The costs of publication of this article were defrayed, in part, by the payment of page charges. This article must, therefore, be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Objectives: The prevalence of childhood overweight in the United States has markedly increased over the last 30 years. We examined differences in the secular trends for BMI, weight, and height among white, black, and Mexican-American children.
Research Methods and Procedures: Analyses were based on nationally representative data collected from 2 to 17 year olds in four examinations (1971–1974 through 1999–2002).
Results: Overall, black children experienced much larger secular increases in BMI, weight, and height than did white children. For example, over the 30-year period, the prevalence of overweight increased ∼3-fold (4% to 13%) among 6- to 11-year-old white children but 5-fold (4% to 20%) among black children. In most sex-age groups, Mexican-American children experienced increases in BMI and overweight that were between those experienced by blacks and whites. Race/ethnicity differences were less marked among 2 to 5 year olds, and in this age group, white children experienced the largest increase in overweight (from 4% to 9%). In 1999–2002, the prevalence of extreme BMI levels (≥99th percentile) reached 6% to 7% among black girls and Mexican-American boys.
Discussion: Because of the strong tracking of childhood BMI levels into adulthood, it is likely that the secular increases in childhood overweight will greatly increase the burden of adult disease. The further development of obesity interventions in different racial/ethnic groups should be emphasized.
The prevalence of obesity in the United States has markedly increased over the last 30 years, with national data indicating that 5% of 2 to 17 year olds were overweight in 1971–1974, but that 10% (ages 2 to 5 years) to 16% (6 to 17 years) of children were overweight in 1999–2002 (1, 2). Although childhood levels of both weight and height have increased over this period, the former have been much larger (3). Similar trends have been seen in other studies, with the proportion of children in Bogalusa, LA, having elevated levels of relative weight and triceps skinfold thickness increasing by >2-fold from 1973 to 1992 (4).
Similar trends have also occurred among adults, with the prevalence of obesity (BMI ≥ 30 kg/m2) reaching 30% in 1999–2002 (2). The prevalence of adult obesity, however, varies markedly by race/ethnicity, with prevalences ranging from 27% (Mexican-American men) to 49% (black women). The prevalence of class 3 obesity (BMI ≥ 40 kg/m2) ranges from 3% (Mexican-American men) to 13.5% (black women).
Despite these differences in the prevalence of adult obesity, few studies have examined the extent to which the secular trends in childhood overweight have varied by race/ethnicity. Analyses from the National Longitudinal Study of Youth indicated that the 1986 to 1998 increase in overweight among 4 to 12 year olds was most pronounced among black and Hispanic children (5). Somewhat similar differences, with white children showing the smallest increases in overweight, were also seen in national data between 1988–1994 and 1999–2000 (1), but these analyses were limited to 12 to 19 year olds. In contrast, the secular trends (1973 to 1992) in relative weight and skinfold thicknesses in Bogalusa did not differ between white and black children (4).
The purpose of this study was to further examine sex and race/ethnicity differences in secular trends for childhood BMI, overweight, weight, and height. These analyses are based on national data from four nationally representative studies conducted from 1971–1974 to 1999–2002.
Research Methods and Procedures
The National Health and Nutrition Examination Survey (NHANES)1 is a series of cross-sectional examinations conducted by the National Center for Health Statistics of the Centers for Disease Control and Prevention. Each sample is representative of the U.S. non-institutionalized civilian population and is selected through a multistage process based on selection of counties, blocks, households, and persons within households.
The current analyses include 2 to 17 year olds who were examined in NHANES I (conducted from 1971 to 1974), NHANES II (1976 to 1980), NHANES III (1988 to 1994), and NHANES 1999–2002. (NHANES became a continuous survey beginning in 1999.) Each examination consists of different individuals. Pregnant women were excluded from the analyses, as were persons with missing information on weight or height. In the last two examinations, race/ethnicity was based on proxy- or self-report and was categorized as non-Hispanic white, non-Hispanic black, or Mexican American; the variables are DMARETHN (race-ethnicity) (1988–1994) and RIDRETH2 (race-ethnicity recode) (1999–2002). Because of small numbers, other categories of race/ethnicity (e.g., other Hispanic) are excluded from the analyses.
In the first two examinations (1971–1974 and 1976–1980), race was classified as white, black, or other, and questions were also asked concerning national origin and ancestry. To derive comparable race/ethnicity categories (non-Hispanic white, non-Hispanic black, and Mexican American) across all examinations, we classified children in the first two examinations as Mexican American if their reported origin/ancestry was Chicano, Mexicano, Mexican, or Mexican American; this category would include children whose race was reported as either white or black. Children whose reported origin/ancestry was South American, Central American, Cuban, or Puerto Rican were excluded from all analyses. It is likely that any misclassification of race/ethnicity would have reduced the observed (cross-sectional) differences between Mexican Americans and other groups.
Sample sizes of the non-Hispanic white, non-Hispanic black, and Mexican-American children, stratified by sex, age, and examination, are shown in Table 1.
Table 1. . Sample sizes by age, sex, race, and examination
Height and weight measurements in each survey were obtained using standardized techniques and equipment. BMI was calculated as weight (in kilograms) divided by height (in meters squared). Sex-specific z-scores and percentiles of weight-for-age, height-for-age, and BMI-for-age were calculated using national data collected from 1963 through 1994 (6, 7). Childhood overweight is defined as a BMI-for-age ≥95th percentile (8, 9). Although there is no standard definition for childhood obesity, we considered children with a BMI-for-age ≥99th percentile to be obese.
It should be realized that, whereas the classification of obesity among adults is based on absolute BMI cut-points, the childhood classification is based solely on the distribution of (sex- and age-specific) BMI levels in national studies. Therefore, we also classified two children who had a BMI ≥30 kg/m2 (the cut-point for adult obesity) but a BMI-for-age <95th percentile as overweight and four children who had a BMI ≥40 kg/m2 (adult cut-point for class 3 obesity) but a BMI-for-age <99th percentile as obese. These six children were 17-year-old girls and represented 0.2% (overweight) to 7% (obese) of the 17-year-old girls in these BMI categories.
Data were analyzed using SAS (version 9.1; SAS Institute, Inc., Cary, NC), S-Plus (version 6.2; Insightful Corp., Seattle, WA), and SUDAAN (version 8.02; Research Triangle Institute, Research Triangle Park, NC). All analyses used sample weights to account for differential probabilities of selection, non-response, and non-coverage. We examined mean levels of BMI, weight, and height across the four surveys and also examined trends within categories of race/ethnicity, sex, and age. We assessed the statistical significance of the observed secular trends using linear or logistic regression models in SUDAAN to account for the complex sampling design. Various interaction (product) terms were included in these regression models, and several secular trends were found to vary by race/ethnicity and age but not by sex.
To illustrate the racial differences in the secular trends, mean weights and heights were calculated within categories of race, sex, and age (1-month age groups) in the first (NHANES I) and last (NHANES 1999–2002) exams. We calculated the difference (at each age) between the initial and final mean levels (a period of ∼30 years). Lowess (locally weighted scatterplot smoother), a non-parametric smoothing technique (10), was used to show the differences in mean levels by race, sex, and age.
Table 2 shows mean levels of BMI, along with changes in the z-scores for BMI, weight, and height over the ∼30-year period. In 1971–1974, Mexican-American children had the highest mean BMI level (18.6 kg/m2), but the subsequent secular increases among black children (from 17.8 to 20.0 kg/m2) were larger than the increases among white (+1.0 kg/m2) and Mexican-American (+1.5 kg/m2) children. (Both the secular trend and the racial difference in the trend were significant at the 0.001 level.) Furthermore, the BMI secular trend began slightly earlier among black girls than among white or Mexican-American girls. Whereas the mean BMI of black girls increased by 0.7 kg/m2 (17.9 to 18.6 kg/m2) from 1971–1974 to 1976–1980, almost no change was seen in the mean BMI of other girls (p = 0.03 for interaction between race and examination). The secular increase in BMI was also larger among older children than younger children (p < 0.01 for interaction) but did not differ between boys and girls.
Table 2. . Mean BMI levels and change in the mean z-scores by race-ethnicity and sex
Increase in mean z-score from 1971–1974 to 1999–2002
Secular trends for BMI, BMI-for-age, weight-for-age, and height-for-age were each statistically significant at the 0.001 level. Trends in BMI, BMI-for-age, and weight also differed (p < 0.001) by race.
Mean BMI levels have been adjusted for differences in age and sex across exams.
Over the 30-year period, the largest increases in the BMI z-score (Table 2) were seen among 12- to 17-year-old blacks (+0.72 SD). Although both weight and height increased over the 30 years (Table 2), most of the increases in weight were substantially larger. Among black 6 to 11 year olds, for example, the mean weight-for-age increased by 0.69 SD (from −0.06 to 0.63) over the four examinations, whereas height-for-age increased by only 0.36 SD. Two- to 5-year-old Mexican-American children were the only subgroup that experienced a larger increase in height than weight over the 30 years (0.43 SD for height vs. 0.29 SD for weight), resulting in unchanged BMI levels over time.
Figure 1 shows the mean changes in weight (top) and height (bottom) that occurred from 1971–1974 to 1999–2002 at various ages. Among boys (Figure 1, left) and girls (Figure 1, right), the weight increases differed only slightly by race before ages 7 to 8 years, but older white children experienced substantially smaller weight increases than did similarly aged black and Mexican-American children. Among boys, the largest weight increases were generally seen among Mexican Americans, whereas among girls, the largest weight increases were seen among blacks. Twelve- to 15-year-old black girls experienced a mean weight increase of ∼10 kg over the 30-year period.
There were also substantial increases in height (Figure 1, bottom). Among boys, the racial differences in trends were somewhat similar to those seen for weight, with larger increases seen among black and Mexican-American children than among whites. Among girls, however, Mexican Americans experienced the largest height increases before age 7 years, but black girls experienced the largest height increases between the ages of 8 and 16 years. With the possible exception of Mexican Americans, there was little change in the heights of the oldest children.
Over the 30-year period, the prevalence of overweight (BMI ≥ 95th percentile) increased 3-fold, and the prevalence of obesity (BMI ≥ 99th percentile) increased ∼4-fold (p < 0.001 for both trends; Table 3). Among 2 to 5 year olds, white children showed slightly larger increases (from 4% to 9%) in overweight than other children, but among older children, the increases in overweight were almost 2-fold greater among black and Mexican-American children than among whites. Among 12 to 17 year olds, for example, the prevalence of overweight increased by 15% among Mexican-American children and by 14% among black children but by only 7% among white children. The highest prevalence of obesity (BMI ≥ 99th percentile) in 1999–2002 was seen among Mexican-American boys (7%) and black girls (6%).
Table 3. . Prevalence of overweight and obesity* among children
Increase in prevalence from 1971–1974 to 1999–2002
Overweight is defined as a BMI ≥ 95th percentile or ≥30 kg/m2; obesity is defined as a BMI ≥ 99th percentile or ≥40 kg/m2.
Values are percentage of overweight children (percentage of obese children).
Our results indicate that black 6 to 17 year olds have experienced much greater increases in BMI and in the prevalence of overweight over the last 30 years than have white children. Furthermore, the secular increases among black girls began during the 1970s, whereas increases among other children were not evident until the 1980s. BMI increases among 2 to 5 year olds have been smaller, and the largest increases have occurred among white children. Mexican-American children had the highest mean BMI level in 1971–1974, but their mean BMI in 1999–2002 was similar to that among black children.
Results of previous studies agree with our findings. Between 1973 and 1994, the mean BMI of children in Bogalusa increased by 1.5 kg/m2 (4), and although the trends did not differ between whites and blacks, this may be unique to that community in Louisiana. The white/black difference in the secular trends for weight that we observed (Figure 1) agrees with longitudinal data indicating that the racial divergence in adiposity occurs during adolescence among girls (11). In addition, analyses of children born to women in the National Longitudinal Survey of Youth (5) found that black and Hispanic 4 to 12 year olds experienced larger increases in overweight from 1986 to 1998 than did white children. Although we also found large increases in the BMI levels of black children and Mexican-American boys, the mean BMI of Mexican-American girls increased only slightly (+0.3 kg/m2) between 1988–1994 and 1999–2002. [Sex-specific analyses were not presented in the National Longitudinal Survey of Youth analyses (5).] These differences may be caused by random variability, the classification of ethnicity (Hispanic vs. Mexican-American), or the requirement for mothers of the children in the National Longitudinal Survey of Youth to have been in the United States in 1978 (12).
Although the weight and height of children in the United States have increased over the last 30 years, the weight changes (as assessed by z-scores) have been substantially larger. As described by Ogden et al. (3), for example, the mean weight of 10-year-old girls increased by 6.2 kg from 1971–1974 to 1999–2002. For BMI to have remained constant, the mean height would have needed to increase by ∼12 cm rather than the actual increase of 3 cm. It should be noted, however, that the large secular increase in the height (0.43 SD) of 2- to 5-year-old Mexican-American children over the study period completely eliminated their 1971–1974 height deficit (mean, 1.6 cm) relative to whites. The relatively low heights of young Mexican-American children in 1982–1984 have been attributed to their greater poverty (13).
The secular increases in BMI that we observed were particularly large at the upper end of the BMI distribution, continuing a trend that was noted from 1976–1980 to 1988–1994 (14). Over the 30-year period, for example, the prevalence of BMI levels above the 99th percentile increased by 6- to 7-fold in some groups, and because many associations with BMI are non-linear, the rapid increase in the number of children with extreme BMI levels may have important health implications. For example, whereas BMI levels among relatively thin children largely reflect levels of fat-free mass, a BMI ≥95th percentile is a very good indicator of excess fat mass (15, 16). Non-linear trends have also been observed between BMI and metabolic risk factors, with the prevalence of multiple risk factors increasing rapidly at BMI levels >95th percentile (17).
The recent increases in obesity likely result from the interaction of biological and social factors with an environment that includes few opportunities for physical activity and an overabundance of high-calorie foods (18, 19). Furthermore, it is reasonable to assume that the influence of these complex changes have varied by race/ethnicity (20, 21). For example, there have been increases in the availability of food energy and processed foods and in the number of meals eaten away from home (22, 23). Some data also indicate that, compared with white girls, black girls have a higher mean energy intake (24) and are more likely to eat food while watching television (25). Furthermore, breast-feeding, which is inversely associated with childhood overweight, is more prevalent among white than black women (26). Compared with whites, black women and adolescents also seem to have a larger ideal body size (27) and are less likely to perceive themselves as overweight (28, 29).
In addition to these environmental and social factors, additional cultural factors may influence childhood obesity among Mexican Americans. Although child fatness has been viewed as a sign of good health among Mexicans (30), given the high prevalence of childhood overweight in the United States, it is not clear whether it continues to be favorably viewed. However, Hispanic high school students consume fewer fruits and vegetables than do whites (31), and physical activity levels tend to be lower among Mexican-American (and black) children than among white children (32). The rapid acculturation (33) of Mexican Americans into an “obesogenic” environment may partly explain why the inverse association between parental socioeconomic status and childhood obesity, which is typically seen among white children, is much weaker (or reversed) among Mexican-American and black children (34, 35).
Because of the strong tracking of high BMI levels from childhood into adulthood, particularly among minorities, the secular trends in childhood overweight and obesity are troubling. Recent longitudinal analyses, for example, indicate that >80% of overweight black 5 to 14 year olds will be obese in early adulthood (36). The racial/ethnic differences in the secular trends that we observed suggest that different factors are operating to promote obesity among white, black, and Hispanic children.
There was no funding/outside support for this study.
Nonstandard abbreviations: NHANES, National Health and Nutrition Examination Survey.