Critical Review

Measurement of Energy Expenditure in Free-LivingHumans by Using Doubly Labeled Water1

DALE A. SCHOELLER?

Clinical Nutrition Research Center, university of Chicago, Chicago, IL 60637

ABSTRACT The doubly labeled water method is a formof indirect calorimetry that has been developed only recentlyto the stage of application to human studies. The methodmeasures integral CO, production for up to 3 wk from thedifference in elimination rates of deuterium and 18O fromlabeled body water. Validations against near-continuousrespiratory gas exchange have demonstrated that the methodis accurate and has a precision of 2—8% depending onthe isotope dose and the length of the elimination period.Although the method has been validated, there is still somedebate on refinements of the kinetic model that may leadto improved accuracy and precision. Because the methodonly requires periodic sampling of body fluids, it is non-restrictive and ideally suited to use in free-living subjects.Recent applications of the method have included obesityresearch, determination of energy requirements in bothdeveloping and developed countries and studies of humangrowth. J. Nutr. 118: 1278-1289, 1988.

INDEXING KEY WORDS:

energy metabolismstable isotopes

deuterium 18Q

Since the time of the early Greek philosophers, it hasbeen recognized that food provides fuel for body heatproduction. In the absence of continued food intake,body energy stores can maintain life, but these storeswill generally be depleted after 60 d. On the other hand,if the body is supplied with food in excess of energyneeds, the excess will be stored as body fat, which leadsultimately to obesity and the associated increase inmorbidity and mortality. Establishing energy requirements thus depends on defining the level of intake thatwill maintain body weight in the ideal range. It remainscontroversial, however, whether the human body canmaintain ideal weight at a single level of intake or ifefficiency can be altered so that ideal weight can bemaintained over a range of intakes. The factor that hasprobably contributed the greatest uncertainty to thedebate has been the absence of an accurate method formeasuring energy expenditure in free-living humans.

The doubly labeled water method for measuring energy expenditure finally provides this missing measure.Development of this method can be traced to a studyperformed by Lifson et al. in the 1940s (1). They administered 18O-labeled water to animals and demonstrated that the oxygen in expired CO2 was derivedfrom body water. This is now known to result from themaintenance of isotopie equilibrium between the oxygen atoms of body water and CO2. On the basis of thisobservation, Lifson, Gordon and McClintock (2) reasoned that total integrated CO2 production could bemeasured from the differential elimination of waterlabeled with both isotopie hydrogen and oxygen. Aftera loading dose of the doubly labeled water the labeledhydrogen would be eliminated as water, whereas theoxygen isotope would be eliminated as water and asCO2. Stated mathematically:

rH2O = NAHand (I)

rH2O + 2rCO2 = (2)

where rH2O and rCO2 are the rates of water and CO2fluxes, respectively, kHand k0 aie the elimination ratesof the hydrogen and oxygen labels, respectively, and Nis the body water pool. Substituting for rH2O in Equation 2 and solving for rCO2 yields the following:

rC02 = |1/2)N|A0 - *„]. (3)

Thus it is theoretically possible to measure CO2 production by measuring the isotopie hydrogen and oxygenremaining in body water after administration of thedoubly labeled water. The labeled water in essence servesas a recorder that monitors the fluxes of water and CO2through the body. Between the initial and final samples, the animal can be set free to engage in normalactivities.

'This work has been partially supported through NIH Grant Nos.

DK30031 and DK26678.2D. A. Schoeller was the recipient of the 1987 Mead Johnson Award

from the American Institute of Nutrition. This article describes theresearch on which his award was based.

0022-3166/88 $3.00 ©1988 American Institute of Nutrition. Received 25 January 1988. Accepted 23 May 1988.

1278

at University of P

ittsburgh HS

LS on M

ay 11, 2007 jn.nutrition.org

Dow

nloaded from

DOUBLY LABELED WATER 1279

TABLE 1

Validations of the doubly labeled water method in small animals

SpeciesMouseMouseMouseMouseRatRatRatMouse

&chipmunkChipmunkSquirrelPigeonSparrowStarlingBudgerygahLocustScorpionBeetleNumber

ofanimals157548666271013599539%

Error(mean±SD)-3-411221-7414-4-70-2361073746645}96143252425

?Reference

method1C02C02C02C02C02C02C02C02C02I/BCO2C02C02C02CO2C02C02Equation2L.34L.34L.34—L.35L.35L.22—L.35—L.6N.2N.2N.2—N.2N.2CitationLifson,

Gordon and McClintock, 1955(2)McClintock& Lifson, 1957(3)McClintock& Lifson, 1958(4)Mullen,

1970|5|McClintock& Lifson, 1958(6)Lee

& Lifson, 1960(7)Lifson& Lee, 1961(8)Randolph,

1980(9)Little& Lifson,1975(10)Karasov,

1981(11)LeFebvre,1964(12)Williams,

1985(13)Williams,1985(13)Buttemer

et al., 1986(14)Buscarlet,Proux and Gester, 1978(15)King&.

Hadley, 1979(16)Cooper,1983 (17)

'Compared with measured CO2 production (CO2) or measured energy intake plus change in body energy stores (I/B).2Doubly labeled water results calculated by equation from Lifson |L) (18) or Nagy (N) (19).

Although the exact relationship in Equation 3 wasmodified to account for isotopie fractionation, the general validity of the theoretical relationship between thekinetics of the hydrogen and oxygen labeled water hasbeen extensively validated in small animals (Table 1).These validations are quite convincing because theyencompass numerous species and involve a number ofinvestigators and laboratories. In general, the methodis accurate to several percent. The coefficient of variation in the small animal studies has been 3-14%. Theonly notable failures have occurred in arthropods (16,17) and this may be due to a deviation from single-compartment kinetics (16).

Although the method has been used in small animalstudies for several decades, the potential for human usewas not realized until 1982 (20).The delay was due tothe high cost of 18O-labeledwater. If the dose of H218Oused in most small animal studies were to be scaledup for humans, the isotope cost alone would exceed$5000.00. Those isotope doses, however, were established according to the analytical requirements of first-generation gas isotope ratio mass spectrometers or proton activation analysis and improvements in gas isotope ratio mass spectrometry obviated the need for thesehigh doses (21).

Given the potential of the doubly labeled water methodfor measuring human energy expenditure, a number ofinvestigators began to probe the assumptions that underpin the method and to validate the method in humans. Because the basic theory had been extensivelyvalidated in animal models, these studies were undertaken not so much to prove the validity of the methodbut to determine the optimal dose and metabolic periodfor human studies and to demonstrate that the reduc

tion in dose did not alter the accuracy and precision ofthe method. Additional studies have been performed torefine the model in an effort to eliminate or identifypotential artifacts when the method is applied underunusual conditions.

OPTIMAL DOSE AND METABOLIC PERIOD

The optimal dose and metabolic period for humanstudies were predicted from a mathematical model basedon propagation of error analysis (22). For that analysis,it was assumed that the error in the isotopie enrichments was random and that any biological variation inthe enrichments was small and covariant for the twoisotopes (22, 23). The optimal dose was predicted to bethat which would give an initial enrichment of 600times the random analytical error. Smaller doses would

TABLE 2

Optimal isotope doses and metabolic period for four hypotheticalsubjects (22)1

SubjectNeonateChild,

8yiMale,25yiFemale,

65 yrWater

half-lifed3.54.58.5102H20H2'8Og/kg

totalbodywater0.4

0.160.30.140.250.120.25

0.12Metabolic

periodd3-103-134-215-25

'Assumes water is labeled to 100% atom percent excess and analytical precisions of 1.7 x 10~5 atom percent (1.1 %) for 2H and3.3 x 10-5 atom percent (0.16 %) for 1SO.

at University of P

ittsburgh HS

LS on M

ay 11, 2007 jn.nutrition.org

Dow

nloaded from

1280 SCHOELLER

lead to precisions that are worse than 5%, whereaslarger doses would raise the isotope cost with minimalimprovement in precision. The optimal period was predicted to be between 0.5 and 3 biological half-lives of

deuterium oxide. Shorter periods yield imprecise results because the change in isotopie enrichment is smallrelative to the analytical error, whereas longer periodsyield imprecise results because the enrichment of thefinal sample above predose isotopie abundance is smallrelative to the analytical error. Table 1 summarizestypical doses and periods for several textbook subjects.

As implied above, the stable isotopes of deuteriumand 18O are naturally occurring isotopes. Unfortu

nately, in reducing the isotope doses to economic levels, the experimental isotopie enrichments above thisnatural abundance are not large, and variations in thenatural abundance cannot be ignored. Fortunately, thenatural variations arise largely as a result of isotopiefractionation during global transport of water, and thevariations in deuterium and 18O tend to be covariant

(24). Theoretically, the error introduced in the doublylabeled water method can thus be minimized if theratio of the final enrichments of deuterium and 18O is

similar to the ratio of the anticipated natural changesin the isotopie abundances (23, 25). This is because thenatural changes in the two isotopie abundances introduce similar errors into both A0 and AH so that theerrors cancel when the difference between the twoelimination rates is calculated. Larger errors may occurif the two isotopes are dosed at ratios outside of theabove range (25).

Our laboratory has tested the predictions of this errormodel and found it to be qualitatively correct. The model,however, underestimated the error for samples collected more than 2 wk after the predose (isotopie baseline) sample. Presumably this results from an underestimate of the effects of natural variations in the isotopie abundances. We have also demonstrated that theerror introduced by fairly large changes in natural abundance is less than 5% when the dose is optimized forthe anticipated change (unpublished data), but may beas large as 10-30% when the changes are not proportional as can happen when subjects are first given is-otopically unusual fluids such as intravenous fluids (26).

TESTING THE ASSUMPTIONS OF THE METHOD

The basic assumptions of the method have been extensively discussed by Lifson and McClintock (18) andNagy (19). In the application of the doubly labeled watermethod to humans, these assumptions have been reex-amined by a number of investigators in an effort toimprove the model and to help predict the magnitudeof error that might arise under nonlaboratory conditions. There are six basic assumptions. In general, theassumptions are those of single-compartment kinetics.Most tests have demonstrated that the assumptions are

reasonable but not perfect because they involve someapproximations.

Constant pool volume. The basic doubly labeled watermodel assumes that the volume of the water pool inwhich the labels are diluted is constant. Because eatingand drinking behavior is episodic rather than constant,this assumption is seldom perfect. Furthermore, subjects who are either growing or losing weight usuallyhave systematic changes in the pool size during themetabolic period. Because the error associated with achange in the pool size is directly proportional to changein the pool size and because changes that occur within1 d in the pool size typically represent only a smallpercentage of the pool size, this error is quantitativelyinsignificant (19). Systematic changes due to growth orweight loss, however, can be larger and should be considered. For example, a growing premature infant mightincrease total body water by 20% in the period of 1 wk.Several models have been developed for these changes,assuming that the change is either linear or exponentialwith time (18, 19). Perhaps the simplest model is thelinear growth model in which it is necessary to useonly the average of the initial and final pool sizes inthe calculation of CO2 production. Unless the changein pool size is larger than 15%, less than a 1 or 2%error is predicted should the actual change in pool sizebe an exponential rather than a linear function of time(19).

Constant water and CO2 flux. Because the model isbased on steady-state kinetics, it is assumed that the

fluxes of water and CO2 are constant. These fluxes,however, are not constant because water intake andphysical activity are episodic. Despite this, the two-point model, in which samples are collected at startand end of the metabolic period still yields an exactaverage of the flux over the metabolic period (18, 27).In contrast, the multipoint regression models (28) thathave also been employed in the calculation of the isotope elimination rates do not yield exact average fluxes;but because day-to-day variations are generally small(Fig. 1 and réf.29), the difference is probably not quantitatively important for most human studies.

Isotopes label body water and CO2 only. This assumption and its corollary "body water acts as a singlecompartment with respect to the labels" have been the

most controversial assumptions. There are two issuesinvolved in the controversies. The first is the questionof whether or not the isotopes exchange with non-aqueous compartments and the second is whether ornot the isotopie disappearance is consistent with a single monoexponential model. Both of these issues impinge on the appropriateness of the single-compartment, steady-state model.

The question of exchange of isotopes between bodywater and body solids has been debated since labeledwaters were first used to measure body water by isotopedilution. In most investigations of the use of isotopiehydrogen, the hydrogen dilution space has been ob-

at University of P

ittsburgh HS

LS on M

ay 11, 2007 jn.nutrition.org

Dow

nloaded from

DOUBLY LABELED WATER 1281

18o_ 25t-,.,•o - ^ » *

£ IUU O.

1.5

1.8

3 C

? È 1.0

o"u

0.6

a:meal

exercisesleep

xiöz

1X10'2

6x1ö3

B

Day

FIGURE 1 (A) Isotopie elimination following a loading doseof 2H2I8O to a single subject living in a respiration chamber.Isotopie enrichment of urinary water is expressed as the atompercent in excess of the abundance in predose urine. (5) CO2production rate measured by respiratory gas exchange. (C)Times of sleep, exercise and meals. [Reprinted, with permission, from Am. J. PhysioL, (37).j

served to be between 2 and 6% larger than the bodywater pool as determined by desiccation (30), althoughlarger differences have been reported (31). In studies inwhich doubly labeled water has been used, the hydrogen dilution has been found to be consistently largerthan the oxygen dilution space (19). This indicates theexistence of a rapidly exchangeable pool of nonaqueoushydrogen, which is probably the pool of acidic hydrogenin protein (32). The difference between the hydrogenand oxygen dilution spaces (±so) has been reported toaverage 2.0 ±1.1% in premature infants (29), 4 ±4 inmalnourished infants and children (33), 3.8 ±1.6% inadolescents (34) and 3.3 ±1.5 in adults (34). These differences are similar to those reported in other species,including 6.9 ±2.6% (2) and 1.6% (19) in mice and3.6% in lizards (19). These observations are consistentwith the hydrogen space being proportionally larger thanthe water space by about 4%. The oxygen space alsoappears to overestimate the water space (2, 19), but theoverestimate appears to be closer to 1% (2, 19, 35).

Because the isotope spaces are larger than the waterspace, several investigators have modified the doublylabeled water model by using the dilution spaces ratherthan total body water for the calculation of the materialflux (36, 37). Our laboratory has used the average difference between these spaces in a manner similar tothat proposed by Lifson and McClintock (18). In thismodel (37), the relationships between the isotope dilution spaces and total body water are assumed to beconstant, i.e., hydrogen dilution space/1.04 = oxygendilution space/1.01 = total body water.

Roberts, Coward and Lucas (36) have also used a modelthat considers the differences in the pool sizes of thetwo labels, but chose to use individually determinedpool sizes rather than average pool sizes. With thatmodel, pool sizes were defined as the isotope dilutionspaces calculated from the zero time intercept of individual plots of isotopie enrichment versus time. Useof the individual values was suggested because randomerrors in the isotopie enrichments would tend to produce partially canceling errors in the dilution space andelimination rate so as to minimize the error in thecalculated CO2 flux (28). The central issue in the debatebetween proponents of use of a constant relationshipbetween the deuterium and oxygen dilution spaces andproponents of use of individually determined dilutionspaces is whether the range of observed values aroundthe mean is due to random measurement error or if itis a real physiological variation. This is not an unimportant debate, because the doubly labeled water methodmeasures CO2 production by difference between thedeuterium and 18O elimination, and thus small differences in the dilution spaces are magnified 3- to 5-fold.

In our laboratory, the measured ratio of the deuteriumto 18O dilution space in adolescents and adults has av

eraged 1.034 with a standard deviation of 0.016 (34).Repeat measures in five of these individuals had thesame average standard deviation, indicating that thevariation was due to measurement error rather than aphysiological phenomenon characteristic of the adultsubject. Further indirect evidence is provided by an increase in the coefficient of variation of the doubly labeled water method when we used individually determined dilution spaces (10-12%) compared with theabove assumed fixed relationship (7%). In contrast, arecalculation of the validation data of Roberts et al. (29)with a fixed relationship between the dilution spacesincreases the coefficient of variation from 5 to 9% indicated that the interindividual variation in the relativeisotopie dilution spaces are physiological. Thus thereis evidence to support both sides of the issue. Shouldone choose to use individually determined dilutionspaces in the flux calculation, it is important to measure the individual dilution spaces with a precision ofgreater than 1%. Lower precisions result in large errorsin estimated CO2 production determination becausethe errors are multiplied by a factor of 3-6 when thedifference between the two isotope fluxes is calculated.

Recently, Speakman (38) questioned the general useof separate dilution spaces. On purely theoreticalgrounds, Speakman argued that the influence of thedifference in dilution spaces could be canceled by theirreversible losses of hydrogen level via nonaqueousroutes. In other words, the slowing of the hydrogenlabel turnover in body water because of the expansionof the pool by rapid isotope exchange could be canceledby an increase in the efflux of the label due to nonex-changeable incorporation into body solids (i.e., de novolipid synthesis) or excretion in the form of solids (i.e.,

at University of P

ittsburgh HS

LS on M

ay 11, 2007 jn.nutrition.org

Dow

nloaded from

1282 SCHOELLER

weak acids). For the influences to cancel, however, thefractional differences between the dilution spaces wouldhave to equal the fractional increase in loss of the hydrogen label via nonaqueous routes.

Although not an unreasonable theoretical argument,all evidence in humans indicates that nonaqueous efflux from the rapidly exchangeable hydrogen pool issmall and that the two-dilution-space model is pre

ferred. Nonaqueous hydrogen label efflux in urine andfeces in adults consuming a western diet is less than0.03% of total aqueous hydrogen efflux (23). Even withallowance for a potential 5-fold increase for a high fiberdiet (39), this would not be a quantitatively significantloss. I estimated irreversible incorporation into bodysolids to be only 0.1% of aqueous turnover on the basisof the small amount of labeled hydrogen found in skinsolids and fat in a man exposed to tritiated water for 8mo (40). Higher values of irreversible incorporation mightbe encountered in rapidly growing infants, but theseare also surmised to be small as the two-dilution-space

model has now been shown to be just as accurate inpremature infants (29) and full-term, postsurgical in

fants (41) as it is in adults (37, 42).The corollary to the above assumption that body water

acts as a single compartment with respect to the twolabels was questioned by Klein et al. (43), who observedlarge deviations from a smooth exponential isotopieelimination in one subject. Because of this, they questioned the use of the two-point method for determining

the elimination rate. We repeated their experiment andreported a smooth decay, with only a small diurnalvariation that reflected day-night differences in waterflux (Fig. 1) (37). Smooth decay curves have also beenreported by Roberts et al. (29), and we have recentlydemonstrated monoexponential decay for subjects followed for as long as 6 wk after the loading dose (44).Although these two studies did not involve analysis ofwithin-day urine samples, they support the assumptionof single-compartment kinetics by urine sampling tomonitor the enrichment of total body water. Becausethe observation has not been reproduced, the variationreported in that single subject by Klein et al. (43) probably reflects analytical or sample handling error.

Since the report by Klein et al. (43), some investigators have begun to calculate the isotopie eliminationrates from regression analysis by using isotopie enrichments of multiple samples collected periodicallythroughout the metabolic period rather than from justthe initial and final samples as has been the approachused in animal studies and by several groups doing human studies. In the initial validation reported by Coward and Prentice (28) use of the regression approachimproved the results compared with those from thetwo-point approach, but it should be noted that theerrors they observed in the CO2 flux calculated withthe two-point approach were two to three times thosereported by our group, indicating a possible problemwith mass-spectrometric performance. Thus it is pos

sible that the regression approach simply improved theresults because of an averaging of random analyticalerror. With the benefit of hindsight, a more statisticallyacceptable comparison of the two methods should haveincluded multiple analyses of the two samples used forthe two-point calculation. More recently, the group from

the Dunn Nutrition Unit has reported that in a furthercomparison of the two-point approach with their slope-

intercept approach with multiple urine samples gavecomparable results with a mean difference of 2% anda coefficient of variation of 7% (45). It should be notedthat one of the limitations of the doubly labeled watermethod is that the isotopie analyses are not routineand must be made with great care to avoid analyticalerrors.

In a study by Stein et al. (46), the authors did raisethe issue of whether recent water intake might temporarily alter the enrichment of plasma water samplesand recommended that samples only be collected aftera period of taking nothing orally. In light of this study,our laboratory calculated daily energy expenditure forthe previously reported study (Fig. 1) (37). Urine samples were collected at regular intervals from this subjectduring the day without any attempt to distance thecollection from periods of fluid intake or mild exercise.Within the uncertainty of the analytical error of theisotopie analysis (22), no deviations from measured dailyCO2 production were noted (Table 3). There was a tendency for greater variability for the period ending afterthe first daily treadmill exercise, but this was not observed after the second exercise period. Thus the rigorous precautions for sampling recommended by Steinet al. (46) are probably not necessary when workingwith subjects under typical conditions of eating andexercise. The power of this comparison, however, istoo low to prove equality between the various samplingintervals.

Enrichments of water and CO2 exiting the body arethe same as body water. The implicit assumption isthat the rate of tracer efflux exactly represents the rateof tracéeefflux. Any isotope fractionation that decreases the relative enrichment of the tracer results in

TABLE 3

Absence of influence of time of urine collection on daily ( »production rate by doubly labeled water1

Collection timeNumber of 24-h

periods rCO,

06301145132020152345

43333

mol/d26.0 ±6.625.8 ±5.623.3 ±11.826.6 ±7.623.4 ±6.0

'Values are means ±SD;value from respiratory gas exchange =28.2 ±1.0.

at University of P

ittsburgh HS

LS on M

ay 11, 2007 jn.nutrition.org

Dow

nloaded from

DOUBLY LABELED WATER 1283

an underestimate of efflux, whereas a relative enrichment above that in body water causes an overestimateof tracéeefflux. Measured isotope fractionation factorsfor deuterium and 18Oindicate that breath water, non-sweat transcutaneous water vapor and CO2 are isotop-ically fractionated relative to body water, whereas otherfluids are not (23, 47). Because of the isotope fractionation, the doubly labeled water model has been modified to correctly calculate the efflux of tracéefrom thetracer data. This modification has been detailed by Lif-son and McClintock (18);however, it should be pointedout that the isotope fractionation values proposed byLifson were determined for 25°Cand thus are in error

because body temperature is higher. More appropriatevalues have been published recently (23, 47). It may bepossible that these values will vary with ambient temperature. Perhaps this would not be important for oxygen fractionation in CO2 because the fractionation occurs at the lung surface, but it may be a factor for watervapor if changes in the temperature of the nasal cavitiesoccur. However, even a change of 5°Cin the temper

ature at the interface at which fractionation of wateroccurs would have less than a 1% effect on calculatedCO2 production for typical adult subjects.

No CO2 or water enters the body via skin or lungs.Because the aim of the doubly labeled water method isto measure the subjects dietary water intake and CO2production and not environmental water or CO2 exchange, these two sources of material flux can causeerrors in the method. Although exchange with environmental water and CO2 has been demonstrated (19,39, 40), the error is usually quantitatively unimportantfor measuring CO2 production.

Pinson and Langham (40) demonstrated that atmospheric water vapor is readily absorbed through the skinand the lungs. If the water vapor is unlabeled, then itwill increase the water flux but not affect the calculation of CO2 production because the elimination ratesof hydrogen and oxygen are affected equally.

Similarly, environmental CO2 is apparently absorbedthrough the lungs and/or the skin, and this does causethe doubly labeled water method to overestimate CO2production (19).This has presented a problem in studiesof small burrowing mammals that live in enclosed,poorly ventilated spaces, but it is hard to picture humansituations that would result in similar increased levelsof environmental CO2. If all of the inspired CO2 in air(0.04%) mixes with the body pool of CO2, it will increase the apparent CO2 production by about 1%. Theerror should increase proportionally with any increasein ambient CO2 concentration. Similarly, we can estimate the error from cigarette smoking. On the assumption that the consumed portion of each cigarettecontains 0.7 g of combustible material and that thesmoker inhales two-thirds of the CO2 produced by thecigarette, smoking three packages per day would produce an error of about 1 mol of CO2 per day or 3-6%of true CO2 production. Thus, heavy smoking or un

usual environments in which ambient CO2 levels exceed 0.2% are expected to result in overestimates ofCO2 production.

CONVERSION OF CO2 PRODUCTION TOENERGY EXPENDITURE

As indicated above, the doubly labeled water methodmeasures CO2 production and thus is a form of indirectcalorimetry. Heat production can be calculated by using standard indirect calorimetrie relationships. This,however, requires knowledge of the metabolic fuel because the heat released per liter of CO2 produced differsby 30% between carbohydrate and lipid. This information could be obtained through continuous monitoring of respiratory gas exchange, but it would defeatthe nonrestrictive character of the doubly labeled watermethod. Thus it is more convenient to estimate themetabolic fuel mix from dietary intake. Black, Prenticeand Coward (48) have recently demonstrated that therespiratory quotient is quite similar to the food quotient and that the former can therefore be predictedfrom the latter. The error in calculating energy expenditure from the food quotient is less than 3% inmost situations, although care must be taken to account accurately for alcohol intake. The authors (48)further pointed out that if energy intake differs fromenergy expenditure, then some adjustment should bemade to the food quotient to correct for body fat utilization or storage. This effect on calculated heat production is not more than 5% unless the difference between intake and expenditure exceeds 20%.

IMPLICATIONS FOR THE DOUBLY LABELEDWATER MODEL

The basic model first visualized by Lifson, Gordonand McClintock (2) and described by Equation 3 wasmodified by the authors with a correction for isotopefractionation. With this correction, the equation describing the model becomesrC02 = (N/2/3) (A0 - *H)

- rH20G(/2 - (4)

where the subscript G indicates water loss via isotop-ically fractionated routes, and /, is the deuterium fractionation between water and water vapor, /2 is the 18Ofractionation between water and water vapor and /3 isthe 18Ofractionation between water and CO2.

The difference between the deuterium and 18Odilution spaces indicates the need for a further change tothe model in which the single body water pool size Nis replaced with individual isotope dilution spaces D0and DH.With this modification the equation describing

at University of P

ittsburgh HS

LS on M

ay 11, 2007 jn.nutrition.org

Dow

nloaded from

1284

the model takes the form

rCO2 = (1/2/a) (Dok0 - £

- rH20G(/2 - Al/2/3 (5)

The investigators at the Dunn (36) have taken a different approach to the mathematical development of themodel, but they included similar considerations aboutisotope fractionation and differences in dilution spaces.

Although most investigators agree on these points,issues that are still being debated include which valuesto use for the isotope pool sizes and which value to usefor the rate of fractionated water loss.

As discussed above, most investigators favor use ofthe two-pool model for the two isotopes; however, some

favor an individualized approach and some a group approach. Among the latter, the average values are believed to differ by 2-4%, but a single value has yet to

be agreed upon.The question of the correction for isotopically frac

tionated water loss has been debated for a long time.In the original work by Lifson (22), fractionated waterloss was assumed to equal the rate of insensible waterloss, which was about 50%. Nagy (19) and Nagy andCosta (39), however, demonstrated that this was anoverestimate and elected not to include any fraction-

ation correction. Recent studies have demonstrated thatisotope fractionation does occur in vivo (23), but theamount of water that is subject to isotopie fractionationis still under debate. Measured values of insensible waterloss in two recent human studies designed to validatethe doubly labeled water method indicate that the insensible water loss is less than 50% of water efflux,specifically, 25% in one adult (43) and 16% in four premature infants (29). Furthermore, recent evidence indicates that not all insensible water loss is isotopicallyfractionated. Specifically, sweat has been shown to befractionated (23). Thus the Lifson estimate of 50% ofwater efflux is much too large for most human studies.

Our laboratory has taken the approach that a physiologically reasonable estimate of fractionated waterloss can be calculated from the isotopie data (37). Onthe assumptions that expired air is saturated with watervapor at 37°Cand that expired air averages 3.5% CO2,we calculated breath water vapor loss from the uncor-rected CO2 production rate (first right-hand term of

Equation 5). Similarly, we can estimate transcutaneouswater from body surface area, the average rate of non-sweat loss [0.18 g/(min-m2)] (49) and that clothing re

duces that rate by 50% in covered areas. Taken together, these estimates can be mathematically expressed as follows:

SCHOELLER

rH20G = 1.05(Dofc0 -

for adults, and

rH20G = 1.45

(6)

(7)

for infants.

The group at the Dunn Nutrition Unit has measuredbreath water loss and other routes of water vapor lossthat are mostly transcutaneous water loss but manyinclude losses through the sweat glands and have foundthat our estimated losses for breath are somewhat higherthan measured losses and those for transcutaneous waterlosses are somewhat lower than measured. The sum ofthe estimates for two routes, however, is quite close tothe measured value (unpublished observation). This observation is consistent with data in the literature thatindicate that breath is not quite unsaturated at 37°(50).

Because the sum of the two routes is close to the measured value, we have retained the estimates shown inEquations 6 and 7.

VALIDATIONS

We have now completed validations of the doublylabeled water method in 33 subjects (Table 4). Thesevalidations were against respiratory gas exchange ordietary intake plus change in body energy stores ascalculated from total body water and body weight. Energy expenditure from doubly labeled water has beencalculated in three ways. The first is the original Lifsonequation (Equation 4) with the original, but now knownto be incorrect, assumptions of rH2OG = 0.5rH2O andthe fractionation factors for 25°C.The second is the

Lifson equation with the average values of fractionatedwater loss of 25% for adults (43) and 16% for infants(29) and isotope fractionation factors for 37°C(37). The

third calculation includes the correction for dilutionspace based on a measure of the oxygen dilution spaceand on the assumption that the deuterium dilution spaceaverages 1.03 times the oxygen dilution space, the estimates for fractionated water loss shown in equations6 and 7 and the fractionation factors for 37°C(23).

Of these three methods of calculation, the third ismost accurate. Both the original Lifson equation(P < 0.01) and the modified Lifson equation (P < 0.01)are systematically in error. This indicates that the original Lifson equation overestimated the isotope fractionation correction but that this error was partiallycompensated for the use of the single-pool model. Thusit is important to incorporate both the more recentvalues for isotope fractionation and the two-pool cor

rection.Validations of the doubly labeled water method in

humans and against respiratory gas exchange have nowbeen reported by four laboratories (Table 5). Two ofthese, our laboratory and that of Westerterp et al. (52)in The Netherlands have used the two-point methodfor determining the isotope elimination rates. The othertwo laboratories, Klein et al. (43) in Houston and Coward et al. (42) in Cambridge, have used the multipointmethod after reporting imprecise values for the two-

point method.The validations shown in Table 5 include a wide

at University of P

ittsburgh HS

LS on M

ay 11, 2007 jn.nutrition.org

Dow

nloaded from

DOUBLY LABELED WATER 1285

TABLE4

Validations of doubly labeled water against respiratory gas analysis or intake/balance

SubjectM17F25F27M27M58MKJDDMABLCDPIDNMED122b3456789123412345meanSDN1mol24531998172626262520238122982587205929202309239225922664113.5144.7161.9144.9111.2115.886.4170.7115.3134.3321016632162259816091445179614741277k0d-¡0.10330.12730.13470.13220.09500.12920.10480.09850.09990.11270.09480.09540.09640.10470.27390.24010.21480.20030.18130.23360.26060.24160.25510.23390.08780.11650.10410.09590.13280.15590.11310.13350.1196k»d-¡0.07790.10230.10980.10640.07260.10380.08360.07180.07580.08810.07070.07000.07170.08060.24280.20390.18000.17210.15570.20280.22710.21370.22370.20210.06320.09340.07890.07270.10360.12600.09330.10000.0939Reí2m/à25.319.619.527.321.724.222.525.320.927.923.625.027.428.31.231.842.251.53.16.250.98.66.46.5132.615.422.727.818.2616.8615.4319.6413.3615.311.11m/d27.120.917.828.424.425.420.530.421.530.724.326.728.027.61.282.082.271.591.111.361.101.741.351.6534.916.123.626.120.118.014.621.514.016.911.4-ifson%37.16.9-8.64.012.44.8-8.720.22.99.93.06.82.2-2.34.412.81.04.0-4.49.012.05.0-7.39.07.14.84.1-6.010.07.0-5.59.64.64.3'6.6rCO2Modifiedm/d28.822.819.530.926.027.622.332.122.933.025.828.229.729.61.612.432.621.891.311.641.332.181.661.9736.817.525.227.821.619.716.122.915.118.212.1Lifson%313.916.20.113.120.014.0-1.126.99.718.29.212.88.34.530.931.916.423.313.331.335.731.013.730.412.713.810.90.218.216.74.216.412.716.0-10.2Equationm/d26.320.417.427.623.724.720.029.520.929.823.625.927.226.91.221.962.141.501.051.291.041.651.281.5633.915.723.025.419.517.614.220.913.616.411.15%34.04.0-11.01.29.22.0-11.216.70.06.90.03.7-0.8-5.1-0.96.5-4.7-1.7-9.63.26.0-0.4-12.13.14.02.01.1-8.66.94.2-7.96.51.70.66.3Citation(51)(37)(26)(20)(53)

'Total body water calculated from the oxygen dilution space divided by 1.01.2Reference value from measured respiratory exchange or calculated from energy intake plus change in body energy stores.^Percent error relative to reference value, |(x-ref)/x]100. 'Significantly different from zero, P < 0.01.

range of subjects and conditions. Subjects range fromadults (37) to premature infants (29) and from healthysubjects eating oral diets (37) to clinical populationsreceiving total parenteral nutrition (51). Some of thesubjects were exercising at very high intensities withdaily energy expenditures of over 35 MJ/d (52),whereasmost others were sedentary or moderately active. Subjects were in positive energy balance weight (51) or innegative energy balance and losing weight (20).Throughout these validations, the method appears tobe equally accurate.

APPLICATIONS

Having demonstrated the general validity of the doubly labeled water method at economical isotope doses

in humans (ca.$300.00), investigators have begun toapply the method to the study of human energy metabolism. The ultimate range of applications of themethod is hard to predict, but most uses to date havetaken advantage of the ability of the method to accurately measure energy expenditure in free-living subjects. Previously available methods have either beentoo restrictive or too dependent on the subject's co

operation and memory to provide definitive results insuch studies. Obviously, most applications of the doubly labeled water method are very recent and have onlyappeared in abstract form. Those results should be considered preliminary.

A number of investigators have applied the methodto the study of obesity. These studies have generallybeen aimed at measuring the energy expenditure of obeseindividuals to determine if they are energy efficient, as

at University of P

ittsburgh HS

LS on M

ay 11, 2007 jn.nutrition.org

Dow

nloaded from

1286 SCHOELLER

TABLE5

Summary of human validations of the doubly labeled water method

Subjects Ref. method Error ±so Cale.' Citation

AdultAdults,n =4Adult,

n =1Adults,n =5Exercising

adults, n =2Prematureinfants, n =4Adults

on TPN4 n =5Adults,n =9Postsurgical

infants, n =9Adults,

n =5Exercising

adults, n = 8I/B2RGERGERGE3RGERGEI/BRGERGERGERGE-0.41.9-41.5-2.5-1.4Ó.Ó1.4-0.95.62.067.64.94.85.97.76.2-1.0

±7.0-1.4

±4.8SCLSLCSSsssSchoeller

and Van Santen, 1982(20]Cowardand Prentice, 1985,(28)Klein

et al., 1984(43)Schoellerand Webb, 1984(51)Westerterp

et al., 1984(52)Robertset al., 1986(29)Schoeller,

Kushner and Jones 1986|53|Schoelleret al., 1986(37)Iones,

étal.,1987(41)Westerterp,Hamers and Brouns,inpress

(54)Westerterp,Hamers and Brouns,inpress

(54)

'Method of calculating the results, L = calculated per Lifson, Equations 3 and 4 (18), S = calculation per equations 5, 6 and 7 detailed herein,C = calculation as detailed by Roberts et al. (36).

2Reference CO2 production from respiratory gas exchange.3Reference CO2 production calculated from intake plus change in body energy stores."Total parenteral nutrition.

some intake studies have suggested. Prentice et al. (55)measured total energy expenditure in obese and leanwomen and reported that the daily energy expenditureof obese women was nearly 30% higher than that ofthe lean women. Moreover, energy expenditure in theseobese women was comparable to that of the lean womenwhen normalized for body composition and size. Theyalso collected energy intake data by self-reported diary.The obese women reported an intake nearly identicalto that of the lean women, but the intakes of the obesewomen were not consistent with the higher expenditure. Changes in body weight indicated that the lowreported intakes were due to a combination of under-eating and under-reporting. In a similar study in adolescents, Bandini et al. (56) reported that obese adolescents had greater daily energy expenditures than leancontrols but that expenditures were similar when normalized for body composition and size. Again, it wasobserved that self-reported intake underestimated habitual energy requirements by 10 and 30% in lean andobese subjects, respectively. In a third study, Schoelleret al. (57)reported that obese patients with Prader-Willisyndrome were not as energy efficient as previous dietary records have indicated, but that they expendedless energy than obese controls. The difference waslargely because of the very small fat-free masses observed in the Prader-Willi subjects, but the Prader-Willisubjects also appeared to be less physically active. Inthe only prospective study to date, Roberts et al. (58)reported that infants of obese mothers who becameoverweight during the first year of life had a tendencyto have lower daily expenditures than infants that didnot become overweight.

Two studies have been reported in which energy requirements of hospitalized subjects were determined.Novick et al. (59) measured the energy expenditures of

surgical patients and reported that energy expenditureincreased by 18% after surgery and that this increasewas not observed by spot measures of resting metabolicrate. Fjeld, Schoeller and Brown (60) determined theenergy requirements for repletion of malnourished infants and children. In this study, requirements per kilogram of body weight were found to decrease duringthe course of recovery, but the decrease was directlyrelated to the decrease in percent fat-free mass.

A novel application of the doubly labeled water methodwas reported by Lucas et al. (61), who determined themetabolizable energy density of breast milk by measuring the energy expenditure of the infant and addingin the energy deposited in body stores. They reporteda density of about 60 kcal/g, but it is likely that theyslightly overestimated the volume of breast milk andthus underestimated energy density because they didnot compensate for environmental water influx (62).

The majority of applications of the doubly labeledwater method have been aimed at determining energyrequirements of healthy individuals. Many of thesestudies have included populations where estimates ofenergy intake are low compared with the requirementestimated from physical activity. Singh et al. (63)measured the energy expenditure of Cambian women andfound that it was about 10 MJ/d or nearly twice thatestimated from dietary records. Stein et al. (64)studiedGuatemalan women and found that daily energy expenditure was about 8 Mf/D, but that this was in agreement with previous estimates from dietary records.

For developed countries, Davies et al. (65) has reported that the Dunn Nutrition Unit has begun to applythe method in studies to determine the energy costs ofpregnancy and lactation. Butte, et al. (66) reported onmeasurements of the energy expenditures of breast-fedand formula-fed infants. Expenditures were not differ-

at University of P

ittsburgh HS

LS on M

ay 11, 2007 jn.nutrition.org

Dow

nloaded from

DOUBLY LABELED WATER 1287

ent between the groups, although the breast-fed groupdid expend a greater proportion of their energy in basalmetabolism.

Two groups have applied the method in very activesubjects. Westerterp et al. (67), measured the energyexpenditures of male cyclists during the Tour de France.Energy expenditure exceeded 35 MJ/d or five times basalmetabolic rate for these athletes, who were probablyworking at their energetic ceiling. Haggarty and McGaw(68) reported that elite female endurance runners wereexpending almost 15 MJ/d or about 2.8 times their calculated basal metabolic rate. They also reported thatthis weight-conscious group severely under-reportedtheir dietary intake.

SUMMARY

The experience with the doubly labeled water methodin humans has confirmed the numerous validations insmall animals. The method is accurate and has a precision between 2 and 8%, depending on the loadingdose, the length of the metabolic period and the numberof samples. Thus there is little question about the validity of the method. At the same time, there is anabsence of consensus in the literature regarding themodel and correction factors to be used. Part of thisreflects the nature of scientific evaluations as each pointis addressed individually. On the basis of evidence todate, future use in humans should include correctionfor isotope fractionation with the fractionation factorsderived at 37°.These corrections should be applied to

only breath water vapor, transcutaneous water vaporloss and CO2. It is also becoming clear that the difference in the two isotope dilution spaces must be considered in the model.

Consensus has yet to reached on the manner for estimating the amount of water that is lost via routessubject to isotopie fractionation. There is also an absence of consensus on the manner in which to incorporate the correction for the difference in the two isotope dilution spaces. Although the estimates used forboth of these factors have been shown to lead to veryaccurate results in our laboratory, equally accurate results have been obtained at the Dunn Nutrition Unitwith their approach. This probably reflects the observation that the mean dilution space and estimated individual correction for fractionated water loss used inour laboratory are in very close agreement with themean of the individual measured dilution space andaverage fractionated water loss values used by the Dunngroup.

The doubly labeled water method is noninvasive andnonrestrictive and is thus ideal for the measurementof total daily energy expenditure in free-living subjects.The number of applications has increased dramaticallyin the last two years and these studies are beginningto provide new answers to old questions about human

energy metabolism. Many of these studies are still inprogress and thus cannot be fully evaluated at this time.Many reports, however, have keyed on consistent differences between measured energy expenditure and intake estimated by dietary record and thus raised concern about the accuracy of dietary records in weight-conscious individuals. Thus it may be shown that anumber of the enigmatic differences between dietaryintake and physical activity in humans are at least partially due to under-reported intakes.

LITERATURE CITED

1. LIFSON,N., GORDON, G. B., VISSCHER,M. B. & NIER, A. O.(1949] The fate of utilized molecular oxygen and the source ofheavy oxygen of respiratory carbon dioxide, studied with the aidof heavy oxygen. /. Biol. Chem. 180: 803-811.

2. LIFSON,N., GORDON,G. B. & MCCLINTOCK,R. (1955) Measurement of total carbon dioxide production by means of D218O. /.Appi. Physiol. 7: 704-710.

3. MCCLINTOCK,R. & LIFSON,N. (1957] Applicability of the D2I8O

method to the measurement of the total carbon dioxide outputof obese mice. /. Biol. Chem. 226: 153-156.

4. MCCLINTOCK,R. &. LIFSON,N. (1958) CO2 output of micemeasured by D2O18under conditions of isotope re-entry into thebody. Am. ]. Physiol. 195: 721-725.

5. MULLEN,R. K. (1970) Respiratory metabolism and body waterturnover rates of Perognathus formosus in its natural environment. Comp. Biochem. Physiol. 32: 259-265.

6. MCCLINTOCK,R. & LIFSON,N. (1958) Determination of thetotal carbon dioxide outputs of rats by the D2O18 method. Am.J. Physiol. 192: 76-78.

7. LEE,]. S. & LIFSON,N. (I960] Measurement of total energy andmaterial balance in rats by means of doubly labeled water. Am.J. Physiol 199: 238-242.

8. LIFSON,N. & LEE,J. S. (1961) Estimation of material balanceof totally fasted rats by doubly labeled water. Am. ]. Physiol.200: 85-88.

9. RANDOLPH,J. C. (1980) Daily energy metabolism of two rodents (Peromyscus leucopus and Tamias striatus] in their naturalenvironment. Physiol. Zool. 53: 70-81.

10. LITTLE,W. S. & LIFSON,N. (1975) Validation study of D218O

method for determination of CO2 output of the eastern chipmunk {Tamias stiaitus}. Comp. Biochem. Physiol. 50A: 55-56.

11. KARASOV,W. H. (1981 ) Daily energy expenditure and the costof activity in a free living mammal. Oecologia (Beri.) 51: 253-259.

12. LEFEBVRE,E. A. (1964) The use of D2O18for measuring energymetabolism in Colomba livier at rest and in flight. Auk 81: 403-416.

13. WILLIAMS,I. B. (1985) Validation of the doubly labeled watertechnique for measuring energy metabolism in starlings and sparrows. Comp. Biochem. Physiol. 80A: 349-353.

14. BUTTEMER,W. A., HAYWORTH,A. M ., WEATHERS,W. W. & NAGY,K. A. (1986) Time budget estimates of avian energy expenditures: physiological and methodological considerations. Physiol. Zool. 59: 131-149.

15. BUSCARLET,L. A., PROUX,J. & GESTER,R. (1978) Utilization dudouble marguage HTI8O dans une étudedu bilan métaboliquechez Locusta migiandia migiatoriodes. J. Insect Physiol. 24: 225-232.

16. KING,W. W. & HADLEY,N. F. (1979) Water flux and metabolicrates of free of free-roaming scorpions using the doubly labeledwater techniques. Physiol. Zool. 52: 176-189.

at University of P

ittsburgh HS

LS on M

ay 11, 2007 jn.nutrition.org

Dow

nloaded from

1288 SCHOELLER

17. COOPER, P. D. (1983) Validation of doubly labeled water|H3H'8O) method for measuring water flux and energy metabolism in lenebronoid beetles. Physiol. Zoo!. 56: 41-46.

18. LIFSON,N. & McCuNTOCK, R. I1966) Theory of use of theturnover rates of body water for measuring energy and materialbalance. /. Theor. Biol. 12: 46-74.

19. NAGY, K. (1980) CO2 production in animals: analysis of potential errors in the doubly labeled water method. Am. /. Physiol.238: 466-473.

20. SCHOELLER,D. A. & VAN SAMTEN,E. (1982) Measurement ofenergy expenditure in humans by doubly labeled water method./. Appi. Physiol. 53: 955-959.

21. LIFSON,N., LITTLE,W. S., LEVITT,D. G. & HENDERSON,R. M.(1975) D2180 method for CO2 output in small mammals andeconomic feasibility in man. /. Appi. Physiol. 39: 657-664.

22. SCHOELLER,D. A. (1983) Energy expenditure from doubly labeled water: some fundamental considerations in humans. Am./. Clin. Nutr. 38: 999-1005.

23. SCHOELLER,D. A., LEITCH,C. A. & BROWN,C. (1986) Doublylabeled water method: in vivo oxygen and hydrogen isotope frac-tionation. Am. /. Physiol. 20: 1137-1143.

24. DANSGAARD,W. (1964) Stable isotopes in precipitation. Tellus16: 436-468.

25. ROBERTS,S. B., COWARD,W. A., EWING,G., SAVAGE,]., COLE,T. J.& LUCAS,A. (1988) Effect of weaning on the accuracy of thedoubly labeled water method in infants. Am. }. Physiol. 254:R622-R627.

26. IONES,P. H. J., WINTHROP,A. L., SCHOELLER,D. A., FILLER,R. M.,SWYER,P. R., SMITH,J. & HEIM,T. (1988) Evaluation of doublylabeled water for measurement of energy expenditure duringchanging nutrition. Am. ]. Clin. Nutr. 47: 799-804.

27. SPEAKMAN,J. R. & RACEY,P. A. (1986) Measurement of CO2production by the doubly labeled water technique. /. Appi. Physiol. 61: 1200-1202.

28. COWARD,W. A. & PRENTICE,A. M. (1985) Isotope method forthe measurement of carbon dioxide production rate in man. Am./. Clin. Nutr. 41: 659-663 (lett.).

29. ROBERTS,S. B., COWARD,W. A., SCHLINGENSEIPEN,H.-H. NOHRIA,V. & LUCAS,A. (1986) Comparison of the doubly labeled water(2H218O)method with indirect calorimetry and a nutrient-balance

study for simultaneous determination of energy expenditure, waterintake and metabolizable energy intake in preterm infants. Am.]. Clin. Nutr. 44: 315-322.

30. SCHOELLER,D. A. & IONES,P. J. H. (1987) Measurement of totalbody water by isotope dilution: a unified approach to calculations. In: In Vivo Body Composition Studies (Ellis, K. J., Yasu-mura, S. and Morgan, W. D. eds.), pp. 131-137, Institute of Physical Sciences in Medicine, London.

31. SHENG,H. P. & HUGGINS,R. A. (1979) A review of body composition studies with emphasis on body water and fat. Am. ¡.Clin. Nutr. 32: 630-647.

32. CULEBRAS,J. M. & MOORE,F. (1977) Total body water and theexchangeable hydrogen. I. Theoretical calculation of nonaqueousexchangeable hydrogen in man. Am. ]. Physiol. 232: 54-59.

33. FIELD,C. R., SCHOELLER,D. A. & BROWN,K.H. (1987) Calculationof percent hydration of fat-free mass during recovery from marasmus. In: In Vivo Body Composition Studies (Ellis, K. ]., Yas-umura, S. and Morgan, W. D., eds.], pp. 138-143, Institute ofPhysical Sciences in Medicine, London.

34. SCHOELLER,D. A., KUSHNER,R. F., TAYLOR,P., DIETZ,W. H. &BANDINI,L. (1985) Measurement of total body water: isotopedilution techniques. In: Sixth Ross Conference on Body Composition Assessments in Youth and Adults (Roche, A. F., ed.),pp. 124-129, Ross Laboratories, Columbus, OH.

35. SCHOELLER,D. A., VANSANTEN,E., PETERSON,D. W., DIETZ,W.,IASPAN,J. & KLEIN,P. D. (1980) Total body water measurement in humans with 18O and 2H labeled water. Am. /. Clin.Nutr. 33: 2686-2693.

36. ROBERTS,S. B., COWARD,W. A. & LUCAS,A. (1987) Reply to

letter by Wong, Butte, Garza, and Klein. Am. /. Clin. Nutr. 45:1545-1547 (lett.).

37. SCHOELLER,D. A., RAVUSSIN,E., SCHUTZ,Y., ACHESON,K. f.,BAETSCHI,P. & JEQUIER,E. (1986) Energy expenditure by doublelabeled water: validation in humans and proposed calculation.Am. /. Physiol. 250: R823-R830.

38. SPEAKMAN,J. R. (1987) Calculation of CO2 production in doubly labelled water studies. /. Theor. Biol. 126: 101-104.

39. NAGY, K. A. & COSTA,D. P. (1980) Water flux in animals:analysis of potential errors in the tritiated water method. Am.J. Physiol 238: 454-465.

40. PINSON,E. A. & LANGHAM,W. H. (1957) Physiology and iicology in man. /. Appi. Physiol 10: 108-126.

41. JONES,P. J. H., WINTHROP,A. L., SCHOELLER,D. A., SWYER,P. R.,SMITH,I., FILLER,R. M. & HEIM,T. (1987) Validation of doublylabeled water for assessing energy expenditure in infants. Pediatr.Res. 21:242-246.

42. COWARD,W. A., PRENTICE,A. M., MURGATROYD,P. R., DAVIES,H. L., COLE,T. J., SAWYER,M., GOLDBERG,G. R., HALLIDAY,D. &MACNAMARA,J. P. (1984) Measurement of CO2 and water production rates in man using 2H18O; comparisons between calo

rimeter and isotope values. In: A Concerted Action Project onNutrition in the European Community (Van Es, A. J. H., ed.),pp. 126-128, Agricultural University, Wageningen, The Netherlands.

43. KLEIN,P. D., JAMES,W. P. T., WONG,W. W., IRVING,C. S., MURGATROYD,P. R., CABRERA,M., DALLOSSO,H. M., KLEIN,E. R. &NICHOLS,B. L. (1984) Calorimetrie validation of the doubly-labelled water method for determination of energy expenditurein man. Hum. Nutr.: Cbn. Nutr. 38C: 95-106.

44. SCHOELLER,D. A. & TAYLOR,P. B. (1987) Precision of the doubly labelled water method using the two-point calculation. Hum.Nutr.: Clin. Nutr. 41C: 215-22.

45. COWARD,W. A. (1988) The 2H218Otechnique-principles and

practice. Proc. Br. [. Nutr. In press.46. STEIN,T. P., HOYT,R. W., SETTLE,R. G., OTOOLE, M. & HILLER,

W. D. B. (1987| Determination of energy expenditure duringheavy exercise, normal daily activity, and sleep using the doubly-labelled-water (2H218O)method. Am. ]. Clin. Nutr. 45: 534-539.

47. WONG,W. W., COCHRAN,W. J., KLISH,W. J., SMITH,E. OU, LEE,L. S. & KLEIN,P. D. (1988) In vivo isotope fractionation factorsand the measurement of deuterium and oxygen-18-dilution spacesfrom plasma, urine, saliva, respiratory water, vapor, and carbondioxide. Am. f. Clin. Nutr. 47: 1-6.

48. BLACK,A. E., PRENTICE,A. M. & COWARD,W. A. (1986) Useof food quotients to predict respiratory quotients for the doubly-labelled water method of measuring energy expenditure. Hum.Nutr.: Cbn. Nutr. 40C: 381-391.

49. KUNO,Y. (1956) Human Perspiration, p. 30, Thomas, Springfield, IL.

50. HEY, E. N. & KATZ,G. (1969) Evaporative water loss in thenew-born baby. /. Physiol. (London/ 200: 605-619.

51. SCHOELLER,D. A. & WEBB,P. (1984) Five-day comparison ofthe doubly labeled water method with respiratory gas exchange.Am. /. Clin. Nutr. 40: 153-158.

52. WESTERTERP,K. R., DE BOER,J. O., SARIS,W. H. M., SCHOFFELEN,P. F. M. & TEN HOOR, F. (1984) Measurement of energy expenditure using doubly labelled water. Int. f. Sports Med. 5: 74-75.

53. SCHOELLER,D. A., KUSHNER,R. F. & JONES,P. J. H. (1986)Validation of doubly labeled water for measuring energy expenditure during parenteral nutrition. Am. f. Clin. Nutr. 44:291 -298.

54. WESTERTERP,K. R., HAMERS,J. & BROUNS,F. Validation of thedoubly-labelled water technique at low and high activity levels.

Proc. Brit. Nutr. Soc. In press (abs.|.55. PRENTICE,A. M., BLACK,A. E., COWARD,W. A. DAVIES,H. L.,

GOLDBERG,G. R., MURGATROYD,P. R., ASHFORD,J., SAWYER,M.

at University of P

ittsburgh HS

LS on M

ay 11, 2007 jn.nutrition.org

Dow

nloaded from

DOUBLY LABELED WATER 1289

& WHITEHEAD,R. (1986| High levels of energy expenditure inobese women. Bi. Med. J. 292: 983-987.

56. BANDINI,L. G., SCHOELLER,D. A., YOUNG,V. R. & DIETZ,W. H.(1987) Basal metabolic rate (BRM) and total daily energy expenditure (TOEE) in obese and nonobese adolescents. Am. ¡.Clin.Ñutí.45: 868 (abs.).

57. SCHOELLER,D. A., LEVTTSKY,L. L., BANDINI,L., DIETZ,W. H. &WALZAR,A. (1988| Energy expenditure and body compositionin Prader-Willi syndrome. Metabolism 37: 109-115.

58. ROBERTS,S. B., SAVAGE,J., COWARD,W. A. & LUCAS,A.(1988) Contributions of low energy expenditure and excessiveenergy intake to the development of fatness during infancy. Proc.Br. Ñutí.Soc. In press (abs.).

59. NOVICK,W. M., NUSBAUM,N. & STEIN,T. P. (1987) The energycosts of elective surgery as measured by the doubly labelled water(2H218O).Fed. Pioc. 46: 581 (abs.).

60. FIELD,C. R., SCHOELLER,D. A. & BROWN,K. H. (1988) Energyexpenditure of malnourished children during catch-up growth.

Pioc. Bi. Nutr. Soc. In press.61. LUCAS,A., EWING,G., ROBERTS,S. B. &. COWARD,W. A. (1987)

How much energy does the breast fed infant consume and expend? Bi. Med. f. 295: 75-77.

62. FJELD,C. R., SCHOELLER,D. A. & BROWN,K. H. (1988) Validationof the deuterium oxide method for measuring average daily milkintake in infants. Am. ]. Clin. Ñutí.In press.

63. SINGH,J., COWARD,W. A., PRENTICE,A. M., ASHFORD,]., SAWYER,M., DIAZ,E. & WHITHEAD,R. G. (1988) Doubly-labelled watermeasurements of energy expenditure in Cambian women duringthe agricultural season. Pioc. Bi. Ñutí.Soc. In press (abs.).

64. STEIN,T. P., JOHNSON,F. E. &. GREINER,L. (1987) Energy expenditure and socioeconomic status in Guatemala as measuredby the doubly-labelled water method. Fed. Pioc. 46: 1184 (abs.).

65. DAVIES,H. L., PRENTICE,A. M., COWARD,W. A., GOLDBERG,G.R., BLACK,A. E., MURGATROYD,P. R., SCOTT,W., ASHFORD,J. &SAWYER,M. (1988) Individual variation in the energy cost ofpregnancy. 2. Doubly-labelled water method. Pioc. Bi. Ñutí.Soc.

In press (abs.).

66. BUTTE,N. F., WONG,W. W., LEE,S. L., GARZA,C. & KLEIN,P. D.(1988) Energy expenditure of 4-month-old breast-fed and formula-fed infants. Pioc. Bi. Ñutí.Soc. In press (abs.).

67. WESTERTERP,K. R., SARIS,W. H. M., VAN Es, M. &. TEN HOOR,F. (1986) Use of the doubly labeled water technique in humans during heavy sustained exercise. /. Appi. Physiol. 61:2161-

2167.

68. HAGGARTY,P. & McGAW, B. A. (1988) Energy expenditure ofelite female athletes measured by the doubly-labelled water

method. Pioc. Bi. Ñutí.Soc. In press (abs.).

at University of P

ittsburgh HS

LS on M

ay 11, 2007 jn.nutrition.org

Dow

nloaded from