The purpose of this study was to examine ethnic differences in adiposity as measured by sum of skinfolds (SKF) and waist circumference (WC) in children and adolescents, after statistical adjustment for the BMI and age. A cross sectional sample of 3,218 (55% white, 49% male) children and adolescents aged 5–18 years who participated in the Bogalusa Heart Study (1992–1994) were included in these analyses. Sex-specific ANOVAs, adjusted for BMI and age, for each 2-year age group compared measures of adiposity (SKF and WC) between ethnic groups. No significant differences in the proportions of children and adolescents who were overweight and obese by ethnicity or sex were found. Mean SKF in normal weight (P < 0.0001) and overweight (P < 0.0001) categories was higher for white than black children of both sexes. Across most age categories, white boys and girls had significantly higher SKF than black boys and girls, respectively (P ≤ 0.05). Across most age categories, white boys had significantly higher WC than black boys (P ≤ 0.05) with no difference in the girls, when adjusted for BMI and age. Measures of adiposity in childhood and adolescence were significantly higher in white children compared to black children, when adjusted for BMI and age. Throughout childhood and adolescence, white boys and girls had higher SKF and white boys had higher WC. Differences in adiposity between ethnic groups should be considered in disease risk assessment and stratification as they are observed even for a given BMI level.
The prevalence of obesity (≥95th percentile of BMI for age (1)) has been steadily increasing in American children and adolescents with an overall increase from 13.9% in the 1999–2000 National Health and Nutrition Examination Survey to 17.1% in 2003–2004 (ref. 2). Furthermore, ethnic differences are apparent, with 20% of black vs. 16.3% of white children and adolescents classified as obese in the 2003–2004 National Health and Nutrition Examination Survey (2). Childhood overweight and obesity tracks into adulthood (3,4) and overweight black children become obese adults more frequently than overweight white children (5). Moreover, childhood overweight and obesity have been associated with increased risk of chronic diseases (6,7) and increased adult morbidity and mortality (8).
Regardless of overweight classifications, studies of adiposity and fat patterning have shown that black children and adolescents have less fat than white children and adolescents and should perhaps have different risk stratifications (9) as disease risk may also differ (10). A few studies have examined ethnic differences in adiposity in children and adolescents using advanced body composition measurement techniques such as dual-energy X-ray absorptiometry (DXA) (11,12,13,14) and densiometry (15). Freedman et al. reported that despite having a higher overall BMI, black children had substantially less fat (∼3%) than white children (12) but that black children also had significantly higher fat-free mass compared to white children at the same BMI for age (11). Further studies are required to determine the clinical application of these results.
The proper identification and classification of children and adolescents that are at increased obesity-related health risk by age is essential to properly direct resources for appropriate interventions. However, research on large biethnic samples of children with sufficient clinical measures of obesity and adiposity to address this issue is limited. Therefore it is necessary to further examine ethnic differences throughout the entire age range of childhood in relation to adiposity and BMI in large samples of children and adolescents using measures likely to be utilized by clinicians. Currently, the BMI is used in the clinical evaluation of children and adolescents, typically by comparison to age- and sex-specific growth charts. As part of the Bogalusa Heart Study, a community-based investigation of the evolution of cardiovascular risk beginning in childhood (16), the present analysis determines whether there are ethnic differences in adiposity as measured by sum of skinfolds (SKF) and waist circumference (WC), for a given level of BMI.
Methods and Procedures
Seven cross-sectional studies were conducted from 1973–1974 through 1992–1994 (16,17) in Bogalusa, Louisiana. Bogalusa is located in Washington Parish and is a semirural (Ward 4 ∼22,000) biethnic (67% white and 31% black) (18) community in Southeastern Louisiana, 70 miles north of New Orleans. This analysis was limited to a sample of 3,218 children, 5–18 years of age, who were examined in the 1992–1994 cross-sectional survey. Informed consent was obtained from all participants, and study protocols were approved by the human subjects review committee of the Louisiana State University School of Medicine and the Tulane University School of Public Health and Tropical Medicine. These analyses were approved by the Institutional Review Board at Pennington Biomedical Research Center.
Standardized techniques and protocols were used by trained examiners. Height and weight were measured in duplicates to the nearest 0.1 cm and 0.1 kg using a manual height board and balance-beam metric scale, respectively. BMI was calculated as weight in kilograms divided by the square of height in meters and used as a measure of overall adiposity. In addition, adiposity was measured in terms of WC and skinfold thicknesses in triplicate. WC was measured midway between the rib cage and the superior border of the iliac crest with a nonstretchable clinician's tape. The subscapular skinfold was measured using Lange calipers (Cambridge Scientific Industries, Cambridge, MD) immediately below the inferior angle of the scapula and the triceps skinfold was measured at the dorsum of the triceps muscle parallel to the long axis, and the two measures were summed (SKF). Reliability quality control checking was completed on a 10% random sample with an intraclass correlation coefficient for height >0.99, weight >0.99, WC >0.97, and SKF thicknesses >0.98 (19). Analyses were performed on averaged anthropometric measures.
Two-year age groups were created for analyses, with the exception of 4, 5, and 6 year olds who were combined into one category. Participants were classified as normal weight, overweight, or obese by sex and age according to the Centers for Disease Control and Prevention BMI percentiles (1). Additionally, participants were classified as having an elevated WC using race-, age-, and sex-specific WC cut-points (≥75th percentile) developed by Fernandez et al. (20).
Descriptive characteristics of the sample were calculated. Sex-specific, age-adjusted correlations were calculated for outcome variables of interest: WC, BMI, and SKF. ANCOVA adjusted or age, race, and sex and χ2 analyses were conducted between ethnic and sex groups for continuous (WC, BMI, and SKF) and categorical (proportion of the sample that was normal weight, overweight, or obese; elevated WC) variables.
The relationship between obesity classifications and adiposity via SKF and WC between two ethnic groups was examined using general linear modeling ANOVA for each 2-year analytic age group adjusted for BMI and age (within each analytical age group). This analysis was repeated for boys and girls separately. The analyses were repeated adjusting for BMI percentiles rather than raw BMI scores and age to compare methodologies. The most conservative statistical approach would be to adjust the level of significance for multiple analyses to P ≤ 0.007. However, as there was no overlap between participants in the age categories in the multiple analyses, results are presented at the traditional level of significance (P ≤ 0.05) in the text and figures. The conservative correction has also been presented in the text. Models were repeated for each of the different adiposity constructs (i.e., SKF and WC).
Descriptive characteristics of the sample are presented in Table 1 by analytic age groups. Overall, 68% of the sample was normal weight (n = 2,171), 16% was overweight (n = 516) and 17% was obese (n = 531). All three outcome variables were significantly correlated for boys and girls when adjusted for age (BMI and SKF r = 0.83 to r = 0.88, P < 0.0001; BMI and WC r = 0.92 to r = 0.93, P < 0.0001; SKF and WC r = 0.85 to r = 0.86, P < 0.0001). Black and white children were significantly different for mean WC (P < 0.008, t = 2.66) and SKF (P < 0.001, t = 4.40) but not BMI (P = 0.95, t = −0.06). A significant sex difference was noted for WC (P < 0.001, t = 5.98) and SKF (P < 0.001, t = −11.16), but not BMI (P = 0.25, t = −1.14). There was no significant difference between ethnic groups or sexes in the proportion of participants classified as normal weight, overweight, or obese. The sample sizes and proportions of overweight and obese did not differ between sex or ethnic groups: white (n = 602) 16.4% and 17.4%; black (n = 445) 15.6% and 15.4%; boys (n = 530) 16.3% and 17.1%; and girls (n = 517) 15.8% and 16.0%, overweight and obese, respectively. Elevated WC did significantly differ between sex (boys 30.6% vs. girls 23.6%, P = 0.02) and ethnic groups (white 28.8% vs. black 24.9%, P < 0.0001). Mean values for BMI, WC, and SKF and the prevalence of overweight, obese, and elevated WC are presented in Table 2 by race-sex specific groups. No difference was noted between groups for BMI. An ethnicity (P < 0.004, F = 8.53) and sex (P < 0.0001, F = 53.21) difference was found for WC and for SKF (P < 0.0001, F = 20.87) and (P < 0.0001, F = 134.64) for ethnic and sex groups, respectively. An ethnicity (P < 0.02, χ2 = 5.93) and sex (P < 0.0001, χ2 = 20.05) difference was noted in the proportion of participants with an elevated WC.
Table 1. Descriptive characteristics of the sample of Bogalusa Heart Study (1992–1994) children and adolescents by 2-year analytic age group
Table 2. Sex and ethnic group mean ± s.e. (adjusted for age, race, and sex) of BMI, WC, and SKF and prevalence of overweight, obese, and elevated WC
The results of the ANOVA examining the relationship between BMI groups and adiposity via SKF between ethnic groups are presented in Figure 1. Analyses were conducted separately for boys and girls and adjusted for BMI and age. There was a significant difference in SKF between black and white children (Figure 1) in the normal weight (P < 0.0001) and overweight (P < 0.003) categories for boys and girls alike. In the obese category, only girls had SKF that was significantly different between ethnic groups (P < 0.05). Results were similar when analyses were adjusted for BMI percentile rather than raw BMI score and age.
When SKF was compared between ethnic groups by age categories (Figure 2) all but one age group (9–10 year olds) were significantly different (P ≤ 0.05) in the boys. Similarly for girls, ethnic groups had significantly different (P ≤ 0.05) SKF for all age groups except 9–10 year olds. When the correction for multiple analyses was considered (P ≤ 0.007) significant differences remained for boys in the 4–6 and 7–8 year olds age groups and for girls in all age groups except 9–10 and 13–14 year olds.
The results of the ANOVA examining the relationship between obesity classifications and adiposity via WC between ethnic groups were conducted separately for boys and girls and adjusted for BMI and age and presented in Figure 3. For the normal weight group, WC was significantly higher in white compared to black boys (P < 0.0001) and in black compared to white girls (P < 0.0001). In the overweight group, white when compared to black boys was also significantly higher (P < 0.0001); however, no relationship was noted between ethnic groups for the girls. In the obese group, white when compared to black boys had a significantly higher WC (P < 0.0001); a similar relationship was noted for girls (P < 0.007). Results were similar when analyses were adjusted for BMI percentile rather than raw BMI score.
When WC was compared between ethnic groups by age categories and adjusted for BMI and age (Figure 4), all age groups were significantly different (P ≤ 0.05) for the boys except 4–6 year olds. For girls, WC was not significantly different between ethnic groups in any age group except 17–18 year olds.
The primary findings of this study support and expand upon previous research(11,12,13,14,21,22) that measures of adiposity are significantly different between black and white children and adolescents for a given BMI classification (normal weight and overweight) such that white children are fatter than black children. A similar study conducted by Freedman et al. (12) in a multiethnic sample in New York, examined adiposity via DXA and controlled for age and BMI percentiles, indicated that body fatness can vary by up to five percentage points across ethnic groups for a given BMI and age percentile. This highlights the importance of including ethnicity whenever possible in the determination of overweight/obesity and disease risk (12). Our study with >3,200 children and adolescents aged 5–18 years of age replicates these findings using a different and larger sample of black and white children and adolescents. Two studies conducted in Europe also showed significant differences in adiposity between ethnic groups, however, no valid comparisons can be made between those studies and ours, as the ethnic classifications were different (21,22). The large sample size in our study allowed for high statistical power for each cell (i.e., black boys, black girls, white boys, white girls). Even though findings were statistically significant, the actual difference in SKF was seemingly small (2–4.5 mm) and may not be clinically relevant. However, with the difference being significant at all but one of the 2-year age groups the likelihood of a statistical anomaly is small; therefore, further investigation into this phenomenon is warranted. Additionally, the utilization of clinical and field methods enhance the generalizability of the findings as BMI, SKF, and WC are likely to be used for the risk assessment of an individual child in a physician's office or school clinic. One of the few contrary reports that showed no difference between black and white children and adolescents when examining adiposity by DXA and SKF (23) had a small sample and limited age range. Another study reported that black girls had higher SKF than white girls at each age. However, when body composition (via bioelectrical impedance) was examined, white girls had higher percent body fat than black girls suggesting that black girls have more subcutaneous fat because overall fatness was higher in white girls (24).
Approximately one-third of the total sample in our study was classified as overweight or obese by the Centers for Disease Control and Prevention BMI-for-age percentiles (1), which is higher than the percentage of overweight in the New York study (ranging from 9 to 24%) (ref. 12). There were no differences between ethnic or sex groups in the proportion of children classified as normal weight, overweight, or obese. This is in contrast to recent data (2003–2004) published from National Health and Nutrition Examination Survey, showing that black children and adolescents have a higher prevalence of overweight than white children and adolescents (2). Freedman et al. reported that in the past several years, the prevalence of overweight and obesity has been increasing more rapidly in black children and adolescents than in white children and adolescents (25). Because our study examined data that was collected in the early 1990s it is plausible that no secular difference in BMI were yet observable in this sample. There have been no new participants added to the Bogalusa Heart Study since the early 1990s so this hypothesis cannot be currently tested. It would be prudent to replicate these analyses with a large biethnic sample of children and adolescents with data more recently collected. However, within all but one of the 2-year age groups in our study, white girls and boys had higher SKF (i.e., increased adiposity) than black girls and boys for a given BMI. It is also interesting to note that for a given BMI, white girls had higher SKF whereas WC was not different from their black counterparts and that white boys had significantly higher WC for a given BMI in all but one age group than black boys.
The results of this study among children and adolescents suggest that for a given BMI, black boys have lower adiposity than white boys, as measured by WC and SKF. The results among girls were not as consistent; black girls had lower SKF but not WC for a given BMI. These results are similar to a number of studies in adults, which have demonstrated that black men have lower visceral adiposity than white men (26,27,28,29), even when adjusting for total body fat (28,29). Racial differences in adiposity are not as clear among women. For example, four studies have reported lower visceral adiposity in black women compared to white women (26,29,30,31), whereas four studies have reported no differences (27,28,32,33).
WC is a viable measure for predicting obesity-related diseases (34), total adiposity, and visceral adiposity in children and adolescents (35) and is used in the classification of elevated abdominal adiposity. A significantly greater percentage (29% vs. 25%) of white than black children had a WC value higher than the 75th percentile. Misclassification of excess abdominal adiposity could potentially be a concern by measures of abdominal adiposity in our study and the elevated classification used. However, the level of elevated WC was based on race-, sex- and age-specific 75th percentiles, which likely indicates true elevated abdominal obesity and disease risk (20). Although the BMI percentiles (1) were not race specific, they were based on a representative sample of children and adolescents from the United States.
The primary strength of this analysis is in the large, biethnic sample size across the growing age spectrum. It allowed for the sex-specific examination of trends in obesity and adiposity as children grow into adolescents rather than simply looking at the differences between ethnic groups with all ages combined. An additional strength of this study is that it examines two indicators of adiposity (SKF and WC) and adjusts for the participant's BMI and age. Results highlight the potential misclassification of “overweight” and “obesity” in the black population because for a given BMI, black children had lower WC and SKF, and presumably lower disease risk according to these measures. The limitations to this study are the clinical and field-type measures of adiposity used rather than more advanced and specific measures of body fatness and fat patterning such as DXA, magnetic resonance imaging, or computed tomography scans. However, epidemiological research and clinical practice typically include field diagnostic measures rather than those that are used in clinical research. BMI and WC are measures that can easily and inexpensively be collected during a regular visit to the doctor whereas DXA or computed tomography scans are more costly, time consuming, and require highly trained technicians. Furthermore, the predictive value of BMI for age (i.e., ability of BMI for age to predict excess adiposity) was 72% to 79% in white and black boys and 100% to 73% in white and black girls (12). Further limitation to the generalizability of this study is that all the participants reside in or around Bogalusa, Louisiana and perhaps are not fully representative of other populations of children.
In conclusion, apparent ethnic and sex differences exist when measuring BMI and adiposity. For the BMI categories of normal and overweight, white boys and girls had higher SKF than blacks. Across childhood and adolescence, white girls and boys had significantly higher SKF than black girls and boys. Across childhood and adolescence, white boys had significantly higher WC than black boys, whereas there was no difference for girls. In addition to the findings that black children and adolescents appear to have less adiposity than whites for a given age and BMI, a significantly higher proportion of white and male children and adolescents were classified as increased risk due to elevated abdominal adiposity. Relative risk for developing chronic disease may be greater for children of certain ethnicities (36,37). Future research should focus not only on the degree to which ethnicity influences disease risk but factors, such as genetic and environmental issues, that lead to these differences in body composition at a young age. An important topic for future research would be to examine whether the higher adiposity of white compared to black children and adolescents for a given BMI increases the risk of chronic disease risk factors such as hypertension and dyslipidemia. When this is considered with the current findings and others that show ethnic differences in childhood and adolescent BMI (2,25), body fat (12), and fat patterning (13,38), the inclusion of a variety of screening measurements such as SKF (39) and WC (20,34) should be considered in pediatric risk stratification.
This research was supported by grant HL38844 from the National Institute on Aging and grant HD043820 from the National Institute of Child Health and Human Development. C.B. is partially supported by the George A. Bray Chair in Nutrition. P.T.K. is partially supported by the Louisiana Public Facilities Authority Endowed Chair in Nutrition. The Bogalusa Heart Study is a joint effort of many investigators and staff members, whose contributions are gratefully acknowledged.