Physical activity related energy expenditure and fat mass in young children

Transcription

1 International Journal of Obesity (1997) 21, 171±178 ß 1997 Stockton Press All rights reserved 0307±0565/97 $12.00 Physical activity related energy expenditure and fat mass in young children MI Goran, G Hunter, TR Nagy and R Johnson Energy Metabolism Research Unit, Department of Nutrition Sciences, University of Alabama at Birmingham, Birmingham, AL; and Department of Nutrition Sciences, University of Vermont, Burlington, VT OBJECTIVE: To examine whether body fat content in pre-pubertal children is in uenced by physical activity related energy expenditure (AEE) and/or more qualitative aspects of physical activity. DESIGN: Cross-sectional study. SUBJECT: 101 pre-pubertal children were examined in Study 1: (age: y; weight: kg). In Study 2: 68 of the original children were re-examined (age: y; weight: y). MEASUREMENT: Fat mass (FM) and fat free mass (FFM) were determined by bioelectrical resistance and skinfolds; AEE was estimated from the difference between total energy expenditure (TEE) by doubly labeled water and postprandial resting energy expenditure (REE) by indirect calorimetry; qualitative information on activity was derived by questionnaire. RESULTS: AEE was signi cantly correlated with FFM (r ˆ 0.32 in both Studies) and body weight (r ˆ 0.28 in Study 1; r ˆ 0.29 in Study 2), but not FM. There were no signi cant relationships between AEE and any of the variables from the activity questionnaire in children (including TV time, playing time, and an accumulated activity index in h/ week). After adjusting for FFM, age, and gender, FM was inversely related to activity time in h/week (partial r ˆ in Study 1; partial r ˆ in Study 2) but not AEE (P > 0.5). CONCLUSION: After adjusting for FFM, age, and gender, a small portion of the variance in body fat mass in children ( 10%) is explained by time devoted to recreational activity, whereas none of the variance is explained by the combined daily energy expenditure related to physical activity. Keywords: obesity; energy expenditure; physical activity; assessment; activity questionnaire; exercise Introduction Numerous studies suggest that physical activity plays an important role in the maintenance of body weight in children. 1±4 However, the exact mechanism of the protective effect of physical activity has not been clearly established. Various elements of physical activity should be considered, including intensity, activity time, metabolic ef ciency, and the overall energy cost, as well as an appreciation of the type of physical activity (for example recreational, occupational, obligatory, and spontaneous movement). Thus, several aspects of physical activity need to be considered, including quantitative (for example the energy cost) and more qualitative aspects (for example type and duration of activity), as well as the effects of exercise on intermediary metabolism 5,6 and appetite. 7 From a quantitative aspect, the importance of physical activity in maintaining energy balance may be attributed to the fact that the energy cost of physical activity is the most variable component of total energy expenditure. 8 Correspondence: Dr MI Goran, The Energy Metabolism Research Unit, Department of Nutrition Sciences, University of Alabama, Birmingham, Alabama 35294, USA. Received 8 March 1996; revised 13 June/15 October 1996; accepted 28 October 1996 Due to the dif culty in measuring activity related energy expenditure in an un-obtrusive manner, most previous studies have used measures of physical activity which focus on recording movement, such as observation analysis, 9 questionnaires, 10 and motion sensors/accelerometry. 11 However, the relationship between simple activity indices and the combined energy cost of physical activity under free-living conditions has not been rigorously examined. This issue may be important since simple activity/inactivity indices (for example time spent performing speci c physical activities, television viewing) and the combined energy cost of physical activity (which combines time and intensity and includes all physical activities) address separate issues which may serve independent functions in maintaining energy balance. The doubly labeled water technique, which has successfully been applied in prior studies in children, 12,13 provides an alternative non-invasive and unobtrusive measure of the energy cost of daily physical activity when combined with measures of resting energy expenditure. Although not a major focus of some of our other reports, we have previously demonstrated that activity related energy expenditure in young children is not related to body fat, gender, or parental weight status. 13 The purpose of this paper is to extend these previous ndings and examine more thoroughly the relationship between activity energy

2 172 expenditure and body composition (fat and fat free mass), and qualitative information on physical activity as determined by an activity questionnaire. The speci c objectives were to examine: (1) the relationships between physical activity indices by questionnaire, physical activity energy expenditure over two weeks by doubly labeled water, and body composition; and, (2) whether the relationship between body fat mass and physical activity is explained by the combined energy cost of daily physical activity, or more qualitative aspects such as time spent performing recreational activities. Methods Subjects Our sample included 74 Caucasian (36 girls/38 boys) and 27 Mohawk children (17 girls/10 boys) aged y (designated as Study 1). In addition, a second set of measurements were obtained after a period of 1 y in 68 of the Caucasian children (designated as Study 2). The Caucasian children were recruited from Burlington, Vermont and the surrounding area (predominantly Chittenden County), and the Mohawk children from Akwesasne in upstate New York. Children were recruited by newspaper advertisements and word of mouth. There were no major inclusion/exclusion criteria other than the absence of major illness since birth. We have previously reported energy expenditure and body composition data in some of these children. 13 All studies were performed during the school year, but not during the winter months (December±February). The nature, purpose and possible risks of the study were carefully explained to both parents before obtaining consent to participate. In order to reduce the potential for behavioral bias with regard to changes in physical activity, subjects were informed that the oral dose of doubly labeled water and urine collections were for the purposes of measurement of body composition rather than energy expenditures of physical activity. The experimental protocol was approved by the Committee on Human Research for the Medical Sciences of the University of Vermont. General outline of protocol All studies were performed at the Obesity/Nutrition Research Center at the University of Vermont. On the evening prior to testing, children came to the laboratory for collection of baseline urine samples and oral dosing with doubly labeled water, and familiarization with the investigators and testing equipment. On the following morning, the children were asked to consume their usual breakfast at home before coming to the laboratory for assessment of: body composition by skinfolds and bioelectrical impedance, post-prandial resting energy expenditure by indirect calorimetry, total energy expenditure by doubly labeled water, and physical activity by questionnaire. Testing was completed by noon. The children returned to the laboratory two weeks after initial testing for repeated measurement of resting energy expenditure and body composition, and the collection of two additional urine samples. Measurement of energy expenditure components Postprandial resting energy expenditure (2±3 h after the children consumed their usual breakfast at home) was measured in duplicate (14 d apart) in children by indirect calorimetry using a portable Deltatrac metabolic monitor as previously described. 13 In our experience postprandial measurements of resting energy expenditure provide the environment for reproducible measures of energy expenditure in children; we have previously reported that the coef cient of variation for 169 repeat measures using this protocol is % for metabolic rate and % for respiratory quotient. 14 We have also previously established that resting energy expenditure is 11% higher using the postprandial protocol as compared to the typical postabsorptive conditions. 14 The postprandial protocol thus includes the average energy cost for mealinduced thermogenesis which occurs 2 h after a meal in children. 15 Total energy expenditure was measured over 14 d under free-living conditions with the doubly labeled water technique, using a protocol that has a theoretical precision of less than 5% for assessment of CO 2 production rate as previously described. 13 Brie y, four timed urine samples were collected after oral dosing with doubly labeled water, two the morning following dosing, and two in the morning 14 d later with a loading dose of 0.15 g of H 18 2 O and 0.12 g of 2 H 2 O per kg body mass. Samples were analyzed in triplicate for H 18 2 O and 2 H 2 O by isotope ratio mass spectrometry. Samples from Study 1 were all analyzed at the Biomedical Mass Spectrometry Facility at the University of Vermont as previously described. 13 The samples from Study 2 were analyzed at The Energy Metabolism Research Unit in The Department of Nutrition Sciences at the University of Alabama at Birmingham. The new facility at UAB consists of a Fisons Optima isotope ratio mass spectrometer and samples are prepared and analyzed in a similar fashion to that previously described, 13 except that carbon dioxide is analyzed for oxygen-18 content by continuous ow, isotope ratio mass spectrometry. The intraassay standard deviation for triplicate analysis of samples at both laboratories is approximately 4 dels/ mille and 0.20 dels/mille for deuterium and oxygen- 18, respectively. In addition, when all samples for deuterium and oxygen-18 were re-analyzed in seven subjects, values of total energy expenditure were in close agreement (4.3%) from both analysis. CO 2 production rate was determined using Equation R2 of Speakman et al, 16 and CO 2 production rate was

3 converted to energy expenditure. Using equation 12 of de Weir 17 and the mean value or the food quotient of the children's diet (0.90) from a food frequency questionnaire. 18 The value of 0.90 is higher than expected and maybe due to inaccuracies in the food frequency questionnaire in children. 19 If the real food quotient is 0.85, total energy expenditure values would be systematically increased by 4.7%. Because samples were analyzed in two different laboratories, a cross-validation study was performed. Urine samples from seven subjects were analyzed in both laboratories and the data were compared by paired t-test. As shown in Table 1, there was excellent agreement in the derived turnover rates and dilution spaces for both isotopes. However, a signi cant difference by paired t-test was observed for the dilution space ratio derived when samples were analyzed in Vermont ( ) vs analyzed in Alabama ( ). Due to the difference in dilution space ratios between laboratories we used two different group mean values of the dilution space ratio for data calculation, as recommended by Speakman et al. 16 The group mean dilution space ratio for all studies analyzed in Vermont was compared to a mean value of for samples analyzed in Alabama. These values were signi cantly different from each other and both were signi cantly different from the previously described population mean value of A 0.01 decrease in the deuterium to oxygen dilution space ratio increases the derived value of total energy expenditure by 3%. However, use of these group speci c ratios improves the overall accuracy of the doubly labeled water technique. 16 Physical activity related energy expenditure was estimated from the difference between total energy expenditure and postprandial resting energy expenditure. No correction for the thermic effect of feeding was necessary, since resting energy expenditure was measured under post-prandial conditions. Unfortunately, estimates based on the difference between two measures are prone to error propagation which can become quite large particularly when the difference between the measures is small. For example, in a very inactive ve year old child, total energy expenditure might be 5.0MJ/d ( 1200 kcal/d) with a resting energy expenditure of 4 MJ/d ( 1000 kcal/d), and an estimated activity energy expenditure of 1 MJ/d ( 200 kcal/d). Assuming that the only source of error in the system is a 10% measurement error in total energy expenditure the projected uncertainty for the difference between total and resting energy expenditure is MJ/d (namely 50% measurement uncertainty). Because of this tremendous error propagation `true scores' were used to improve the precision of activity related energy expenditure estimates, according to the description of Crocker and Algina. 20 True scores take into account the measurement reliability of total and resting energy expenditure before subtracting them from one another. The reliability score is based on a Pearson correlation between two repeat measures. For resting energy expenditure a correlation of 0.8 was used based on previous data in children. 14 For total energy expenditure we assumed a correlation of 0.6 which is based on a reliability study that we have previously performed in adults under test-retest conditions. 21 The use of `true scores' has no effect on group mean data, but improves the precision of an estimate (activity energy expenditure) that is based on the difference between two measurements (total and resting energy expenditure). Physical activity questionnaire Qualitative information on physical activity patterns in children were estimated using the structured activity questionnaire of Kriska et al 10. Mothers were interviewed in the presence of the child. The questionnaire is designed to assess hours per day sleeping, viewing television and performing various recreational physical activities. Leisure time physical activity was also assessed in both parents using the Minnesota structured interview 22 as previously described. 23 Measurement of body composition Fat and fat free mass were estimated using an equation developed in children with dual energy X-ray absorptiometry as a criteria method. 24 This equation estimates fat and fat free mass from triceps skinfold, subscapular skinfold, body weight, height 2 /resistance and gender with an R 2 of Body weight, skinfold thickness, and whole body resistance (RJL 101A) were measured in duplicate as previously described Table 1 Cross-validation of doubly labeled water between two laboratories Parameter Value obtained by analysis invermont Value obtained by analysis in Alabama Turnover for deuterium (days-1) Turnover for oxygen-18 (days-1) Dilution space for deuterium (moles) Dilution space for oxygen-18 (moles) Deuterium : oxygen-18 dilution ratio * Total energy expenditure (MJ/d) * Signi cant difference by paired t-test (P < 0.05).

4 174 Statistics Data are presented as mean values standard deviation (s.d.), unless stated otherwise. Pearson product moment correlations, partial correlations, and multiple regression techniques were used to examine relationships among activity energy expenditure and body weight, fat free mass, fat mass, and activity indices from the questionnaire. Data for Study 1 and Study 2 were analyzed separately and were handled as two discrete studies. The intention of the analysis was not to examine the reproducibility of data over time. We feel that such an analysis is not justi ed because of the signi cant (P < 0.05) increases in body weight ( kg), fat free mass ( kg), and fat mass ( kg) that occurred over time. Rather, our intention was to examine the relationships under investigation at two discrete time-points. Data were analyzed on a personal microcomputer using SAS for Windows version 6.08 (Carey, North Carolina). Table 3 Summary of signi cant correlations between activity energy expenditure (AEE) and other variables Correlation with AEE STUDY 1: Fat free mass (kg) 0.32 (n ˆ 101) Weight (kg) 0.28 (n ˆ 101) Activity Index (h/week) NS STUDY 2: Fat free mass (kg) 0.32 (n ˆ 65) Weight (kg) 0.29 (n ˆ 65) Activity Index (h/week) NS Correlations were signi cant at P < 0.05; other variables with no signi cant correlations were: sleep time (h/day); play time (h/ day); time spent watching television (h/day); activity index; body fat mass (kg); leisure time physical activity in parents. Results Descriptive statistics for variables being examined are shown in Table 2. Because of missing data we did not have complete data in all subjects for some variables and the sample sizes for each variable are given in the Table. In Study 2, we observed three values of AEE that were negative and these numbers were reproducible when all doubly labeled water samples were reanalyzed. Data from these three subjects were excluded from further analysis. A summary of the correlations and partial correlations between activity energy expenditure and other variables is shown in Table 3. There were signi cant (P < 0.05) and reproducible correlations between activity energy expenditure and fat free mass and body weight (Table 3 and Figure 1). There were no signi cant correlations (P > 0.05) in either Study 1 or Study 2 between activity energy expenditure and any other variable examined including body fat mass, activity index (Figure 2), time spent watching television, and leisure time physical activity in parents. Figure 1 Activity energy expenditure (AEE) by doubly labeled water plotted versus fat free mass in Study 1 ( lled circles) and Study 2 (open circles). Solid lines are regression lines. Figure 2 Activity energy expenditure (AEE) by doubly labeled water plotted versus activity index (hours/week) by questionnaire in Study 1 ( lled circles) and Study 2 (open circles). No signi cant correlation in either Study 1 or Study 2. Table 2 Descriptive statistics Variable Study1: mean s.d. (n) Study 2: mean s.d. (n) Age (y) (101) (65) Weight (kg) (101) (65) FFM (kg) (101) (65) FM (kg) (101) (65) Play time (h/d) (82) (59) TV time (h/d) (83) (58) Activity Index (h/week) (83) (58) AEE (MJ/d) (101) (65) TEE (MJ/d) (101) (65) REE (MJ/d) (101) (65) FFM ˆ fat free mass; FM ˆ fat mass; AEE ˆ physical activity related energy expenditure over 14 days; TEE ˆ total energy expenditure over 14 days; REE ˆ resting energy expenditure; see Methods for measurement details.

5 Table 4 Summary of correlations and partial correlations between body fat mass and activity energy expenditure (AEE) and activity indices by questionnaire 175 Correlation with fat mass Partial correlation with fat mass (adjusting for FFM, gender, and age) STUDY 1: AEE NS (n ˆ 101) NS (n ˆ 83) Activity index (h/week) NS (n ˆ 83) (P ˆ 0.03; n ˆ 83) STUDY 2: AEE NS (n ˆ 65) NS (n ˆ 58) Activity index (h/week) (P ˆ 0.01; n ˆ 58) (P ˆ 0.02; n ˆ 58) We next examined the relationship of body fat mass with physical activity related energy expenditure and other activity indices derived from the questionnaire. The results of this analysis are summarized in Table 4. Body fat mass was not signi cantly related to physical activity related energy expenditure over two weeks derived from the doubly labeled water technique. These relationships remained similar after adjusting for variables such as fat free mass, age, and gender. After adjusting for fat free mass, gender and age, body fat mass was signi cantly and inversely related to physical activity in hours per week derived for the questionnaire in Study 1 (partial r ˆ 70.27, P ˆ 0.03) and Study 2 (partial r ˆ 70.32, P ˆ 0.02), as shown in Table 4. Discussion Many studies have examined the relationship between body composition and physical activity in children, 1±4 but none have examined the effects of both activity energy expenditure and activity time. Therefore, this study is unique in that it is the rst to explore the relationships among activity energy expenditure, activity time, and body composition in children. Our data show that activity energy expenditure is highly variable, and only weakly related to fat free mass (Figure 1) and body weight. In addition, the daily energy cost of physical activity is unrelated to time spent performing physical activities (Figure 2). Moreover, we demonstrate that body fat mass is more related to activity time than to the combined energy cost of physical activities (Table 4). These ndings were reproducible in two separate studies. Collectively, these results suggest that time devoted to recreational physical activity may be a more important factor in the maintenance of whole body energy stores than the combined daily energy cost of physical activity. Previous studies in children have demonstrated negative relationships between body fatness and physical activity using methods based on observation and questionnaires. 1±3 In addition, the epidemiological study by Dietz et al 4 demonstrated that increased television viewing was a major risk factor for the development of obesity in children. These prior studies imply that decreased physical activity plays a key role in the development of obesity. However, this hypothesis remains questionable because of the crude measures of physical activity that were used in previous studies. We examined the relationship between body fatness and activity using a more rigorous measure of the energy cost of physical activity based on the doubly labeled water technique. However, the two measures of physical activity (time of activity by questionnaire and the combined energy cost by doubly labeled water) were unrelated to each other (Figure 2). Moreover, differences in body fatness were explained by differences in time spent performing activities and not the combined energy cost (Table 4). The two indices of activity that were examined in this study measure different parameters of physical activity (namely combined daily energy cost, vs time spent in physical activity) and must be considered separately. The activity index derived by the questionnaire is simply a re ection of time devoted to recreational activities. It is not dependent on factors such as intensity or ef ciency. Activity energy expenditure by doubly labeled water re ects time and intensity of all physical activities and also has a built in ef ciency factor. The ef ciency factor may be important in children since ef ciency is known to improve during maturation. 25 The nding that time devoted to physical activity is more important than the combined energy cost implies that long bouts of physical activity (which can be sustained at low intensity) may be more protective than shorter bouts of high intensity activity. These two scenarios may have the same energy cost, but extended periods of low intensity exercise may be more bene cial because of the promotion of an active lifestyle and the avoidance of a sedentary lifestyle that has been implicated to promote food intake, especially snacking. 26 Previous studies in adults however refute this idea. For example Trembley et al, 27 showed that men and women who routinely perform vigorous exercise (above 9 METS) were leaner than those routinely exercising at lower intensities (below 5 METS). Nevertheless, there are several possible reasons why increased bouts of physical activity do not necessarily ensure an increase in activity related energy expenditure. First, we have previously noted that total energy expenditure was unchanged in elderly subjects during

6 176 Figure 3 Propagated error in TEE-REE as a function of the ratio of TEE : REE at 3 different levels of precision for TEE. TEE is total energy expenditure and REE is resting energy expenditure. intense endurance training, due to a compensatory reduction in activity energy expenditure (increased sedentary activities) during the remainder of the day. 23 Thus, physical activity energy expenditure is a source of `plasticity' that allows changes in total energy expenditure to occur in the face of external stimuli such as intense exercise. 23 Second, activities such as resistance exercise and maintenance of posture during sitting and standing have a low energy cost. Thus, physical activity energy expenditure may not always be representative of time spent exercising, because the daily energy expended in physical activity includes the combined energy cost of all physical activities including sedentary activities. Our initial hypothesis was that the relationship between body fat mass and physical activity would be stronger when using doubly labeled water to measure the energy cost of physical activity. We postulated a role for the energy cost of physical activity because it is the most variable component of total energy expenditure and thus plays a key role in determining energy balance. However, our data provide further evidence against an inverse association between physical activity energy expenditure and body fat. The lack of this relationship was also apparent in the study by Livingstone in children. 12 Although the correlation was not reported in the original paper, the published individual data were used to compute a correlation coef cient of r ˆ 0.05 (not signi cant) between activity related energy expenditure and percent body fat. These ndings in children are in agreement with a meta-analysis showing no signi cant relationship between total energy expenditure and fat mass in adults after adjusting for resting energy expenditure. 28 Using a similar approach in the current data set, we also did not observe any relationships between total energy expenditure and fat mass after adjusting for resting energy expenditure in either Study 1 or Study 2. Thus, crosssectional data in children and adults suggest that physical activity related energy expenditure is not necessarily a predictor of energy imbalance. In interpreting these data we also considered why there was no relationship between body fat mass and physical activity energy expenditure. The lack of relationship was evident even after adjusting for fat free mass, age, time of activity, and gender in a multiple regression model. There are several possible reasons why activity energy expenditure may be unrelated to body fatness. First, although activity energy expenditure is determined over two weeks, this measure does not take seasonal variation into account. We avoided this potential confounding factor by not conducting studies in the summer months or the severe winter months, periods of time in which seasonality is likely to in uence activity in children. In addition, when season of study was included in the multiple regression model described above, there was still no relationship between fat mass and activity energy expenditure. Second, activity energy expenditure in children is in uenced by ef ciency which improves during childhood growth. 25 We did not perform any measure of ef ciency in this study but the lack of relationship between fat mass and activity energy expenditure remained apparent after adjusting for fat free mass and age (which may be proxy measures of the development change in ef ciency). The third possible explanation is that activity energy expenditure may be an imprecise measure. There are no test/re-test studies of physical activity energy expenditure in children but precision can be estimated using a propagation of error approach. 8 As shown in Figure 3, the projected error in total energy expenditure minus resting energy expenditure approaches 50% when the ratio of these two components to one another is 1.4±1.6, as seen in several studies in children. 13,29 Because of this limited precision, we suggest correcting values of total and resting energy expenditure for their `true score' as described in the Methods section; this approach improves the precision of an estimate that is derived from the difference between two independent measures. 20 In addition, repeat measures of physical activity energy expenditure should be performed to improve precision. In this study we performed duplicate observations showing that the main relationships observed (or lack of) were persistent in two discrete studies performed one year apart. Only a few studies have measured the energy cost of physical activity in children under free-living conditions, 12,13,29 and several other ndings merit discussion. First, all of the studies that have been performed in children are unanimous in showing that the energy expended in physical activity is 400 kcal/d (1.67 MJ/d). This low level of physical activity energy expenditure is consistent with several studies showing that children only spend a small fraction of their active time in high intensity activities. 9,30 In

7 Table 5 Reliability of total, resting and activity energy expenditure (kcal/d) in 7 subjects 177 Subject TEE 1 TEE 2 TEE CV (%) REE 1 REE 2 REE CV (%) AEE 1 AEE 2 AEE CV (%) TEE is total energy expenditure; reliability was assessed by repeated analysis of all doubly labeled water samples and standards. REE is resting energy expenditure; reliability was assessed by repeated measures 14 days later under similar conditions. AEE is activity energy expenditure estimated from the difference between TEE and REE. Reliability is expressed as the coef cient of variation (CV). addition, physical activity energy expenditure is only weakly related to fat free mass (r ˆ 0.32) and body weight (r ˆ 0.28±0.29). These observations may seem counter-intuitive since the energy cost of physical activity is usually thought to be a function of body weight (at least for weight bearing exercise). However, the physical activity being measured combined all physical activities throughout the day (including exercise, spontaneous and obligatory physical movement) and is a combination of weight-bearing and non-weight bearing activities. The implications of this nding are that both body weight and fat free mass only explain a small amount of the variance in physical activity related energy expenditure. Physical activity questionnaire recall methods have proven useful for large scale epidemiological studies. Studies using a simple assessment of television viewing has led to the hypothesis that increased sedentary behavior is a risk factor for obesity. 4 For this study we used the questionnaire derived by Kriska et al. 10 As with other physical activity questionnaire techniques, the Kriska questionnaire reliability (Spearman rankorder correlation for test/re-test was 0.35) and validity (0.8 for correlation of data with caltrac accelerometer) are limiting. The lack of relationship between activity questionnaire data and measurements of physical activity energy expenditure in two separate studies, has several important implications. First, the data suggest that simple indices of inactivity (for example time spent in physical activity, time spent viewing television) may not be representative of the daily energy expended in physical activity by children. Second, simple activity questionnaires will need to be modi ed in order to be of value in estimating the daily energy expended in physical activity. However, as our data point out, future development of physical activity instruments may need to consider quantitative as well as qualitative information on physical activity. Because of the dif culty in accurately measuring energy expenditure components in children we have paid particular attention to reliability issues. Firstly, we have previously shown that our measurements of resting energy expenditure are highly reliable to within 5.4% on average. 14 Secondly, for this study we examined the reliability of our doubly labeled water analysis by re-analyzing all samples from seven subjects and showed that the data were reproducible to within 4.3% on average. These data are summarized on Table 5. For this analysis we chose to re-analyze samples from subjects with high, medium and low total energy expenditure values. As seen on Table 5, the data are reproducible with the exception of one of the subjects with a low total energy expenditure. Also, shown on Table 5 are the two separate measures of resting energy expenditure from these subjects. Finally, we have included the estimates of activity energy expenditure (derived from total minus resting) on Table 5, and demonstrate the error propagation that occurs during the procedure. Because of overall concerns with the reliability estimates for activity energy expenditure, not only due to the error propagation but also due to the potential for this variable to change over time, we repeated our entire experiment one year later and found that our overall ndings were entirely reproducible (for example see Table 4). Thus despite the intrinsic measurement dif culties associated with these types of studies, we feel that we have taken adequate measures to ensure the reliability of our ndings. Conclusions We have examined the relationship of physical activity relate energy expenditure with body composition and simple activity indices. We show that the energy expended in physical activity is low in children ( 400 kcal or 1.67 MJ/d on average, or 20% of total energy expenditure) and is weakly related to fat free mass and body mass. Although, our study is limited by the use of a crude physical activity questionnaire, the analysis shows that simple indices of activity in children, such as activity by questionnaire (including time spent watching television and a cumulative estimate of time spent in structured activities) are not related to the daily energy cost of physical activity measured over two weeks under free living conditions. Furthermore, after adjusting for fat free mass, age, and gender, a small portion of the variance in body fat mass ( 10%) is related to activity time, whereas none of the variance is explained by the combined energy cost of physical activities measured over two weeks. These data suggest that children with

8 178 greater fat mass devote less time to recreational physical activity, but this is not re ected in differences in the combined daily energy cost of physical activity. Thus, activity time, rather than activity energy expenditure may be a more signi cant factor in the maintenance of whole body energy stores. Acknowledgements We thank the following for their support, The American Diabetes Association (MIG), the United States Department of Agriculture (MIG), The National Institute of Child Health and Human Development (MIG; R29 HD 32668), and, in part by the General Clinical Research Center (National Institutes of Health Grant No. R-109). References 1 Bar-Or O. Physical activity and physical training in childhood obesity. J Sports Med Phys Fitness 1993; 33: 323± Obarzanek E, Schreiber GB, Crawford PB, Goldman SR, Barrier PM, Frederick MM, Lakatos E. Energy intake and physical activity in relation to indexes of body fat: the National Heart, Lung, and Blood Institute Growth and Health Study. Am J Clin Nutr 1994; 60: 15±22. 3 Wilkinson PW, Parkin JM, Pearlson G, Strong H, Sykes P. Energy intake and physical activity in obese children. Br Med J 1977; 6063: Dietz WH, Gortmaker SL. Do we fatten our children at the television set? Obesity and television viewing in children and adolescents. Pediatrics 1985; 75: 807± Poehlman ET. A review: exercise and its in uence on resting energy metabolism in man. Med Sci Sports Exer 1989; 21: 515± Forbes GB. Exercise and body composition. J Appl Physiol 1991; 70: 994± Kissileff HR, Pi-Sunyer FX, Segal K, Meltzer S, Foelsch PA. Acute effects of exercise on food intake in obese and nonobese women. Am J Clin Nutr 1990; 52: 240± Goran MI, Beer WH, Poehlman ET, Wolfe RR, Young VR. Variation in total energy expenditure in young, healthy free living men. Metabolism 1993; 42: 487± Bailey RC, Olson J, Pepper SL, Porszasz J, Barstow TJ, Cooper DM. The level and tempo of children's physical activities: an observational study. Med Sci Sports Exer 1995; 27: 1033± Kriska AM, Knowler WC, LaPorte RE, Drash AL, Wing RR, Blair SN, Bennett PH, Kuller LH. Development of questionnaire to examine relationship of physical activity and diabetes in Pima Indians. Diabetes Care 1990; 13: 401± Freedson PS. Electronic motion sensors and heart rate as measures of physical activity in children. J Sch Health 1991; 61: 220± Livingstone MBE, Coward WA, Prentice AM, Davies PSW, Strain JJ, McKenna PG, Mahoney CA, White JA, Stewart CM, Kerr M-JJ. Daily energy expenditure in free-living children: Comparison of heart-rate monitoring with the doubly labeled water ( 2 H 2 18 O) method. Am J Clin Nutr 1992; 56: 343± Goran MI, Carpenter WH, McGloin A, Johnson R, Hardin M, Weinsier RL. Energy expenditure in children of lean and obese parents. Am J Physiol 1995; 268: E917±E Goran MI, Nagy TR. Effect of the pre-testing environment on measurement of metabolic rate in children. Int J Obes 1996; 20: 83± Maffeis C, Schutz Y, Zoccante L, Micciolo R, Pinelli L. Meal-induced thermogenesis in lean and obese prepubertal children. Am J Clin Nutr 1993; 57: 481± Speakman JR, Nair KS, Goran MI. Revised equations for calculating CO 2 production from doubly labeled water in humans. Am J Physiol 1993; 264: E912±E de Weir JB. New methods for calculating metabolic rate with special reference to protein metabolism. J Physiol 1949; 109: 1±9. 18 Willett WC, Sampson L, Stampfer MJ, Rosner B, Bain C, Witschi J, Hennekens CH, Speizer FE. Reproducibility and validity of a semiquantitative food frequency questionnaire. Am J Epidemiol 1985; 122: 51± Kaskoun MC, Johnson R, Goran MI. Comparison of energy intake by food frequency questionnaire with total energy expenditure by doubly labeled water in young children. Am J Clin Nutr 1994; 60: 43± Crocker L, Algina J. In: Introduction to Classical and Modern Test Theory. Harcourt Brace Jovanovich College Publishers: Fort Worth, 1986, pp 146± Goran MI, Poehlman ET, Danforth EJ. Experimental reliability of the doubly labeled water technique. Am J Physiol Endocrinol Metab 1994; 266: E510±E Taylor HL, Jacobs DR, Schucker B. Questionnaire for the assessment of leisure time physical activities. J Chron Dis 1978; 31: 741± Goran MI, A Poehlman ET. Endurance training does not enhance total energy expenditure in healthy elderly persons. Am J Physiol 1992; 263: E950±E Goran MI, Driscoll P, Johnson R, Nagy TR, Hunter GR. Cross-calibration of body composition techniques against dual-energy X-ray absorptiometry in young children. Am J Clin Nutr 1996; 63: 299± Forster M, Hunter GR, Hester DJ, Dunaway D, Shuleva A. Aerobic capacity and grade-walking of children 5±9 years: A longitudinal study. Paediatr Exer Sci 1994; 61: 31± Taras HL, Sallis JF, Patterson TL, Nader PR, Nelson JA. Television's in uence on children's diet and physical activity. J Dev Behav Pediatr 1989; 10: 176± Tremblay A, Despres JP, Leblanc C, Craig CL, Ferris B, Stephens T, Bouchard C. Effect of intensity of physical activity on body fatness and fat distribution. Am J Clin Nutr 1990; 51: 153± Carpenter WH, Poehlman ET, O'Connell M, Goran MI. In uence of body composition and resting metabolic rate on variation in total energy expenditure: A meta analysis. Am J Clin Nutr 1995; 61: 4± Fontvieille AM, Kriska AM, Ravussin E. Decreased physical activity in Pima Indian compared to Caucasian children. Int J Obes 1993; 17: 445± Gilliam TB, Freedson PS, Geenen DL, Shahrary B. Physical activity patterns determined by heart rate monitoring in 6±7 year-old children. Med Sci Sports Exer 1981; 13: 65± 67.