点击显示 收起
1 From the Cincinnati Children's Hospital Medical Center (BES, KvS, LFD, and KJO) and University of Cincinnati College of Medicine (BES, RJS, and LFD)
2 Supported by National Institutes of Health grants DK60476 and General Clinical Research Center grant M01 RR08084 from the National Center for Research Resources. 3 Reprints not available. Address correspondence to BE Saelens, University of Washington, Children’s Hospital and Regional Medical Center, 6200 NE 74th Street, Suite 210, Seattle, WA 98115. E-mail: bsaelens{at}u.washington.edu.
ABSTRACT
Background: Abdominal fat is more related to health risk than is whole-body fat. Determining the factors related to children's visceral fat could result in interventions to improve child health.
Objective: Given the effects of physical activity on adults' visceral fat, it was hypothesized that, after accounting for whole-body fat, physical activity would be inversely related to children's visceral (VAT), but not to subcutaneous (SAT), abdominal adipose tissue.
Design: In this cross-sectional observational study conducted in forty-two 8-y-old children (21 boys, 21 girls) at risk of obesity [>75th body mass index (BMI) percentile, with at least one overweight parent], familial factors (eg, maternal BMI), historic weight-related factors (eg, birth weight), and the children's current physical activity (self-reported and measured with accelerometry) and diet were examined as potential correlates of the children's whole-body composition (measured with BMI and dual-energy X-ray absorptiometry) and abdominal fat distribution (measured by magnetic resonance imaging).
Results: Accelerometer-measured physical activity was related to whole-body fat (r = –0.32, P < 0.10), SAT (r = –0.29, P < 0.10), and VAT (r = –0.43, P < 0.05). In regression models, whole-body fat was positively associated with and the only significant correlate of SAT. Whole-body fat was positively related and accelerometer-measured physical activity was negatively and independently related to the children's VAT.
Conclusions: Both SAT and VAT in 8-y-old children at risk of obesity are most closely associated with whole-body fat. However, after control for whole-body fat, greater physical activity is only associated with lower VAT, not SAT, in these children.
Key Words: Visceral fat • physical activity • children • obesity • abdominal fat
INTRODUCTION
The prevalence of overweight in children remains high in the United States (1). Similar to adults, children's abdominal fat distribution, particularly visceral or intraabdominal fat deposition, appears more strongly related to various cardiovascular and diabetes risk factors than does whole-body fat (2-5). Physical activity is a critical determinant of adults' visceral fat, even independent of relations between physical activity and whole-body fat and between whole-body fat and visceral adiposity (6-8). Adults with similar body mass indexes (BMIs) with higher fitness levels have lower abdominal fat than do their less fit counterparts (9), although adults matched only on BMI that differ in fitness likely have different whole-body fat levels. Physical activity interventions, even without concurrent weight loss, lead to visceral fat reductions in adults (10-12).
Little is known about the determinants of visceral fat accumulation in children (13). Among US nonobese prepubertal (average age 10-y-old) and pubertal (average age 13-y-old) youth, one study found negative associations between the children's self-reported activity assessed by a 7-d recall and visceral fat in boys (r = –0.43) and across sex among prepubertal children (r = –0.38), but adjustment for age and total fat mass substantially diminished the strength of these associations (14). Twelve-y-olds from eastern France with greater structured physical activity (>140 compared with >0–140 and 0 min/wk) have lower waist circumferences (P < 0.0001) (15), and overweight or obese 6- to 13-y-old children from Greece with higher cardiorespiratory fitness have lower subscapular skinfolds (P < 0.01) (16). However, evidence is limited, and most research on behavioral correlates of children's abdominal fat distribution is marked by poor physical activity measurement (eg, self-report; 17), imprecise abdominal fat distribution measurement (eg, waist circumference as a proxy for visceral fat), and are often confounded by other factors that differ between the participants, including weight status and age. There is scant investigation on whether other factors, such as weight and feeding history or current dietary intake, are related to abdominal fat distribution in children (13). In one of the few studies to examine dietary factors, Ku et al (18) found no significant associations between dietary variables and abdominal subcutaneous or visceral fat deposition in a 4-to-10 y old child sample that was approximately two-thirds African-American and one-third white.
The present study examined the associations between abdominal fat distribution and familial factors (eg, maternal BMI), historic weight-related factors (eg, child's birth weight), and current body composition and energy balance behaviors (ie, children's physical activity and diet) among a sample of 8-y-old children at risk of obesity. Whole-body fat was expected to be a strong correlate of the children's visceral and subcutaneous fat accumulation. On the basis of evidence from adult samples, it was hypothesized that physical activity would be negatively related to visceral fat and explain variance in visceral fat above and beyond that explained by whole-body fat.
SUBJECTS AND METHODS
Subjects
Fifty-five families responded to a mass mailing describing the study sent to families in Hamilton County, OH (greater Cincinnati area). Forty-two child-parent pairs were interested and met the eligibility criteria. Children met the following inclusion criteria: >75th BMI percentile for age and sex, within 3 mo of their 8th birthday, had at least one overweight (BMI > 25 kg/m2; 19) parent, were not receiving any treatment that would interfere with their growth, and did not have any known medical reason for being at a high weight status. The children were 50% girls. The children were 95.2% non-Hispanic, 66.7% white, 28.6% black or African-American, and 4.8% multiple or other races. This is consistent with the 2000 Census data for the county, which found Hamilton County to be 98.9% non-Hispanic, 72.9% white, 23.4% black or African-American, and 3.7% other or multiple races. The families had a median family income between $50 000 and $60 000, similar to the Hamilton county–level 2000 Census estimate of family income of $53 449. The children were mostly prepubertal (90.5%), with 2 boys and 2 girls considered to be in early puberty. The present study was approved by the Cincinnati Children's Hospital Medical Center (CCHMC) Institutional Review Board. The parents provided consent, and the children provided verbal assent for their participation. The families were recruited and participated from November 2002 to August 2004.
Methods
The children, accompanied by a parent, were brought individually into the CCHMC General Clinical Research Center (GCRC) in the morning after an overnight fast.
Weight and height
Child weight was measured in triplicate to the nearest 0.1 kg by using a Scaletronic electronic scale, with children wearing hospital scrubs. Further measurements were obtained if necessary until 3 of 4 consecutive values were within 0.1 kg of each other. Weight was the average of these 3 proximal values. Child height was measured by Holtain stadiometer to the nearest 0.1 cm, with further measurements obtained if necessary to obtain 3 of 4 consecutive values within 0.5 cm of each other. Height was the average of these 3 proximal values. Nurses trained in anthropometric measurement conducted weight and height assessments using the same scale and stadiometer for all subjects. BMI was calculated as weight (in kg)/height2 (in m) with the use of the Box-Cox transformation (LMS method) and Centers for Disease Control and Prevention National Center for Health Statistics 2000 growth curves (20). Weight and height were measured for the accompanying parent or parents. Because the accompanying parent was usually the child's mother, only maternal BMI is reported herein. For the 2 instances in which the mothers did not accompany the child, maternal self-reported weight and height were obtained. For the 4 instances in which the mother was not overweight, paternal self-reported weight and height were obtained to confirm that at least one parent had a BMI > 25 (to meet the study inclusion criteria).
Pubertal assessment
The children's pubertal status was determined through a physical exam given by the study staff, who were trained by an expert in pubertal assessment. This assessment included assessments of breast and pubic hair development in girls (21) and testicular size and pubic hair development in boys (22). A same-sex study staff member assessed all children of a given sex.
Whole-body fat and soft lean mass
A whole-body fan-beam dual-energy X-ray absorptiometry (Hologic QDR 4500A; Bedford, MA) scan was obtained to estimate the children's whole-body fat, percentage body fat, and soft lean mass (subtracting bone mineral content). This is a noninvasive procedure, requiring a minimal radiation dose (3.6 µSv) and a short scan time (3 min), and it provides accurate and precise (error estimates <1% to 2%) measurements of whole-body fat and soft lean and bone masses (23, 24).
Abdominal fat distribution
Magnetic resonance imaging was used to assess the children's visceral fat and subcutaneous abdominal fat, by using a 1.5 T scanner (Signa; General Electric, Milwaukee, WI). The children were positioned within a torso-array send-receive coil placed overlying the abdomen. Initial 3-dimensional localization images were obtained by using a fast gradient echo sequence. Axial T1-weighted images were then obtained throughout the abdomen and pelvis, with the following imaging parameters: 10-mm slice thickness (zero gap), repetition time = 400, time to echo = 20, 256 x 128 pixel resolution, number of excitations = 2, and a 256 x 128 matrix. To avoid artifacts, images were acquired sequentially every other slice with a 10-mm gap and interleaved (eg, first acquired slices 1,3,5, etc, then acquired slices 2,4,6, etc). The fat contained within a slice was determined by Children's Hospital Imaging Processing Software developed at CCHMC with the use of a K-means algorithm (25). Total abdominal adipose tissue (TAT) and subcutaneous abdominal adipose tissue (SAT) volumes were derived by outlining the area of total and subcutaneous regions, respectively, in the 10 slices above the iliac crest. The total abdominal fat volume (in cm3) was calculated by summing 10 individual slice fat volumes (10-mm slice thickness). The volume of visceral adipose tissue (VAT) was calculated by subtracting the total volume of subcutaneous abdominal fat from total abdominal fat volume. The same technologist conducted all scans, and 1 person processed all the images. Images from a randomly selected subset of subjects (9 of 42) were processed again by an independent processor, with reliability estimates between raters > 0.94.
Self-reported activity
Activity was assessed in 2 ways. At the GCRC, the child and a parent were interviewed about the child's habitual moderate-to-vigorous physical activity during the past year by using the Past Year Leisure-Time Physical Activity questionnaire (26). This questionnaire queries about children's participation in 26 common child physical activities and whether they engaged in them 10 times in the past year. The number of months in the past year, days per week, and minutes per day the child engaged in each activity was assessed. The total physical activity hours per week was calculated (26). The child and parent also reported the child's time spent, in number of days per week and average minutes per day, in 7 sedentary activities (on the computer or internet, playing video games, watching television or videos, doing homework, reading, talking with friends, and listening to music) during the past 7 d. This sedentary recall questionnaire was created for the present study, but sedentary behavior recall assessments have shown good reliability and concurrent validity with sedentary behavior logs (27).
Accelerometer-measured activity
Before the GCRC visit, children were asked to wear an accelerometer (Computer Sciences Applications Inc activity monitor model 7164; Manufacturing Technology Incorporated, Fort Walton, FL). The accelerometer was worn on a belt and oriented above the right hip and was set to record activity in 1-min epochs. The children were instructed to wear the accelerometer during nonschool hours for 7 d, including at least 2 weekend days, when not engaged in water activities (eg, swimming or showering). The children and parents indicated the days during which the accelerometer was worn.
To calculate valid hours of accelerometer wearing, the log was examined for days in which the accelerometer was worn. On those days, valid hours were considered those hours in which there were no 30-min blocks of consecutive 0 activity counts (ie, assumed that the child was not wearing the accelerometer during that hour). Children had to have 2 valid hours per day on 5 d and to have worn the accelerometer on 1 weekend day to be considered to have complete accelerometer data. Children without complete accelerometer data were asked to rewear the accelerometer (n = 2), with all accelerometer data for each of these children combined for analysis. The average (±SD) number of valid hours was 60.5 ± 21.4, with a range of 29–108 h. Minute-by-minute activity counts during valid hours were converted to minute estimates of sedentary (<800 activity counts) and light (801–3199 activity counts), moderate (3200–8199 activity counts), and vigorous (8200 activity counts) intensity physical activity with the use of published thresholds for this accelerometer and age group (28). Total physical activity was the summation of light + moderate + vigorous physical activity. For comparability across children, these estimates were divided by the child's number of valid hours in which the accelerometer was worn. For comparability with the self-reported measures, hours per week estimates are reported. Accelerometer data were not available for one participant due to accelerometer malfunction.
Diet
Both children and parents were provided with instructions on how to keep a 3-d food diary in which to record the children's food and beverage consumption on 2 weekdays and 1 weekend day. The diaries were evaluated for completeness on their return, and the parents and children queried about any incomplete information. The food records were analyzed through Nutritionist-V software (29). Similar to a prior investigation regarding adults' dietary intake and visceral fat (30), daily estimates of total energy, protein, carbohydrate, fat, and saturated fat intakes were derived. Dietary data were not available for one participant due to substantial incomplete recording.
Family history and early child feeding
Parents reported birth and breastfeeding history for the participating children along with the family history of type 2 diabetes extending back to the participating child's grandparents.
Data analysis
The total sample size allowed adequate power (>80%) to detect correlations of approximately r 0.38 or r –0.38 with a 5% level of significance < 0.05 (r 0.51 or r –0.51 within sex) with the use of one-tailed tests, because this association is considered at least "medium" in size (31). Data were examined for normality. Self-reported physical activity was log transformed for correlational analyses (skewness > 1.5). Pearson product-moment correlations were conducted among and between body composition, and weight or feeding history and current child weight-related behaviors for continuous variables. The correlations were derived for the entire sample and separately by sex. Student's t tests were used to evaluate differences in body composition for dichotomous variables (eg, pubertal status) and to examine differences in correlations by sex. Continuous factors related to or dichotomous factors that showed differences in subcutaneous or visceral fat at the zero-order level of P < 0.10 when examined within the whole sample or within either sex when examined separately by sex were entered into separate linear regression models. The collective and unique amount of variance accounted for in subcutaneous and visceral fat was evaluated after accounting for the best correlate of whole-body composition. Analyses were conducted by using SPSS version 12 software for WINDOWS (SPSS Inc, Chicago, IL).
RESULTS
The children's current body composition and historic and family weight-related characteristics are provided in Table 1. More than 90% of the children were >85th BMI percentile and 64.3% were >95th BMI percentile. Likewise, 90.5% of the mothers had BMI values > 25. Information regarding the children's current weight-related behaviors is provided in Table 2. Self-reported physical activity, which includes only moderate and vigorous activity estimates, was more than twice the amount of accelerometer-measured moderate and vigorous physical activity. The only significant sex difference for variables presented in Table 1 and 2 was that the girls had higher average percentage body fat than did the boys.
View this table:
TABLE 1. Current body-composition and historic and family weight-related factors in the children
View this table:
TABLE 2. Current weight-related behaviors in the children1
Shown in Table 3 are the correlations, for the entire sample and separated by sex, among various measurements of the children's body composition. The associations among whole-body composition (eg, BMI, z BMI, and whole-body fat) and total abdominal and subcutaneous abdominal fat were generally higher (all r 0.71) than were the correlations between whole-body composition factors and visceral fat (all r 0.62). This pattern of findings was similar for the boys and girls (see Table 3).
View this table:
TABLE 3. Associations among body-composition variables in the children1
The correlations between the children's body composition, the children's birth weight, maternal BMI, and current child weight-related behaviors are provided in Table 4. Accelerometer-measured physical activity was negatively related to whole-body fat, SAT, and VAT for the whole sample, although only the relation with VAT was statistically significant. This relation seemed particularly strong for the boys, although there were no significant differences in associations by sex. Child body composition by ethnicity, pubertal status, history of breastfeeding, and family history of type 2 diabetes are provided in Table 5. No significant differences in the children's whole-body or abdominal fat composition were observed based on these factors.
View this table:
TABLE 4. Correlations between body-composition measures in the children and historic and family weight-related factors and current child weight-related behaviors1
View this table:
TABLE 5. Body-composition measures of the children across various current and past weight-related factors1
Because whole-body fat was the best whole-body composition correlate of SAT and VAT for the entire sample and across sex, whole-body fat was entered first into the regression models for these fat depots. Whole-body fat (positive correlate) accounted for a substantial amount of the variance in SAT. Whole-body fat, but not accelerometer-measured total physical activity, was independently related to SAT. Collectively, whole-body fat and accelerometer-measured total physical activity accounted for 93% of variance in this fat depot (Table 6). Both whole-body fat (positive correlate) and accelerometer-measured physical activity (negative correlate) were independent correlates of visceral fat, even though they were correlated with each other (see Table 4). Carbohydrate intake was not an independent correlate of VAT in the model. Collectively, these factors accounted for 52.5% of the variance in the children's VAT. Interactions between sex and correlates in each model were nonsignificant.
View this table:
TABLE 6. Regression models for subcutaneous (SAT) and visceral (VAT) abdominal adipose tissue
DISCUSSION
In the present study, physical activity was an independent correlate of the children's visceral fat, with greater physical activity associated with lower fat accumulation in this depot. The association between physical activity and visceral fat was maintained even after accounting for the strong relation between whole-body fat and visceral fat and for the negative association between physical activity and whole-body fat. These findings in children are consistent with those found among adults, wherein greater physical activity or fitness is related to lower visceral adiposity levels (6-8, 30, 32). In contrast, there was an expected strong relation between whole-body fat and subcutaneous abdominal fat, but no independent association between physical activity and subcutaneous abdominal fat after accounting for whole-body fat.
Relations between the children's physical activity and both whole-body fat and visceral fat were only evident with the use of a more objective measure of activity (ie, an accelerometer) than with a self-report measure. Findings for relations between the children's self-reported physical activity and whole-body weight or fat in prior research are mixed (33, 34). Yet, recent studies that used more objective measures of physical activity appear more consistent with energy balance expectations, finding lower physical activity levels among obese youth than in their lean counterparts (35) and negative relations between physical activity and whole-body fat (36, 37), as seen in the present study. Lower physical activity measured by accelerometry early in childhood was related to greater body fat later in childhood (38). Analogous to the need for more sophisticated methods to assess visceral fat (magnetic resonance imaging compared with waist circumference measures), especially among children (39), more reliable and valid measures of physical activity may be needed to specify relations between physical activity and children's body composition (37) and health risk (40).
There was some suggestion of the relation between accelerometer-measured physical activity and visceral fat being stronger in boys than girls, although when considered in the model that included whole-body fat as a covariate, the interaction between physical activity and sex was a not a significant correlate of visceral fat. Treuth et al (41) recently found that accelerometry-measured activity was significantly related to whole-body fat mass only in girls and not boys, but that study and the present study differed in sample ages and weight status. A recent prospective study found that physical activity,measured via self-report, was inversely related to fat mass development only in boys and not girls (42). Others have found that declining physical activity from childhood into adulthood increases the risk of high waist circumference in women but not men (43). Dietary factors have been implicated in the development of the metabolic syndrome in children (44), and there was some suggestion in the present study that dietary factors were related to the girls' visceral fat accumulation, although not in the regression model that accounted for whole-body fat. Clearly, more investigation is needed into sex differences in the relations among diet, physical activity, and abdominal fat accumulation in children.
Consistent with present findings, structured physical training appears to prevent further visceral fat accumulation in obese 7–11-y-olds (45). Reductions in visceral adiposity levels in obese adolescents resulting from structured physical training occur regardless of the prescribed physical training intensity (46). Improvements in children's estimated cardiorespiratory fitness are related to decreases in visceral adiposity (46). However, improvements in fitness, which are best attained through increases in vigorous intensity activity, may not be required to reduce or prevent visceral fat accumulation. Even increases in moderate intensity physical activity, such as walking, are associated with decreases in visceral fat in adults (47), although the effects of intervention with moderate intensity physical activity on visceral fat have not been evaluated in children.
In the present study, the average accelerometer-measured moderate-to-vigorous physical activity of 4.8 h/wk was higher than that obtained in another accelerometer-based study with similarly aged children (48). The major difference was that the present study extrapolated to a general weekly estimate based on wearing the accelerometer during more discretionary hours (ie, after school and on weekends), whereas the prior study collected child accelerometer data during all waking hours. The present study also used a very conservative approach to considering an hour as valid for analysis. In both studies, self-reported physical activity was more than twice the accelerometer estimates of moderate-to-vigorous intensity physical activity, which is not unexpected given the prevalence of high overreporting in self-reported physical activity assessments (17). The present study suggests that greater precision in physical activity measurements may be particularly important when examining body composition or health effects that are perhaps more influenced by total activity or light and moderate activity than by vigorous intensity physical activity.
The factors that influence children's weight-related health risks change throughout childhood. Current whole-body fat was clearly an important correlate of visceral fat accumulation among the children examined in the present study, as seen in prior research (49-52). Even among early school-aged children, a child's weight status appears to be an important determinant of their adult health risk and obesity (53), perhaps even more critical than some historic (eg, child's birth weight) or early feeding practices. Parent-child BMI correlations obtained were similar in magnitude to those observed in prior research (54), but visceral fat estimates among the parents of the children in the present study were not available. Prior studies suggest some heritability of abdominal fat distribution (54, 55), but more studies are needed to specify genetic compared with environmental influences on abdominal fat distribution in children.
Limitations of the present study include the cross-sectional design, the relatively small sample size and corresponding inability to examine moderator effects (eg, by sex or ethnicity), and the use, in part, of self-reported measures of physical activity and dietary intake. The recall physical activity measurement used has documented reliability and validity in children >12 y old (26, 56), but has not been evaluated for psychometrics in younger children. Self-report for physical activity in the present study's age group is suspect, with the present study findings highlighting the need to obtain more objective physical activity assessments in this age group. Shorter accelerometer epoch periods should also be considered (eg, 15 or 30 s instead of 1 min) in future research, particularly for children, to better characterize moderate-to-vigorous physical activity (57). Dietary self-reports have well-documented problems with underreporting and inaccuracy (58). The present study only examined children within 3 mo of their 8th birthday. Relations between physical activity and visceral adiposity could be more complicated in older children, particularly during the changing hormonal milieu of puberty. Less than 10% of the children in the present study were pubertal, and most were white. The present study was not able to detect potentially important interactions among sex, ethnicity, physical activity, and dietary intake for visceral fat accumulation. In prior studies, sex and ethnicity were not the most potent correlates of children's visceral fat levels when including general anthropometric measures, but sex and ethnicity are not insignificant factors (59). The relation between activity and visceral fat may only be present in overweight children, who show a greater variability in visceral adiposity than do their leaner counterparts. Although 7 d of accelerometry measurement appears to be a reliable indicator of children's regular physical activity (60), there are likely seasonal and other factors that affected this measurement. Not all children wore the accelerometer for 7 d, with a broad range of valid hours of accelerometry; however, others have found little bias due to incomplete accelerometer data in children (61).
The increasing rates of childhood obesity and especially the trends toward increasing central adiposity in children (62) are alarming. Improvements in children's cardiovascular risk factors through physical activity may not require changes in whole-body fat or body weight (63), although some effects of physical activity on health risks may be mediated through changes in whole-body fat or fat distribution (64). The present findings support prior results documenting the strong positive relation between whole-body fat and visceral and subcutaneous abdominal fat accumulation. Consistent with the assertion that physical activity likely has both direct effects and indirect effects through absolute weight on health (65), the present study suggests that physical activity may be a critical target for the specific prevention of visceral fat accumulation and corresponding health risks in overweight children.
ACKNOWLEDGMENTS
We thank the families who participated in this study and Carrie Pugh and Monica J Smith for their assistance in conducting the study.
BES and RJS designed the study. BES, KvS, LFD, and KJO collected and processed the data. BES analyzed the data. BES, RJS, KVS, LFD, and KJO participated in data interpretation and manuscript preparation. None of the authors has any personal or financial conflicts of interest.
REFERENCES