Body Mass Index in Childhood— Measuring a Moving Target

Abstract

The perplexing matter of the measurement of a moving target, relative fatness in children and adolescents, is addressed in this issue by Lindsay et al. (1) from the NIH population study of diabetes in the Gila River Indian Community in Arizona. They conclude from their analysis of over 900 Pima children, ages 5-20 yr, that body mass index (BMI) is a good measure of adiposity, and that BMI and dual-energy x-ray absorptiometry (DXA)-derived fat variables (percent body fat and fat mass) have broadly similar relationships to cardiovascular risk factors. The relationship of BMI to DXA variables in their report is in agreement with earlier studies in African American and Caucasian children (2, 3). The increased prevalence of childhood overweight and obesity, as well as the recognition that members of minority groups are at particular risk for morbidities associated with obesity, make investigation of measures and metabolic implications of adiposity in specific ethnic groups, such as in this study, especially important (4, 5). BMI is readily obtained in clinical and field settings. It has established validity in the definition of adult obesity by health outcome, although its association with more precise measures of fat also varies with age in adults (6, 7). The normal changes in BMI with growth and maturation, generally reflecting increase in fat free mass more than fat mass, complicate the use of absolute values for BMI as a measure of fatness in children (2, 8). In 1982, Rolland-Cachera et al. (9) compared various measures of weight for height and, with qualifications, identified BMI as a reasonable index of pedi-atric adiposity and proposed reference values for BMI by age and sex for French children. Other national reference standards have also been published, most recently from the National Center for Health Statistics as BMI percentile charts (10). An expert committee recommended guidelines for screening overweight adolescents based on BMI percentiles for age and sex ("at risk for overweight" 85%, "over-weight" 95%) (11). These guidelines had low false positive rates, but also low sensitivity for detecting a high percentage of body fat (12, 13). The International Obesity Task Force extrapolated the WHO standards for adult obesity (over-weight, 25 kg/m 2 ; obese, 30 kg/m 2) to children and adolescents 6-18 yr of age to establish an international definition of pediatric obesity (14). Because these two definitions of pediatric overweight are not based on health outcome and remain controversial, investigations of the cardiovascular risk associations of BMI and other measures of fatness, such as that presented by Lindsay et al. (1), are important contributions. But can the conclusions about the associations of BMI as a continuous variable to DXA measures and to cardiovas-cular risk factors from this study be applied to the general population? As noted by the authors, the Pima Indians of Arizona suffer unprecedented high levels of obesity in childhood and later life. The mean BMI values for this study population were "at risk for overweight" (5-9 yr), "over-weight" (10-14 yr), and "obese" (15-19 yr) by both the BMI percentile and international definitions; thus, the associations with BMI are all based on comparisons within a predominantly overweight population. Unquestionably, this information is important for clinicians evaluating obese children; BMI has now been shown to correlate with cardio-vascular risk factors in groups of overweight children from three backgrounds in the United States (African American, Caucasian, and Native American; Refs. 15 and 16). However, the observed associations of BMI and DXA variables with cardiovascular risk may only be applicable to subjects with high BMI for age, not to subjects across all BMI ranges. What does BMI measure in children and adolescents? The components of BMI (body mass/stature 2) are fat mass/ stature 2 and fat free mass/stature 2. Pediatric studies, including that of Lindsay et al. (1), show a strong correlation between BMI and fat mass, but there is also correlation between BMI and fat free mass (8). Whether fat or fat free mass is the dominant determinant of BMI, or the driving force for increase in BMI, depends on age, sex, pubertal stage, and ethnicity (2, 3, 8, 17). The stronger relationship of BMI to fat mass than to percent body fat observed in the current study by Lindsay et al. (1) is consistent with other studies. All reports that include a measure of percent body fat demonstrate that, for a given BMI, percent body fat is lower in boys than girls, lower with increasing age and more specifically with more advanced puberty, and lower in African Ameri-cans than Caucasians. What a given BMI measures depends on the body composition of the subject, and for a given BMI there is a range of percent body fat (3). This is why use of BMI in childhood is an attempt to measure a moving target-weight/stature 2-that increases and also changes in composition with normal growth. Nonetheless, it is highly correlated with specific measures of adiposity in children, possibly more so in overweight subjects, and, as now demonstrated by Lindsay et al. (1), in Pima Indians in addition to African Americans and Caucasians (2, 3, 8, 13). Lindsay et al. (1) did not explore the strength of the relationship of BMI and DXA fat variables to cardiovascular risk factors in detail. The current high prevalence of pediatric obesity, its consequences during childhood, and its implications for adulthood make extension of this analysis the next task, with the goal of establishing health risk-based definitions of pediatric obesity (4, 5). This is imperative for clinical care and research and needs to be done in a population with a representative range of adiposity. An example is the recently proposed cutoffs for both percent body fat by DXA Abbreviations: BMI, Body mass index; DXA, dual-energy x-ray ab-sorptiometry.