PURPOSE OF REVIEW: Although the prevalence of childhood obesity, as assessed by BMI (kg/m2), has tripled over the last 3 decades, this index is a measure of excess weight rather than excess body fatness. In this review we focus on the relation of BMI to body fatness and health risks, particularly on the ability of BMI for age ≥95th Centers for Disease Control and Prevention [CDC] percentile to identify children who have excess body fatness. We also examine whether these associations differ according to race/ethnicity and whether skinfold and circumference measurements provide additional information on body fatness or health risks.
RESULTS: The accuracy of BMI varies according to the degree of body fatness. Among relatively fat children, BMI is a good indicator of excess adiposity, but differences in the BMIs of relatively thin children can be largely due to fat-free mass. Although the accuracy of BMI in identifying children with excess body fatness depends on the chosen cut points, we have found that a high BMI-for-age has a moderately high (70%–80%) sensitivity and positive predictive value, along with a high specificity (95%). Children with a high BMI are much more likely to have adverse risk factor levels and to become obese adults than are thinner children. Skinfold thicknesses and the waist circumference may be useful in identifying children with moderately elevated levels of BMI (85th to 94th percentiles) who truly have excess body fatness or adverse risk factor levels.
CONCLUSION: A BMI for age at ≥95th percentile of the CDC reference population is a moderately sensitive and a specific indicator of excess adiposity among children.
Although BMI(kg/m2) is widely used as an index of body fatness, it is a measure of weight relative to height rather than of adiposity. BMI cannot distinguish between body fatness, muscle mass, and skeletal mass, and its use can result in large errors in the estimation of body fatness.1 The weight and height changes that occur during growth (ages 5–18 years) result in substantial (∼50%) increases in BMI, further complicating the interpretation of this index among children and adolescents. For example, increases in the BMI levels of boys during adolescence seem to be largely the result of increases in fat-free mass rather than body fatness.2
BMI has been characterized as being “almost useless as an estimator of percentage of body fat in normal-weight children,”3 but much evidence indicates that a child with a high BMI relative to his or her gender and age peers (BMI for age) is likely to have excess body fatness. In this review we focus on the relation of childhood BMI to body fatness and health risks, particularly on the ability of a high BMI level to correctly identify a child with excess body fatness. In addition, we examine whether these associations differ according to race/ethnicity and whether skinfold and circumference measurements provide additional information on body fatness or health risks.
The term “obesity” is frequently used to refer to a high BMI among adults and children,4,5 and the adult cut point for obesity (BMI ≥ 30 kg/m2) is approximately the 95th percentile of BMI for age among 19-year-old boys and 17-year-old girls in the Centers for Disease Control and Prevention (CDC) reference population6 (an SAS program for the CDC growth charts is available at www.cdc.gov/nccdphp/dnpa/growthcharts/resources/sas.htm). Although there are several reasons to prefer this terminology,5 it can create confusion between high levels of BMI and high levels of body fatness. Because a main goal of this review is to quantify the ability of a high BMI (excess weight) to identify children who have a high level of body fatness, we specify the BMI cut points (eg, BMI for age ≥95th CDC percentile) that are used to classify overweight and obese children in addition to the descriptive terms. A high level of body fatness is termed “excess body fatness.”
CHILDHOOD BMI AND BODY FATNESS
Measurement of Body Fatness
Although an ideal measure of body fatness would be accurate, precise, and accessible, with widely available population data, no existing measure satisfies all these criteria.7 Reference methods are expensive and can be inaccessible, and the simpler, widely used anthropometric methods can be inaccurate.3 The 4-compartment model, which incorporates independent estimates of mineral, total body water, body density, and body weight, is widely viewed as the most accurate method for calculating body fatness,8 but it is time consuming, expensive, and available at few locations.
Although the initial costs are high, the speed and low operating costs of dual-energy radiograph absorptiometry (DEXA) has led to its use as a reference method in many studies that have examined the screening ability of BMI. These estimates of fat mass are highly correlated with those from 4-compartment models, but can be influenced by several factors including the hydration of the subjects and the equipment manufacturer, which creates the potential for systematic biases and large errors for a single subject.3,8–10 A bias would not influence the relative ranking of subjects, but it could influence the proportion of children who are classified as having excess body fatness; this, in turn, would influence the ability of a high BMI to correctly identify children with excess fatness (positive predictive value). In addition, random errors in DEXA-estimated body fatness would reduce the (apparent) screening performance of BMI.
Body fatness is usually standardized for body weight and expressed as percent body fat [fat mass (kg)/body weight (kg)], but an alternative is to express fat mass relative to height squared. This leads to the use of fat mass index (FMI) (fat mass/height squared) and the fat-free mass index (FFMI) (fat-free mass/height squared).11,12 Because FMI + FFMI = BMI, the use of these indices allows one to easily assess whether BMI differences (between subjects, ages, or time periods) are a result of fat or fat-free mass.
Although excess body fatness among children would ideally be defined on the basis of disease risk, this may be difficult. Almost all studies of health outcomes have examined the effects of excess weight rather than body fatness, and longitudinal studies spanning many decades would be necessary to have childhood body-fatness cut points to be based on associations with clinical outcomes. Furthermore, because the relation of childhood body fatness to disease outcomes is likely to be influenced by adult levels of body fatness, it would be necessary to have multiple measurements of body fatness throughout life. Even among adults, a World Health Organization expert committee concluded that “there is no agreement about cutoff points for the percentage of body fat that constitutes obesity.”13 The derivation and use of various cut points for excess body fatness are discussed below in “Classification of Excess Body Fatness.”
Anthropometry is a widely used, inexpensive method for assessing body fatness, but it can be inaccurate. The most frequent measurements are weight and height, and these are usually combined as weight/heightp; the power “p” can be chosen to minimize the correlation with height or to maximize the correlation with body fatness. BMI is a power index with p = 2.
Various skinfold and circumference (particularly the waist) measurements have also been used to assess fatness or fat distribution, but the errors associated with these measurements are larger than for weight and height.14,15 (The use of self-reported weight and height may result in biased and imprecise estimates of BMI16 and will not be discussed in this review.) Furthermore, it is uncertain if skinfold or circumference measurements provide a substantial amount of additional information on body fatness or health risks if BMI for age is already known (see “Additional Information From Skinfolds and Circumferences”).
Despite its widespread acceptance among adults, the use of BMI among children has sometimes been questioned because of its moderate association with height during growth and development (r ∼ 0.3 after adjustment for age). However, height is correlated with the body fatness of children,7,17 and the higher BMIs of taller children correctly identify their increased fatness. Furthermore, the correlation between childhood BMI and body fatness is close to the maximum that is possible for any power index,17,18 and childhood BMI is more strongly correlated with adult skinfold thicknesses than are other weight-height indices.17,19
Although BMI cut points of 25 and 30 kg/m2 are used to identify adults who are at increased risk for various diseases, the large increases in BMI that occur during growth and development require a “high” BMI to be defined relative to other children of the same gender and age. (Because body fatness also changes substantially with age, “excess body fatness” is also typically defined relative to a child's peers.) The use of a reference population to classify gender- and age-specific levels of BMI allows one to examine secular trends and compare the prevalence of overweight and obesity among children and adolescents across populations.
There is a “confusing multiplicity of child and adolescent obesity rates” in the literature,20 largely based on the use of different reference populations and different percentile (or z-score) cut points. The 2 most widely used classification systems for BMI levels among children are the CDC growth charts21 and the International Obesity Taskforce (IOTF) cut points.4,22 Both classifications are based on cross-sectional, gender-specific distributions of BMI levels according to age, with the cut points for high BMI levels based on somewhat arbitrary decisions. The 85th and 95th percentiles of BMI for age are used in the CDC growth charts, and the IOTF classification used BMI levels at age 18.0 years (rather than at other ages) to represent levels among adults.
The CDC growth charts were based on data from 4 nationally (US) representative surveys of children and adolescents conducted from 1963–1965 to 1976–1980; 2- to 5-year-olds who were examined in 1988–1994 were also included in these growth charts.7,21 BMI levels were expressed as gender-specific BMI-for-age z scores (SD scores) and percentiles, which allow the BMI of any child to be expressed relative to this reference population.
Most investigators now classify children who have a BMI for age at ≥95th percentile of this reference population as obese, and children with BMI levels between the 85th and 94th percentiles as overweight. (These 2 BMI-for-age categories had originally been termed “overweight” and “at risk for overweight.”23,24) Older adolescents with a BMI at ≥25 kg/m2 (∼85th percentile at ages 16–17 years in the reference population) are also considered to be overweight, whereas those with a BMI at ≥30 kg/m2 (∼95th percentile at ages 17–19 years) are also considered to be obese. These cut points were not based on associations with disease outcomes, but subsequent studies have shown that a high BMI for age is associated with increased body fatness, adverse levels of metabolic risk factors, and high levels of adult BMI.25,26 On the basis of these cut points, 17% of 6- to 19-year-olds in the United States were obese in 2005–2006, and another 16% were overweight.
The IOTF cut points, based on data from 6 countries (including the United States), linked BMIs of 25 kg/m2 (adult overweight) and 30 kg/m2 (adult obesity) at age 18.0 years to BMIs at younger ages.4 An 18-year-old boy in the United States with a BMI of 30 kg/m2, for example, was at the 96.7th percentile, and the corresponding z score (1.84) was used as the cut point for obesity between the ages of 2.0 and 17.5 years. The estimated cut points for overweight (IOTF-25) and obesity (IOTF-30) were then averaged across the 6 countries. Although this classification linked childhood BMI cut points to the adult classifications, if the distribution of BMIs at ages 25 or 30 years (rather than age 18.0 years) had been used, the prevalence of childhood overweight and obesity would be increased, because the adult BMI distribution would have been shifted toward higher values. Furthermore, the data used to derive these BMIs are largely from high-income countries.
Despite differences in the development of these 2 classification systems, the BMI levels corresponding to the IOTF-25 and 85th CDC percentile cut points are similar between the ages of 6 and 17 years. The IOTF-30 cut points are higher than the 95th CDC percentile and roughly correspond to the 97th CDC percentile among school-aged children. At younger ages (2–5 years), the CDC cut points are lower than the IOTF cut points, possibly reflecting the smaller secular increase in the BMI levels of preschool-aged (versus school-aged) children in the United States.27 Differences in the BMI cut points used in these 2 classification systems to identify the highest BMI category (IOTF-30 versus 95th CDC percentile) can complicate comparisons across studies.20,22,28
The 99th percentile of BMI for age in the CDC growth charts has been suggested as a possible cut point for severe obesity,26 identifying children who are at very high risk for severe adult obesity, excess body fatness, and biochemical abnormalities. However, the underling parameters29 of the CDC growth charts were based on 10 selected percentiles of BMI for age (according to gender) between the 3rd and 97th percentiles, and it can be problematic to use these parameters to estimate the 99th percentile. For example, as compared with the 99th percentile of BMI in the CDC reference population, the 99th percentile estimated from the growth-chart parameters may be too low (∼98th percentile in this reference population) among younger children and too high (∼99.5th percentile) among older adolescent girls. It may be best to classify severe obesity among children by a BMI at or above (1) either the CDC 99th percentile or 35 kg/m2 (class 2 adult obesity), (2) 120% of the CDC 95th percentile, or (3) 112% of the CDC 97th percentile.
Classification of Excess Body Fatness
Because of the lack of longitudinal data on the relation of body fatness among children to disease risk in adulthood, some investigators have derived cut points for excess body fatness among children from cross-sectional associations with metabolic risk factors.30–32 (Associations with metabolic risk factors are discussed below in “Childhood BMI and Health Risks.”) Of these classifications, the most widely used is based on the relation of skinfold-estimated percent body fat to adverse levels (upper quintile for a child's race, gender, and age) of lipids and blood pressures in the Bogalusa Heart Study.30,33 The proposed cut points for excess body fatness differed according to gender (25% body fat among boys, 30% among girls) but not according to age among these 5- to 18-year-olds.
Although it is appealing to use a single cut point to identify children who have excess body fatness at all ages, a single cut point is unlikely to be optimal. An age-invariant cut point cannot account for the changes in body fatness that occur during growth and development and for age differences (interactions) in the relation of body fatness to risk factors. On the basis of the data from girls in the Pediatric Rosetta Study, 30% body fat is at ∼95th percentile at the age of 7 years but is only slightly greater than the 50th percentile at the age of 15 years. Although the health effects of 30% body fat are likely to differ between 7- and 17-year-old girls, the possibility of age modification was not considered in the analyses that derived the 25% and 30% cut points.30
Despite the added complexity, it is likely that excess body fatness should be defined relative to a child's gender and age peers in much the same way that BMI cut points among children have been constructed. However, there is little agreement on the cut points that should be used to classify excess body fatness.13
Various techniques and samples have been used to estimate gender- and age-specific percentiles (eg, 85th, 90th, and 95th) of body fatness,18,34–37 as well as age-specific levels that correspond to the percentage body fat of a typical 18-year-old with a BMI of 30 kg/m2.36,38 Children with body fatness above these cut points have been classified as having excess body fatness. Other studies39,40 have used “prevalence matching,” in which cut points were chosen so that the prevalence of children with excess body fatness (within gender and age groups) would be approximately equal to the prevalence of children considered to be obese or overweight on the basis of their BMI.
It has been noted36 that an important limitation of all published cut points for excess body fat is the lack of clinical correlates underlying the classification systems. However, because the relation of childhood body fatness to disease is likely to be influenced by the adult level of adiposity, it may be difficult to derive childhood cut points from longitudinal studies that would last many decades.
RELATION OF BMI TO BODY FATNESS
Among adults, BMI is moderately correlated with adiposity, frequently accounting for ∼75% of the variability (multiple R2) across subjects. The magnitudes of the observed associations among children have varied considerably, and relatively weak associations have been reported in several subgroups.41–44 It is possible that these differences reflect the difficulties involved in using BMI to estimate the body fatness of relatively thin children. Although BMI is “almost useless as an estimator of percentage of body fat in normal-weight children,”3 its accuracy increases with the degree of body fatness. One study,42 for example, found the relation of BMI to skinfold thicknesses to be nonexistent (r = 0.01) among relatively thin boys but moderate (r = 0.58) among fatter boys. The differing associations between BMI and body fatness across studies may largely reflect differences in the fatness of examined subjects.
Analyses of the DEXA-estimated body fatness of children have confirmed that associations with BMI for age vary considerably according to the degree of body fatness. Fig 1 shows the estimated (cross-sectional) relation of BMI for age to levels of FMI and FFMI at ages 6 and 17 years among 1196 children in the Pediatric Rosetta Study.25 The relation of BMI for age to FMI was nonlinear, with fat mass increasing substantially only at BMI-for-age z scores of >1.0 (∼85th CDC percentile). In contrast, the relation of BMI for age to FFMI was fairly linear, with a positive association seen even among children who had a BMI at <50th CDC percentile.
Stratified analyses confirmed that the degree of fatness modified the relation of BMI for age to body fatness.25 BMI levels among thin 5- to 8-year-old boys, for example, showed a much stronger association with FFMI (r = 0.83) than with FMI (r = 0.22), whereas BMI levels were strongly associated with body fatness (r = 0.96) among relatively heavy (BMI ≥ 85th CDC percentile) boys. Even in this relatively heavy subgroup, however, BMI for age was moderately correlated with FFMI (r = 0.57).
Longitudinal analyses have also shown that increases in the BMI of adolescent boys, but not girls, largely reflect increases in FFMI rather than in FMI.2 Furthermore, these increases in FFMI were seen even among boys who had a BMI between the 85th and 94th CDC percentiles. BMI differences among normal-weight children (and BMI changes among these children) can largely reflect differences in fat-free, rather than fat, mass.
Identification of Excess Body Fatness by BMI
Although several investigators have examined the ability of a high BMI to identify children with excess body fatness,18,35,45,46 the results are difficult to compare. There are differences in the (1) methods (skinfolds, DEXA, etc) used to estimate body fatness, (2) cut points used to classify high levels of both BMI and body fatness, and (3) statistics used to summarize the screening performance of BMI.
The accuracy of a high BMI in identifying children with excess body fatness has frequently been summarized by calculating the (1) sensitivity (proportion of children with excess body fatness who have a high BMI), (2) positive predictive value (proportion of children with a high BMI who have excess body fatness), and (3) specificity (proportion of children without excess body fatness who do not have a high BMI). It should be realized, however, that the cut points chosen for high levels of BMI and body fatness can strongly influence estimates of screening performance. For example, even if BMI for age and body fatness were perfectly correlated, if the prevalence of excess body fatness was 2 times higher than the prevalence of a high BMI, the maximum sensitivity would be 50% rather than 100%. In contrast, if the prevalence of a high BMI was twofold higher than that of excess body fatness, the positive predictive value could be no higher than 50%.
Although it is frequently stated that a high BMI is not a sensitive indicator of excess body fatness,35,45–49 a low sensitivity may reflect a low prevalence of children with high BMI levels (relative to those considered to have excess body fatness) rather than the poor performance of BMI in detecting excess body fatness. For example, reported sensitivities of almost 0 in some subgroups are almost certainly a result of large differences in the prevalences of high BMI levels (<10%) and excess body fatness (up to 66% in some groups on the basis of a body-fatness cut point of 30%) among examined subjects.45 Furthermore, differences in the performance of various BMI cut points,49 including population-specific versus international cut points,50 may also be largely due to differing prevalences of high BMI levels based on different classification systems. The potential influence of the cut points used for levels of BMI and body fatness on the screening performance of BMI can be large and must be considered when interpreting the results of various studies. It would be helpful if investigators consistently reported the prevalences of children with high BMIs and those with excess body fatness.
Analyses from the Pediatric Rosetta Project have indicated that a high BMI for age is a good index of excess body fatness,37,51 and Fig 2 shows levels of percent body fat versus age for the 626 boys (left) and the 570 girls (right) in this study. Because ∼15% of the children were obese (≥95th CDC percentile), the 85th percentile of percent body fat (in the sample) was used for the classification of excess body fatness; these percent-body-fat cut points were estimated at each age by using quantile regression.52 If a high BMI could perfectly classify excess body fatness, all obese children (triangles) would be above the smoothed lines and all nonobese children (gray points) would be below the lines. Despite the obvious errors, the classification is fairly good: most children with a BMI at ≥95th CDC percentile had excess body fatness, and most nonobese children did not. Additional analyses (not shown) indicated that ∼80% of the false-positives (obese children who did not have excess body fatness) had a BMI for age between the 95.0th and 96.9th percentiles of the CDC reference population.
The ability of obesity (BMI ≥ 95th CDC percentile) to correctly identify children with excess body fatness can be summarized in 2 × 2 tables (Table 1). Of the 103 obese boys (2 upper-left cells), 75 had excess body fatness (positive predictive value = 73%), whereas the sensitivity was 74% (75 of 102). There were similar estimates (75%) of the positive predictive value and sensitivity among girls, and the specificity was high (∼95%) among both boys (496 of 524) and girls (465 of 486). Analyses based on the 85th percentile of FMI, rather than percent body fat, yielded slightly higher (81%–86%) positive predictive values but similar estimates of sensitivity and specificity (not shown). The use of the 99th CDC percentile of BMI for age (bottom of Table 1) increased the specificity and positive predictive value to 100% (all children with these high BMI levels, represented by the open triangles in Fig 2, had excess body fatness). However, most children with excess body fatness did not have a BMI for age at ≥99th CDC percentile, resulting in a sensitivity of only 20% (22 of 102 and 17 of 84; Table 1).
In contrast to obese children, most children who have a BMI for age between the 85th and 94th percentiles (overweight) do not have excess body fatness.40 For example, of the 200 overweight children in the Pediatric Rosetta Study, only 18% (positive predictive value) had a percent-body-fat level at or above the age-specific 85th percentile. (The corresponding positive predictive value for the 95th CDC percentile was >70%.) Another 50% of overweight children in the Pediatric Rosetta Study had a moderately elevated body-fatness level, with levels of percent body fat corresponding to their moderately elevated BMI levels (85th–94th percentiles), whereas ∼30% of overweight children had a body-fatness level that was comparable to that among normal-weight children.40 Analyses from the Bogalusa Heart Study have also shown that relatively few (13%) overweight children have elevated skinfold thicknesses.26 It is likely that BMI levels considered to be “overweight” on the basis of the 2000 CDC growth charts can be a result of moderate increases in levels of either fat or fat-free mass.25
The screening accuracy of BMI can also be summarized by the area under the receiver operator characteristic curve (AUC).53 This statistic accounts for the inherent trade-offs between sensitivity and specificity obtained with various BMI cut points and can be interpreted as the probability that the BMI for age of a randomly selected child who has excess body fatness would be higher than the BMI for age of a child without excess body fatness. On the basis of this statistic, BMI for age was a good indicator (AUC ∼ 0.95; the maximum is 1.0) of excess body fatness among both boys and girls in the Pediatric Rosetta Study. Although the AUC can be somewhat difficult to interpret, it provides a more complete representation of the screening performance of BMI for age than do estimates of sensitivity, positive predictive value, and specificity obtained from a single BMI cut point.
Racial Differences in BMI and Body Fatness
The body fatness of adults differs according to race/ethnicity: at equivalent levels of BMI and age, white adults have less body fat than Asian adults54 but more than black adults.55 On the basis of these findings, along with possible differences in the relation of body fatness to disease risk, it has been suggested that BMI cut points be lowered among Asian adults56 and raised among black adults.55 This could result in BMI cut points identifying similar levels of body fatness (and possibly, risk) across race/ethnicity groups. There have been fewer studies of differences in the body composition of children, but black/white differences in body fatness have been noted by several investigators.55,57
Analyses from the Pediatric Rosetta Study58 have confirmed that there are differences in the DEXA-estimated body fatness of white, Asian, and black children (Fig 3). At comparable levels of BMI for age, the estimated body fatness of black girls (and boys) was ∼3% less than that of white children, but the white/Asian difference varied substantially across levels of BMI for age (Fig 3, right). Among relatively thin (BMI < 50th CDC percentile) girls, for example, Asians had a mean (adjusted) level of body fatness that was ∼2% higher than that of whites. In contrast, obese white girls had ∼2% to 3% more body fatness (at the same BMI level) than did obese Asian girls. Some results have also suggested that the increased body fatness of Asian (versus white) adults may be most evident at low BMI levels.59
These racial differences in body fatness, however, are much smaller than is the mean (15%–20%) difference in percent body fat between nonobese and obese children. Therefore, the influence of race/ethnicity on the ability of high BMI levels to identify excess body fatness is uncertain. (An additional complication is that the association between child and adult BMI levels may differ according to race/ethnicity.) The interaction between BMI for age and race/ethnicity in the estimation of body fatness, however, suggests that it may be difficult to identify equivalent levels of body fatness by simply adjusting BMI for the average difference in body fatness across race/ethnicity groups.
CHILDHOOD BMI AND HEALTH RISKS
Cross-Sectional Associations With Cardiovascular Disease Risk Factors
Several reviews have focused on the short-term and long-term consequences of childhood obesity,19,60–62 and high BMI levels have consistently been found to be associated with cardiovascular disease risk factors such as insulin resistance, dyslipidemia, and increased blood pressure. The relation of BMI for age to the clustering of multiple risk factors was recently analyzed in a large (n ∼ 10 100) sample of children and adolescents from the Bogalusa Heart Study.26 On the basis of these associations, the proportion of children with at least 2 (of 5) risk factors increased from 5% (<25th CDC percentile) to 59% (≥99th CDC percentile) with increasing levels of BMI for age (Fig 4). Furthermore, the observed associations were markedly nonlinear, with substantial increases in the prevalence of multiple risk factors seen only at high levels of BMI for age. Of the 226 children who had a BMI for age at ≥99th CDC percentile, 84% had an adverse level of at least 1 risk factor.
Tracking of Childhood BMI Into Adulthood
An important criterion of the validity of childhood BMI is its relation to adult obesity, and almost all longitudinal studies have found that BMI levels track throughout life. Children with high BMI levels are more likely to become obese adults than are thinner children.60–64 Although some negative findings have been reported, these may be a result of the relative thinness of the examined children (eg, children born in the United Kingdom in 194765), and it is likely that these results are not applicable to the current levels of body fatness seen among US children. Differences in the BMIs of relatively thin children are likely to reflect differences in fat-free mass rather than fat mass.25
The strength of the association between levels of BMI in childhood and adulthood increases with childhood age,19 and positive predictive values of childhood obesity (≥95th CDC percentile) for adult obesity have ranged from ∼0.35 (BMI assessed at the age of 5 years) to 0.57 (assessed at the age of 15 years).66 These estimates, however, are strongly influenced by the prevalences of obesity among children and adults and were based on only 347 subjects. A much larger longitudinal study67 revealed that, even when childhood BMI was assessed between the ages of 6 and 8 years, the prevalence of adult obesity was >10 times greater (78% vs 6%) among obese (≥95th CDC percentile) children than among those with a BMI at ≤50th percentile. As assessed by the correlation between childhood BMI for age and adult BMI, the magnitude of the association was only slightly weaker among subjects who were initially examined at the ages of 6 to 8 years (r ∼ 0.6) than at the ages of 15 to 17 years (r ∼ 0.7).
Other characteristics such as the length of follow-up, race/ethnicity, parental fatness, birth weight, timing of sexual maturation, socioeconomic status, and physical activity may influence the tracking of childhood BMI.63 Although an early adiposity rebound has also been suggested to increase the risk for adult obesity, a young age at the BMI nadir may simply identify children whose BMI for age is high or moving upward.68 It is likely that equivalent information can more easily be obtained from a single BMI measurement at the age of 7 years.69
ADDITIONAL INFORMATION FROM SKINFOLD AND CIRCUMFERENCE MEASUREMENTS
Skinfold thicknesses have often been considered an attractive, noninvasive tool for estimating the amount of subcutaneous fat. Although a single measurement (or set of skinfolds) provides a more accurate estimate of body fatness than does BMI for age, relatively small differences in the relation of percent body fat to BMI versus skinfold thicknesses have been reported.17 Furthermore, skinfolds can be subject to large measurement errors, which makes their interpretation problematic if examiners have not been trained extensively.14 The optimal sites for skinfold-thickness measurements are also uncertain, and some sites require disrobing. These measurements have not been recommended as part of a routine examination,5 but it has been suggested23,70 that skinfolds may help determine if an overweight child may have excess body fatness.
Recent results, however, suggest that skinfold measurements may only slightly improve the estimation of body fatness among children who are obese (BMI ≥95th CDC percentile).37,71 For example, information on the triceps and subscapular skinfolds significantly improved the prediction of DEXA-calculated body fatness in the Pediatric Rosetta Study among all children, but there were only relatively small (8%) improvements among the obese children.71 Approximately 75% of children with a BMI at ≥95th CDC percentile have excess body fatness (Fig 2, Table 1), and this positive predictive value can be increased by simply using a higher BMI cut point. Other analyses indicate that BMI levels among children are as strongly correlated with various cardiovascular disease risk factors (lipid and insulin levels, blood pressure), as is the sum of the subscapular and triceps skinfold thicknesses.72 Although skinfold thicknesses may be useful in identifying nonobese children who truly have excess body fatness, the large measurement errors may preclude their widespread use.
Body-fat patterning influences the risk of type 2 diabetes and cardiovascular disease among adults, and abdominal obesity is associated with metabolic disorders among children and adolescents. However, because of the strong intercorrelation between fat patterning and the amount of body fatness, the additional information conveyed by body-fat distribution remains uncertain.
Several recent studies of children and adolescents focused on the waist-to-height ratio (WHtR), an index of abdominal obesity that seems to be (at least) as strongly correlated with various metabolic complications as is BMI.73,74 The use of WHtR also has the potential to simplify the assessment of obesity-related risk.75 Whereas a child's BMI must be expressed relative to his or her gender and age peers, it may be possible to use a single WHtR cut point (0.5) to identify children and adults at increased risk, because levels vary only slightly according to gender and age. It is also possible that the concept of a large waist relative to height, both of which are expressed in the same units, may be easier to explain to children and parents than is the division of weight (kg) by height (m)2 (or 703 times the division of pounds by inches squared).
There are, however, several limitations of WHtR: there have been few longitudinal studies on the relation of childhood WHtR to adult levels; numerous locations have been recommended for the measurement of the waist circumference15,76; and the measurement error for waist circumference is greater than those for weight and height.14,74 In addition, we have found that WHtR provides relatively little additional information (among all children) on risk-factor levels if BMI is already known.74
The accuracy of BMI as an indicator of adiposity varies substantially according to the degree of body fatness. Among relatively fat children, BMI is a good indicator of excess adiposity, but differences in the BMIs of relatively thin children (eg, BMI for age <85th CDC percentile) can be largely due to differences in fat-free mass. If appropriate cut points for both BMI and body fatness are selected, so that the prevalences of high levels of each characteristic are approximately equal, a BMI for age at ≥95th CDC percentile has moderately high (70%–80%) sensitivity and positive predictive value, along with high specificity (95%), for identifying children with excess body fatness. As compared with thinner children, overweight children (85th–94th CDC percentiles) are also more likely to have adverse levels of multiple risk factors and to become obese adults; their risks, however, are lower than those of obese children.
Skinfold-thickness and circumference measurements require additional training and may be difficult to standardize, and there is little information to support their use in the general assessment of body fatness and health risks. The additional information supplied by these measurements among overweight children, beyond that conveyed by BMI for age, is uncertain.
- Accepted April 29, 2009.
- Address correspondence to David S. Freedman, PhD, Centers for Disease Control and Prevention, 4770 Buford Hwy, Mailstop K-26, Atlanta, GA 30341-3717. E-mail:
Financial Disclosure: The authors have indicated they have no financial relationships relevant to this article to disclose.
- ↵Demerath EW, Schubert CM, Maynard LM, et al. Do changes in body mass index percentile reflect changes in body composition in children? Data from the Fels Longitudinal Study. Pediatrics.2006;117 (3). Available at:www.pediatrics.org/cgi/content/full/117/3/e487
- ↵Bray GA, DeLany JP, Volaufova J, Harsha DW, Champagne C. Prediction of body fat in 12-y-old African American and white children: evaluation of methods. Am J Clin Nutr.2002;76 (5):980– 990
- ↵Cole TJ, Bellizzi MC, Flegal KM, Dietz WH. Establishing a standard definition for child overweight and obesity worldwide: international survey. BMJ.2000;320 (7244):1240– 1243
- ↵Krebs NF, Himes JH, Jacobson D, Nicklas TA, Guilday P, Styne D. Assessment of child and adolescent overweight and obesity. Pediatrics.2007;120 (suppl 4):S193– S228
- ↵Kuczmarski RJ, Ogden CL, Grummer-Strawn LM, et al. CDC growth charts: United States. Adv Data.2000;(314):1– 27
- ↵Cole TJ, Rolland-Cachera MF. Measurement and definition. In: Lissau I, Burniat W, Poskitt EME, et al, eds. Child and Adolescent Obesity: Causes, Consequences, Prevention and Management. Cambridge, United Kingdom: Cambridge University Press; 2002:1– 27
- ↵Fields DA, Goran MI. Body composition techniques and the four-compartment model in children. J Appl Physiol.2000;89 (2):613– 620
- Wong WW, Hergenroeder AC, Stuff JE, Butte NF, Smith EO, Ellis KJ. Evaluating body fat in girls and female adolescents: advantages and disadvantages of dual-energy x-ray absorptiometry. Am J Clin Nutr.2002;76 (2):384– 389
- ↵Williams JE, Wells JC, Wilson CM, Haroun D, Lucas A, Fewtrell MS. Evaluation of Lunar Prodigy dual-energy x-ray absorptiometry for assessing body composition in healthy persons and patients by comparison with the criterion 4-component model. Am J Clin Nutr.2006;83 (5):1047– 1054
- ↵VanItallie TB, Yang MU, Heymsfield SB, Funk RC, Boileau RA. Height-normalized indices of the body's fat-free mass and fat mass: potentially useful indicators of nutritional status. Am J Clin Nutr.1990;52 (6):953– 959
- ↵Wells JC. A critique of the expression of paediatric body composition data. Arch Dis Child.2001;85 (1):67– 72
- ↵World Health Organization, Expert Committee on Physical Status. Physical Status: The Use and Interpretation of Anthropometry. Geneva, Switzerland: World Health Organization; 1995:420. WHO technical report 854
- ↵Wang J, Thornton JC, Bari S, et al. Comparisons of waist circumferences measured at 4 sites. Am J Clin Nutr.2003;77 (2):379– 384
- ↵Mei Z, Grummer-Strawn LM, Pietrobelli A, Goulding A, Goran MI, Dietz WH. Validity of body mass index compared with other body-composition screening indexes for the assessment of body fatness in children and adolescents. Am J Clin Nutr.2002;75 (6):978– 985
- ↵Whitlock EP, Williams SB, Gold R, Smith PR, Shipman SA. Screening and interventions for childhood overweight: a summary of evidence for the US Preventive Services Task Force. Pediatrics.2005;116 (1). Available at:www.pediatrics.org/cgi/content/full/116/1/e125
- ↵Ogden CL, Kuczmarski RJ, Flegal KM, et al. Centers for Disease Control and Prevention 2000 growth charts for the United States: improvements to the 1977 National Center for Health Statistics version. Pediatrics.2002;109 (1):45– 60
- ↵Himes JH, Dietz WH. Guidelines for overweight in adolescent preventive services: recommendations from an expert committee. The Expert Committee on Clinical Guidelines for Overweight in Adolescent Preventive Services. Am J Clin Nutr.1994;59 (2):307– 316
- ↵Kuczmarski RJ, Flegal KM. Criteria for definition of overweight in transition: background and recommendations for the United States. Am J Clin Nutr.2000;72 (5):1074– 1081
- ↵Flegal KM, Tabak CJ, Ogden CL. Overweight in children: definitions and interpretation. Health Educ Res.2006;21 (6):755– 760
- ↵Lohman TG, Going SB. Assessment of body composition and energy balance. In: Lamb D, Murray R, eds. Exercise, Nutrition, and the Control of Body Weight. Carmel, IN: Cooper Publishing Group; 1998:61– 99
- ↵Lazarus R, Baur L, Webb K, Blyth F. Body mass index in screening for adiposity in children and adolescents: systematic evaluation using receiver operating characteristic curves. Am J Clin Nutr.1996;63 (4):500– 506
- ↵Zimmermann MB, Gubeli C, Puntener C, Molinari L. Detection of overweight and obesity in a national sample of 6–12-y-old Swiss children: accuracy and validity of reference values for body mass index from the US Centers for Disease Control and Prevention and the International Obesity Taskforce. Am J Clin Nutr.2004;79 (5):838– 843
- ↵Mei Z, Grummer-Strawn LM, Wang J, et al. Do skinfold measurements provide additional information to body mass index in the assessment of body fatness among children and adolescents? Pediatrics.2007;119 (6). Available at:www.pediatrics.org/cgi/content/full/119/6/e1306
- ↵Taylor RW, Jones IE, Williams SM, Goulding A. Body fat percentages measured by dual-energy x-ray absorptiometry corresponding to recently recommended body mass index cutoffs for overweight and obesity in children and adolescents aged 3–18 y. Am J Clin Nutr.2002;76 (6):1416– 1421
- ↵Flegal KM, Shepherd JA, Looker AC, et al. Comparisons of percentage body fat, body mass index, waist circumference, and waist-stature ratio in adults. Am J Clin Nutr.2009;89 (2):500– 508
- ↵Freedman D, Wang J, Thornton J, et al. The classification of body fatness by BMI-for-age categories among children. Arch Pediatr Adolesc Med.2009; In press
- ↵Daniels SR, Khoury PR, Morrison JA. The utility of body mass index as a measure of body fatness in children and adolescents: differences by race and gender. Pediatrics.1997;99 (6):804– 807
- Bray GA, DeLany JP, Harsha DW, Volaufova J, Champagne CC. Evaluation of body fat in fatter and leaner 10-y-old African American and white children: the Baton Rouge Children's Study. Am J Clin Nutr.2001;73 (4):687– 702
- ↵Kerruish KP, O'Connor J, Humphries IR, et al. Body composition in adolescents with anorexia nervosa. Am J Clin Nutr.2002;75 (1):31– 37
- ↵Neovius MG, Linne YM, Barkeling BS, Rossner SO. Sensitivity and specificity of classification systems for fatness in adolescents. Am J Clin Nutr.2004;80 (3):597– 603
- ↵Fletcher RW, Fletcher SW. Clinical Epidemiology: The Essentials. 4th ed. Baltimore, MD: Lippincott, Williams & Williams; 2005
- ↵Wagner DR, Heyward VH. Measures of body composition in blacks and whites: a comparative review. Am J Clin Nutr.2000;71 (6):1392– 1402
- Reilly JJ, Methven E, McDowell ZC, et al. Health consequences of obesity. Arch Dis Child.2003;88 (9):748– 752
- ↵Wright CM, Parker L, Lamont D, Craft AW. Implications of childhood obesity for adult health: findings from thousand families cohort study. BMJ.2001;323 (7324):1280– 1284
- ↵Guo SS, Wu W, Chumlea WC, Roche AF. Predicting overweight and obesity in adulthood from body mass index values in childhood and adolescence. Am J Clin Nutr.2002;76 (3):653– 658
- ↵Freedman DS, Khan LK, Serdula MK, Dietz WH, Srinivasan SR, Berenson GS. The relation of childhood BMI to adult adiposity: the Bogalusa Heart Study. Pediatrics.2005;115 (1):22– 27
- ↵Barlow SE, Dietz WH. Obesity evaluation and treatment: Expert Committee recommendations. The Maternal and Child Health Bureau, Health Resources and Services Administration and the Department of Health and Human Services. Pediatrics.1998;102 (3). Available at:www.pediatrics.org/cgi/content/full/102/3/e29
- ↵Freedman DS, Katzmarzyk PT, Dietz WH, Srinivasan SR, Berenson GS. Relation of body mass index and skinfold thicknesses to cardiovascular disease risk factors in children: the Bogalusa Heart Study. Am J Clin Nutr.2009;90 (1):210– 216
- ↵Freedman DS, Kahn HS, Mei Z, et al. Relation of body mass index and waist-to-height ratio to cardiovascular disease risk factors in children and adolescents: the Bogalusa Heart Study. Am J Clin Nutr.2007;86 (1):33– 40
- Copyright © 2009 by the American Academy of Pediatrics