Influence of body composition and resting metabolic rate on variation in total energy expenditure: a meta-analysis13

Transcription

1 Original Research Communications Influence of body composition and resting metabolic rate on variation in total energy expenditure: a meta-analysis13 William H Carpenter, Eric T Poehiman, Maureen O Connell, and Michael I Goran ABTRACT The influence of body composition, resting metabolic rate (RMR), and sex on variation in total energy expenditure (TEE) was examined by using meta-analysis (13 published studies in a total of 162 adults) TEE (109 ± 32 MJ/d) was significantly correlated with body mass (BM; weighted mean converted r = 068, P < 005) and resting metabolic rate (RMR; weighted mean converted r = 066, P < 005) The correlation coefficients, regression coefficients, and intercepts for the regression between TEE and RMR and between TEE and BM were highly variable between studies When TEE was adjusted for RMR, there was no significant correlation with body fat, and females had a significantly lower TEE than males by 11 MJ/d (P = 0005) We conclude that 1) the relationships between TEE and RMR and TEE and BM are highly variable between studies with nonzero intercepts, precluding use of the traditionally used ratios of TEE to RMR or TEE to BM for data comparison; 2) after adjustment of TEE for RMR, TEE is not associated with adiposity, and women have a significantly lower TEE than men; and, 3) there are insufficient published data to develop prediction models for TEE in adults Am J Clin Nutr 1995;61:4-1O KEY WORD Physical activity, obesity, doubly labeled water, fat-free mass, adiposity, body weight, sex, fat mass Introduction The doubly labeled water technique is the first noninvasive means to measure total daily energy expenditure in free-living humans (1) The technique was introduced by Lifson et al (2) in the 1950s as an isotopic technique for measuring energy expenditure in animals, but was not used in human subjects for another 30 y (3) The delay in the application of the technique to humans was due to the time required for necessary advances in the sensitivity of isotope ratio mass spectrometry, which made it possible to reduce the isotope loading dose, thus making it economically feasible to perform studies in humans The energy expenditure data that has subsequently been gathered in free-living humans has proved invaluable in understanding the regulation of human energy metabolism in general Despite many narrative reviews on data collected by using the doubly labeled water technique (4-7), there have been no prior attempts to statistically review the data in a meta-analytic fashion None of the prior reviews have attempted to systematically examine the determinants of total energy expenditure The meta-analysis was prompted by several recent findings First, in previous work from our laboratory we showed that resting metabolic rate, which in absolute terms explains 60-70% of total energy expenditure, explains <50% of the vanation in total energy expenditure among healthy children (8), young adults (9), and elderly people (10) In addition, several studies from other laboratories suggest that total energy expenditure is not affected by body fatness (1 1, 12) although there is disagreement in the literature concerning this effect (13, 14) Finally, recent studies have suggested sex differences in resting metabolic rate (15-17), but this effect has not yet been exammed for total energy expenditure Knowledge of the major factors that influence total energy expenditure has important implications when considering the variables that need to be taken into account when adjusting total energy expenditure data to allow for comparison between subjects For example, total energy expenditure is traditionally expressed as a ratio, with either resting metabolic rate or body weight as the denominator The use of a ratio to express data that is body-size dependent assumes that the relationship between the dependent variable being adjusted (ie, total energy expenditure) and the variable being adjusted for (ie, resting metabolic rate or body mass) is linear and passes through the origin However, the relationship between total energy expenditure and either resting metabolic rate or body mass has never been examined in a large enough data set to test these underlying assumptions The purpose of this paper was therefore to examine the relationships among total energy expenditure, body composition, and resting metabolic rate Because there are heavy economical restraints in conducting such a study, our objectives were achieved by pooling data published in the literature and performing a meta-analysis The specific objectives for this 1 From the Division of Endocrinology, Metabolism and Nutrition, Department of Medicine, and the ims Obesity/Nutrition Research Center, College of Medicine, niversity of Vermont, Burlington 2 upported by The American Diabetes Association, the nited tates Department of Agriculture ( ), and the National Institute of Child Health and Human Development (R55 HD28720) (MIG); a grant from the National Institute of Aging (AG-07857), an RCDA from the National Institute of Aging, and the American Association of Retired Persons Andrus Foundation (ETP); and by the General Clinical Research Center (NIH RR-109) Address reprint requests to MI Goran, Energy Metabolism Research nit, Department of Nutrition ciences, The niversity of Alabama at Birmingham, 1675 niversity Boulevard, Birmingham, AL Received eptember 1, 1993 Accepted for publication July 12, 1994 Downloaded from wwwajcnorg at Norris Med Lib erials ect on July 11, Am J Clin Nutr 1995;61:4-1O Printed in A 1995 American ociety for Clinical Nutrition

2 TEE, RMR, AND BODY COMPOITION 5 analysis were as follows: 1) to examine the nature of the relationship between total energy expenditure and resting metabolic rate and body composition, thereby assessing the assumptions inherent in the use of ratios to adjust total energy expenditure; 2) to establish the regression coefficient and intercept for the relationship between total energy expenditure and resting metabolic rate and body composition; 3) to examine the effects of body fat and sex on variation in total energy expenditure; and 4) to examine whether existing data can be used to develop practical models for predicting total energy expenditure in adults Methods Description of data The data for this study were compiled from peer-reviewed journal articles that were identified from a Medline literature search, and cross-referencing For this study we were specifically interested in gathering data from free-living adult humans Therefore, the following criteria were used to select studies to be included in this analysis: 1) studies of free-living adult humans (ages 17 y); 2) studies of subjects with no disease states other than obesity; 3) no data from pregnant or lactating women; 4) measurement of free-living total energy expenditure by the doubly labeled water method; 5) no data from subjects undergoing atypical periods of physical activity (eg, soldiers and athletes); 6) no data from subjects living in developing countries; and 7) only studies published in peerreviewed journals, to ensure quality of the data A total of 13 articles met these criteria and are summarized in Table 1 Data were extracted from each study and compiled into a database For each individual value of total energy expenditure, the following information was also extracted into the database when reported: sex, body mass, standing height, resting metabolic rate, total body water determined by isotopic dilution, fat mass, fat-free mass, and percent body fat (from total body water), and fat-free mass determined by using techniques other than total body water (ie, underwater weight and bioelectrical impedance analysis) Information on some of these variables TABLE 1 Overview of total energy expenditure (TEE) studies used in the analysis was not always available; therefore, missing values existed in the data set The primary variables extracted were used to derive body mass index (wt/ht2), body surface area by using the equation of Dubois and Dubois (27), and predicted resting metabolic rate from height, weight, and age by using the equations of the World Health Organization (28) tatistical methods The data were analyzed by using meta-analysis techniques, which treat each study as an individual data point and weights each study according to the sample size Data from each study were analyzed separately to obtain Pearson product r values with total energy expenditure as the dependent variable and the physiological indexes under investigation as the independent variables The r values were transformed to a z statistic by using the following equation (29): z = 05 X log{(1 + r)/(1 - r)} (1) and the z statistic was weighted for sample size (n) by using this equation (29): z+ = w1z1 + w2z WkZk (2) where w, = (n1-3)/(n - 3k), i = 1 to k; n is the number of subjects in study i; N is the total number of subjects; and k is the total number of studies The 95% CIs for the weighted z values were calculated as ± 196/(N - 3k)#{176}5The z statistic and the upper and lower 95% CIs were then reconverted to Pearson r values by using this equation (29): r - (e22 1)/(e2z + 1) (3) To test the homogeneity of the correlation coefficients from the various studies, a Q statistic was calculated by using the following equation (29): Q = (n - 3)(z1 - z )2 (4) The Q statistic was used to test whether the Pearson r coefficients from each study were consistent with the assump- Downloaded from wwwajcnorg at Norris Med Lib erials ect on July 11, 2008 Reference TEE measurement period Age range Weight status d y chulz et al (18) (,z = 4 M, 2 F) Normal Welle et al (11) (,i = 38 F) Normal, 26 obese Livingstone et al (19) (n = 9 M, 5 F) Normal Wcsterterp et al (20) (ii 6 M, 9 F) Low to obese cale Ct al (21) (n = 4 M) Normal Roberts et al (22) (n = 14 M) Normal Prentice et al (23) (ft = 10 F) Normal to obese tein et al (24) (n = 15 F) Normal to obese Riumallo et al (25) (n = 6 M) nderweight Goran and Poehlman (10) (n = 7 M, 6 F) Normal to obese Goran et al (9) (n = 17 M) Normal choeller and van anten (3) (n = 3 M, 1 F) 14 NA Normal to obese Drews and tein (26) (n = 6) Normal NA, not available 14

3 6 CARPENTER ET AL tion that they came from a single, underlying population correlation and could therefore be pooled (29) The Q value was compared with the 5% level of the chi-square distribution with (k - 1) degrees of freedom (where k is the number of studies pooled) To test the significance of the pooled r values the null hypothesis tested was that the population correlation coefficient equalled zero The test statistic used to examine the null hypothesis was calculated as [z + X (N - 3k)#{176}5] (5) where N is the total number of persons per variable and k is the number of studies involved The null hypothesis was rejected if this test statistic was 196 Weighted mean regression equations were created by pooling both the regression coefficients and the intercept from each study by using the same methods as those described for averaging the r values Results Table 1 shows an overview of the studies that were used for this analysis A total of 13 studies were selected based on the previously outlined criteria These studies were published between 1982 and 1992 and originated from nine different research facilities Table 2 summarizes the physical characteristics of the subjects in the various studies A total of 162 individual measurements of total energy expenditure were reviewed, but there were numerous missing data for some of the associated variables The mean value for total energy expenditure was 113 ± 32 MJ/d, with a range of MJ/d As summarized in Table 2, the data analyzed represented a heterogenous group of 86 women and 70 men (sex was not reported for 6 subjects) The mean data in Table 2 are not weighted for sample size Table 3 presents the results of the weighted mean correlation coefficients for the studies examined All the independent variables showed significant mean weighted correlations with total energy expenditure The correlations with age, fat mass, and fat-free mass were not pooled because they failed the Q test for homogeneity; therefore, these values should be interpreted TABLE 2 Physical characteristics of study subjects with caution The Pearson product correlation coefficients were highly variable between studies, ranging from 042 to 089 for resting metabolic rate and from 018 to 10 for body mass The regression equations relating total energy expenditure to resting metabolic rate and body mass in each study are shown in Tables 4 and 5, respectively Weighting each of these regression equations for sample size the mean equation relating total energy expenditure (TEE) to resting metabolic rate (RMR) is as follows: TEE = (148 x RMR) + 14 Mi/d (6) and is shown in Figure 1 imilarly, the mean weighted equation relating total energy expenditure (TEE) to body mass (BM) is as follows: TEE = (0103 x wt, kg) + 40 Mi/d (7) and is shown in Figure 2 When total energy expenditure was adjusted for resting metabolic rate by using the above regression equation, the pooled partial correlation with body-fat content was not significant The data set was next used to examine sex differences in total energy expenditure Absolute total energy expenditure was significantly higher in males than in females by 3 1 Mi/d (102 ± 21 Mi/d in females, 132 ± 3 1MJ/d in males; P < 0001 by one-way analysis of variance) When adjusted for resting metabolic rate, total energy expenditure remained higher in men by 11 Mi/d (adjusted total energy expenditure in men = 119 ± 25 vs 108 ± 15 Mi/d in women; P = 0005 by one-way ANOVA) Discussion We analyzed the relationship among total energy expenditure, body composition, and resting metabolic rate in freeliving adult humans The main purpose of performing this analysis was to examine the factors influencing variation in total free-living energy expenditure in a large heterogenous database of healthy subjects Our major findings were that the relationships between total energy expenditure and resting metabolic rate and between total energy expenditure and body Downloaded from wwwajcnorg at Norris Med Lib erials ect on July 11, 2008 Variable (n = number of data points) Age (y) (n = 151) 336 ± 127 (17-78) Weight (kg) (n = 162) 722 ± 228 ( ) Height (cm) (n = 156) 1677 ± 106 ( ) Body mass index (kg/rn2) (n = 156) 252 ± 70 ( ) Body surface area (m2) (n = 156) 179 ± 028 ( ) Resting metabolic rate (Mild) (n = 113) 66 ± 09 (50-94) Predicted resting metabolic rate (Mild) (n 156) 63 ± 18 (36-166) Fat-free mass (kg)2 (n = 124) 479 ± 1 10 ( ) Fat mass (kg)2 (n = 124) 225 ± 182 ( ) Percent body fat (n 124) 298 ± 123 (70-617) Total energy expenditure (Mild) (n = 162) 113 ± 32 (45-217) Fat-free mass by other method3 (n = 66) 524 ± 127 ( ) ; ± D; range in parentheses n number of data points (different sample sizes reflect missing values) 2 Values determined by the isotopic dilution method 3 Body-composition data determined by techniques other than total body water

4 TABLE 3 Weighted mean correlations between total energy expenditure and various physiological variables TEE, RMR, AND BODY COMPOITION 7 Independent variable z to r (range) Clforr value Body mass (n = 13) 068 ( ) 057, 078 Body surface area (n = 12) 068 ( ) 057, 078 Resting metabolic rate (n = 7) 066 ( ) 053, 076 Predicted resting metabolic rate (n = 1 1) 061 ( ) 049, 071 Height (n = 12) 036 ( ) 019, 050 Fat-free mass from total body water (n = 10) 0752 ( ) 065, 083 Fat-free mass from alternative technique (n 4) 070 ( ) 080, 091 Age (n = 1 1) ( ) -038, 0 Fat mass from total body water (n = 10) 0452 ( ) 027, 059, Total energy expenditure is the dependent variable n = number of studies All correlations are significant 2 Correlations failed the Q test for sample homogeneity mass were highly variable between studies with variable nonzero intercepts This variability precludes use of the traditionally used ratios for comparing total energy expenditure data among individuals Finally, our analysis revealed that after adjustment for resting metabolic rate, total energy expenditure is not significantly influenced by adiposity and is significantly lower in women than men study Despite the fact that resting metabolic rate constitutes the largest component of total daily energy expenditure, the reviewed data spanning 113 measurements across 7 studies from various laboratories revealed that resting metabolic rate accounted for <50% of the variance in total energy expenditure in adults This finding confirms previous findings from our laboratory (9, 10) The new finding from the present analysis relates to the variability of the correlation coefficients, regression coefficients, and intercepts for the relationship between total energy expenditure and resting metabolic rate between studies (Table 4) We identified only seven studies in the literature that reported both total energy expenditure and resting metabolic rate in free-living adults Among these studies, the correlation coefficient between total energy expenditure and resting metabolic rate ranged from r = 042 (study 10) in a study of 13 healthy elderly men and women (10) to r = 089 (study 7) in a study of 10 normal-weight to obese women (23) The correlation coefficient between total energy expenditure and resting metabolic rate is an index of the heterogeneity of TABLE 4 Regression equations relating total energy expenditure to resting metabolic rate (RMR) in the various studies tudy Regression equation EE2 Mild chulz et al (18) (152 x RMR)+28 ±21 Welle et al (11) (137 X RMR)+23 ±14 Westerterp et al (20) (120 X RMR)+35 ±21 Prentice et al (23) (169 X RMR)+21 ±24 Riumallo et al (25) (172 X RMR)-22 ±08 Goran and Poehlman (10) (093 X RMR)+38 ± 17 Goran et al (9) (214 X RMR)-34 ±18 Describes the relationship between total energy expenditure (Mild) and RMR (Mild) within each study Weighted mean equation: (148 X RMR) tandard error of the estimated total energy expenditure within each the study group with respect to activity-related energy expenditure For example, at the one extreme in highly sedentary, hospitalized, and immobilized bum patients, the correlation between total and resting energy expenditure was found to be 096 (30) The high correlation coefficient is explained by the fact that the contribution of activity-related energy expenditure to total energy expenditure is relatively constant from one subject to another in these patients At the other extreme, the correlation between total energy expenditure and resting metabolic rate was 042 in a group of healthy elderly subjects (10) This relatively low correlation coefficient reflects that the contribution of activity-related energy expenditure to total energy expenditure in that population was highly variable from one subject to another Thus, the variability of the correlation coefficients describing the relationship between total energy expenditure and resting metabolic rate reflects the variable degree to which activity-related energy expenditure contributes to total energy expenditure from one subject to another within each particular subgroup of subjects studied A similar pattern was apparent for the correlation coefficients between total TABLE 5 Regression equations relating total energy expenditure to body mass (BM) in the various studies study tudy Regression equation EE2 Mild chulz et al (18) (020 x BM)-12 ±16 Welle et al (11) (0076 X BM)+49 ±13 Livingstone et al (19) (017 X BM)+06 ±27 Westerterp et al (20) (0064 X BM)+68 ±25 cale et al (21) (0017 X BM)+ 120 ±08 Roberts et al (22) (012 X BM)+63 ±26 Prentice et al (23) (0056 X BM)+45 ± 13 tein et al (24) (0078 x BM)+44 ±09 Riumallo et al (25) (024 X BM)- 18 ±07 Goran and Poehlman (10) (045 X BM)+68 ±02 Goran et al (9) (017 X BM)-001 ± 17 choeller and van anten (3) (021 X BM)-25 ±03 Drews and tein (26) (015 X BM)+21 ±37 Describes the relationship between total energy expenditure (Mild) and BM (kg) within each study Weighted mean equation: (0103 X BM) tandard error of the estimated total energy expenditure within each Downloaded from wwwajcnorg at Norris Med Lib erials ect on July 11, 2008

5 8 CARPENTER ET AL > 0 O - F- z MEAN RMR (MJ/day) FIG 1 Weighted mean regression equation between total energy expenditure (TEE) and resting metabolic rate (RMR) Each point represents the mean value of total energy expenditure and resting metabolic rate in each of the studies reporting data The line shown is the mean regression line among the studies (see Table 4) energy expenditure and body mass For body mass the correlation coefficient with total energy expenditure ranged from 018 (study 5) in a study of four men (21) to 10 (study 12) in a study of three men and one woman (3) Both of these extremes are explained by very low sample sizes in those studies The presence of significant and variable intercepts in the regression equations relating total energy expenditure to either resting metabolic rate (Table 4) or body mass (Table 5) invalidates the use of the traditionally used ratios (ie, TEEIRMR or TEE/body mass) for expressing total energy expenditure data This phenomenon has been similarly described for the relationship between resting metabolic rate and fat-free mass (31), and between peak oxygen consumption (y02) and both body mass and fat-free mass (32, 33) In both of these cases, significant intercepts in the regression equations do not statistically support the validity of using a mathematical ratio to express data as previously suggested (22, 28) On the basis of our findings we recommend that individual values of total energy expenditure can be compared by using a regression-based approach in which total energy expenditure is examined as a function of resting metabolic rate Resting metabolic rate is the preferred covariate because it serves as a proxy indicator of fat-free mass and represents the largest component of total energy expenditure When pooling the data from the various studies the correlations with fat-free mass failed the Q test for homogeneity In most studies the only available data on fat-free mass was that derived from information on total body water, and in this situation fat-free mass and energy expenditure are codependent variables because both are derived from total body water data Thus, pooling of the fat-free mass data may have failed the Q test because of the artificial nature of the association between total energy expenditure and fat-free mass Only four studies reported fat-free mass data using techniques other than total body water However, these correlations also failed the Q test, thus invalidating the pooling of data into a weighted mean The potential utility of fat-free mass to adjust total energy expenditure may therefore have to be reevaluated when additional data becomes available in which body-composition techniques other than body water dilution are used The lack of association between adjusted total energy expenditure and adiposity implies that total energy expenditure is not affected by adiposity, confirming other reports that have compared total energy expenditure in lean and obese subjects as a function of fat-free mass (1 1, 23) Interestingly, when total energy expenditure was adjusted for body mass, we found a significant inverse correlation with percent body fat (weighted mean partial correlation = -049) This may be an artifact because when using this approach both the dependent and independent variables are related to body mass The reported sex effect suggesting a lower adjusted total energy expenditure in women compared with men is supported by other studies showing sex differences in resting energy expenditure: 1) a 051-Mild lower resting metabolic rate in women than in men, independent of differences in body composition and the confounding effects arising from the menstrual cycle (16); 2) a significant 021-Mi/d lower resting metabolic rate, independent of differences in fat-free mass, fat mass, and Downloaded from wwwajcnorg at Norris Med Lib erials ect on July 11, 2008

6 TEE, RMR, AND BODY COMPOITION > 0 V -, F- z I I I I I I MEAN WEIGHT (kg) FIG 2 Correlation between total energy expenditure (TEE) and body mass Each point represents the mean value of total energy expenditure and body mass in each of the studies reporting data The line shown is the mean regression line among the studies (see Table 5) physical activity in women than in men (15) These sex differences are probably not due to differences in sex hormone concentrations because we have shown that resting energy expenditure is significantly higher in prepubescent boys than in girls, independent of body composition (17) We also examined the data set to see whether there was a combination of variables that could be of practical use for accurately predicting total energy expenditure on an individual basis After total energy expenditure was adjusted for resting metabolic rate, the partial correlations with body mass were nonsignificant Thus, body mass did not explain any of the unique variance in total energy expenditure after variation due to resting metabolic rate was taken into account In multipleregression analysis, we were unable to identify any combination of variables that could explain >50% of the variation in total energy expenditure between individuals Therefore, at present the data in the literature cannot be used in any practical manner to formulate accurate equations for predicting total energy expenditure in adults An obvious limitation in the data set pooled from the literature was the absence of markers of activity-related energy expenditure, the most variable component of total energy expenditure under both sedentary (31) and free-living (10) conditions In preliminary studies in both elderly adults (10) and children (8), we have shown that the inclusion of activity-related variables such as leisure time activity by questionnaire, V02 max, and heart rate act as proxy indicators of individual levels of activity-related energy expenditure, and improved the ability to predict individual levels of total energy expenditure We therefore suggest that identification of reliable proxy indicators of physical activity are essential for the modeling of total energy expenditure into practical equations for predicting energy requirements from doubly labeled water data A major limitation of this study relates to the possible inconsistencies between methods when combining data across laboratories However, as stated at the outset there are heavy economical constraints in conducting large-scale studies of total energy expenditure Ideally, cross-calibration of instrumentation and methods should be a prerequisite of crosslaboratory pooling of data In summary, the relationship between free-living total energy expenditure and both body composition and resting metabolic rate has been analyzed in adults by pooling data from the literature From this analysis we conclude that 1) the relationships between total energy expenditure and either resting metabolic rate or body mass have variable nonzero intercepts, thus precluding the use of the traditional ratio approach for data comparison; 2) after adjustment for resting metabolic rate, total energy expenditure is not associated with adiposity and is lower in women than in men; and 3) there is insufficient data in the literature to develop practical models for predicting total energy expenditure in adults References 1 choeller DA Measurement of energy expenditure in free-living humans by using doubly labeled water I Nutr 1988;118: Lifson N, Gordon GB, McClintock R Measurement of total carbon dioxide production by means of D2180 i Appl Physiol 1955;7: choeller DA, van anten E Measurement of energy expenditure in Downloaded from wwwajcnorg at Norris Med Lib erials ect on July 11, 2008

7 10 CARPENTER ET AL humans by doubly labeled water method I Appl Physiol 1982;53: choeller DA, Fjeld CR Human energy metabolism: what have we learned from the doubly labeled water method Annu Rev Nutr 1991; 11: Roberts B se of the doubly labeled water method for measurement of energy expenditure, total body water, water intake, and metabolizable energy intake in humans and small animals Can J Physiol Pharmacol 1989;67: Goran MI Application of the doubly labeled water technique for studying total energy expenditure in young children: a review Pediatr Exerc ci 1994;6: Goran MI, Poehlman ET, Johnson RK Energy requirements across the life span: new findings based on measurement of total energy expenditure with doubly labeled water Nutr Res (in press) 8 Goran MI, Carpenter WH, Poehlman ET Total energy expenditure in 4 to 6 year old children Am i Physiol 1993;264:E Goran MI, Beer WH, Poehlman ET, Wolfe RR, Young VR Variation in total energy expenditure in young, healthy free living men Metabolism 1993;42: Goran MI, Poehlman ET Total energy expenditure and energy requirements in healthy elderly persons Metabolism 1992;41: Welle 5, Forbes GB, tatt M, Barnard RR, Amatruda JM Energy expenditure under free-living conditions in normal-weight and overweight women Am i Clin Nutr 1992;55: Prentice AM, Black AE, Coward WA, et al High levels of energy expenditure in obese women Br Med i 1986;292: Roberts B, Young yr Fuss P, et al What are the dietary energy needs of elderly adults? Int i Obes 1992;16: Rising R, Harper RT, Fontveile AM Determinants of total daily energy expenditure: variability in physical activity Am i Clin Nutr 1994;59: Arciero Pi, Goran MI, Poehlman ET Resting metabolic rate is lower in women compared to men, i AppI Physiol 1993;75: Ferraro RT, Lillioja 5, Fontvieille AM, Rising R, Bogardus C, Ravussin E Lower sedentary metabolic rate in women compared to men i Clin Invest 1992;90: Goran MI, Kaskoun MC, Johnson RK Determinants of resting energy expenditure in young children I Pediatr 1994;125: chulz, Westerterp KR, Bruck K Comparison of energy expenditure by the doubly labeled water technique with energy intake, heart rate, and activity recording in man Am J Clin Nutr 1989;49: Livingstone MBE, Prentice AM, Coward WA, et al imultaneous measurement of free-living energy expenditure by the doubly labeled water method and heart-rate monitoring Am J Clin Nutr 1990; 52: Westerterp KR, Meijer GAL, aris WHM, oeters PB, Winants Y, Hoor Fl Physical activity and sleeping metabolic rate Med ci ports Exerc 1991;23: cale IL, Rumpler WV, Conway JM, Miles CW Comparison of doubly labeled water, intake-balance, and direct- and indirect-calorimetry methods for measuring energy expenditure in adult men Am I Clin Nutr 1990;52: Roberts B, Heyman MB, Evans WI, Fuss P, Tsay R, Young VR Dietary energy requirements of young adult men, determined by using doubly labeled water method, Am I Clin Nutr l991;54: Prentice AM, Davies HL, Black AE, et al, nexpectedly low levels of energy expenditure in healthy women Lancet 1985: tein 1?, Johnston FE, Greiner L Energy expenditure and socioeconomic status in Guatemala as measured by the doubly labelled water method Am I Clin Nutr 1988;47: Riumallo IA, choeller D, Barrera G, Gattas V, auy R Energy expenditure in underweight free-living adults: impact of energy supplementation as determined by doubly labeled water and indirect calorimetry Am I Clin Nutr 1989;49: Drews D, tein 1? Effect of bolus fluid intake on energy expenditure values as determined by the doubly labeled water method I AppI Physiol 1992;72: Du Bois D, Du Bois EF A formula to estimate the approximate surface area if height and weight are known, Arch Intern Med 1916;17: World Health Organization Energy and protein requirements WHO Tech Rep er 1985; Hedges LV, 01km I Combining estimates ofcorrelation coefficients in statistical methods for meta-analysis Orlando, FL: Academic Press, Goran MI, Peters EJ, Herndon DN, Wolfe RR Total energy expenditure in burned children using the doubly labeled water technique Am I Physiol 1990;259:E Ravussin E, Lillioja 5, Anderson TE, Christin L, Bogardus C Determinants of 24-hour energy expenditure in man Methods and results using a respiratory chamber I Clin Invest 1986;78: Buskirk ER, Taylor HL Maximal oxygen intake and its relation to body composition, with special reference to chronic physical activity and obesity I Appi Physiol 1957;11: Toth Mi, Goran MI, Ades PA, Howard DB, Poehlman ET An examination of data normalization procedures for expressing peak V02 data I Appl Physiol 1993;75: Downloaded from wwwajcnorg at Norris Med Lib erials ect on July 11, 2008