| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research Communications |
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Objective: We developed equations to predict adult energy requirements from simple anthropometric and laboratory measures by using the doubly labeled water method to determine each subject's total energy expenditure (TEE), which is equal to usual energy requirements in weight-stable individuals.
Design: This was a cross-sectional study conducted with 93 healthy, free-living adults [44 men and 49 women; body mass index range (in kg/m2): 18.4–31.8] aged 18–81 y. Body fat and fat-free mass were measured by underwater weighing, physical activity was estimated by using activity monitors, and resting energy expenditure was determined by indirect calorimetry. Information on anthropometric variables and reported strenuous activity was also collected.
Results: Three regression equations were developed and verified for accuracy by using bootstrap analysis and doubly labeled water data published by other research groups. The first equation used information on only age, weight, height, and sex and had an SEE for prediction of TEE of 1.80 MJ/d. The second and third equations used different combinations of basic and laboratory data and had SEEs of 1.55 and 1.65 MJ/d, respectively. With use of the same analytic approaches, the RDAs for energy were shown to significantly underestimate usual energy needs by 
10%; the extent of underestimation was significantly greater for subjects with high TEEs than for subjects with low TEEs.
Conclusion: Regression equations based on doubly labeled water measurements of TEE appear to be more accurate than the current RDAs for predicting energy requirements in healthy, nonobese adults living in affluent countries.
Key Words: Energy requirements • energy expenditure • isotopes • body composition • exercise • regression equations • doubly labeled water method • recommended dietary allowances • adults
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Recently, the accuracy of the RDAs has been questioned on the grounds that measurements of total energy expenditure (TEE) obtained by using the accurate doubly labeled water technique (3, 4) in healthy, weight-stable subjects are typically considerably higher than anticipated on the basis of the RDAs (5–10). There is therefore a need for equations that can be used to predict the energy requirements of individuals and groups more accurately than the current RDAs. Two studies reported relations between TEE determined by doubly labeled water and simple laboratory measures in older adults and children (11, 12) (n = 13 and 30, respectively), which suggests the feasibility of developing a regression equation based on this approach. The numbers of subjects in those studies, however, were too small for development or cross-validation of generally applicable equations. The study described here was designed to develop and cross-validate regression equations to predict TEE and energy requirements in a larger group of healthy adults.
|
SUBJECTS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. The studies were conducted at the Jean Mayer US Department of Agriculture Human Nutrition Research Center on Aging at Tufts University, Boston, and the Clinical Research Center at the Massachusetts Institute of Technology, Boston, with permission from the respective institutes. Informed consent was obtained from all volunteers before the start of the study.
|
Measurement of total energy expenditure
At the start of each TEE measurement, subjects were given a mixed dose of doubly labeled water (2H218O) containing 0.15 g H218O/kg body wt and 0.07 g 2H2O/kg body wt. The dose was given early in the morning after subjects had fasted overnight and after the collection of 2 baseline urine specimens. Subjects then fasted for 4 h while hourly urine specimens were collected. Urine specimens at the second void of the day were then collected daily at home by the subjects, at about the same time of day that the isotope dose was administered. The final samples were collected at the research center when subjects returned for testing. Isotope analyses were performed by using isotope ratio mass spectrometry (SIRA-10; VG Isogas, Middlewich, United Kingdom). Samples were prepared for 1H:2H analysis by using the zinc-reduction reaction and for 16O:18O analysis in carbon dioxide by using the H2O-CO2 equilibrator system (14). Analyses for 1H:2H abundances were conducted in triplicate on each urine or standard sample and analyses for 16O:18O abundances were conducted in duplicate.
TEE was calculated by using standard equations (15). In these calculations, the values for respiratory quotient used to convert measured values for carbon dioxide production rate to TEE were obtained by using a food quotient. For the women, the food quotient was calculated from 7-d weighed food intake records kept by the subjects during the doubly labeled water measurement period, as described elsewhere (7). For the men, the food quotient was calculated from the diet provided to them by the research center because they were participating concomitantly in a study in which they were provided with all food and were required to maintain a constant body weight. For all subjects, food quotients were modified to take into account small changes in energy balance during the measurement period, assuming proportional changes in body weight and body fat (16).
Measurement of resting energy expenditure
Each subject's REE was determined by indirect calorimetry for 30 min after he or she had fasted overnight for 12–13 h. Before the REE measurement, subjects rested quietly for 
30 min under thermoneutral conditions. They were familiarized with the calorimetry procedure before the measurement. Routine alcohol burn experiments in our laboratory showed that the calorimeter measurements were accurate to ±1%. Rates of energy expenditure were calculated from oxygen consumption and carbon dioxide production by using de Weir's equation (17). The REE of each subject was also predicted (pREE) by using the FAO/WHO/UNU equations (1) on the basis of age group, sex, and body weight.
Estimation of physical activity
Activity monitors were worn like a pager at the waist to estimate physical activity. The monitors are accelerometers and detect motion in the horizontal plane. When a subject moves, a cantilevered beam in the monitor (supported at one end) bends and emits a current proportional to the force acting on it. A small computer in the unit then plots an acceleration curve and integrates the area under that curve for the estimation of the amount of physical activity. Activity monitors predict REE by using an equation that includes age, sex, height, and weight (18) and give estimates of TEE by combining the estimate for REE with an estimate for energy expended in physical activity. Subjects were instructed how to use the monitors while they were resident in our research facility and were instructed to record readings in the morning immediately after waking and at night immediately before sleeping for 7 consecutive 24-h periods. Readings of daily energy expenditure from the activity monitors were used to develop the equations.
Self-reported strenuous activity was also assessed. Sixty-four subjects reported durations and types of strenuous leisure time activity daily for 7 consecutive days, and the subset of these reported activities with an estimated energy expenditure of >5 x pREE (1) were judged to be truly strenuous and averaged per day. In the remaining 29 subjects, the Minnesota Leisure Time Activity Questionnaire (19) was completed; again, reported activities with an estimated energy cost of >5 x pREE were used for data analysis.
Measurement of body composition
Standard anthropometric measurements (including weight, height, and waist, hip, and thigh circumferences) were obtained on the first and last days of the study period. Body fat and fat-free mass were also determined by underwater weighing (20) with a correction for measured residual lung volume (21) on the first study day after subjects had fasted overnight. Measurements were repeated until 4 determinations of body fat content were within 1% of each other. Residual lung volume was measured by nitrogen washout [model 505 Nitralyzer; Med Science, St Louis (22)].
Statistical analysis
Data are expressed as means ± SDs unless specified otherwise. Differences between subject groups were assessed by one-way analysis of variance with Tukey's honestly significant difference adjustment for multiple comparisons. Multiple regression analysis was performed to determine prediction equation 1, and stepwise multiple regression analyses were performed to determine prediction equations 2 and 3. Cross-evaluation (23) was conducted by calculating predicted TEE with prediction equation 1 and comparing this value with TEE measured by the doubly labeled water method in subjects from 3 different studies (11, 24, 25). Bootstrap analysis with 100 bootstrap samples was performed according to Efron and Tibshirani (26) to estimate the true prediction error of prediction equation 1. A bootstrap sample is a random sample of the original data set, with the same number of observations as the original data set, drawn with replacement. For prediction equation 2, stepwise regression analysis was performed with 100 bootstrap samples to determine the retention of variables in a set that included age, sex, height, weight, activity monitor activity, percentage body fat, reported minutes of strenuous activity, and REE. Statistical analyses were done with SPSS (version 7.5 for WINDOWS; SPSS Inc, Chicago) and SYSTAT (version 7.01 for WINDOWS; SPSS Inc).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
is a summary of the doubly labeled water and other measured and reported energy expenditure data for subjects divided into 5 groups by age and sex. Young subjects had significantly higher TEEs and REEs than older subjects and men had significantly higher TEEs and REEs than women. The TEE and REE of middle-aged women was not significantly different from that of young women, but TEE in these 2 groups was significantly higher than that in older women. In addition, young men had a significantly higher physical activity level (equal to the ratio of TEE to REE) than older men and than all age groups of women, and the older men had a significantly higher physical activity level than the older women. Measured PAL ranged from 1.21 to 2.57 in individual subjects with a mean of 1.80. This value is significantly higher than the mean adult energy requirements of 1.55 and 1.67 times metabolic rate as proposed by the FAO/WHO/UNU (1) and the National Research Council (2).
|
|
| (1) |
where TEEeq1 is predicted TEE in MJ/d, age is in y, wt is body weight in kg, ht is standing height in cm, and sex is 0 for men and 1 for women [r2 = 0.64, SEE = 1.80 MJ/d (430 kcal/d), P < 0.001]. Plots of TEEeq1 versus TEE measured by the doubly labeled water method and of TEEeq1 versus the residuals (the difference between measured and predicted TEE) are shown in Figure 1. The plot of predicted TEE versus the residuals shows that the 80% limit of agreement was 0 ± 2.25 MJ/d, indicating that in 80% of the cases individual TEE could be predicted within ±2.25 MJ/d of measured TEE.
|
The validity of this equation was also evaluated by using published TEE data on 30 nonobese, free-living men and women (11, 24, 25). Reported values for age, weight, height, and sex in those studies were used to calculate TEE by using equation 1 and the values were compared with data on measured TEE also given in the reports. The equation accounted for 66% of the variance in TEE. A plot of TEEeq1 versus measured TEE in this data set and a plot of TEEeq1 versus the residuals showed that 80% prediction limits were from -1.8 to 1.8 MJ/d (Figure 2).
|
|
| (2) |
where TEEeq2 is predicted TEE in MJ/d and activity monitor activity and REE are also in MJ/d [r2 = 0.78, SEE = 1.55 MJ/d (370 kcal/d), P < 0.001]. Plots of TEEeq2 versus measured TEE and of the residuals versus TEEeq2 are shown in Figure 1
. The plot of the residuals versus TEEeq2 showed that the 80% limit of agreement was 0 ± 1.95 MJ/d, indicating that in 80% of the cases, individual TEE could be predicted within ±1.95 MJ/d of true TEE. Bootstrap analysis was performed to evaluate equation 2 because there were no data available in the literature to perform a direct validation. Again, 100 bootstrap samples were generated, and stepwise regression with backward elimination was performed on each sample by using the same methods as in the development of equation 2 to check the stability of the model. In 67 of 100 bootstrap samples, the same model with the same set of independent variables (percentage body fat, REE, and activity monitor activity) was kept in the final regression model. Activity monitor activity was one of the independent variables in 96 bootstrap samples, percentage body fat in 92 samples, and REE in 75 samples. All other independent variables tried in the original stepwise regression were retained as significant independent variables in <50% of the samples (between 17 and 47 times), indicating that the model produced from the original data set was reasonably stable and would be found in different comparable samples regularly.
Because the activity monitor data were such strong predictors of TEE in prediction equation 2, and are a measure that is feasible to incorporate in field studies, we evaluated a third model including the independent variables of equation 1 (age, sex, height, and weight) and activity monitor data. The equation from this analysis was as follows:
|
| (3) |
where TEEeq3 is predicted TEE in MJ/d (r2 = 0.76, SEE = 1.65 MJ/d, P < 0.001). This model was less precise than equation 2, but more precise than equation 1, indicating that by adding a measure of energy expenditure to a field study, the accuracy of prediction of average individual energy requirements could be somewhat improved. Plots of TEEeq3 versus measured TEE and of the residuals versus TEEeq3 are shown in Figure 1. The plot of the residuals versus TEEeq3 shows that the 80% limit of agreement was 0 ± 2.04 MJ/d, indicating that in 80% of cases individual TEE could be predicted within ±2.04 MJ/d of true TEE.
Data on measured TEE and predicted TEE by equations 1, 2 and 3, and energy requirements calculated by using the RDAs are summarized in Table 2 (2). As expected, there were no significant differences between measured TEE and predicted TEE calculated by using equation 1, 2, or 3 in any of the age and sex groups. There were, however, significant differences between the RDAs (2) and measured TEE in 4 of the 5 groups (all except the older women), with measured TEE being higher than predicted TEE. These differences in 4 of the 5 groups were due to differences in energy expended in physical activity because, although the RDAs underpredict TEE, the predicted REEs on which the RDAs are based were actually higher than measured REE (Figure 3).
|
. The RDAs underestimated measured TEE, with increasing underestimation with increasing TEE. In the validation data set (11, 24, 25), the RDA underestimated measured TEE as well (Figure 4).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
In the current study we used data from 93 healthy men and women to develop equations for predicting TEE based on both demographic data and simple laboratory measures. Prediction equation 1, which used data on age, sex, height, and weight, was comparable in terms of its precision to the equation developed by Black et al (23) and was shown to be valid for similar populations in a cross-validation conducted by using published TEE values from other research groups. Although the SEE of the equation, 1.80 MJ/d, was relatively large and comparable with the SEE for the relation of TEE to energy requirements predicted by using current RDAs (1.74 MJ/d), there was no demonstrable bias in predicted energy requirements over the range of TEEs tested. In contrast, the RDAs for energy were shown to generally underpredict true TEE in our 93 subjects, with the extent of underestimation increasing with increasing TEE. This underestimation was not found in REE, suggesting that the multiplication factor in the RDAs to allow for physical activity is too low, especially in physically active subjects. The reason the current RDAs may underestimate usual energy requirements is unknown. We speculated previously (5, 6), however, that the use of theoretical predictions of usual activity patterns to generate energy RDAs (1, 2) may be an important factor. This is because such predictions are more are more likely to underestimate total activity than to overestimate it because of the difficulty of accounting for activity (and hence energy expenditure) unrelated to specific tasks.
To investigate whether more precise estimates of TEE could be obtained by using information on physical activity and body composition, we developed 2 additional equations. Equation 2 included the variables REE, percentage body fat, and physical activity as determined by using activity monitors. The agreement of equation 2 with TEE measured by doubly labeled water was improved over that of equation 1, with an SEE of 1.55 MJ/d. Equation 3 used data on age, sex, height, weight, and physical activity as determined by using activity monitors and yielded an SEE of 1.65 MJ/d. Both equations 2 and 3 thus somewhat improved the estimate of TEE over equation 1 and, as with equation 1, predicted TEE with equivalent accuracy over the range of TEE measured.
There are several potential uses for these new equations. In particular, equation 1 may provide an alternative to the current RDAs for estimating group energy requirements in healthy adults when the mean BMI of the group is within the range of 18–31. Equations 2 and 3 can also be used for the same purpose when more information than just age, weight, and height is available and will provide slightly more precise values. In addition, all 3 equations may be useful in research studies requiring estimations of individual energy requirements as a starting point for other measurements. For example, studies needing to provide weight-maintenance diets as a starting point for nutrition interventions may be more accurate if they make use of equations such as those described here rather than the current RDAs. In addition, equations predicting usual energy needs also can be used to screen for inaccurate records in dietary surveys by providing a value for expected energy intake against which reported values can be compared.
In summary, we showed the feasibility and improved accuracy of predicting usual energy requirements from simple equations rather than from current RDAs in healthy, nonobese adults aged 18–81 y. Additional research is needed to develop equivalent equations suitable for other groups, including children and adolescents, obese adults, adults with chronic diseases, and persons in developing countries.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
2 Supported by the US Department of Agriculture, Agricultural Research Service, under contract 53-3K06-5-10 and by NIH grants AG12829, NIH DK46124, and 2P30DK46200.
3 Address reprint requests to SB Roberts, Energy Metabolism Laboratory, Jean Mayer USDA Human Nutrition Research Center on Aging at Tufts University, 711 Washington Street, Boston, MA 02111. E-mail: roberts_em{at}hnrc.tufts.edu.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  M. St-Onge, D. Mignault, D. B Allison, and R. Rabasa-Lhoret Evaluation of a portable device to measure daily energy expenditure in free-living adults Am. J. Clinical Nutrition, March 1, 2007; 85(3): 742 - 749. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. J Muller, A. Bosy-Westphal, S. Klaus, G. Kreymann, P. M Luhrmann, M. Neuhauser-Berthold, R. Noack, K. M Pirke, P. Platte, O. Selberg, et al. World Health Organization equations have shortcomings for predicting resting energy expenditure in persons from a modern, affluent population: generation of a new reference standard from a retrospective analysis of a German database of resting energy expenditure Am. J. Clinical Nutrition, November 1, 2004; 80(5): 1379 - 1390. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. A. McCrory, V. M.M. Suen, and S. B. Roberts Biobehavioral Influences on Energy Intake and Adult Weight Gain J. Nutr., December 1, 2002; 132(12): 3830S - 3834. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. K Horner, R. E Patterson, M. L Neuhouser, J. W Lampe, S. A Beresford, and R. L Prentice Participant characteristics associated with errors in self-reported energy intake from the Women's Health Initiative food-frequency questionnaire Am. J. Clinical Nutrition, October 1, 2002; 76(4): 766 - 773. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. L. Birmingham and P. J. Jones Clinical nutrition: 5. How much should Canadians eat? Can. Med. Assoc. J., March 1, 2002; 166(6): 767 - 770. [Full Text] [PDF]
|
|
|
|
 |
|
 G. P. Bathalon, N. P. Hays, S. N. Meydani, B. Dawson-Hughes, E. J. Schaefer, R. Lipman, M. Nelson, A. S. Greenberg, and S. B. Roberts Metabolic, Psychological, and Health Correlates of Dietary Restraint in Healthy Postmenopausal Women J. Gerontol. A Biol. Sci. Med. Sci., April 1, 2001; 56(4): 206M - 211. [Abstract] [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
