European Journal of Clinical Nutrition (1997) 51, 856±863 ß 1997 Stockton Press. All rights reserved 0954±3007/97 $12.00 Comparison of energy expenditure measurements by diet records, energy intake balance, doubly labeled water and room calorimetry
JL Seale1 and WV Rumpler2
1Diet and Human Performance Laboratory, Beltsville Human Nutrition Research Center, 2Agricultural Research Service, United States Department of Agriculture
Objectives: The purpose of this study was to compare estimates of daily energy expenditure (EE) using energy intake from self reported diet records, metabolizable energy intake balance, doubly labeled water and room calorimetry methods. Design: Cross sectional design. Setting: Beltsville Human Nutrition Research Center, Beltsville, MD USA. Interventions: Energy intake was measured using seven-day self reported diet records (EI), and metabolizable energy (ME) intake balance. EE was measured using doubly labeled water (TEE) and 24 h indirect room calorimetry (24 EE). Body composition was measured using stable isotope dilution and DEXA. Results: EI measured by self reported diet records was 22% less than ME intake balance, 23% less than TEE by doubly labeled water and 8% less than 24 EE by room calorimetry. 24 EE was 16% less than TEE and 16% less than ME. TEE was not signi®cantly greater than ME (0.3%). While mean ME, TEE and 24 EE measurements were signi®cantly lower in female compared to male subjects, mean EI and the mean percent difference between measurement methods were not. Conclusions: Direct comparison of these methods indicate self reported diet records and room calorimetry underestimate daily energy expenditure. While EI balance accurately estimates energy expenditure, EE measured by doubly labeled water is a more direct approach. Sponsorship: US Department of Agriculture, Agricultural Research Service. Descriptors: diet records; energy intake; energy expenditure; indirect calorimetry
Introduction food item consumed. The metabolizable energy intake balance method (ME) involves exclusively feeding subjects The energy requirement of a weight stable adult population an experimental diet where the amount of each food item can be determined from energy intake (EI) or energy on the menu is measured. A complete duplicate of the diet expenditure (EE) (FAO/WHO/UMU, 1985). EI is the as well as all urine and feces are collected and composites energy contained in the food consumed in the diet that is of the experimental diet and excreta are analyzed for available for metabolism. EE is the heat released and energy and nutrient content. The energy available for mechanical work performed by the body that is necessary metabolism from the entire diet is computed by subtracting to sustain life and lifestyle. Quantitatively the EE of an the energy in excreta from the energy in the food con- individual is equal to EI when the individual is weight and sumed. body composition stable or EI minus the change in body Energy expenditure can be assessed directly by measur- energy stores. ing average daily EE using doubly labeled water (TEE) or When energy expenditure is assessed by EI methods, all 24 h EE using room calorimetry (24 EE). The doubly food consumed must be accounted for during a period when labeled water (2H 18O) method is used to determine TEE body weight and composition are constant or when changes 2 from the disappearance rates of 18O(k ) and 2H(k ) non- in body energy stores are determined by measuring weight O H radioactive, naturally occurring, stable isotopes from the and body composition at the beginning and end of the total body water pool (Lifson, 1966). 18O is eliminated assessment period. EI determined from self reported dietary from the water pool as water and carbon dioxide while 2H intake records requires that subjects record the type and is eliminated only as water. The difference in the elimina- amount of all food they consumed during the assessment tion rates of 18O and 2H(k 7k ) is related to the carbon period. The energy and nutrient intake is evaluated using a O H dioxide production rate (rCO ) from which TEE can be food database such as the Minnesota Nutrition Data System 2 determined (Weir, 1949). Indirect room calorimeters are (Nutrition Coordinating Center, University of Minnesota, used to accurately determine 24 EE in the controlled Minneapolis, MN). The energy content of the entire diet is environment of a metabolic chamber (Seale et al, 1991). computed as the sum of the energy available from each Indirect calorimetry is the measurement of the respiratory gas exchange, oxygen consumption and carbon dioxide production, which is used to calculate 24 EE (Weir, 1949). Correspondence: Dr JL Seale, Bldg. 308, Rm 212, BARC-East, 10300 Baltimore Avenue, Beltsville, MD 20705-2350, USA. The measurement of self reported dietary intake has Received 16 April 1997; revised 8 August 1997; accepted 18 August 1997 played a central role in accessing the nutrient requirements Comparison of energy intake and expenditure JL Seale and WV Rumpler 857 of populations including energy requirements. Dietary a ten day period. During this ten day period subject's EI intake records have been compared to EI balance (Mertz from self reported diet records were determined for seven et al, 1991) and EE measured by doubly labeled water days. EE was also determined using the ME balance (Black et al, 1993; Livingstone et al, 1990; Livingstone method during the ®rst seven weeks of each eight week et al, 1992; Martin et al, 1996; Prentice et al, 1986; Sawaya period subjects were fed the experimental diet. Composite et al, 1996; Schoeller, 1995; Schoeller et al, 1990; Schulz samples of total diet, urine and feces collected during week et al, 1989) and room calorimetry (Drougas et al, 1992) to ®ve were used to measure ME. 24 EE was measured for a determine the reliability of self reported dietary intake one day period in a room calorimeter during the eighth methods. week of both experimental diet phases. The overall experi- The purpose of this paper is to compare the results of EE mental protocol is outlined in Figure 1. measurements from EI using seven-day self reported diet- This study was approved by the Agricultural Research ary intake records, ME intake balance, TEE using doubly Service Human Studies Review Committee of the US labeled water and 24 EE using room calorimetry in the Department of Agriculture and the Institutional Review same individuals. These results may provide some insight Board of Georgetown University. All subjects were medi- into the bias and limitations of EE measurements with EI cally screened and found free of metabolic disease. All using dietary intake records, ME from energy intake subjects were normotensive, nondiabetic and had normal balance, TEE using doubly labeled water and 24 EE from blood chemistry values. All subjects gave informed consent room calorimetry and depending on which particular to participate in this study after the procedures were method is used what the likely inaccuracies will be. explained to them. All subjects were ®nancially compen- sated for participating in this study. Subjects were fed at the BHSF, Beltsville Human Materials and methods Nutrition Research Center for both eight week diet periods. Diets were formulated to meet the recommended daily Twenty four healthy adult female (14) and male (10) allowances for nutrient intake and provide suf®cient subjects were recruited for this experiment from partici- energy to maintain subject weight. Diets consisted of pants of a 22 week diet study at the Beltsville Human Study foods normally consumed in human diets and no test Facility (BHSF). The 22 week study (two eight week chemicals were added. feeding periods with a six week break between periods) was designed to investigate the effects moderate alcohol consumption coupled with high 40% or low 20% fat diets Dietary intake records on several metabolic parameters (Rumpler et al, 1996). EI was estimated from 7 d self recorded dietary intake During initial screening subject's height, weight, body records. Subjects were required to record the type and mass index (BMI; kg/m2) and percent body fat measured amount of all food consumed during a weight stable using bioelectric impedance analysis were determined. period. Subjects were trained by registered dietitians to Subjects also completed questionnaires on alcohol con- use scales, measuring spoons and cups and how to estimate sumption and physical activity. Subjects selection was portion sizes when they are unable to actually measure the based on BMI, percent body fat and daily activity patterns. food consumed. The diet records were periodically Because the difference between EI and TEE is effected by reviewed in the presence of the subjects to resolve any obesity (Prentice et al, 1986), ®ve subjects were excluded uncertainty in the entries and to assess the completeness of because they had a BMI which placed them in an obese the record. Subjects were instructed not to change their classi®cation (3 female BMI > 27.3 kg/m2 and 2 male behavior during the recording period including physical BMI > 27.8 kg/m2) (Najjar & Rowland, 1987). Final data activities or eating habits. analysis included data from nineteen healthy adult female The completed diet records were evaluated for energy (11) and male (8) (ages 40±62 y) (BMI 17.9±27.6 kg/m2). and nutrient content under contract by the Nutrition Co- Free-living TEE and EI from self reported diet records ordinating Center, University of Minnesota. Each food item were measured in the subjects participating in this experi- was coded from a descriptive list in a food database ment during the two week period prior to the start of the (Minnesota Nutrition Data System, Nutrition Coordinating second eight week phase of the crossover diet study. Free- Center, University of Minnesota, Minneapolis, MN). The living TEE was measured with doubly labeled water during data base uses an average value for the nutrient content and
Figure 1 Time line of study protocol including period for measurements of TEE doubly labeled water, EI 7-day diet records, 24 EE room calorimetry and ME intake balance with 7 d food, urine and feces composites. Comparison of energy intake and expenditure JL Seale and WV Rumpler 858 metabolizable energy of the food consumed. The energy accuracy for the given instrument sensitivity for determin- 2 18 content of the entire diet is computed as the sum of the ing H2O clearance for 10 d (Seale et al, 1989). The O energy available in each food item consumed. concentration was determined using the CO2 equilibrium method and an isotope ratio mass spectrometer by a Metabolizable energy intake balance commercial laboratory (Metabolic Solutions, Inc. Merri- Subjects were fed experimental diets at the BHSF at a mack, NH, USA) Blind standards derived from a serial 18 weight maintenance level for two eight-week periods. dilution of dose H2 O were included with the samples to Subjects ate breakfast and evening meals at the facility validate the laboratory results. Linear regression on the and were given a packed lunch, snack and evening bev- standard concentration versus analyzed values (r2 erage during weekdays. Weekend meals were packed in 0.99999; slope 0.995 Æ 0.001; intercept 0.000 Æ 0.000) coolers and taken home by subjects on Friday evening. indicate accurate 18O analysis. Diets contained 14% of calories from protein, either 20% or The 2H and 18O zero time intercepts and clearance rates 40% from fat, either 61% or 41% carbohydrate and 5% (kH and kO) were calculated using least square linear ethanol or soluble carbohydrate powder (Polycose, Ross regression on the natural logarithm of isotope concentration Laboratory, Columbus, OH, USA). Each food item on the as a function of elapsed time from dose administration menu was measured to within 1 g at the BHSF according to (Corel Quatro Pro 7, Corel Corporation, Ottawa, Ontario, intake level. Subjects were weighed in a laboratory coat Canada). The zero time intercepts were used to determine weekly and intake level was adjusted if a change in weight the isotope pool sizes at the time of dose (Seale et al, 1989). persisted. Body composition was measured using dual- The 18O pool size was used to estimate total body water energy X-ray absorptiometry (DEXA, Lunar DPX, Lunar (N 18O pool size/1.01). The equation used to calculate Radiation Corporation, Madison, WI, USA) at the begin- rCO2 (Schoeller et al, 1986) from the isotope clearance (kH ning (week 1) and at the end (week 7) of each dietary and kO) rates and total body water (N) is shown in equation phase. The change in body weight and composition were 1. The respiratory quotient determined from the self used to estimate changes in total body energy stores which recorded diet records were used to estimate EE from rCO2. were subtracted from ME intake to determine energy expenditure. rCO2 N=2f3 1:01 kO 1:04 kH The ME intake was determined during week ®ve of each Â1 1:05 f2 f1 l=d 1 eight week diet phase. Total fecal and urine excretion and duplicate meals prepared in the BHSF were collected Where constant isotope fractionation factors are: during the seven day period. Fecal collections were f 0:941; f 0:992 and f 1:039 frozen, pooled, weighted, homogenized and freeze-dried 1 2 3 before analysis. Urine samples collected each day were The application of the doubly labeled water method used 2 pooled, weighted and 10% of each days urinary output was in this study was to measure H2O isotope concentration in 18 added to a composite sample for the entire week. Duplicate body water by infrared spectrophotometry and H2 O con- meals were prepared from each days menu food and centration by isotope ratio mass spectrometry. Infrared beverages were combined, weighted, homogenized and spectrophotometry is not as precise as isotope ratio mass 2 freeze-dried. The energy content of the food, feces and spectroscopy and requires a larger dose of H2O, but is urine were analyzed using an adiabatic bomb calorimeter typically less dif®cult and less expensive (Stansell & (Parr Instrument Corporation, Moline, IL USA). The ME Mojica, 1968; Karasov et al, 1988). Coupled with a 2 for each subject was calculated as the difference between larger initial dose of H2O the precision of infrared spectro- total energy intake and the energy excreted in the urine and photometry provides the accuracy required for implemen- feces. tation of the doubly labeled water method. In addition, 18 2 H2 O does not interfere with the measurement of H2O by Doubly labeled water IR absorbance (Karasov et al, 1988). The dosing ratio 2 18 Average daily rCO2 and EE were determined using doubly ( H O) is well above shifts in background abundance for labeled water for a 10 d period. Subjects reported to the these isotopes which was theorized to be a potential source BHSF at 5:00 pm on the day prior to the start of the 10 d of error in determining rCO2 (Schoeller, 1983; Roberts et period, a baseline urine sample was collected and the al, 1988). Variable shifts in the background abundance of isotope dose was administered. Subjects were given an 2H and 18O which indicate that a ®xed isotope dose ratio 18 oral dose of doubly labeled water (H2 O: 0.14 g/kg body will not reduce error have also been observed (Jones et al, 2 2 18 weight and H2O: 0.70 g/kg body weight). Subjects were 1988). When within-subject variation of H2O and H2 O given labeled containers and instructions for collecting natural abundance was determined over six consecutive urine specimens and released. Second void urine samples days in seven human subjects, no signi®cant relationship 2 18 were collected daily for ten days following administration was found between H2O and H2 O natural abundance of the isotope. Baseline urine samples collected before the (Ritz et al, 1996). In ®eld studies where there is a potential dose and second void urine samples collected on days 1, 2, for background shifts in isotope abundance and correspond- 3, 8, 9, 10 of the 10 d measurement period were analyzed ing error rCO2 it is important to account for any changes in for 2H and 18O concentrations. background isotope abundance in the study design. While 2 2 The H concentration was determined using duplicate increasing the H2O dose does not affect the error caused 18 analysis with an infra-red spectrophotometer (MIRAN 1A- by H2 O shifts, it may reduce the potential for error 2 FF, Foxboro/Wilks Inc, South Norwalk, CT, USA) on introduced by background shifts in H2O (Seale et al, 2 duplicate vacuum sublimations of urine (Stansell & 1989). This indicates that a higher dose of H2O is appro- Mojica, 1968). The coef®cient of variation for repeated priate when subjects remain within a geographic location 2 18 values from multiple deuterium analysis of body water and the natural abundance of H2O and H2 O in drinking samples was 1.71% (PROC GLM, SAS Institute Inc. Cary, water and diet remain relatively constant. Validation of NC, USA). The 2H dose was calculated to provide the best these analytical techniques and computational methods Comparison of energy intake and expenditure JL Seale and WV Rumpler 859 have an accuracy of 1.6% Æ 2.6% in nine subjects (Seale et Results al, 1993). The physical characteristics of the female (11) and male (8) subjects are shown in Table 1. Values in the table represent Calorimetry the mean Æ standard deviation for age, height, weight, body Subjects 24 h EE was measured in a room calorimeter mass index (BMI), fat mass and fat free mass. Female (Seale et al, 1991) during a one day period between day 60 subjects had signi®cantly less height, weight and fat free and 67 of the study. Subjects reported to the BHSF by mass than male subjects. Fat free mass was determined by 7:00 am in the morning. Subjects entered the calorimeter at the isotope dilution of 18O (FFM 18O pool size/1.01/ 8:00 am on the day of the measurement and exited 7:30 am 0.723). Fat mass was determined by the difference between the following day. While in the calorimeter subjects fol- total body weight and fat free mass. lowed a standard activity schedule (Rumpler et al, 1990). Results (mean Æ standard deviation) from measurements This activity schedule included two exercise periods, three of EI by self reported diet records, ME intake balance, TEE meals, and a minimum of six and one half hours of sleep. from doubly labeled water and 24 EE from room calori- The meals consumed while in the calorimeter were con- metry are shown in Table 2. Results (mean Æ standard sistent with the study diet. 3 deviation) for the difference between EI by self reported A room sized (20.39 m ) calorimeter was used to diet records, ME intake balance, TEE from doubly labeled determine 24 h total EE (EE-24 h) in a controlled environ- water and 24 EE from room calorimetry are shown in ment (Seale et al, 1991). This calorimeter continuously Table 3. measures respiratory gas exchange to within 1.3% over a The EI from diet records for female, male and all one day period (Seale et al, 1991). The repeatability of subjects were signi®cantly less than ME intake balance human EE-24 h measurements reported for this chamber is (17%, 30% and 22% respectively) and EE by doubly 5% (Rumpler et al, 1990). During the calorimeter experi- labeled water (18%, 30% and 23% respectively). EI from ment the rate of CO2 production (rCO2) and O2 consump- diet records were also signi®cantly less than EE by calori- tion (rO2) were measured continuously while the subjects metry in male and all subjects (16% and 8% respectively) were in the chamber. EE-24 h values were calculated based but not for female subjects (2%). Figures 2, 3 and 4 on the 23.5 h measurements corrected for the 30 min represent the difference between EI from diet records for period not spent in the chamber to re¯ect a full 24 h female and male subjects graphed as a function of the measurement. The average EE measured during the average value of EI and ME intake balance, TEE by doubly waking hours of the day in the calorimeter were used for labeled water and 24 EE by calorimetry respectively. These this correction. 24 h urine composites were collected graphs not only illustrate the consistent underestimation of during calorimetry experimental period. Urinary nitrogen energy requirement by EI from diet records compared to levels were determined using a combustion technique the other methods but also show the degree of variation (CHN-600, LECO Corporation, St Joseph, MI) in order to between subjects. calculate daily nitrogen loss (UN). The 24 h EE was While mean EI was not signi®cantly different between calculated using equation 2 with the respiratory gas totals female and male subjects, mean TEE and 24 EE were and urinary nitrogen production values (Weir, 1949).
EE 16 500 rO2 4630 rCO2 9080 UN J 2 Table 1 Physical characteristics of female, male and all subjects including number of subjects, age, height, weight, body mass index (BMI), fat mass and fat free mass Statistical analysis An analysis of variance `a priori' was used to determine Female Male All signi®cant differences in physical characteristics between Number 11 8 19 female and male subjects (PROC GLM, SAS Institute Inc. Mean Æ Standard Deviation Cary, NC, USA). An analysis of variance was used to Age (y) 51.9 Æ 4.9 49.5 Æ 7.2 50.9 Æ 6.1 a b determine signi®cant difference between energy measure- Height (m) 1.66 Æ 0.04 1.79 Æ 0.08 1.72 Æ 0.09 Weight (kg) 62.4 Æ 6.4a 82.4 Æ 7.0b 70.8 Æ 11.9 ment methods using a subject identi®cation number as a BMI (kg/m2) 22.6 Æ 2.5a 25.7 Æ 1.3b 23.9 Æ 2.6 covariate (PROC GLM, SAS Institute Inc. Cary, NC, Fat mass (kg)c 20.2 Æ 4.6 21.0 Æ 2.2 20.5 Æ 3.8 USA). An analysis of variance was used to determine Fat free mass (kg)d 42.2 Æ 2.6a 61.4 Æ 7.2b 50.3 Æ 10.7 signi®cant difference between the difference in energy a,b Superscripts indicate signi®cant difference (P < 0.05) within a row measurement methods and zero using a subject identi®ca- between female and male subjects. c 18 tion number as a covariate (PROC GLM, SAS Institute Inc. Fat mass determined by isotope dilution (H2 O). d 18 Cary, NC, USA). Fat free mass determined by isotope dilution (H2 O).
Table 2 Mean values for female, male and all subjects of energy intake from dietary records (EI), metabolizable energy intake balance (ME), total energy expenditure by doubly labeled water (TEE) and 24h energy expenditure room calorimetry (24EE)
Female Male All
Mean Æ Standard Deviation Diet Records EI (MJ/d) 7.88 Æ 1.941 8.96 Æ 1.381 8.34 Æ 1.811 Metabolizable energy intake balance ME (MJ/d) 9.57 Æ 0.60a2 12.79 Æ 1.54b2 10.93 Æ 1.932 Doubly labeled water TEE (MJ/d) 9.57 Æ 0.47a2 12.91 Æ 1.15b2 10.97 Æ 1.842 Indirect calorimetry (24EE (MJ/d) 8.07 Æ 0.68a1 10.72 Æ 0.85b3 9.19 Æ 1.513
a,bSuperscripts indicate signi®cant difference (P < 0.05) within a row between female and male subjects. 1,2,3Superscripts indicate signi®cant difference (P < 0.05) within a column between methods used to measure energy intake and expenditure. Comparison of energy intake and expenditure JL Seale and WV Rumpler 860 Table 3 Mean difference for female, male and all subjects between energy intake from dietary records (EI), metabolizable energy intake balance (ME), total energy expenditure by doubly labeled water (TEE) and 24h energy expenditure room calorimetry (24EE)
Difference Female Male All
Mean Æ Standard Deviation Diet records (ME±EI) (MJ/d) 1.69 Æ 2.10a1 3.83 Æ 1.28b1 2.59 Æ 2.091 (TEE±EI) (MJ/d) 1.68 Æ 1.92a1 3.95 Æ 0.93b1 2.64 Æ 1.941 (24EE±EI) (MJ/d) 0.19 Æ 1.90a 1.75 Æ 1.63a1 0.85 Æ 1.951 Metabolizable energy intake balance (TEE±ME) (MJ/d) 70.01 Æ 0.58a 0.11 Æ 0.76a 0.04 Æ 0.66 (24EE±ME) (MJ/d) 71.50 Æ 0.65a1 72.08 Æ 1.39a1 71.74 Æ 1.071 Doubly labeled water (24EE±TEE) (MJ/d) 71.49 Æ 0.76a1 72.19 Æ 1.20a1 71.79 Æ 1.031
a,b Superscripts indicate signi®cant difference (P < 0.05) within a row between female and male subjects. 1Superscripts indicate signi®cant difference (P < 0.05) of mean difference from 0.
Figure 2 Graph of difference against mean for energy intake from dietary intake records (EI) and metabolizable energy intake balance (ME).
Figure 3 Graph difference against mean for energy intake from dietary intake records (EI) and energy expenditure measured with doubly labeled water (TEE). Comparison of energy intake and expenditure JL Seale and WV Rumpler 861
Figure 4 Graph of difference against mean for energy intake from dietary intake records (EI) and 24 h energy expenditure measured with indirect room calorimetry (24 EE).
Figure 5 Graph of difference against mean for metabolizable energy intake balance (ME) and energy expenditure measured with doubly labeled water (TEE). signi®cantly less for female subjects than for male subjects. measured by room calorimetry in female, male and all Additionally, the difference between EI and ME and EI and subjects (16%, 15% and 16% respectively) and (15%, 17% TEE was signi®cantly greater for male than for female and 16% respectively). A graph of the difference between subjects. However, when the difference between EI by diet values for ME intake balance and TEE by doubly labeled records and ME by intake balance, TEE by doubly labeled water and 24 EE vs the average value is presented in water and 24 EE by calorimetry were calculated on a Figures 6 and 7. percent basis the difference between male and female subjects was not signi®cant. Discussion Free-living EE measured by doubly labeled water was not signi®cantly different than ME intake balance for A common health concern of the adult population in the female, male or all subjects (70.2%, 1.0% and 0.3% United States is obesity. Obesity occurs when excess respectively). A graph of the difference between values energy is consumed in the diet relative to energy expendi- for ME intake balance and TEE vs the average value is ture. Methods to accurately estimate energy expenditure are presented in Figure 5. necessary to aid in combating and avoiding obesity. Four ME intake balance and free-living EE measured by methods used to estimate energy expenditure by measuring doubly labeled water were signi®cantly greater than EE EI or EE were compared within the same subjects in this Comparison of energy intake and expenditure JL Seale and WV Rumpler 862
Figure 6 Graph of difference against mean for metabolizable energy intake balance (ME) and 24 h energy expenditure measured with indirect room calorimetry (24 EE).
Figure 7 Graph of difference against mean for energy expenditure measured with doubly labeled water (TEE) and 24 h energy expenditure measured with indirect room calorimetry (24 EE).
study. A comparison of the results from this study indicate Schultz et al, 1989 ). Self reported diet records were on that some methods are better suited for determining energy average 8% less than 24 h EE measured by room calori- expenditure than others. metry. These results are consistent with reported results EI as determined by self reported dietary records was which indicate that EI measured by dietary intake records found to underestimate energy expenditure as measured by were 10% less than EE measured by calorimetry (Drougas ME intake balance by an average of 22% and by doubly et al, 1992). Results from this study indicate that dietary labeled water by an average of 23% in 19 adult female and intake records consistently underestimate energy expendi- male subjects. The results from this study are consistent ture measured by ME intake balance or doubly labeled with published results where EI by diet records was 18% water methods. lower than EI by intake balance in 266 adult female and 24 EE measured in a room calorimeter was found to be male subjects (Mertz et al, 1991). These results are also 16% (women: 15%, men: 17%) less than free-living TEE consistent with reported results where EI by dietary intake measured with doubly labeled water in the 19 subjects. In a records were 20% lower than EE by doubly labeled water previous study EE measured by room calorimetry and by in men and women between 20 and 65 y of age (Living- doubly labeled water in nine subjects (5 men, 4 women) stone et al, 1990; Martin et al, 1996; Sawaya et al, 1996; was similar while subjects were within the calorimeter Comparison of energy intake and expenditure JL Seale and WV Rumpler 863 chamber, but under free-living conditions EE measured by Lifson N (1966): Theory of use of the turnover rates of body water for doubly labeled water was 13% greater (Seale et al, 1993). measuring energy and material balance. J. Theor. Biol. 12, 46±74. Livingstone MBE, Prentice AM, Coward WA, Black AE, Barker ME, In another study free-living EE measured by doubly labeled McKenna PG & Whitehead RG (1990): Accuracy of weighed dietary water was 15% greater than EE measured by calorimetry records in studies of diet and health. Br. Med. J. 300, 708±712. but was not signi®cantly different than ME intake in four Livingstone MBE, Prentice AM, Coward WA, Strain JJ, Black AE, Davies men (Seale et al, 1990). Results indicate that the energy PSW, Stewart CM, McKenna PG & Whitehead RG (1992): Validation of estimates of energy intake by weighted dietary record and diet history expenditure of free-living adults is signi®cantly greater in children and adolescents. Am. J. Clin. Nutr. 56, 29±35. than energy expenditure within a calorimeter chamber. Martin LJ, Su W, Jones PJ, Lockwood GA, Tritcher DL & Boyd NF These results suggest that doubly labeled water and ME (1996): Comparison of energy intakes determined by food records and intake balance may be better methods for determining doubly labeled water in women participating in a dietary-intervention energy expenditure of a normal free-living population. trail. Am. J. Clin. Nutr. 63, 483±490. Mertz W, Tsui JC, Judd JT, Reiser S, Hallfrisch J, Morris ER, Steele PD & Two independent methods were used to accurately Lachley E (1991): What are people really eating? The relationship measure free-living energy expenditure in this study. The between energy intake from estimated diet records and intake deter- ME balance method was used for two seven week periods. mined to maintain body weight. Am. J. Clin. Nutr. 54, 291±295. In each period total energy intake, excretion and the change Najjar MF & Rowland M (1987): Anthropometric reference data and in body energy stores were carefully measured. The aver- prevalence of overweight, United States, 1976±1980. Vital Health Stat. 11, No. 238. age value from both seven week periods was equated to Prentice AM, Black AE, Coward WA, Davies HL, Goldberg GR, Murga- energy expenditure. The doubly labeled water method was troyd PR, Ashford J, Sawyer M & Whitehead RG (1986): High levels of used to measure free-living energy expenditure over a ten energy expenditure in obese women. Br. Med. J. 292, 983±987. day period. The mean value for energy expenditure mea- Prentice AM, Leavesley K, Murgatroyd PR, Coward WA, Schorah CJ, Bladon PT & Whitehead RG (1986): Is severe wasting in elderly mental sured by doubly labeled water was 0.3% greater than by patients caused by an excessive energy requirement. Age Ageing 18, ME intake balance and was not signi®cantly different. The 158±167. percent difference between EE measured by doubly labeled Ritz P, Cole TJ, Couet C & Coward WA (1996): Precision of DLW energy water and ME intake balance was determined for each expenditure measurements: contribution of natural abundance varia- subject (100(EE-ME)/EE). The mean of the percent differ- tions. Am. J. Physiol. 270, E164±E169. Roberts SB, Coward WA, Ewing G, Savage J, Cole TJ & Lucas A (1988): ence was 70.17% Æ 6.03% (mean Æ standard deviation), Effect of weaning on accuracy of doubly labeled water method in for women and 1.01% Æ 6.03% for men. The percent infants. Am. J. Physiol. 254, R622±R627. difference was not signi®cantly different between genders Rumpler WV, Rhodes DG, Baer DJ, Conway JM & Seale JL (1996): and not signi®cantly different from zero. The agreement Energy value of moderate alcohol consumption by humans. Am. J. Clin. Nutr. 64, 108±114. between the ME intake balance and doubly labeled water Rumpler WV, Seale JL, Conway JM & Moe PW (1990): Repeatability of results serves to cross-validate both methods as an accurate 24-h energy expenditure measurements in humans by indirect calori- assessment of the subject's energy expenditure. metry. Am. J. Clin. Nutr. 51, 47±52. While ME intake balance methods during a control Sawaya AL, Tucker K, Tsay R, Willet W, Saltzman E, Dallal GE & feeding study accurately estimate energy expenditure, the Roberts SB (1996): Evaluation of four methods for determining energy intake in young and older women: comparison with doubly labeled study protocols can be cumbersome and may confound water measurements of total energy expenditure. Am. J. Clin. Nutr. 63, results by interfering with subjects normal activities and 491±499. eating habits. EI measurements with dietary intake records Schoeller DA (1983): Energy expenditure from doubly labeled water: interfere less with normal eating habits but have been some fundamental considerations in humans. Am. J. Clin. Nutr. 38, 999±1005. shown to be inaccurate and to underestimate energy Schoeller DA (1995): Limitations in the assessment of dietary energy expenditure. Room calorimeters are the most precise and intake by self-report. Metabolism 44 (2,Suppl 2), 18±22. accurate tools measuring 24 EE, but results are an indica- Schoeller DA, Bandini LG & Dietz WH (1990): Inaccuracies in self- tion of the energy expenditure of a subject restricted in reported intake identi®ed by comparison with the doubly labeled water activity to a small room sized chamber. In general TEE method. Can. J. Physiol. Pharmacol. 68, 941±949. Schoeller DA, Ravussin E, Schultz Y, Acheson KJ, Baertschi P & Jaquier measurements using doubly labeled water is a more direct E (1986): Energy expenditure by doubly labeled water: validation in approach to determining free-living energy expenditure humans and proposed calculation. Am. J. Physiol. 250, R823±R830. than intake balance methods or calorimetry. Schulz S, Westerterp KR & Bruck K (1989): Comparison of energy expenditure by the doubly labeled water technique with energy intake, heart rate, and activity recording in man. Am. J. Clin. Nutr. References 49, 1146±1154. Seale JL, Conway JM & Canary JJ (1993): Seven-day validation of doubly Black AE, Prentice AM, Goldberg GR, Jebb SA, Bingham SA, Living- labeled water method using indirect calorimetry. J. Appl. Physiol. 74(1), stone MBE & Corward WA (1993): Measurements of total energy 402±409. expenditure provide insights into the validity of dietary measurements Seale JL, Miles CW & Bodwell CE (1989): Sensitivity of methods for of energy intake. J. Am. Diet. Assoc. 93, 572±579. calculating energy expenditure by use of doubly labeled water. J. Appl. Drougas HJ, Reed G & Hill JO (1992): Comparison of dietary self-reports Physiol. 66, 644±653. with energy expenditure measured using a whole room indirect calori- Seale JL, Rumpler WV & Moe PW (1991): Description of a direct-indirect meter. J. Am. Diet. Assoc. 92, 1073±1077. room-sized calorimeter. Am. J. Physiol. 260, E306±E320. FAO/WHO/UMU Expert Consultation (1985): Energy and Protein Seale JL, Rumpler WV, Conway JM & Miles CW (1990): Comparison of Requirements. Technical Report Series 724, World Health Organiza- doubly labeled water, intake balance, and direct- and indirect-calori- tion, Geneva. metry method for measuring energy expenditure in adult men. Am. J. Jones PJH, Winthrop AL, Schoeller DA, Filler RM, Sawyer PR, Smith J & Clin. Nutr. 52, 66±77. Heim T (1988): Evaluation of doubly labeled water for measuring Stansell MJ & Mojica Jr L (1968): Determination of body water content energy expenditure during changing nutrition. Am. J. Clin. Nutr. 47, using trace levels of deuterium oxide and infrared spectrophotometry. 799±804. Clin. Chem. 14, 1112±1124. 2 Karasov WH, Han LR & Munger JC (1988): Measurement of H2O by IR Weir JB DeV (1949): New methods for calculating metabolic rate with absorbance in doubly labeled H2O studies of energy expenditure. Am. J. special reference to protein metabolism. J. Physiol. Lond. 109, 1±9. Physiol. 24, R174±E177.