly different than those reported elsewhere in the text.
After BAI, height was the second strongest correlate with PBF in sex-pooled data (r = −0.54). However, height lost its correlation with PBF in sex-stratified analyses (Figure S1); conversely, waist, hip, weight, and BMI all strengthened in their correlations with PBF in sex-stratified analyses.
Discussion
We found that the cardiometabolic disease risk factors were more consistently correlated with BMI, waist circumference, and weight than with BAI. In cases wherein significant correlations were found, BAI was either similar to, or weaker than, BMI in the strength of these associations. Thus, our results show that with regards to its utility as a correlate of the cardiometabolic risk factors measured in our study, BAI is similar or inferior to the currently widely used BMI.
Our study is one of a growing number that have assessed the strength of BAI’s association with cardiometabolic risk factors [9], [13], [14], [15], [16], [17], [18], [29]. We included three variables that have not yet been studied with regards to their correlation with BAI, namely, carotid IMT, plasminogen activator inhibitor-1, and metabolic clearance rate of insulin from the euglycemic clamp. Only three other studies besides ours utilized a statistical test to compare correlation coefficients between anthropometric measures of body adiposity and cardiometabolic risk factors [9], [17], [18]. One found that waist circumference and BMI were more strongly correlated than BAI with OGTT-derived insulin sensitivity and type 2 diabetes risk [9]. The second found that BMI and waist circumference were superior to BAI in the strength of their correlations with most of the seven cardiovascular risk variables studied (LDL-C, HDL-C, TG, insulin, fasting glucose, SBP, and DBP); this superiority of BMI and waist circumference was upheld also when data was stratified by sex and age [17]. There was no case in which BAI was found to be superior to BMI or waist with regards to its correlative strength with cardiovascular risk variables [17]. Finally, the third of these studies found that while BAI was superior to BMI in its correlation with measures of leptin, BMI was more strongly correlated than BAI with adiponectin levels, HDL-C, TG, and glucose homeostasis traits (insulin, fasting glucose, and homeostasis model assessment of insulin resistance) [18].
Additional studies compared correlations of BAI and BMI with cardiometabolic traits, without formal statistical comparison of the correlation coefficients. One reported that regardless of glycemic status (euglycemic, impaired fasting glucose, impaired glucose tolerance, or type 2 diabetes mellitus), BAI had the weakest correlation with fasting glucose, 2-hour glucose, SBP, and DBP, as compared with other indices of adiposity (waist circumference, waist-to-hip ratio, waist-to-height ratio, and BMI) [14]. A study of obese post-menopausal women examined BAI’s ability to detect changes in cardiometabolic risk factors after a weight loss intervention [13]. The study found that the percent change in BAI following weight loss was significantly associated with percent changes in CRP and leptin, but not with percent change in any of the other cardiometabolic risk factors examined (total cholesterol, HDL-C, TG, fasting glucose, clamp-derived insulin sensitivity, SBP, or DBP) [13]. A study examining correlation coefficients between several anthropometric measures and metabolic risk factors (HDL-C, LDL-C, TG, fasting glucose, SBP, and DBP) concluded that BMI, waist circumference, waist-to-height ratio, and waist-to-hip ratio were all better correlated than BAI with the risk factors measured [16]. In a biracial cohort, anthropometric measures, including BAI, were found to be similarly correlated with most cardiovascular risk factors; however, BAI lacked correlation with LDL-C and total cholesterol in African-American women, and with SBP and DBP in African American men [15]. Finally, in a cohort of obese individuals classified as insulin resistant (homeostasis model assessment of insulin resistance (HOMA-IR) >2.5) and insulin sensitive, BAI and BMI differed in their correlations with cardiometabolic risk factors [29]. In each individual group, as well as combined, BAI was correlated with serum adiponectin, leptin, and CRP levels; however, no correlation was found between BAI and insulin levels or HOMA-IR. BMI, on the other hand, was not correlated with adiponectin levels in any group, but was correlated with leptin, CRP, insulin, and HOMA-IR in the insulin-resistant group [29].
In our study, BAI was significantly associated with several of the cardiometabolic risk factors, and in several cases performed similarly to BMI in the strength of its correlations with these risk factors (Table 3). Within the four categories examined (glucose homeostasis traits, lipid parameters, cardiovascular traits, and biomarkers), there was no clear pattern of association, or lack thereof, with respect to BAI. Further, with certain variables studied (e.g. LDL-C, TG, fasting glucose, carotid IMT, SBP, and adiponectin), no correlation was found with either BAI or PBF when analyzing sex-pooled data; in most cases, correlations of BAI and PBF with these traits were only revealed when the data was sex-stratified. This may be explained by differences between men and women with respect to these variables, as is demonstrated in the clinical characteristics of our cohort (Table 1). Nevertheless, comparisons of BMI and BAI with respect to their correlations with these variables within sex-stratified data showed that BMI was superior to BAI in its associations with SBP and TG in both men and women, as well as adiponectin and carotid IMT in women. BMI and BAI were similar in their associations with LDL-C and fasting glucose in sex-stratified analyses. Therefore, variation between the data in men and women with respect to BAI is unlikely to fully account for the relative weakness of its associations with cardiometabolic risk factors.
To understand the differences between BAI and BMI in the strength of their correlations with cardiometabolic risk factors, we compared hip and weight, the two variables that differ in the calculations of these adiposity indices. We found that weight was more associated with many of the cardiometabolic risk factors examined, consistent with the findings comparing BAI and BMI.
The results of our study illustrate the complexity of the task of identifying accurate markers of cardiometabolic risk factors associated with obesity. We found that PBF was less consistently correlated with several of the cardiometabolic risk factors than BMI. This suggests that BMI’s relationship with total adiposity may not be the sole determinate of its relationship with cardiometabolic risk. Further research is needed to clarify the characteristics of anthropometric variables that determine their associations with cardiometabolic risk factors and disease.
We also aimed to identify a single anthropometric measurement or index that most accurately predicted body adiposity as measured by DXA. While BAI did correlate most strongly with PBF in sex-pooled data analyses, BMI proved the most accurate tool to predict PBF in men and women in sex-stratified analyses, a finding supported in a recent study performed on a large German cohort [9]. These findings indicate that BMI is better able to account for the differences in body fat content and distribution in men and women, a conclusion that conflicts with recent data in a cohort of women with familial lipodystrophy, which suggested that BAI may be a more sensitive mode of estimating adiposity [11]. Our findings suggest that sex-adjustment of current BMI cutoffs for defining normal weight, overweight, and obese, might yield a more accurate assessment of body adiposity than either BMI or BAI as currently used.
BAI was designed specifically as a single equation that could predict body adiposity in pooled analyses of both genders. It was not designed as a tool to predict cardiometabolic risk. Our current data verifies that BAI is a stronger correlate of PBF than BMI in sex-pooled data, providing further support that BAI achieves what it was designed to do. Recent literature has raised questions about sex-specific bias in the variables used to derive BAI [9]. We found that among the characteristics considered for inclusion in the BAI equation (i.e. waist, hip, height, and weight), hip and height did, in fact, show the strongest correlation with PBF in sex-pooled data, in agreement with Bergman et al. [7]. However, as described in recent articles and the results herein, the correlation between height and adiposity is lost in sex-stratified analysis, suggesting that the correlation between height and PBF may be driven by sex-differences in these variables [9], [19] (Figure S1). However, the significance of this finding with respect to the comparative analysis of BAI and BMI is unclear, given that both indices incorporate height into their calculations.
We acknowledge limitations in our study, including the lack of racial and ethnic diversity in our cohort. We studied Mexican Americans, as did Bergman et al. [7], which provided us with the ability to test BAI’s performance in a population representative of its derivation group. Nevertheless, the results herein may not apply to other ethnic groups. Additionally, our study did not have access to longitudinal follow-up data on study subjects, which would have provided us with cardiovascular and metabolic disease outcomes, such as incident myocardial infarction and diabetes. Further studies are needed to examine BAI as it relates to disease outcomes, and to clarify any other clinical role that BAI may play in the evaluation of obesity. At the present time, if the goal is assessment of cardiometabolic risk, BMI is a more suitable tool than BAI.
Supporting Information
Figure S1.
Relationship between height and percent body fat (PBF) in sex-pooled (A) and sex-stratified (B) analyses.
https://doi.org/10.1371/journal.pone.0065954.s001
(TIFF)
Table S1.
Correlations of anthropometric measurements with cardiometabolic risk factors taking family relationships into consideration
https://doi.org/10.1371/journal.pone.0065954.s002
(DOCX)
Table S2.
Comparison of correlation coefficients between cardiometabolic risk factors and hip circumference versus weight.
https://doi.org/10.1371/journal.pone.0065954.s003
(DOCX)
Author Contributions
Conceived and designed the experiments: CTL JC XG YIC WAH JIR MOG. Performed the experiments: XG YIC WAH JIR MOG. Analyzed the data: CTL JC MOG. Wrote the paper: CTL MOG. Revised the manuscript critically for important intellectual content: JC XG YIC WAH JIR. Gave final approval of the manuscript: CTL JC XG YIC WAH JIR MOG.
References
- 1. Flegal KM, Carroll MD, Ogden CL, Curtin LR (2010) Prevalence and trends in obesity among US adults, 1999–2008. JAMA 303: 235–241.
- 2. Must A, Spadano J, Coakley EH, Field AE, Colditz G, et al. (1999) The disease burden associated with overweight and obesity. JAMA 282: 1523–1529.
- 3. Eknoyan G (2008) Adolphe Quetelet (1796–1874)–the average man and indices of obesity. Nephrol Dial Transplant 23: 47–51.
- 4. Clinical guidelines on the identification, evaluation, and treatment of overweight and obesity in adults: executive summary. Expert Panel on the Identification, Evaluation, and Treatment of Overweight in Adults. Am J Clin Nutr 68: 899–917.
- 5. Gallagher D, Visser M, Sepulveda D, Pierson RN, Harris T, et al. (1996) How useful is body mass index for comparison of body fatness across age, sex, and ethnic groups? Am J Epidemiol 143: 228–239.
- 6. Jackson AS, Stanforth PR, Gagnon J, Rankinen T, Leon AS, et al. (2002) The effect of sex, age and race on estimating percentage body fat from body mass index: The Heritage Family Study. Int J Obes Relat Metab Disord 26: 789–796.
- 7. Bergman RN, Stefanovski D, Buchanan TA, Sumner AE, Reynolds JC, et al. (2011) A better index of body adiposity. Obesity (Silver Spring) 19: 1083–1089.
- 8. Johnson W, Chumlea WC, Czerwinski SA, Demerath EW (2012) Concordance of the recently published body adiposity index with measured body fat percent in European-American adults. Obesity (Silver Spring) 20: 900–903.
- 9. Schulze MB, Thorand B, Fritsche A, Haring HU, Schick F, et al. (2012) Body adiposity index, body fat content and incidence of type 2 diabetes. Diabetologia 55: 1660–1667.
- 10. Barreira TV, Harrington DM, Staiano AE, Heymsfield SB, Katzmarzyk PT (2011) Body adiposity index, body mass index, and body fat in white and black adults. JAMA 306: 828–830.
- 11. Godoy-Matos AF, Moreira RO, Valerio CM, Mory PB, Moises RS (2012) A new method for body fat evaluation, body adiposity index, is useful in women with familial partial lipodystrophy. Obesity (Silver Spring) 20: 440–443.
- 12.
Geliebter A, Atalayer D, Flancbaum L, Gibson CD (2012) Comparison of Body Adiposity Index (BAI) and BMI with estimations of % body fat in clinically severe obese women. Obesity (Silver Spring) https://doi.org/10.1038/oby.2012.187
- 13. Elisha B, Rabasa-Lhoret R, Messier V, Abdulnour J, Karelis AD (2013) Relationship between the body adiposity index and cardiometabolic risk factors in obese postmenopausal women. Eur J Nutr 52: 145–151.
- 14. de Lima JG, Nobrega LH, de Souza AB (2012) Body adiposity index indicates only total adiposity, not risk. Obesity (Silver Spring) 20: 1140.
- 15. Barreira TV, Staiano AE, Harrington DM, Heymsfield SB, Smith SR, et al. (2012) Anthropometric correlates of total body fat, abdominal adiposity, and cardiovascular disease risk factors in a biracial sample of men and women. Mayo Clin Proc 87: 452–460.
- 16. Snijder MB, Nicolaou M, van Valkengoed IG, Brewster LM, Stronks K (2012) Newly proposed body adiposity index (BAI) by Bergman et al. is not strongly related to cardiovascular health risk. Obesity (Silver Spring) 20: 1138–1139.
- 17.
Freedman DS, Blanck HM, Dietz WH, Dasmahapatra P, Srinivasan SR, et al.. (2012) Is the body adiposity index (hip circumference/height1.5) more strongly related to skinfold thicknesses and risk factor levels than is BMI? The Bogalusa Heart Study. Br J Nutr: 1–8.
- 18.
Melmer A, Lamina C, Tschoner A, Ress C, Kaser S, et al.. (2012) Body Adiposity Index and other indexes of body composition in the SAPHIR study: association with cardiovascular risk factors. Obesity (Silver Spring) https://doi.org/10.1038/oby.2012.160
- 19. Schulze MB, Stefan N (2011) The body adiposity index and the sexual dimorphism in body fat. Obesity (Silver Spring) 19: 1729.
- 20. Goodarzi MO, Guo X, Taylor KD, Quinones MJ, Saad MF, et al. (2004) Lipoprotein lipase is a gene for insulin resistance in Mexican Americans. Diabetes 53: 214–220.
- 21. Goodarzi MO, Guo X, Taylor KD, Quinones MJ, Samayoa C, et al. (2003) Determination and use of haplotypes: ethnic comparison and association of the lipoprotein lipase gene and coronary artery disease in Mexican-Americans. Genet Med 5: 322–327.
- 22. DeFronzo RA, Tobin JD, Andres R (1979) Glucose clamp technique: a method for quantifying insulin secretion and resistance. Am J Physiol 237: E214–223.
- 23. Goodarzi MO, Taylor KD, Guo X, Quinones MJ, Cui J, et al. (2005) Variation in the gene for muscle-specific AMP deaminase is associated with insulin clearance, a highly heritable trait. Diabetes 54: 1222–1227.
- 24. Goodarzi MO, Taylor KD, Guo X, Quinones MJ, Cui J, et al. (2005) Association of the diabetes gene calpain-10 with subclinical atherosclerosis: the Mexican-American Coronary Artery Disease Study. Diabetes 54: 1228–1232.
- 25. Wang D, Yang H, Quinones MJ, Bulnes-Enriquez I, Jimenez X, et al. (2005) A genome-wide scan for carotid artery intima-media thickness: the Mexican-American Coronary Artery Disease family study. Stroke 36: 540–545.
- 26. Selzer RH, Mack WJ, Lee PL, Kwong-Fu H, Hodis HN (2001) Improved common carotid elasticity and intima-media thickness measurements from computer analysis of sequential ultrasound frames. Atherosclerosis 154: 185–193.
- 27. Li X, Quinones MJ, Wang D, Bulnes-Enriquez I, Jimenez X, et al. (2006) Genetic effects on obesity assessed by bivariate genome scan: the Mexican-American coronary artery disease study. Obesity (Silver Spring) 14: 1192–1200.
- 28. Zeger SL, Liang KY (1986) Longitudinal data analysis for discrete and continuous outcomes. Biometrics 42: 121–130.
- 29.
Stepien M, Stepien A, Wlazel RN, Paradowski M, Rizzo M, et al.. (2012) Predictors of insulin resistance in patients with obesity: A pilot study. Angiology https://doi.org/10.1177/0003319712468291