Kidney International, Vol. 58 (2000), pp. 408–416

Comparing different methods of assessing body compositionin end-stage renal failure

BRUCE A. COOPER, ALIREZA ASLANI, MURIELLE RYAN, FIONA Y-P. ZHU, LLOYD S. IBELS,BARRY J. ALLEN, and CAROL A. POLLOCK

Department of Renal Medicine, Royal North Shore Hospital, University of Sydney, Sydney, New South Wales, Australia

Comparing different methods of assessing body composition Patient outcome in end-stage renal failure has beenin end-stage renal failure. shown to correlate with the adequacy of dialysis delivery.

Background. Accurate measurement of nutritional status in To allow for the comparison of dialytic doses in continu-patients with end-stage renal disease is important because ofous ambulatory peritoneal dialysis (CAPD) patients ofits clear association with prognosis. Total body water (TBW)different size, dialytic clearances are frequently normal-has additionally been recently recognized as an independent

prognostic value because of its relationship with hypertension ized to either body surface area or, more commonly, theand cardiac morbidity. The current study was designed to assess volume of distribution of urea (Vdurea). In CAPD, thethe utility of surrogate markers of nutritional state and TBW most commonly used estimate of dialytic clearance,in patients with end-stage renal disease.

Kt/V, the K (clearance), and t (time) are measured,Methods. Fifty-four patients with renal disease were studied.whereas the V is estimated, usually using anthropometri-TBW obtained using the deuterium dilution technique was com-

pared with estimates derived from anthropometric measures cally measured data that include body weight to deriveof TBW, including 58% body weight, Watson equations, and the Vdurea. Similarly, in hemodialysis (HD), Kt/V deter-bioelectrical impedance analysis (BIA). Anthropometrically mines dialytic adequacy, with derived formulae usingderived fat-free mass (FFM) was compared with BIA-derived

estimated dry body weight as a marker of Vdurea. As ureaestimates. Total body nitrogen (TBN) measurements were cor-is uniformly distributed in body water, the estimated Vrelated with TBW estimates and BIA-derived resistance.

Results. TBW was significantly underestimated by the Wat- is equal to total body water (TBW). Thus, the accuracyson equation (mean difference, 21.751 L, P 5 0.01) and the of the method used to determine V or dry body weight58% body weight approximation significantly overestimated influences the reliability of the Kt/V calculation. Errorsit (mean difference, 1.792 L, P 5 0.04). The Kushner BIA

in the estimation of body water will inversely alter theestimation of TBW did not significantly differ from that of thegold standard determined from D2O dilution (mean difference, final Kt/V result; that is, underestimation of V will give an21.221 L, P 5 0.12) and was also the method that showed the inflated Kt/V and vice versa. This may have a significantbest agreement with the D2O estimate. However, the limits of impact on dialysis prescription and subsequent clinicalagreement were large. Accurate prediction equations for FFM

outcome.(FFM 5 221.768 1 0.001 3 ht2 1 6630.669 3 1/R 1 0.312 3Although the recent focus on the estimation of TBWwt, R2 5 0.95) and TBN (TBN 5 2668.324 2 3.963 3 age 1

10.133 3 wt 1 0.045 3 ht2 1 32141.457 3 1/R, R2 5 0.91) were has been because of its relevance to the normalizationderived from BIA obtained resistance. of clearance data in CAPD patients, TBW in itself is

Conclusions. The estimation of TBW varies significantly de- an independent predictor of outcome [1]. Low TBW ispending on the method of calculation. BIA is the most accurate

associated with a worse outcome in dialysis patients,surrogate marker for the measurement of both TBW and othercaused partly due to its correlation with fat-free massparameters of body composition.(FFM) and hence nutritional status. Patients with lowTBW can, in fact, have a high urea reduction rate (URR),which may not be predictive of a good outcome. It hasbeen suggested that V should be considered indepen-dently of Kt, which also is an independent predictorKey words: deuterium oxide, Fourier transform infrared analysis, totalof mortality [2]. Indeed, more recent guidelines of thebody water, fat-free mass, bioelectrical impedance analysis, total body

nitrogen. adequacy of dialysis have suggested that maintenance ofa normal extracellular fluid volume should be an addi-Received for publication September 8, 1999tional indicator of the adequacy of dialysis.and in revised form January 10, 2000

Accepted for publication February 5, 2000 In dialysis populations, V has been most commonlyestimated using 58% of total body weight (58%bwt) or 2000 by the International Society of Nephrology

408

Cooper et al: Body composition in end-stage renal failure 409

the Watson formulae. The former method does not take a volume correction factor was made derived from theamount of weight loss achieved by ultrafiltration to ob-into consideration any changes in body fat mass. The

Watson equations were derived from pooled data from tain their previously determined ideal dry weight.many trials in normal healthy volunteers using deute- Plasma water was separated out from the samples us-rium, tritium, or antipyrine dilution as their gold stan- ing an ultrafiltration technique [4] that was performeddard estimates of TBW [3]. Excluded from the analysis by centrifugation at 2110 3 g for two hours in centrifugewere any studies of patients who had an abnormal fluid tubes with a 5000 D cut-off membrane (Centrisartt I,state. Hence, the suitability of using these estimates of Sartorius AG. Gottingen, Germany). The plasma waterTBW derived from normal individuals and extrapolated obtained was then analyzed using the Fourier transformto a dialysis population has not been validated. infrared analysis (FTIR) technique, which has been pre-

The aim of this study was to compare V determined viously validated [5].by different methods in patients with renal disease. The The volume of TBW was calculated as previously de-methods studied were TBW measured by deuterium di- scribed by Blagojevic et al [5]:lution, bioelectrical impedance analysis (BIA), 58% body

TBW 5 1/NS(M 2 Lt)/[D2O]twt, and the Watson formulae.An additional aim of the study was to determine whether where M equals the ingested mass of D2O (g), [D2O]

accurate and easily attainable methods for assessing equals the serum concentration of D2O at time t, L equalsbody composition are valid in patients with renal disease. the sum of urine loss at time t, and N equals the numberAll methods of TBW assessment were each compared of samples taken after the four-hour equilibration pe-in regard to their relationship to directly measured body riod.protein by in vivo neutron capture analysis [total body The Watson formulae were used as previously de-nitrogen (TBN)] and with anthropometrically derived scribed [3]:estimates of FFM. Comparisons between BIA-derived

Males: V 5 2.447 2 (0.09516 3 age years)FFM assessments and anthropometrically derived esti-mates of FFM were also made. Prediction equations 1 (0.1074 3 ht cm) 1 (0.3362 3 wt kg)for TBN and FFM were derived using BIA-obtained

Females: V 5 22.097 1 (0.1069 3 ht cm)resistance and simple anthropometric measurements us-ing multiple regression techniques. 1 (0.2466 3 wt kg)

Body water was additionally estimated using 58% ofMETHODS each patient’s actual dry weight (58% body wt).

Bioelectrical impedance analysis was performed (N 5Fifty-four patients with renal disease were studied [35patients on CAPD, 14 patients on HD, and 5 patients 40) with the patient in the supine position using two

leads on the nondominant hand and two leads on thesubsequent to receiving a transplant (Tx)] at their “dryweight.” “Dry weight” was determined clinically by a ipsilateral foot (tetra-polar placement). The patients were

rested in the supine position for five minutes, and thetrained nephrologist (B.A.C.) in conjunction with thepatient’s usual treating physician. measurements were performed with their arms parallel

but separate to the trunk, and their legs apart far enoughThe deuterium volume of distribution was used as thegold standard for Vdurea. Each patient was given an oral so that the thighs were not touching. Three resistance

measurements were obtained using a swept frequencysolution of deuterium (D2O, isotopic purity $ 99.9%;Australian Nuclear Science and Technology Organisa- bioimpedance meter (SEAC SFB2 multifrequency bio-

impedance meter). TBW estimates were calculated fromtion, Lucas Heights, Sydney, Australia) at a dose of 1g/kg of lean body weight, estimated from anthropometric the average of these resistance measurements using the

equations described by Fredrix et al [6], Pullicino etmeasurements. Blood samples were taken when steadystate was reached after greater than four hours. Urine al [7], and Kushner and Schoeller [8]. All of the BIA

measurements in the CAPD patients were performedwas simultaneously collected to determine net loss ofheavy water. In CAPD patients, the heavy water solution with the peritoneum empty and, in the HD patients, at

least 1.5 hours postdialysis.was ingested immediately following a bag exchange withthe subsequent blood sample being taken immediately Fat-free mass was calculated using anthropometric

measurements, including body weight and skinfold thick-before the next exchange, when equilibration of plasmaand dialysate D2O concentration was demonstrated to nesses, using the method of body density assessment [9],

which was then used to calculate percentage body fatoccur (data not shown). In the CAPD patients, a volumecorrection was made depending on the size of the ex- using the equation derived by Siri [10]. FFM was calcu-

lated to equal the body weight minus the product of thechange volume of their dialysate. In HD patients, as theblood samples were collected prior to the HD session, percentage body fat and body weight. FFM estimates

Cooper et al: Body composition in end-stage renal failure410

Table 1. Patient characteristicswere also calculated from BIA-derived equations [11–13],again using the average of the three resistance measures. Characteristic Mean6SE Median Range

Total body nitrogen analysis was performed by in vivo For all patients (N 5 55)Age years 65.161.49 67.8 32.3 to 79.2neutron capture analysis, as previously described [14].Wt kg 66.462.55 63.4 41.6 to 113.7The observed TBN results were compared with that ofBMI kg/m2 24.560.82 22.2 16.0 to 39.3

the expected TBN for normal sex- and height-matched Body fat % 27.360.94 27.6 12.4 to 43.2TBN g 1534653 1537 936 to 2982volunteers and recorded as a nitrogen index (NI) [15].NI % 90.461.80 88.2 69.2 to 133.1Residual renal function (RRF) was determined in eachRRF CCr mL/min 1.360.25 0.2 0.0 to 5.6

patient calculated as the creatinine clearance (CCr) in TBW L 36.7861.485 36.09 19.83 to 62.21FFM kg 46.7761.398 46.20 31.05 to 80.11milliliters per minute from 24-hour urine collections.

For HD patients (N 5 14)Ethical approval to conduct the study was obtainedAge years 67.562.65 69.3 43.2 to 79.2

from the Royal North Shore Hospital Human Medical Wt kg 71.065.22 65.0 53.3 to 133.7BMI kg/m2 25.261.54 23.0 19.5 to 39.3Research Ethics Committee.Body fat % 25.561.85 26.1 14.4 to 35.4Student’s paired t-test was used for determining theTBN g 16756123 1576 1083 to 2982

significance between two estimates of TBW or FFM. To NI % 92.564.80 88.3 69.8 to 133.1RRF CCr mL/min 0.260.12 0.0 0.0 to 1.6assess the agreement between two estimates, limits ofTBW L 38.4163.084 36.16 22.20 to 66.21agreement were derived using the method described byFFM kg 49.9462.933 47.82 39.50 to 80.11

Bland and Altman [16]. To assess the differences in mea- For PD patients (N 5 35)Age years 66.461.63 68.2 32.3 to 76.5sures of body composition between the end-stage renalWt kg 65.263.26 63.2 41.6 to 49.9failure (ESRF) subgroups, one-way analysis of varianceBMI kg/m2 24.361.07 22.2 16.0 to 38.3

(ANOVA) was performed. To assess the relationship Body fat % 27.861.13 27.8 12.4 to 43.2TBN g 1483663 1543 936 to 2175between FFM, TBN, and body water measurements, re-NI % 88.261.80 87.1 69.2 to 118.8gression analyses were also undertaken. Statistical analy-RRF CCr mL/min 1.760.31 1.2 0.0 to 5.6

ses were done using Statviewt 5.0 (SAS Institute Inc., TBW L 36.9461.909 37.49 19.83 to 60.13FFM kg 45.9261.783 47.21 31.05 to 63.77Cary, NC, USA).

For Tx patients (N 5 5)Age years 48.764.58 47.4 38.7 to 59.5Wt kg 62.365.34 56.6 51.8 to 78.0RESULTSBMI kg/m2 24.262.37 21.3 19.3 to 31.2Body fat % 28.464.01 22.9 20.5 to 40.1The characteristics of the patients studied are pre-TBN g 14986111 1424 1233 to 1827sented in Table 1. The only significant difference be-NI % 99.965.80 95.8 90.0 to 121.5

tween the different patient groups was that the transplant RRF CCr mL/min 53.666.28 56.5 34.1 to 72.2TBW L 31.0061.590 30.40 26.64 to 36.09patients were significantly younger, and in the dialysisFFM kg 43.7961.358 45.00 40.20 to 46.72patients, those on HD had significantly less RRF thanAbbreviations are: Wt, weight; BMI, body mass index; TBN, total body nitro-that of the CAPD patients. Twenty-two of the dialysis

gen; NI, nitrogen index; RRF, residual renal function; TBW, total body water;patients were anuric, and the remaining patients had FFM, free-fat mass; HD, hemodialysis; PD, peritoneal dialysis; Tx, transplant.

minimal RRF (HD mean 5 0.2 mL/min and median 50 mL/min, CAPD mean 5 1.7 mL/min and median 51.2 mL/min). When analyzed by ANOVA, there was no

pared with the D2O dilution (mean difference, 1.792 L,significant difference in mean D2O-derived TBW (P 5P 5 0.04, CI, 0.086 to 3.498 L).0.4), percentage of body fat (P 5 0.5; Fig. 1), or body

No significant difference was found between the TBWmass index (BMI; P 5 0.9) between the patient sub-estimate by D2O compared with the BIA-derived TBWgroups HD, CAPD, or those with a functioning renalby Kushner (mean difference, 21.221 L, P 5 0.12, CI,transplant.22.754 to 0.312 L) and Fredrix (mean difference,

Total body water 21.723 L, P 5 0.06, CI, 23.544 to 0.098 L) estimates.The influence of obesity (as determined by percentageIn this population of patients with renal disease, there

body fat .30%) on estimates of TBW compared withwas a significant difference in the estimation of TBWD2O dilution was further assessed. The estimate mostby some of the methods when compared with the D2Oeffected by obesity was that using 58% body wt, whichdilution results. The Watson equation significantly un-found no significant difference in the nonobese individu-derestimated the TBW (mean difference, 21.751 L, P 5als (N 5 38, mean difference, 0.001, P 5 1.0); however,0.01, CI 23.129 to 20.373 L) compared with the D2Oin obese individuals, significant overestimation occurreddilution. The Pullicino-derived TBW also significantly(N 5 13, mean difference, 6.789, P 5 0.001).underestimated the TBW (mean difference, 24.443 L,

The TBW results are summarized by a cell plot inP , 0.0001, CI, 26.152 to 22.734 L). The 58% body wtestimate significantly overestimated TBW when com- Figure 2.

Cooper et al: Body composition in end-stage renal failure 411

Fig. 1. Differences in body composition byhemodialysis (HD), peritoneal dialysis (PD),and transplantation (Tx). (A) Total body wa-ter, P , 0.04. (B) Percent body fat, P , 0.05.

Fig. 2. Estimates of total body water (TBW)compared with gold standard of deuteriumoxide dilution (D2O). Error whiskers are 6SE.

Table 2. Total body water (TBW) estimates relative to deuterium Watson (P 5 0.5) and D2O/58% body wt (P 5 0.4)oxide (D2O) dilution

methods (Fig. 4).Bias Limit of agreement Using simple regression, there were strong correla-

Measurement liter tions between all of the estimates of TBW and FFM, asassessed by anthropometric measurements. The estimateKushner 21.221 210.809 to 8.367

Watson 21.751 211.699 to 8.197 of TBW using Kushner BIA equation explained thisPullicino 24.443 215.127 to 6.241 relationship the best (R2 5 0.92, P , 0.0001; Fig. 5).Fredrix 21.723 213.103 to 9.657

Similar findings were also seen when studying the rela-58% body weight 11.792 210.522 to 14.106tionship between TBN and the estimates of TBW withthe Kushner BIA method again explaining this relation-ship the best (R2 5 0.89, P , 0.0001; Fig. 6).

There was no correlation found between RRF in theWhen studying the agreement between the estimatesdialysis patients and TBW (D2O dilution) or with nutri-of TBW using the method described by Bland and Alt-tional state assessment (TBN).man, it can be seen the there are large limits of agreement

between all methods of body water assessment whenFat-free masscompared with the gold standard (Table 2). The method

Fat-free mass estimated by BIA using the Segal equa-that gave the narrowest limits of agreement was thetion and the Van Loan equation significantly underesti-Kushner BIA estimate of TBW. However, these limitsmated anthropometrically derived estimates of FFMwere wide and clinically significant (210.809 to 8.367 L);(mean difference, 21.252 kg, P 5 0.01, CI, 22.203 tothat is, they underestimate TBW by as much as 10.8 L

or overestimate it by as much as 8.4 l (Fig. 3). There was 20.298 kg; mean difference, 24.496 kg, P , 0.0001,CI, 25.544 to 23.448 kg, respectively). No significantalso no significant difference between the three sub-

groups when comparing the difference between the D2O/ difference was found between the anthropometric FFM

Cooper et al: Body composition in end-stage renal failure412

Fig. 3. D2O versus Kushner bioelectrical impedance analysis (BIA) (A) shown as a Bland Altman plot (B).

Fig. 4. Differences in estimates of body waterby hemodialysis (HD), peritoneal dialysis(PD), and transplantation (Tx). (A) Deute-rium oxide dilution (D2O)/Watson difference,P , 0.05. (B) D2O/58% body weight differ-ence, P , 0.04.

Fig. 6. Comparison of Kushner-derived total body water (TBW)Fig. 5. Comparison of Kushner-derived measure of total body waterand directly measured total body nitrogen (TBN). Y 5 279.435 1(TBW) and fat-free mass (FFM). Y 5 2.848 1 1.265 · X; R2 5 0.92.46.721 · X; R2 5 0.89.

Cooper et al: Body composition in end-stage renal failure 413

Fig. 7. Comparison of estimates of FFM. Er-ror whiskers are 6 SE.

Table 3. Fat-free mass (FFM) estimates relative to anthropometryand the Lukaski BIA method (mean difference, 20.723kg, P 5 0.25, CI, 21.979 to 0.533 kg; Fig. 7). Bias Limit of agreement

When studying the reliability of estimations of FFM Measurement kgusing the Bland and Altman method, it can be seen that

Segal 21.252 27.215 to 4.713the limits of agreement between all methods of FFM Van Loan 24.496 211.048 to 2.056

Lukaski 20.723 28.579 to 7.133assessment are still large when compared with the an-thropometric FFM assessment (Table 3). The methodthat gave the narrowest limits of agreement was the SegalBIA estimate of FFM. However, these were wide and

there was a significant overestimation of body water.clinically significant (27.215 to 4.713 kg; Fig. 8).Even though the mean estimate of TBW as assessed byMultiple regression techniques were used to assess theFredrix and Kushner using BIA did not significantly dif-ability to predict anthropometric FFM and TBN mea-fer from the mean D2O value, the reliability of the resultssurements in these patients by using the BIA-derivedvaried enormously, as shown by the subsequent Blandresistance and other parameters such as height (cm),and Altman analyses (Table 2). It is clear that with allweight (kg), and age (years).estimates used there are large limits of agreement whenHeight squared, weight, and the reciprocal of resis-compared with the gold standard TBW estimate. Thesetance were able to significantly predict FFM (R2 5 0.95)results indicate that there is significant error in the rou-in all patient groups. From this, the following equationtine estimation of CAPD dialysis adequacy because ofcould be derived:the use of these inaccurate methods of TBW assessment.

FFM 5 221.768 1 0.001 3 ht2This is likely to result in clinically significant errors inKt/V assessment and dialysis prescription in CAPD pa-1 6630.669 3 1⁄R 1 0.312 3 wttients.

Similar significant predictions (R2 5 0.91) in TBN could The variation in body composition (that is, percentagealso be obtained using the previously mentioned parame- of body fat), independent of FFM, is likely to accountters but also including the patients age, represented in for the bias in the TBW estimates as well as the largethe following equation: limits of agreement found by some of the methods stud-

ied. Variations in body fat appear to stretch the limits ofTBN 5 2668.324 2 3.963 3 age 1 10.133 3 wtthe previously defined equations. Thus, in obese patients

1 0.045 3 ht2 1 32141.457 3 1⁄R (body fat .30%), indirect methods of assessment ofbody water, especially 58% body wt, should be appliedwith caution. Significant gains in body weight caused by

DISCUSSION an increase in body fat, which is relatively anhydrous,will result in a higher estimate of body water if the 58%It is clear from these results that there is a significant

variation in the TBW estimation in renal patients, de- body wt assumption is used. This obviously results inerrors of TBW estimation and was confirmed by thispending on which method of calculation is applied. The

use of the Watson equations and one of the BIA-derived study. Errors in estimations of body water using theWatson formulae are predictable, as the regression equa-estimates (Pullicino) significantly underestimated TBW.

Whereas using the 58% body wt assumption of TBW, tion derivations obtained in the original article [3] for

Cooper et al: Body composition in end-stage renal failure414

Fig. 8. Anthropometric cFFm versus Segal-derived FFM (A) and shown as a Bland Altman plot (B).

both males and females did not predict all variations on apparent edema at the time of study. Intraperitonealfluid does not appear to influence the estimates in bodyTBW (males R2 5 0.70 and females R2 5 0.74).

The reliability of BIA to determine body water repre- water composition, as measured by BIA [23, 24]. How-ever, this potential confounder was avoided in the pres-sents a considerable advance in its assessment, as the

electrical current is conducted through the fluid compart- ent study by uniform assessment at the time of an emptyperitoneum.ments and is therefore less affected by the presence of

body fat. The utility of many equations previously de- Wide limits of agreement in body water estimationhave also been found when comparing other dilutionrived from BIA to predict body water is limited in pa-

tients with renal disease. This is because the body compo- techniques as reported in an article by de Fijter et al,who studied antipyrine distribution volume (ADV) insition of patients studied to derive these equations differs

significantly from patients with renal disease. The Fred- CAPD patients [25]. They found that anthropometricand BIA estimates (RJL) of TBW compared with ADVrix equation is derived from 33 patients with cancer [6];

the Kushner equation used 40 patients, some of whom had wide limits of agreement (25.9 to 11.6 and 211.4to 8.6, respectively).had diabetes and significant obesity [8], and Pullicino et

al studied BIA in normal subjects [7]. As the majority Studies have previously been performed in assessingbody water in patients with renal disease. Wong et alof patients in the present study were on CAPD and likely

to have increased fat mass at the expense of FFM, the studied 20 CAPD patients and also found that TBW wassignificantly underestimated by the Watson equation andbetter correlation with the Kushner equation is not unex-

pected. Similarly, the BIA-derived estimations of FFM overestimated by 58% body wt, and that error wasgreater in obese patients [26]. Even when the best esti-were also obtained from nonrenal populations.

One reason for the lack of correlation of BIA esti- mates of body water (BIA) were compared with D2O,there were large limits of agreement with this techniquemates of body water in normal compared with renal

populations is the potential alteration in the ratio of (213.75 to 10.32 L), which were narrower than that ob-tained by the Watson equation or 58% body wt. Theyintracellular to extracellular water [17]. This is well

known to occur and results in an overestimation of ultra- suggested that if clinical measures of V require a mea-surement precision of ,5%, then BIA does not providefiltrative losses in HD patients if measurements are taken

immediately postdialysis [18–21]. Additionally, Foster results close enough to those of D2O. Unlike their study,we also investigated other markers of body composition,and Lukaski showed that the largest contributors to

whole body resistance are the forearms (28%) and the including TBN, and were able to further define the asso-ciation of BIA with body composition measures, as seenlegs (33%) compared with the trunk (9%) [22]. Hence,

alterations in body water associated with dependent by our derived prediction equations for TBN and FFM.Woodrow et al also compared different methods ofedema may result in significant body resistance changes.

In the present study, this was unlikely to account for TBW assessment using both patients with chronic renalfailure and controls [27]. Even with the Kushner BIAthe measured differences, as no patients had clinically

Cooper et al: Body composition in end-stage renal failure 415

equation, which in our study also showed the best corre- has already been shown in the normal population. Theequations derived for the prediction of FFM and TBNlation with body water, wide limits of agreement were

uniformly present. In the controls, there was a bias of in the present study, which are specific for patients withrenal disease, are likely to improve its applicability and3 L with limits of agreement of 21.9 to 8.0 L, and in the

chronic renal failure patients (HD, PD, and undialyzed reliability in clinical practice. In patients in whom theapplication of surrogate markers of body water and bodychronic renal failure patients), there was a bias of 2.6 L

and limits of agreement of 23.4 to 8.6 L. composition are likely to be unreliable, more specificdirect measures of TBW using deuterium dilution andIn a second article by Woodrow et al, TBW measure-

ments were reported (W, 58% body wt, BIA and D2O) TBN using in vivo neutron capture analysis should beundertaken.in a subgroup of 20 PD patients drawn from their initial

article using the same control group [28]. This subanal-ysis showed that in the PD group, the three estimates ACKNOWLEDGMENTS(W, 58% body wt, and BIA) all significantly overesti- The authors acknowledge the support of the Baxter Extramural

grant program, the National Health and Medical Research Council ofmated TBW when compared with the D2O methodAustralia, and the Royal North Shore Hospital. Preliminary results of(mean difference, 1.7, P , 0.05; mean difference, 5.6,the body water assessment were presented as a poster at the American

P , 0.0001; and mean difference, 3.2, P , 0.0001, respec- Society of Nephrology Annual Scientific Meeting, Philadelphia, PA,USA, October 23–28, 1998 (B.A. Cooper, C.A. Pollock, L.S. Ibels, M.tively). In the control group, all methods significantlyWarnant, F-Y. Zhu, A. Aslani, B.J. Allen: Kt/V: As good as it gets?overestimated TBW except the Watson formulae, whichOr is it? Department of Renal Medicine, Royal North Shore Hospital,

was not significantly different. Their results were consis- NSW, Australia).tent with ours in that in the PD group, the most accurate

Reprint requests to Dr. Carol Pollock, Department of Medicine,of the three methods was again BIA.Wallace Freeborn Building, Royal North Shore Hospital, St. Leonards,

In an additional small study in 10 CAPD patients by NSW 2065, Australia.E-mail: carpol@medicine.usyd.edu.auArkouche et al, TBW was measured using a double-

labeled water technique and analyzed using isotopicmass spectrometry of the patients’ saliva samples [29]. REFERENCESA comparison was also made with that of FFM derived 1. Owen WF Jr, Chertow GM, Lazarus JM, Lowrie EG: Dose of

hemodialysis and survival: Differences by race and sex. JAMAby skinfold thickness estimation and BIA of FFM and280:1764–1768, 1998TBW. Although their methods of analysis of TBW dif-

2. Lowrie EG, Zhu X, Lew NL: Primary associates of mortalityfered from those in the present study, their results were among dialysis patients: Trends and reassessment of Kt/V and urea

reduction ratio as outcome-based measures of dialysis dose. Amagain consistent with our data.J Kidney Dis 32(Suppl):S16–S31, 1998Recently, a large study of peritoneal dialysis patients

3. Watson PE, Watson ID, Batt RD: Total body water volumeswas reported by Johansson, who found that the Watson for adult males and females estimated from simple anthropometricequation again significantly underestimated TBW com- measurements. Am J Clin Nutr 33:27–39, 1980

4. Aslani A: The prediction of body composition in assessing dosepared with that obtained by a tritiated water-dilutioneffects in cancer chemotherapy. Masters of Science in Medicine,technique (abstract; Johansson, Perit Dial Int 19(Suppl); Sydney, University of Sydney, 1997

S19, 1999). To our knowledge, no other methods of TBW 5. Blagojevic N, Allen BJ, Gaskin KJ, Baur LA: Determinationof total body water by Fourier transform infrared analysis. Aus-estimation have been reported thus far.tralas Phys Eng Sci Med 13:110–116, 1990Few studies have correlated anthropometrically de- 6. Fredrix EW, Saris WH, Soeters PB, Wouters EF, Kester AD,

rived FFM with BIA and validated the techniques with von Meyenfeldt MF, Westerterp KR: Estimation of body compo-sition by bioelectrical impedance in cancer patients. Eur J ClinTBN. Rayner et al measured TBN at one site and, extrap-Nutr 44:749–752, 1990olating it to estimate whole body nitrogen, found strong 7. Pullicino E, Coward WA, Stubbs RJ, Elia M: Bedside and field

correlations with anthropometrically derived FFM [30]. methods for assessing body composition: Comparison with thedeuterium dilution technique. Eur J Clin Nutr 44:753–762, 1990This relationship was greater in controls than in patients

8. Kushner RF, Schoeller DA: Estimation of total body water bywith renal failure (r 5 0.951 and 0.876, respectively). bioelectrical impedance analysis. Am J Clin Nutr 44:417–424, 1986However, no study has compared BIA and TBN, as was 9. Durnin JV, Womersley J: Body fat assessed from total body

density and its estimation from skinfold thickness: Measurementsdone in our present study. The predictive equation thaton 481 men and women aged from 16 to 72 years. Br J Nutrhas been derived appears as though it will provide a32:77–97, 1974

reliable estimation of TBN and hence nutritional status. 10. Siri WE: The gross composition of the body. Adv Biol Med Phys4:239–280, 1956As a result of the more recent recommendations that

11. Lukaski HC, Bolonchuk WW, Hall CB, Siders WA: Validationadequate fluid removal, in addition to dialysis adequacyof tetrapolar bioelectrical impedance method to assess human body

and nutritional maintenance, should be specific goals in composition. J Appl Physiol 60:1327–1332, 198612. Segal KR, Van Loan M, Fitzgerald PI, Hodgdon JA, Vanthe treatment of end-stage renal failure, accurate and

Itallie TB: Lean body mass estimation by bioelectrical impedanceeasily reproducible methods of TBW estimation areanalysis: A four-site cross-validation study. Am J Clin Nutr 47:7–14,

needed. BIA appears to be a very useful tool in assessing 198813. Van Loan M, Mayclin P: Bioelectrical impedance analysis: Is itboth TBW and body composition in renal patients, as

Cooper et al: Body composition in end-stage renal failure416

a reliable estimator of lean body mass and total body water. Hum 22. Foster KR, Lukaski HC: Whole-body impedance: What does itmeasure? Am J Clin Nutr 64:S388–S396, 1996Biol 59:299–309, 1987

23. Schmidt R, Dumler F, Cruz C, Lubkowski T, Kilates C: Im-14. Allen BJ, Blagojevic N, Delaney I, Pollock CA, Ibels LS,proved nutritional follow-up of peritoneal dialysis patients withAllman MA, Tiller DJ, Gaskin KJ, Baur LA, Waters DL,bioelectrical impedance. Adv Perit Dial 8:157–159, 1992Cowell C, Ambler G, Quigley C, Fletcher JP: The role of body

24. Rallison LR, Kushner RF, Penn D, Schoeller DA: Errors inprotein studies in clinical trials. Basic Life Sci 55:155–169, 1990estimating peritoneal fluid by bioelectrical impedance analysis and15. Harrison JE, McNeill KG, Strauss AL: A nitrogen index—totaltotal body electrical conductivity. J Am Coll Nutr 12:66–72, 1993body protein normalized for body size—for diagnosis of protein 25. de Fijter WM, de Fijter CW, Oe PL, ter Wee PM, Donker

status in health and disease. Nutr Res 4:209–224, 1984 AJ: Assessment of total body water and lean body mass from16. Bland JM, Altman DG: Statistical methods for assessing agree- anthropometry, Watson formula, creatinine kinetics, and body

ment between two methods of clinical measurement. Lancet 1:307– electrical impedance compared with antipyrine kinetics in perito-310, 1986 neal dialysis patients. Nephrol Dial Transplant 12:151–156, 1997

17. Deurenberg P, van der Kooy K, Leenen R, Schouten FJ: Body 26. Wong KC, Xiong DW, Kerr PG, Borovnicar DJ, Stroud DB,impedance is largely dependent on the intra- and extra-cellular Atkins RC, Strauss BJ: Kt/V in CAPD by different estimationswater distribution. Eur J Clin Nutr 43:845–853, 1989 of V. Kidney Int 48:563–569, 1995

27. Woodrow G, Oldroyd B, Turney JH, Davies PS, Day JM, Smith18. Chertow GM, Lowrie EG, Wilmore DW, Gonzalez J, Lew NL,MA: Measurement of total body water by bioelectrical impedanceLing J, Leboff MS, Gottlieb MN, Huang W, Zebrowski B: Nutri-in chronic renal failure. Eur J Clin Nutr 50:676–681, 1996tional assessment with bioelectrical impedance analysis in mainte-

28. Woodrow G, Oldroyd B, Turney JH, Davies PS, Day JM, Smithnance hemodialysis patients. J Am Soc Nephrol 6:75–81, 1995MA: Measurement of total body water and urea kinetic modelling19. de Lorenzo A, Barra PF, Sasso GF, Battistini NC, Deurenbergin peritoneal dialysis. Clin Nephrol 47:52–57, 1997P: Body impedance measurements during dialysis. Eur J Clin Nutr

29. Arkouche W, Fouque D, Pachiaudi C, Normand S, Laville M,45:321–325, 1991 Delawari E, Riou JP, Traeger J, La Ville M: Total body water20. de Lorenzo A, Deurenberg P, Andreoli A, Sasso GF, Palestini and body composition in chronic peritoneal dialysis patients. J Am

M, Docimo R: Multifrequency impedance in the assessment of Soc Nephrol 8:1906–1914, 1997body water losses during dialysis. Ren Physiol Biochem 17:326–332, 30. Rayner HC, Stroud DB, Salamon KM, Strauss BJ, Thomson1994 NM, Atkins RC, Wahlqvist ML: Anthropometry underestimates

21. Nyboer J, Sedensky JA: Bioelectrical impedance during renal body protein depletion in haemodialysis patients. Nephron 59:33–40,1991.dialysis. Proc Clin Dial Transpl Forum 4:214–219, 1974