Surveillance results from 2003 to 2006 indicated that 31.9% of US youth had a body mass index (BMI) ≥ 85th percentile (34). Despite recent data indicating a plateau in the previously upward trend in overweight and obesity in US youth, obesity still represents a significant health risk (33,34). Elevated BMI is related to an increased risk for type 2 diabetes, elevated cholesterol, and high blood pressure in youth (23,38). These findings, coupled with the fact that obesity has been shown to track into adulthood (44), place overweight and obese youth at a significant health risk as they progress into adulthood.
There is evidence of an inverse relationship between weight status and physical activity (PA) in youth (37). For example, BMI increases were associated with more television viewing and declines in moderate-to-vigorous PA (MVPA) in longitudinal studies (3,31). There is also evidence for a decline in PA with age, particularly in adolescent females (32,42). Energy intake may also be associated with weight status and PA (13,22). For example, energy intake was negatively correlated with fat mass index and was positively correlated with MVPA in a diverse sample of youth (21).
However, findings on race/ethnic differences in PA are less clear. Whereas a previous study has reported lower levels of PA (measured via self-report) in non-Hispanic black youth compared with their non-Hispanic white counterparts (39), another that measured PA via accelerometry has reported that non-Hispanic white youth are less physically active than other race/ethnic groups (35). The objective of this study was to report differences in PA by race/ethnicity, age, gender, and weight status in a nationally representative sample of US youth. Strengths of this analysis include the large sample size, the national representativeness of the data set, the objectively measured PA data, and the availability of data on total energy intake obtained from two 24-h recalls.
Data for this study are from the National Health and Nutrition Examination Survey (NHANES), a cross-sectional health interview survey representative of the US civilian, noninstitutionalized population. The NHANES is administered by the National Center for Health Statistics of the Centers for Disease Control and Prevention (CDC). Data are collected year-round using a complex, stratified, multistage probability cluster sampling design strategy. Details of sampling and data collection have been reported previously (10,42). Public-use data are released in 2-yr cycles. Accelerometer measurement was included in both the 2003-2004 and 2005-2006 survey cycles.
The present study focused on youth age 6-19 yr. The sample consisted of 1508 participants from the 2003-2004 cycle and 1598 participants from the 2005-2006 cycle. Preliminary data analysis indicated no statistically significant differences in the outcome variables of interest between the 2003-2004 and 2005-2006 samples. Therefore, the samples were combined. Participants were included if they had no missing demographic or anthropometric data and at least 4 d with 10 or more hours of accelerometer data (42). The final analytic sample consisted of 3106 participants between 6 and 19 yr with complete accelerometer, demographic, and anthropometric data. On the basis of survey design characteristics, results for non-Hispanic white, non-Hispanic black, and Mexican American race/ethnic groups were estimable. The National Center for Health Statistics ethics review board approved the study, and written informed consent was obtained from all participants. The University of Southern California institutional review board did not require review for this analysis.
Age (yr) was calculated as the time between birth and examination date. Race/ethnicity was self-reported by participants and categorized as non-Hispanic white, non-Hispanic black, Mexican American, other Hispanic, and other. Age and racial/ethnic groups were categorized according to the NHANES analytic guidelines (11). Socioeconomic status was calculated using the poverty-to-income ratio (PIR), which is a ratio of the household income to the Census Bureau poverty threshold that is adjusted for family size and updated annually to adjust for inflation.
Standing height (cm) and weight (kg) were used to calculate BMI (kg·m−2). BMI percentiles were defined for normal, overweight, and obese categories according to the age- and gender-specific growth charts published by the CDC (26). Normal weight is defined as between the 5th and <85th percentiles, overweight is defined as between the 85th and <95th percentiles, and obese is defined as ≥95th percentile.
All ambulatory participants older than 6 yr were asked to wear an ActiGraph (Model 7164; ActiGraph, LLC, Fort Walton Beach, FL) on the right side of the hip. The uniaxial accelerometer measures acceleration intensity as "counts" in response to body movement. Details on the accelerometer protocol, data reduction process, and the definition of wear time have been reported previously (30,42). Similar to previous studies using accelerometer data, a valid day was considered to be 10 or more hours of wear time, and participants with at least four valid days were included in the present study (7,42). Accelerometer data are presented here as (a) mean counts per minute and (b) as estimates of mean minutes per day in sedentary behavior, moderate, vigorous, and moderate-to-vigorous PA (MVPA). The thresholds for moderate (4 METs) and vigorous (7 METs) PA were age-adjusted using the criteria from Trost et al. (43) for participants between the ages of 6 and 17 yr. The thresholds for moderate (3 METs) and vigorous (6 METs) PA were used for participants older than 18 yr and are defined at 2020 and 5999 counts per minute, respectively (5,20,28,42,47). A sedentary behavior cut point of 100 counts per minute was applied to all age groups (30).
Diet was assessed by up to two interviewer-administered 24-h dietary recalls. The first recall was obtained in person; the second (if available) was obtained via telephone. Measurement tools such as food workbooks, bowls, glasses, and plastic foods were used to help participants give more accurate estimates of intake (12). Proxy interviews were conducted for children younger than 9 yr, and children age 9-11 yr were permitted to receive assistance from an adult. Participants were included in the dietary analyses if they had at least 1 d of valid diet data. There were 131 participants excluded because of unreliable or missing dietary data.
All analyses were conducted in SAS 9.1 (SAS Institute, Inc., Cary, NC) using specialized procedures to account for the sample weighting and complex multistage probability design. The NHANES-provided sample weights were adjusted to account for the combination of data from 2003-2004 and 2005-2006 and for the subsampling of individuals with at least four valid days of accelerometer data. Time spent in each activity level (i.e., moderate, vigorous, and MVPA) was calculated per person by summing each minute with a count above the threshold for that activity level. Per-person mean counts per minute were calculated by dividing the sum of counts each day by the wear time minutes each day across all valid days. Population mean estimates and associated SE of minutes and counts, adjusted for categorized PIR, were calculated with the SAS SURVEYREG procedure using the Taylor series linearization method (45). Planned comparisons of mean counts per minute between each data collection year and between subpopulations were made with pairwise contrasts.
To further identify differences in mean minutes spent in MVPA within subgroups, data were stratified by gender, and a three-way race/ethnic-age group-BMI percentile interaction was examined using the SURVEYREG procedure (adjusting for categorized PIR). Linear regression using age, BMI percentile, and PIR as continuous variables was conducted to understand the magnitude of the effect of each variable on mean minutes per day spent in MVPA. To predict the odds of reaching the 2008 PA guidelines of 60 min·d−1 of MVPA for youth aged 6-17 yr or 21 min·d−1 for older youth aged 18-19 yr (on the basis of the recommendation that adults perform 150 min·wk−1 MVPA or an average of 21 min·d−1) (44), logistic regression was conducted using the SURVEYLOGISTIC procedure adjusting for categorized PIR. Significance was set at α = 0.05 and df = 30, which is the number of primary sampling units (11).
Additional analyses were conducted to examine whether dietary intake might confound the relationship between activity levels and weight status. Crude assessment was performed by adding individual mean reported energy intake as an additional covariate to the SURVEYREG models of accelerometer minutes on weight status, race/ethnicity, and categorized PIR. Because the distribution of mean energy intake was right-skewed, the same models were run using individual averages of Box-Cox transformed energy intakes. The Box-Cox parameter value of 0.21 (the fifth root) approximately normalized the distribution of the residuals of energy intake regressed on weight status, race/ethnicity, and categorized PIR.
Recognizing that the mean of one or two 24-h recalls does not reflect long-term intake (16), a full-regression calibration approach (9) was used to fit regression models of accelerometer minutes on weight status, race/ethnicity, categorized PIR, and predicted long-term average transformed energy intake. The calibration model for long-term average transformed intake included weight status, race/ethnicity, categorized PIR, 24-h recall sequence (in person vs by telephone), and recall day (weekend vs weekday). Both recall sequence and day assist with modeling long-term energy intake but were not required in the models for accelerometer minutes because the PA data were not modeled on the day level. Because the regression calibration approach required an additional modeling step to predict energy intake, the SURVEYREG procedure could not be used to obtain correct SE or significance tests. Therefore, SE for the calibrated models were computed using balanced repeated replication (25), with 32 replicate weight sets poststratified by age, gender, and race/ethnicity on the basis of the original analytic sample weights. The perturbation factor was f = 0.7 (25).
The characteristics of the samples are presented in Table 1 by race/ethnicity and gender. Of the 5687 participants with accelerometer data, 3698 met the inclusion criteria of having 4 d (10 h·d−1) of valid data. Of these, 99 underweight youth were excluded, 453 had invalid demographic data, and 40 had invalid anthropometric data. The participants with invalid PA data significantly differed from those with valid PA data on demographic characteristics; however, the sample was reweighted on the basis of the inclusion criteria of four or more 10-h days of valid data. The analytical sample (n = 3106) was 49.5% female, 69.9% non-Hispanic white, 16.6% non-Hispanic black, and 13.5% Mexican American. Similar to previous studies using NHANES data (34), 19.2% of youth aged 6-19 yr were obese and 17.0% were overweight. The highest prevalence of obesity by race/ethnic group and gender was in non-Hispanic black females, Mexican American males, and Mexican American females. Obesity prevalence was highest in 16- to 19-yr-olds and lowest in 6- to 11-yr-olds. For the analyses including total energy intake, 2975 had one or more days of valid dietary data. Irrespective of form (individual means or calibrated), energy intake was not statistically significant in the regression equations for accelerometer minutes. The energy-adjusted mean minutes spent in each activity level by weight status differed by <4% from the unadjusted means (data not shown).
Mean Activity Counts per Minute
Non-Hispanic white youth recorded fewer mean counts per minute than non-Hispanic black and Mexican American youth, and counts per minute consistently declined with age. Youth in the 6- to 11-yr-old group demonstrated higher mean counts per minute than youth in both the 12- to 15- and 16- to 19-yr-old groups (P < 0.001 for both), and youth in the 12- to 15-yr-old group had higher mean counts per minute than youth in the 16- to 19-yr-old group (P < 0.001; Table 2). Females always had lower mean counts per minute than males independent of race/ethnicity and weight status (P < 0.001). Overweight non-Hispanic black males recorded 67 more counts per minute per day than overweight Mexican American males (P = 0.037) and 104 more counts per minute per day than overweight non-Hispanic white males (P = 0.002; Table 3). Normal-weight youth of all race/ethnic groups had, on average, 87 more counts per minute than obese youth (P < 0.001). This difference was constant across all race/ethnic groups, with the exception of overweight and obese Mexican American females who recorded similar counts per minute.
Mean Minutes in Activity Levels
The 6- to 11-yr-old group recorded 88 min of MVPA per day, whereas youth in the 12- to 15- and 16- to 19-yr-old groups recorded 33 and 26 min of MVPA per day, respectively (Table 4). The 6- to 11-yr-old group also spent fewer minutes in sedentary behavior and more minutes in moderate PA and vigorous PA per day than the 12- to 15- and the 16- to 19-yr-old groups across all race/ethnic groups (P < 0.001 for both). Non-Hispanic black 16- to 19-yr-olds were the most inactive race/ethnic subgroup, spending 520 min·d−1 in sedentary behavior. Non-Hispanic white youth spent fewer minutes in vigorous PA than both non-Hispanic black and Mexican American youth (P < 0.001 and P = 0.004, respectively; Table 5). Females spent fewer minutes in MVPA than males (P < 0.001) and 20 more minutes per day in sedentary behavior than males (P < 0.001). Normal-weight youth spent more minutes in moderate PA, vigorous PA, and MVPA (P < 0.001 for all) than obese youth (Table 6). Normal-weight non-Hispanic white youth spent 34 fewer minutes in sedentary behavior per day than normal-weight non-Hispanic black youth (P = 0.001). Also, the non-Hispanic white youth were the only race/ethnic subgroup with differences in sedentary behavior between the normal weight and overweight and obese groups (P = 0.008 and P = 0.026, respectively). Multivariable linear regression with continuous variables indicated that age was inversely associated with MVPA in all race/ethnic groups for both genders (P < 0.001 for all; Table 7). Similarly, BMI percentile was negatively associated with MVPA in all race/ethnic groups for both genders (P < 0.050 for all); however, the effect was not as strong as age. Age, BMI percentile, and PIR accounted for 44% of the variance in males and 49% of the variance in females.
There was a statistically significant three-way interaction for mean minutes per day spent in MVPA between age group, BMI percentile category, and race/ethnic group for both males and females (P < 0.001; Figs. 1 and 2). In the 12- to 15-yr-old Mexican American race/ethnic group, male and female overweight youth spent significantly fewer minutes in MVPA than normal-weight youth (P = 0.022 and P = 0.027, respectively); however, this consistent difference was not seen within the other race/ethnic groups.
Race/Ethnic PA Differences
Table 4 presents differences in PA levels by age group. Non-Hispanic white 6- to 11- (P = 0.018) and 12- to 15-yr-olds (P = 0.005) spent fewer minutes in MVPA than non-Hispanic black youth of the same age groups. Non-Hispanic black 6- to 11-yr-olds spent 12 more minutes per day in MVPA than Mexican American 6- to 11-yr-olds (P = 0.007). Non-Hispanic black 12- to 15-yr-olds spent 25 more minutes in sedentary behavior than non-Hispanic white 12- to 15-yr-olds (P < 0.001), whereas non-Hispanic black 6- to 11-yr-olds spent three more minutes in vigorous PA than Mexican American 6- to 11-yr-olds (P = 0.007). Although differences in PA were seen in younger age groups, PA levels declined in the oldest age group, so that youth aged 16-19 yr of all race/ethnic groups spent the same amount of time (between 24 and 29 min) in MVPA (P > 0.050).
Table 5 presents differences in activity levels by gender. In non-Hispanic white and Mexican American youth, females spent approximately 20 more minutes per day in sedentary behavior than males (P = 0.002 and P = 0.029, respectively). Non-Hispanic white males spent fewer minutes per day in vigorous PA than both non-Hispanic black (P < 0.001) and Mexican American males (P = 0.004). With the exception of non-Hispanic black youth, females of all race/ethnic groups spent significantly more time in sedentary behavior (P < 0.050) and less time in PA behavior (P < 0.050) than males.
By BMI percentile.
Table 6 presents the PA differences by weight status. Normal-weight and overweight non-Hispanic whites spent 9 and 18 fewer minutes in MVPA per day than normal-weight and overweight non-Hispanic blacks (P = 0.043 and P = 0.001, respectively). There were no race/ethnic differences within the obese group for any activity level. However, the only differences in minutes spent in moderate PA, vigorous PA, and MVPA between the overweight and obese groups were seen in the non-Hispanic black race/ethnic group, where the overweight group recorded more minutes in each of the activity levels than obese youth (P < 0.001 for all activity levels).
Age-gender-BMI interaction effects within race/ethnic groups.
In 6- to 11- and 16- to 19-yr-old Mexican American males and females, obese youth spent fewer minutes in MVPA than normal-weight youth (P < 0.016), but in the 12- to 15-yr-old Mexican American males and females, obese and normal-weight youth spent statistically equivalent amounts of time in MVPA per day (P > 0.379; Figs. 1 and 2). In 6- to 11-yr-old non-Hispanic black males and females, obese youth recorded less time in MVPA than both normal-weight and overweight youth. In non-Hispanic white females, 6- to 11- and 12- to 15-yr-old normal-weight youth recorded more minutes in MVPA per day than obese youth (P < 0.019).
Age-gender-BMI interaction effects between race/ethnic groups.
In 6- to 11-yr-old females, normal-weight youth spent more minutes per day in MVPA than obese youth in all race/ethnic groups (P < 0.050; Figs. 1 and 2). However, in 6- to 11-yr-old males, this difference was seen in the non-Hispanic black (P < 0.001) and Mexican American (P < 0.001) race/ethnic groups but not in the non-Hispanic white race/ethnic group. Most of the interaction effects were seen in the youngest age group. Within the 6- to 11-yr-old females, the non-Hispanic white and Mexican American race/ethnic groups had the greatest differences in MVPA between the normal-weight and obese groups; however, the non-Hispanic black race/ethnic group had the greatest differences in MVPA between the overweight and obese groups.
Meeting PA Guidelines
Table 8 shows the results from the logistic regression analysis to determine which groups were more likely to meet the 2008 PA guidelines. In this sample, 41.4% met the recommendations. Females were less likely to adhere to the 2008 PA guidelines than males (odds ratio = 0.41, 95% confidence interval = 0.30-0.54). For males, those most likely to meet the 2008 PA guidelines were non-Hispanic black, aged 6-11 yr, and of normal weight. Because females were found to be profoundly inactive, race/ethnicity did not predict whether they would meet the guidelines. Females in the older age groups who were obese and had a higher poverty level were less likely to meet the guidelines.
On the basis of the objectively measured levels of PA, non-Hispanic white youth were the least active race/ethnic group and non-Hispanic black youth were the most active race/ethnic group. Previous findings using self-report measures of PA in US samples showed that non-Hispanic black youth were the least physically active race/ethnic group (1). However, using accelerometers to measure PA, Owen et al. (35) found that blacks recorded four more minutes per day in MVPA than whites did. In our sample, non-Hispanic black youth spent about eight more minutes per day in MVPA than non-Hispanic white youth.
Troiano et al. (42) previously reported an age-related decline in PA in US youth on the basis of the results from NHANES 2003-2004. For the 2003-2004 and 2005-2006 survey combined samples, we also found an age-related decline in PA. Pate et al. (35a) reported a 4% annual decline in MVPA in adolescent girls (35). Brodersen et al. (6) found that PA dropped off between ages 11 and 12 yr. In our sample, 6- to 11-yr-olds participated in twice as much MVPA than the older age groups, consistent with the observation that the most dramatic age-related decline in PA may occur at the start of puberty.
To better understand the age-related decline in MVPA observed in this sample, the three-way interaction among age, BMI category, and race/ethnic group was examined for each gender. Non-Hispanic whites and blacks in higher BMI categories spent less time in MVPA than normal-weight non-Hispanic white and non-Hispanic black youth, whereas obese 12- to 15-yr-old Mexican American youth recorded the same amount of time in MVPA per day as normal-weight youth. A previous study in adolescent Hispanic females found a trend for normal-weight Mexican American females spending fewer minutes in MVPA than overweight females (8). This suggests that BMI may interact with PA levels differently in Mexican American youth than in other race/ethnic groups. In the non-Hispanic white race/ethnic group, the largest difference in MVPA is between the normal-weight and overweight groups, with the overweight and obese groups participating in approximately the same amount of MVPA. However, in the non-Hispanic black race/ethnic group, the largest difference in MVPA is between the overweight and obese youth. Thus, the findings indicate that different race/ethnicities have different thresholds of BMI percentile past which MVPA declines.
The large difference in PA between males and females is particularly striking. Normal-weight females of all race/ethnic groups achieved less PA than obese males of all race/ethnic groups. A recent study conducted in adolescents using accelerometers found that, although the rates of decline of PA were the same for both genders, females participated in significantly fewer minutes of MVPA than males and MVPA dropped below 60 min·d−1 1 yr earlier in females than in males (32). Non-Hispanic black youth had the largest gender differences in counts per minute and minutes spent in MVPA: females recorded about 140 counts per minute and 27 min·d−1 less than males. Using previous prediction equations on the basis of overweight youth (18), this deficit is broadly similar to 600 kcal·d−1. The difference in PA levels between males and females in this sample may contribute to the fact that the non-Hispanic black females had the highest prevalence of obesity.
Overall, the inverse association between PA levels and BMI percentile in this sample is consistent with previous findings (38). Contrary to our expectations, higher levels of PA were not associated with lower prevalence of obesity across the race/ethnic groups. Non-Hispanic white youth had lower mean counts per minute and spent fewer minutes per day in MVPA than non-Hispanic black and Mexican American youth, yet they had a lower prevalence of obesity than the other race/ethnic groups. This paradox may be accounted for by the fact that non-Hispanic white youth may spend more time in activities not captured well by accelerometry such as swimming or bicycling. These differences could also be attributed to the higher socioeconomic status found in the non-Hispanic white youth because socioeconomic status has been inversely related to obesity and positively related to PA (24). However, socioeconomic status was controlled for in all analyses; other factors may contribute to the pattern of obesity and PA in non-Hispanic white youth.
Genetic predisposition to obesity, socioeconomic status, and cultural differences in behavior may play a role in the race/ethnic differences found in this sample and elsewhere (46). Non-Hispanic black and Mexican American adults have the highest prevalence of overweight and obesity in the US population (33). Children of overweight and obese parents have been shown to have higher rates of obesity than children of normal-weight parents (4). Furthermore, non-Hispanic black and Mexican American females have been shown to have lower basal metabolic rates and expend less activity energy than non-Hispanic white females, which may put them at higher risk for overweight and obesity (17). Dietary intake may also account for the differences in the obesity prevalence between non-Hispanic white youth and the minority race/ethnic groups, particularly given that there are race/ethnic differences in the consumption of unhealthy foods. Arcan et al. (2) found that non-Hispanic black high school students were more likely to consume sugar-sweetened beverages and high-fat foods than other race/ethnic groups. The higher rate of obesity in non-Hispanic blacks may be explained by a higher intake of unhealthy foods, particularly in non-Hispanic black females who have been found to have the lowest levels of PA and highest intakes of unhealthy foods (27).
In this study, adjustment for reported energy intake did not moderate the associations between total PA and age, race/ethnicity, or weight status and PA levels. Non-Hispanic white youth reported significantly higher energy intake than the other race/ethnic groups, and normal-weight youth reported significantly higher energy intake than overweight or obese youth. Differences across weight status may be due to the differential underreporting of energy intake (29,41), whereas differences across race/ethnicity may be due to a combination of differential misreporting and actual differences in true intake. Although the observed differences suggest that energy intake was a potential confounder, analysis determined that it was not confounding the relationship between weight status and race/ethnicity and PA levels. Further analysis of the association between patterns of food intake and PA could be of interest because of the high-quality data available in this survey.
No other past study that we are aware of has described race/ethnic differences in objectively measured PA in a large representative sample of US youth; however, several limitations to the present study merit discussion. First, this is a cross-sectional analysis; thus, we cannot determine any causal associations. However, the large sample size allows robust estimates of associations between variables of interest and may help inform future longitudinal studies. Second, accelerometers do not capture all types of PA (40). However, accelerometers are considered to be an excellent objective measurement of PA in youth because they minimize self-report bias and eliminate human error in recalling previous PA (14). Third, the accelerometers do not record the type of PA as do self-report measures, which prevents us from exploring the frequency of specific behaviors (i.e., TV viewing) that could explain the observed differences in PA levels among race/ethnic groups. Fourth, NHANES is designed to sample the three largest race/ethnic groups in the United States and, therefore, does not provide data sufficient for a national estimate for other minority groups such as Asians or other non-Mexican Hispanic populations that comprise a significant and growing proportion of the US population. Finally, BMI percentile category is used here as a proxy measure of adiposity. Although some findings indicate that it is not an accurate measure of body fat for all race/ethnic groups, Flegal et al. (19) recently demonstrated that it corresponds well with percent body fat in an adult sample, and BMI has been shown to be significantly correlated with percent body fat in youth (36). Furthermore, BMI percentile is a cost-efficient and feasible measure in a large population-based study (15).
As measured by accelerometry, non-Hispanic white youth engaged in less PA than both non-Hispanic black and Mexican American youth, yet they had the lowest prevalence of obesity in this sample. Also, non-Hispanic black females are the least physically active and have the highest prevalence of obesity in this sample. Mexican American 12- to 15-yr-old obese and normal-weight youth had the same amount of MVPA. Explanations for differences in obesity rates between youth of different race/ethnic groups must be influenced by factors other than variations in PA levels.
This work was supported by the University of Southern California Center for Transdisciplinary Research on Energetics and Cancer (NCI U54 CA 116848). The National Cancer Institute reviewed and approved this article before submission.
No author reports any conflicts of interest.
The findings and conclusions in this article are those of the authors and do not necessarily represent the views of the funding agency.
The results of the present study do not constitute endorsement by the American College of Sports Medicine.
1. Anderson S, Economos C, Must A. Active play and screen time in US children aged 4 to 11 years in relation to sociodemographic and weight status characteristics: a nationally representative cross-sectional analysis. BMC Public Health
2. Arcan C, Kubik MY, Fulkerson JA, Story M. Sociodemographic differences in selected eating practices among alternative high school students. J Am Diet Assoc
3. Berkey C, Rockett H, Field A, Gillman M, Frazier A, Camargo C. Activity, dietary intake, and weight changes in a longitudinal study of preadolescent and adolescent
boys and girls. Pediatrics
4. Boney C, Verma A, Tucker R, Vohr B. Metabolic syndrome in childhood: association with birth weight, maternal obesity, and gestational diabetes mellitus. Pediatrics
5. Brage S, Wedderkopp N, Franks P, Bo A. Reexamination of validity and reliability of the CSA monitor in walking and running. Med Sci Sports Exerc
6. Brodersen N, Steptoe A, Boniface D, Wardle J, Hillsdon M. Trends in physical activity and sedentary behaviour in adolescence: ethnic and socioeconomic differences [commentary]. Br J Sports Med
7. Butte N, Puyau M, Adolph A, Vohra F, Zakeri I. Physical activity in nonoverweight and overweight Hispanic children and adolescents. Med Sci Sports Exerc
8. Byrd-Williams C. Influence of gender, BMI and Hispanic ethnicity on physical activity in children. Int J Pediatr Obes
9. Carroll R. Measurement Error in Nonlinear Models: A Modern Perspective
. 2nd ed. Boca Raton, FL: CRC Press; 2006. p. 65-95.
10. Center for Disease Control and Prevention (CDC). National Health and Nutrition Examination Survey Analytic and Reporting Guidelines
. Hyattsville (MD): US Department of Health and Human Services, Centers for Disease Control and Prevention; 2006. Available from: http://www.cdc.gov/nchs/data/nhanes/nhanes_03_04/nhanes_analytic_guidelines_dec_2005.pdf
. Accessed June 22, 2010.
11. Center for Disease Control and Prevention (CDC). National Health and Nutrition Examination Survey Analytic Guidelines
. Hyattsville (MD): US Department of Health and Human Services, Centers for Disease Control and Prevention; 2006. Available from: http://www.cdc.gov/nchs/data/nhanes/nhanes_general_guidelines_june_04.pdf
. Accessed June 22, 2010.
12. Center for Disease Control and Prevention (CDC). MEC In-Person Dietary Interviewers Procedures Manual
. Hyattsville (MD): US Department of Health and Human Services, Centers for Disease Control and Prevention; 2006. Available from: http://www.cdc.gov/nchs/data/nhanes/nhanes_05_06/DIETARY_MEC.pdf
. Accessed June 22, 2010.
13. Davis JN, Alexander KE, Ventura EE, et al. Associations of dietary sugar and glycemic index with adiposity and insulin dynamics in overweight Latino youth. Am J Clin Nutr
14. de Leeuw E, Borgers N, Smits A. Pretesting questionnaires for children and adolescents. In: Presser S, Rothgeb J, Couper M, et al, editors. Methods for Testing and Evaluating Survey Questionnaires
. Malden MA: Wiley-Interscience; 2004. p. 409-29.
15. Dietz W, Robinson T. Use of the body mass index (BMI) as a measure of overweight in children and adolescents. J Pediatr
16. Dodd K, Guenther P, Freedman L, et al. Statistical methods for estimating usual intake of nutrients and foods: a review of the theory. J Am Diet Assoc
17. Dugas L, Ebersole K, Schoeller D, et al. Very low levels of energy expenditure among pre-adolescent
Mexican-American girls. Int J Pediatr Obes
18. Ekelund U, Särnblad S, Brage S, Ryberg J, Wareham N, Åman J. Does physical activity equally predict gain in fat mass among obese and nonobese young adults? Int J Obes
19. Flegal K, Shepherd J, Looker A, et al. Comparisons of percentage body fat, body mass index, waist circumference, and waist-stature ratio in adults. Am J Clin Nutr
20. Freedson P, Melanson E, Sirard J. Calibration of the Computer Science and Applications, Inc. accelerometer. Med Sci Sports Exerc
21. Fulton JE, Dai S, Steffen LM, Grunbaum JA, Shah SM, Labarthe DR. Physical activity, energy intake, sedentary behavior, and adiposity in youth. Am J Prev Med
. 2009;37(1 suppl):S40-9.
22. Hill JO. Understanding and addressing the epidemic of obesity: an energy balance perspective. Endocr Rev
23. Jago R, Harrell J, McMurray R, Edelstein S, El Ghormli L, Bassin S. Prevalence of abnormal lipid and blood pressure values among an ethnically diverse population of eighth-grade adolescents and screening implications. Pediatrics
24. Janz K, Burns T, Levy S. Tracking of activity and sedentary behaviors in childhood: the Iowa Bone Development Study. Am J Prev Med
25. Korn E, Graubard B. Analysis of Health Surveys
. Malden MA: Wiley-Interscience; 1999. p. 31.
26. Kuczmarski RJ, Ogden CL, Guo SS, et al. 2000 CDC Growth Charts for the United States: methods and development. Vital Health Stat 11
27. Kumanyika S. Environmental influences on childhood obesity: ethnic and cultural influences in context. Physiol Behav
28. Leenders J. Evaluation of methods to assess physical activity in free-living conditions. Med Sci Sports Exerc
29. Lissner L, Troiano R, Midthune D, et al. Open about obesity: recovery biomarkers, dietary reporting errors and BMI. Int J Obes
30. Matthews CE, Chen KY, Freedson PS, et al. Amount of time spent in sedentary behaviors in the United States, 2003-2004. Am J Epidemiol
31. McMurray RG, Ward DS, Elder JP, et al. Do overweight girls overreport physical activity? Am J Health Behav
32. Nader P, Bradley R, Houts R, McRitchie S, O'Brien M. Moderate-to-vigorous physical activity from ages 9 to 15 years. JAMA
33. Ogden C, Carroll M, Curtin L, McDowell M, Tabak C, Flegal K. Prevalence of overweight and obesity in the United States, 1999-2004. JAMA
34. Ogden CL, Carroll MD, Flegal KM. High body mass index for age among us children and adolescents, 2003-2006. JAMA
35. Owen CG, Nightingale CM, Rudnicka AR, Cook DG, Ekelund U, Whincup PH. Ethnic and gender differences in physical activity levels among 9-10-year-old children of white European, South Asian and African-Caribbean origin: the Child Heart Health Study in England (CHASE Study). Int J Epidemiol
35a. Pate RR, Stevens J, Webber LS, et al. Age-Related Change in Physical Activity in Adolescent
Girls. J Adolesc Health
36. Pietrobelli A, Faith M, Allison D, Gallagher D, Chiumello G, Heymsfield S. Body mass index as a measure of adiposity among children and adolescents: a validation study. J Pediatr
37. Reichert F, Menezes A, Wells J, Dumith S, Hallal P. Physical activity as a predictor of adolescent
body fatness: a systematic review. Sports Med
38. Rubin D, McMurray R, Harrell J, Hackney A, Thorpe D, Haqq A. The association between insulin resistance and cytokines in adolescents: the role of weight status and exercise. Metab Clin Exp
39. Sirard JR, Pfeiffer KA, Dowda M, Pate RR. Race differences in activity, fitness, and BMI in female eighth graders categorized by sports participation status. Pediatr Exerc Sci
40. Spruijt-Metz D, Berrigan D, Kelly L, et al. Measures of physical activity and exercise. In: Allison D, Baskin M, editors. Handbook of Assessment Methods for Eating Behaviors and Weight-Related Problems: Measures, Theory, and Research
. Thousand Oaks: Sage Publications Inc., in press.
41. Tooze J, Subar A, Thompson F, Troiano R, Schatzkin A, Kipnis V. Psychosocial predictors of energy underreporting in a large doubly labeled water study. Am J Clin Nutr
42. Troiano RP, Berrigan D, Dodd KW, Masse LC, Tilert T, McDowell M. Physical activity in the United States measured by accelerometer. Med Sci Sports Exerc
43. Trost SG, Pate RR, Sallis JF, et al. Age and gender differences in objectively measured physical activity in youth. Med Sci Sports Exerc
44. US Department of Health and Human Services (USDHHS). 2008 Physical Activity Guidelines for Americans
. Washington, (DC): USDHHS; 2008. Available from: http://www.health.gov/paguidelines/default/aspx
. Accessed June 22, 2010.
45. Woodruff R. A simple method for approximating the variance of a complicated estimate. J Am Stat Assoc
46. Yang W, Kelly T, He J. Genetic epidemiology of obesity. Epidemiol R
47. Yngve A, Nilsson A, Sjostrom M, Ekelund U. Effect of monitor placement and of activity setting on the MTI accelerometer output. Med Sci Sports Exerc