- Top of page
- Limitations and Strengths
J Clin Hypertens(Greenwich). 2010;12:636–644. © 2010 Wiley Periodicals, Inc.
The objective of this study was to determine independent and joint association of body mass index (BMI) percentile and leisure time physical activity (LTPA) with continuous metabolic syndrome (cMetS) risk score in 12- to 17-year-old American children. The 2003 to 2004 US National Health and Nutrition Examination Survey data were used for this investigation. LTPA was determined by self-report. cMetS risk score was calculated using standardized residuals of arterial blood pressure, triglycerides, glucose, waist circumference, and high-density lipoprotein cholesterol. Multiple linear regression analysis was used to evaluate association of BMI percentile and LTPA with cMetS risk score, adjusting for confounders. Increased BMI percentile and LTPA were each associated with increased and decreased cMetS risk score, respectively ((P<.01). There was a gradient of increasing cMetS risk score by BMI percentile cutpoints, from healthy weight (−0.77) to overweight (3.43) and obesity (6.40) ((P<.05). A gradient of decreasing cMetS risk score from sedentary (0.88) to moderate (0.17) and vigorous (−0.42) LTPA levels was also observed (P<.01). The result of this study suggests that promoting LTPA at all levels of weight status may help to reverse the increasing trends of metabolic syndrome in US children.
The metabolic syndrome (MetS) is characterized by a constellation of metabolic factors that confers an increased risk for cardiovascular disease–associated morbidity and mortality as well as all-cause mortality.1–3 In adults, the core variables in MetS are well defined and include central obesity, elevated blood pressure (BP), dyslipidemia (elevated triglycerides and low levels of high-density lipoprotein cholesterol [HDL-C]) and hyperglycemia.1–3
In children, MetS is poorly defined, and in the United States, the prevalence of MetS in 12- to 19-year-old adolescents (using a modified adult definition) is 4%, and an increasing trend has been noted.4–6 The association of MetS with obesity and other factors has been studied extensively in adults. However, the association of MetS with obesity among children remains poorly understood. Because of the increasing prevalence of obesity and MetS in US children,6,7 and the link of the two comorbidities with risks of cardiovascular diseases, diabetes, and other chronic diseases, understanding the epidemiology of MetS in children is imperative in developing strategies for forestalling its increasing prevalence in the United States. Indeed, some studies have shown that MetS and components of MetS track from childhood to adulthood.8,9 In addition, cohort studies have shown that a pronounced increase in body mass index (BMI) during adolescence persists into adulthood and that overweight in early childhood is associated with a greater risk of MetS in adulthood.10
One fundamental methodologic pitfall in the study of the association of MetS in children is the lack of unanimity in the definition of MetS in this population. The lack of consensus in the definition may be responsible for varying findings across epidemiologic studies.4,5,7 Due to the low prevalence of MetS in children,4,5 a large sample size is necessary for association studies using clinical cutpoints of factors that have been proposed for adults. Using the current prevalence estimate of 4% in adolescents, 40 patients would meet the criteria for MetS in a random sample of 1000 US children. Thus, the ability to determine association between MetS (defined as a categoric variable) and risk factors (eg, diet, physical activity) using discriminate analysis or multivariate logistic regression analysis is limited. The American Diabetes Association and the European Association for the Study of Diabetes has consequently recommended using a continuous value of MetS risk score for investigating the association of MetS with potential risk factors in children and adolescents.11
In children, the relative importance of the joint occurrence of obesity and leisure time physical inactivity in the development of MetS is unclear. However, in adults, some evidence suggests that MetS is modified by physical activity.12,13 In this study, MetS is defined using a continuous metabolic syndrome (cMetS) risk score in a representative sample of 12- to 17-year-old American children. The objective is to clarify the association of BMI percentile and leisure time physical activity (LTPA) with MetS in children. We hypothesize that increase in BMI percentile and LTPA is associated with high and low cMetS risk scores, respectively. We also hypothesize that LTPA will attenuate the association of BMI percentile with cMetS risk score in 12- to 17-year-old American children.
- Top of page
- Limitations and Strengths
The basic characteristics of eligible children (n=655) by levels of BMI percentile are presented in Table I. Overall, 32.1%, 44.8%, and 23.1% of participants reported vigorous, moderate, and sedentary LTPA, respectively. There were statistically significant differences between healthy-weight, overweight, and obese children in age, weight, height, waist circumference, and educational attainment. Analysis of post hoc tests indicated pair-wise differences between healthy-weight, overweight, and obese children with respect to these variables (P<.05). Obese children were older, heavier, taller, and had larger waist circumference compared with healthy-weight and overweight children (P<.01). Distribution of race/ethnicity, sex, and annual household income did not vary across levels of BMI percentile.
Table I. Basic Anthropometric and Demographic Characteristics of Studied Population
|Variables||Healthy Weight||Overweight||Obese||P Value|
|No. (BMI percentile)||466 (<85)||107 (85–94.9)||82 (≥95)|| |
|Waist circumference, cm||73.0±6.9a||90.6±6.9b||107.5±12.2c||<.001|
|Race/ethnicity, %|| || || ||.500|
| Non-Hispanic white||26.0||25.2||24.4|| |
| Non-Hispanic black||43.3||37.4||48.8|| |
| Mexican American||30.7||37.4||26.8|| |
|Education, %|| || || || .001|
| Elementary school||91.5||86.5||81.5|| |
| Middle school+||8.5||13.5||18.5|| |
|Sex, %|| || || ||.667|
| Male||56.4||55.1||51.2|| |
| Female||43.6||44.9||48.8|| |
|Annual household income, %|| || || || .675|
| <$25,000||38.7||38.4||32.5|| |
| $25,000–$74,000||40.7||41.1||46.3|| |
| >$75,000||20.6||20.5||21.2|| |
|Leisure time physical activity, %|| || || || .002|
| Sedentary||8.2||20.7||60.5|| |
| Moderate||31.8||47.1||20.1|| |
| Vigorous||60.0||32.2||19.4|| |
Table II shows average standardized scores of individual components of cMetS according to BMI percentile cutpoints. Participants who were obese presented with higher standardized scores of MAP, triglycerides, glucose, waist circumference, and HDL-C compared with healthy-weight and overweight children (P<.01).
Table II. Mean Values of Components of Continuous Metabolic Syndrome Score in US Children Stratified by Adiposity Level
|Healthy Weight||Overweight||Obese||P Value|
cMetS risk score stratified by BMI percentile and LTPA levels are presented in the Figure. Test of linear trend showed a gradient of increasing cMetS risk score by BMI percentile cutpoints, from healthy weight, to overweight, and obesity (P<.05). The mean cMetS risk scores were −0.77, 3.45, and 6.40 for healthy weight, overweight, and obese children, respectively. A gradient of increasing cMetS risk score from sedentary, to moderate, to vigorous LTPA levels was also observed (P<.001). The mean cMetS risk scores were 0.88, 0.17, and −0.42 for sedentary LTPA, moderate, and vigorous LTPAs, respectively. Assessment of cMetS risk scores by BMI percentile cutpoints across LTPA levels showed that vigorous LTPA was associated with significantly lower cMetS risk score at each level of BMI percentile defined by healthy weight, overweight, and obesity. In healthy-weight children, mean cMetS risk scores were −0.45, −0.63, and −0.96, for sedentary, moderate, and vigorous physical activities, respectively. The corresponding values for overweight children were 3.87, 3.33, and 1.87, respectively. A similar gradient of decreasing mean cMetS risk score with increasing LTPA was also observed in obese children, with values of 5.73, 5.12, and 4.67 for children reporting sedentary, moderate, and vigorous LTPAs, respectively.
Figure Figure. Continuous metabolic syndrome (CMetS) risk score by (A) body mass index (BMI) percentile cut points, (B) leisure time physical activity levels, and (C) across BMI percentile by leisure time physical activity levels.
Download figure to PowerPoint
In order to determine whether BMI percentile was associated with cMetS risk score and components of cMetS independent of other factors, BMI percentile, race/ethnicity household income, and child education were tested in multiple linear regression models using cMetS risk score and each component of cMetS as dependent variables. We compared model I and model II, representing unadjusted and LTPA-adjusted models, respectively (Table III). In both models, BMI percentile was positively associated with cMetS risk score and components of cMetS, independent of other variables (P<.01). Comparative analysis of model I and model II indicated that adjustment for LTPA significantly attenuated the association of BMI percentile with cMetS risk score as well as components of cMetS. There was a negative interaction between BMI percentile and LTPA (model III) (P<.01).
Table III. Effect of LTP Activity in the Association of BMI Percentile with Standardized Components of a Continuous Metabolic Syndrome Score in US Children
|Model 1: unadjusted for LTP activity|
| BMI percentile||0.179a||0.257a||0.628a||0.922a||0.289a||0.691a|
| Household income||0.049||0.068||0.068b||0.080||0.024||0.068b|
| Child education||−0.111a||0.060||0.046||0.040||0.030||−0.084|
|Model 2: adjusted for LTP activity|
| BMI percentile||0.162a||0.262a||0.619a||0.912a||0.271a||0.687a|
| Household income||0.050||0.070||0.069b||0.080||0.020||0.081b|
| Child education||−0.102b||0.020||0.040||0.030||−0.031||−0.041|
| LTP activity||−0.152a||−0.165a||−0.162a||−0.192a||−0.157a||−0.151a|
|Model 3: adjusted for LTP activity, and BMI×LTP activity|
| BMI percentile||−0.157a||0.347a||0.449a||0.851a||0.249b||0.567a|
| Household income||0.049||0.071||0.057b||0.060||0.024||0.070|
| Child education||−0.104a||0.008||0.068b||0.090||0.050||−0.016|
| Physical activity||−0.339b||−0.168a||−0.298b||−0.111b||−0.073||−0.130b|
| BMI percentile×LTP activity||−0.382b||−0.118b||−0.280b||−0.126b||−0.056||−0.227a|
Finally, to determine the effect of LTPA on the associations of BMI percentile with cMetS risk score, we performed LTPA level–specific multivariable regression analyses (Table IV). As shown, BMI percentile was positively associated with increase in cMetS risk score adjusting for race/ethnicity, household income, and child education in participants with sedentary, moderate, and vigorous LTPAs (P<.01). Adjusting for race/ethnicity, household income, and education, a 1-kg/m2 increase in BMI percentile was associated with 0.701, 0.691, and 0.652 increases in cMetS risk score in children reporting sedentary, moderate, and vigorous LTPAs, respectively (P<.01). The results of the regression model showed that the proportion of the variance of cMetS risk score explained by the model that included BMI percentile, age, race/ethnicity, household income, and child education were 46.4%, 48.8%, 59.1% in children reporting sedentary, moderate, and vigorous leisure physical activities, respectively.
Table IV. Association of BMI Percentile With a Continuous Metabolic Syndrome Across Leisure Time Physical Activity Levels in US Children
|Variables||Leisure Time Physical Activity Intensity Levels|
|Model summary (R2)||0.464||0.488||0.591|
- Top of page
- Limitations and Strengths
The role of LTPA in the association between BMI and MetS in children has not been well defined. To our knowledge, we are unaware of published studies that examined the association of BMI and LTPA with MetS using a cMetS risk score in a nationally representative sample of American children. The emphasis of this study is on the role of LTPA in modifying the association between the BMI and cMetS.
Although it is well-known that BMI is related to MetS in youth,10 many studies evaluating the relationship between BMI and MetS in children have methodological problems, including lack of unified definition of MetS and the use of adult clinical cutpoints for the definition of childhood MetS. Increasing evidence supports the use of continuous values such as cMetS risk score, instead of the binary classification for epidemiologic analyses.11,18,19 As a continuous variable, cMetS risk score is a more robust measure of MetS than a categoric measure, with these advantages: (1) since cardiovascular risk is a progressive function of several MetS risk factors, the use eliminates the need to dichotomize these factors,21,22 (2) cMetS risk score is more sensitive and less error prone compared with categoric measures of MetS, and (3) statistical power is increased with the use of cMetS.21 The use of cMetS risk score has been validated in adults against MetS defined using International Diabetes Federation guidelines.23 However, we do not yet know which cMetS component confers increased risk of cardiovascular disease in youth.
The use of NHANES for this study is appropriate because the sampling scheme is representative of the national population of non-Hispanic whites, non-Hispanic blacks, and Mexican Americans. The training program and quality-control procedures instituted in the surveys give additional credibility to the data.
The result of this study clearly demonstrates significant independent and combined influences of BMI percentile and LTPA in cMetS in children. The results of this study indicate that BMI percentile is positively associated with elevated cMetS risk score in American children, controlling for concomitant variables including race/ethnicity, household income, and child education. The results of this study show a gradient of increasing cMetS risk score by increasing BMI percentile. The mean cMetS risk scores increased as BMI percentile category increased by more than 4 times for healthy-weight compared with obese children. The mean cMetS risk scores decreased by more than 4 times as LTPA increased from sedentary to vigorous LTPAs. Sedentary obese children had much higher mean cMetS risk scores compared with sedentary overweight and sedentary healthy-weight children. A similar gradient was observed for both moderate and vigorous leisure time physically active obese, overweight, and healthy-weight children.
The result of this study also shows that LTPA modifies the relationship between BMI percentile and CMetS. LTPA was inversely associated with cMetS risk score. Indeed, vigorous LTPA attenuated cMetS risk score within BMI percentile categories. Our finding of an inverse association between LTPA and cMetS score is consistent with findings of other investigators using maximum cycle-ergometer and aerobic fitness as measures of physical activity.24,25 Findings from the European Youth Heart Study26 also showed a similar inverse relationship between fitness and the clustering of cardiovascular disease risk factors in children.
Another important finding of this study is that at a fixed level of BMI percentile, increase in LTPA was associated with reduced cMetS risk score as well as each component of cMetS (P<.01). Overall, levels of LTPA (moderate and vigorous) were associated with a decreased association of BMI percentile with cMetS risk scores compared with sedentary LTPA. The difference was most prominent in the obese group. This finding is similar to the Québec Family Study in which adolescents with low BMI and high fitness levels were found to have the lowest MetS score, whereas those with high BMI and low fitness levels had the highest MetS score.27 The significant association of race/ethnicity with cMetS in this study is noteworthy and was driven by racial/ethnic differences in triglyceride level. Triglyceride levels in this study were much lower in non-Hispanic blacks compared with non-Hispanic whites and Mexican American participants.
Limitations and Strengths
- Top of page
- Limitations and Strengths
Some important limitations must be taken into account in the interpretation of the results from this study. One, this study did not differentiate between unfit healthy-weight, fit-fat, and unfit-fat participants. Several studies have shown that the association between cardiovascular risk factors vary by fat and fitness status.27–29 In those studies, fat/low-fit participants had the highest MetS risk scores while obese-unfit participants possessed the worst MetS risk score. Two, as a cross-sectional study, causal inferences must be made with caution. However, improvement in metabolic risk profile due to LTPA is biologically plausible. Three, in this study, physical activity status in children was based on self-report compared with more objective measures such as pedometers or accelerometers. Self-report of physical activity suffers from reporting bias attributable to social desirability and the cognitive problem associated with estimating frequency and duration of physical activity in children.30 However, a recent study of physical activity in 6- to 19-year-old US children and adolescents from the survey for this study indicates that subjective measures of physical activity give qualitatively similar results as objective measures using accelerometers.31 It is also worth mentioning that it would have been more appropriate if the volume of physical activity was used in this study since many studies report that health and function outcomes are related to the total moderate to vigorous physical activity. Estimating health status based on a minimum of 10 minutes of activity per month is less than ideal. In the interpretation of the role of physical activity, there is also the need to recognize differences between physical activity and fitness. Certainly these two variables influence each other and both can be modified. However, fitness (peak aerobic capacity) is determined by physical activity, genetics, and other factors. There are situations where exercise training and physical activity level are weakly related to peak oxygen uptake. Thus, the terms fitness and physical activity cannot be used interchangeably. Four, although we controlled for the confounding effect of race/ethnicity, income, and education, other unmeasured factors such as genetics and dietary factors could explain our findings.
The mechanism linking LTPA to cMetS is unclear but may be related to association of physical activity with fitness. Although physical activity and fitness are a separate construct that may have separate mechanisms for influencing MetS, they both have an effect on health and thus are interrelated in a reciprocal manner.32 Some studies have shown that glucose transport can be improved by physical fitness,33,34 while other studies in adults have shown that increased physical fitness can delay or prevent diabetes.35,36 Also it has been shown that physical fitness is associated with increased capillarization, leading to increased blood flow and oxygen supply to the muscle tissue and ultimately improved fat metabolism, higher HDL-C concentration, and decreased BP.29,37 Additionally, physical fitness is associated with a much improved overall sympathetic tone, leading to reduced BP through efficient use of the motor units in the muscle.38
In agreement with other studies,39,40 our results showed that the association of BMI percentile with MetS is stronger in sedentary children and may represent a subgroup of youths who carry a greater risk for cardiovascular disease and type 2 diabetes. Our findings of the effects of LTPA in the association of BMI percentile with cMetS risk score have clinical and public health significance. From a clinical standpoint, this information could be helpful to health care providers when determining treatment options for at-risk children. The results of this study suggest that children who are sedentary and obese are those in need of aggressive lifestyle modifications, and recommending increased LTPA may be a significant modifier of disease. Children should be encouraged to engage in healthy lifestyle habits, including LTPA and obtaining or maintaining a healthy weight in order to improve their metabolic health. Promoting LTPAs during childhood is a vital public health agenda in reducing the risk of clustering cardiovascular disease risk factors. Some studies have shown that physical inactivity may independently contribute to metabolic disease risk in youth independent of total body fat in children.38–40 Clear-cut claims about the effect of physical inactivity on MetS must be supported by fitness data.