点击显示 收起
1 From the Department of Physical Education, Sports Science, and Recreation Management, Loughborough University, Leicestershire, United Kingdom; the School of Physical Education, the National Institute of Education, Nanyang Technological University, Singapore; and the School of Sport, Performing Arts, and Leisure, University of Wolverhampton, Walsall Campus, United Kingdom.
2 Supported by the Academic Research Fund, the National Institute of Education, Nanyang Technological University, Singapore. 3 Address reprint requests to DJ Stensel, Department of Physical Education, Sports Science, and Recreation Management, Loughborough University, Loughborough, Leicestershire, LE11 3TU United Kingdom. E-mail: d.j.stensel{at}lboro.ac.uk.
ABSTRACT
Background: Previous studies investigating the hypothesis that a low resting metabolic rate (RMR) is a cause of obesity yielded discrepant findings. Two explanations for these findings are the use of imprecise methods to determine obesity and a failure to control for differences in fat mass (FM) and fat-free mass (FFM) when comparing RMR values.
Objective: This study tested the hypothesis that RMR is lower in obese than in nonobese boys (with the use of precise methods to quantify body fatness and with adjustment for differences in both FM and FFM).
Design: Forty Chinese Singaporean boys aged 12.8–15.1 y were recruited. Boys were classified as obese (n = 20) or nonobese (n = 20) on the basis of their adiposity index (ratio of FM to FFM: >0.60 = obese, <0.40 = nonobese) determined by dual-energy X-ray absorptiometry. RMR was determined by using indirect calorimetry. RMR values were compared by using both linear (analysis of covariance) and log-linear (analysis of covariance with log-transformed data) regression to control for differences in FM and FFM.
Results: Age, height, and FFM did not differ significantly between groups. Body mass was 13 kg greater and FM was 16 kg greater in the obese boys than in the nonobese boys (P < 0.001). After control for FFM and FM, RMR did not differ significantly between the groups.
Conclusion: When body composition is appropriately controlled for, RMR does not differ significantly between obese and nonobese boys.
Key Words: Resting metabolic rate • obesity • Chinese Singaporean boys • fat mass • fat-free mass • analysis of covariance • log-linear regression • dual-energy X-ray absorptiometry • indirect calorimetry
INTRODUCTION
Cross-sectional studies examining resting metabolic rate (RMR) in obese and nonobese children have yielded discrepant findings. Usually, RMR is found to be higher in obese than in nonobese children when absolute values (kJ/d) are compared (1–7), but there are exceptions (8, 9). After control for fat-free mass (FFM), most studies reported that RMR did not differ significantly between obese and nonobese children (3, 4, 6, 7, 9) although some studies still report higher RMR values in the obese children (1, 5).
Explanations for the disparate findings presented above include 1) the methods used to measure body composition, 2) the criteria used to determine obesity, and 3) the means of controlling for differences in body size or composition when RMR values are compared. Most studies measured skinfold thickness to assess body composition (3–6, 9), but weight and height (2), total body water (1), and dual-energy X-ray absorptiometry (DXA) (7, 8) have also been used. Criteria used to classify children as obese included weight-for-height (2, 3, 5, 7, 9), body mass index (4, 6), skinfold thickness (8), and percentage of ideal body weight determined from measurements of total body water (1). The predominant method used to adjust for differences in body size or composition in comparisons of RMR values was analysis of covariance with FFM as the covariate (1, 3, 4, 6, 7, 9). This practice was promoted by Poehlman and Toth (10) in a paper published in this journal. It is now clear, however, that fat mass (FM) also contributes significantly to RMR (11, 12) and that differences in FM should be controlled for in comparisons of RMR between groups. Only one of the studies cited above (5) controlled for both FM and FFM when comparing RMR in obese and nonobese children. This study found no difference in RMR between groups, but the imprecise method used to determine body composition (skinfold thickness) and the criterion for obesity (weight-for-height) mean that the findings require verification.
Therefore, we decided to reexplore the relation between RMR and obesity in 40 Chinese Singaporean boys by using DXA to determine body composition and both linear and log-linear regression to control for FM and FFM. We hypothesized that with accurate determination of body composition and appropriate adjustment for both FM and FFM, RMR would be significantly lower in obese than in nonobese Chinese Singaporean boys.
SUBJECTS AND METHODS
Subjects
The subjects in this study were 40 Chinese Singaporean boys aged 12.8–15.1 y (Table 1). The boys were year 1 and year 2 students at Catholic High School, Singapore. Written informed consent was obtained from the boys and from their parents before the start of the study. The Ethical Advisory Committee of the School of Physical Education (Nanyang Technological University) also gave its permission for the study. Boys were classified as obese (n = 20) or nonobese (n = 20) on the basis of their adiposity index (ratio of FM to FFM). Boys with an adiposity index >0.60 were classified as obese whereas those with an adiposity index <0.40 were classified as nonobese. These values were chosen arbitrarily to create 2 distinctly different groups with respect to body fatness. Boys with an adiposity index of 0.40–0.60 were excluded from the study. The pubertal status of the boys was not determined, but the groups did not differ significantly with respect to age, height, or FFM.
View this table:
TABLE 1. Age, height, weight, and adiposity of obese and nonobese boys1
Body composition
Height was measured to the nearest 0.1 cm by using a wall-mounted stadiometer (Holtain, Dyfed, United Kingdom). The mass of the subjects, who were wearing socks and shorts, was measured to the nearest 0.01 kg with an electronic weighing scale (IDL Plus; Mettler Toledo, Giessen, Germany). Body mass index was calculated as weight (kg)/height2 (m). The waist-to-hip ratio was determined from the waist (minimal circumference of the abdomen) and hip (maximal circumference over the greater trochanters) circumferences measured to the nearest 1.0 cm by using a plastic measuring tape. A DXA instrument (LUNAR DPX-L model 1.31; Lunar Corp, Madison, WI) was used to measure each subject's FM, FFM, and percentage body fat. The DXA scans were performed in the Orthopaedic Diagnostic Centre at the National University Hospital, Singapore.
Resting metabolic rate
RMR was determined by using indirect calorimetry. The boys were asked to avoid strenuous exercise for 1 d before the test and to fast for 12 h the night before the test. On the morning of the test, the boys were collected from school at 0700 and transported by car to the laboratory of the School of Physical Education. On arrival at the laboratory, the boys sat quietly for 30 min before RMR was measured by using a metabolic cart (model 2900Z; SensorMedics, Yorba Linda, CA). This metabolic cart was calibrated each morning before the tests began. A half-face mask with a 2-way breathing valve (model 2700; Hans Rudolph Inc, Kansas City, MO) was fitted to the subjects' mouths and connected by respiratory tubes to the metabolic cart. Testing was conducted with the boys in a seated position. Expired air was collected for 30 min. RMR (kJ/d) was calculated from oxygen consumption and carbon dioxide production (13) without measurement of protein oxidation. The values used to calculate RMR included initial steady state values and all values after this. Steady state was defined as ventilation oxygen consumption and the respiratory exchange ratio each varying by <3% during 2 consecutive minutes. Steady state was reached generally between 5 and 15 min of the collection period. If a steady state was attained after 15 min, the duration of the collection period was extended. If a steady state was not reached at any point during the collection period, then the first 5 min of data were discarded; RMR was calculated from the remaining data. The mean duration of data collection used to calculate RMR for both groups was 20 min.
Statistical analysis
SPSS (version 9.0; SPSS, Inc, Chicago) was used to analyze the data. RMR values were divided by body mass and FFM so that values could be expressed as kJ•kg bodywt-1•d-1 and kJ•kg FFM-1•d-1, respectively. These values were then compared with the use of two-tailed t tests for independent samples. In addition, analysis of covariance was used to compare RMR in the obese and nonobese boys with the use of only FFM as a covariate and then with both FFM and FM as covariates. Finally, log-linear regression was used to compare the RMR values of the groups. For this comparison, analysis of covariance was again used to control for FFM alone and for FFM in combination with FM but using log-transformed data (eg, ln RMR, ln FFM, and ln FM). This approach was based on the assumption that RMR was likely to be proportional to the subject's body size (similar to the relation between peak oxygen consumption and body mass), and hence, best expressed by the allometric model Y = amb, described previously (14, 15). Pearson product-moment correlation coefficients were used to examine relations between variables. Significance was set at P < 0.05. Data are given as means ± SDs.
RESULTS
Body-composition variables for the obese and nonobese boys are shown in Table 2. All variables were significantly different between groups except for FFM. The minimum and maximum values for each group for each body-composition variable shown in Table 2 indicated that there was some overlap between groups for each variable with the exception of percentage body fat, for which the maximum value in the nonobese group (28.3%) was 11% less than the minimum value in the obese group (39.0%). Conversely, minimum and maximum values for FFM indicated considerable overlap between groups.
View this table:
TABLE 2. Body-composition variables for obese and nonobese boys1
The relations between RMR and body mass, RMR and FFM, and RMR and FM are shown in Figures 1–3. All relations were significant. The highest correlations were between RMR and body mass. Correlations between RMR and FFM were only slightly higher than those between RMR and FM in each group.
FIGURE 1. . Relation between body mass and resting metabolic rate (RMR) in 20 obese boys (r = 0.76, P < 0.0001) and 20 nonobese boys (r = 0.83, P < 0.0001).
FIGURE 2. . Relation between fat-free mass and resting metabolic rate (RMR) in 20 obese boys (r = 0.71, P < 0.001) and 20 nonobese boys (r = 0.74, P < 0.0001).
FIGURE 3. . Relation between fat mass and resting metabolic rate (RMR) in 20 obese boys (r = 0.69, P < 0.001) and 20 nonobese boys (r = 0.70, P < 0.001).
RMRs are expressed in 7 different ways in Table 3. Absolute values did not differ significantly between groups. When absolute values were divided by body mass (kJ•kg bodywt-1•d-1), they were significantly lower in the obese than in the nonobese boys; significantly higher values were obtained in the obese boys when absolute values were divided by FFM (kJ•kg FFM-1•d-1). The use of linear or log-linear regression yielded significantly higher RMR values in the obese boys when FFM was controlled for. The fitted exponent for FFM based on the log-linear model was ß = 0.50 (95% CI: 0.34, 0.67). Conversely, when both FFM and FM were controlled for by using either linear or log-linear regression, RMR values did not differ significantly between groups. The fitted exponents for FFM and FM based on the log-linear model were ß = 0.39 (95% CI: 0.20, 0.56) and ß = 0.15 (95% CI: 0.03, 0.28), respectively. Interestingly, the covariates of FFM and FM contributed significantly to the prediction of RMR in the above regression analyses.
View this table:
TABLE 3. Resting metabolic rates (RMR) in obese and nonobese boys1
DISCUSSION
The main finding of this study was that after FM and FFM were controlled for, RMR did not differ significantly between the obese and nonobese boys. This is consistent with previous results of cross-sectional studies that did not show a lower RMR in obese than in nonobese children (1–9). It is also consistent with the finding that infants (16) and children (17–21) who are genetically predisposed to obesity do not have lower RMR values than infants and children who are not predisposed to obesity.
Although the results of the present study are consistent with data from previous cross-sectional studies in showing that obese children do not have lower RMRs, the findings of the present study differ in several ways. Most previous comparisons observed that absolute RMR values were higher in obese than in nonobese children (1–7). In the present study, absolute RMR values did not differ significantly between groups. A likely explanation for this is that FFM did not differ significantly between obese and nonobese boys. In fact, the mean FFM value of the obese group was slightly lower than that of the nonobese group (40.5 compared with 44.1 kg; NS). All previous studies found that FFM was higher in obese than in nonobese children. In some studies these differences were significant (1, 4–7) and in others they were not (3, 8, 9).
It is possible that the lower FFM values in the obese boys were a peculiarity of the selection criteria that were used (ie, adiposity index >0.60). This may have encouraged recruitment of boys with a particularly low FFM to the obese group. Another possibility is that DXA did not accurately measure FFM differences between the obese and nonobese boys. There is evidence that tissue thickness affects the accuracy of DXA measurements (22); therefore, it is possible that FFM was underestimated in the obese boys or overestimated in the nonobese boys. However, Svendsen et al (23) found excellent agreement between DXA (Lunar DPX-L, version 1.31) measures of FFM and values determined by chemical analysis in 7 pigs weighing 35–95 kg and ranging in body fat from 10% to 50%. A further possibility is that the hydration status of the obese and nonobese boys differed before DXA analysis. Hydration influences DXA measurements of FFM because DXA assumes that water is a constant fraction of bone-free lean tissue mass (0.73 mL/g). Hence, dehydration results in a reduction in FFM measured by DXA (24). However, we have no reason to suspect that the obese boys were dehydrated at the time of measurement. Furthermore, although the water content of lean tissue changes during childhood, any difference between groups in the present study should have had a small effect on the assessment of FFM by DXA (22).
When analysis of covariance was used to control for FFM, some studies reported that RMR values were still higher in obese children (1, 5) whereas others did not (3, 4, 6, 7, 9). In the present study, RMR values were higher in the obese boys when we controlled for FFM using either linear or log-linear regression. This is consistent with the results of the study of Bandini et al (1) and that of Molñar and Schutz (5) and is most likely due to the difference in FM between the obese and nonobese groups. This difference was 16 kg on average in the present study. In the study of Bandini et al, FM values were not reported but were based on body mass and percentage body fat data; the difference between the obese and nonobese groups was 29 kg. In the study of Molnar and Schutz (5), the difference was 18 kg. Although FM makes only a small contribution to RMR, it is an important contribution (12). Thus, a large FM would be expected to elevate RMR. This is supported by the fact that both FFM and FM predicted RMR when analysis of covariance was performed in the present study. In studies that did not show elevated RMR values in obese children after controlling for FFM (3, 4, 6, 7, 9), differences in FM were much smaller—ranging from 6.0 to 10.4 kg and, hence, would have had a smaller effect on RMR.
The preceding discussion highlights the importance of 2 factors when conducting studies of this nature: 1), the criteria used to determine obesity, and 2) more important, the method of expressing RMR. To our knowledge, only one previous study controlled for both FM and FFM when comparing the RMR of obese and nonobese boys (5). Our findings confirm that with control for FM and FFM there is no difference in RMR between groups. Thus, obese Chinese Singaporean boys who did not differ in age or height from their nonobese counterparts but carried 16 kg of excess fat appeared to have normal RMRs. This finding is perhaps not surprising in light of obesity-prevalence data. The prevalence of obesity (defined by weight-for-height with regard to age and sex) in Singaporean schoolchildren increased from 5.4% in 1980 to 15.1% in 1991 (25). It is unlikely that such a sharp increase in a short amount of time could be explained by a sudden alteration in RMR.
It is possible that RMR was low in the obese boys in the present study before they became obese. Such a hypothesis is supported by the results of Griffiths and Payne (26). Moreover, the results of a study by Roberts et al (27) suggest that low total energy expenditure in infants is predictive of future weight gain. However, more recent cross-sectional (16–21) and prospective (28, 29) studies do not support the hypothesis that a low RMR is predictive of obesity, and there are limitations to the earlier work. Griffiths and Payne's (26) study of 2 groups of children matched for body size found lower RMRs per kilogram body mass in the group of 3–5-y-old children of obese parents than in the group of children of nonobese parents. However, a follow-up study conducted 12 y later showed no difference in percentage body fat between subgroups of the original children (30). The study of Roberts et al (27) examined infants predisposed to obesity by virtue of having an overweight mother. It was observed that infants who became overweight by 1 y of age had a low total energy expenditure (measured by using doubly labeled water) at 3 mo of age. However, RMR was not measured in this study and the postprandial metabolic rate did not differ significantly between groups. Thus, the evidence that a low RMR in infancy or early childhood is a cause of obesity is limited.
In conclusion, the present study addressed the limitations of previous cross-sectional studies examining RMR in obese and nonobese children by using more appropriate methods and criteria to determine body composition and obesity and by controlling for FM as well as for FFM when comparing RMR between groups. Our findings indicate that RMR is neither lower nor higher in obese boys than in nonobese boys and therefore suggest that the RMR of obese children is not abnormal.
ACKNOWLEDGMENTS
We thank the students from Catholic High School who participated in the study; the staff at Catholic High School, particularly Parameswari Thambusamy and Benjamin Kwok for allowing us into their school and assisting in the administration of this study; and Joyce Tan Bee Lian and Gillian Ng Bee Kit for their help with the RMR measurements.
REFERENCES