DIALYSIS OUTCOMES Quality Initiative

Bipedal Bioelectrical Impedance Analysis Reproducibly Estimates Total Body Water in Hemodialysis Patients Robert F. Kushner, MD, and David M. Roxe, MD Formal kinetic modeling for hemodialysis patients requires accurate determination of in vivo dialyzer clearance, careful measurement of the actual dialysis time, and calculation of the patient s total body water (V) or urea space. Bipedal (BP) bioelectrical impedance analysis (BIA) is a simple method for the determination of V at the same time the patient is weighed. Because of better correlation with the Watson formula volume and a smaller coefficient of variation than urea kinetic volumes, BP-BIA estimates of V may be preferred in dialysis units that meet guidelines by following formal kinetic modeling data or use V estimates in calculating treatment time. 2002 by the National Kidney Foundation, Inc. INDEX WORDS: Kinetic modeling; total body water (V); urea space; bioelectrical impedance; hemodialysis (HD). DIALYSIS OUTCOMES Quality Initiative (DOQI) guidelines specify that The delivered dose of hemodialysis in adult and pediatric patients should be measured using formal kinetic modeling, employing the single-pool, variable volume model. 1 Many dialysis units substitute use of the urea reduction ratio as a surrogate for formal modeling because of difficulties associated with the procedure. Difficulties include knowing actual dialyzer clearance, accurately measuring the true dialysis time, and quantifying the urea distribution space or total body water (V). Of these, perhaps the most difficult to define for a given patient is the volume of V. The purpose of this study is to investigate the utility and practicality of bipedal (BP) bioelectrical impedance analysis (BIA) to estimate the volume of V (BP-BIA-V) in dialysis patients and compare that value with volumes estimated from a standard urea kinetic (UK) technique (UK-V) and the Watson formula (W-V). METHODS Subjects Fourteen adult patients (7 men, 7 women) with end-stage renal disease undergoing chronic outpatient hemodialysis From the Northwestern University Medical School, Chicago, IL. Received January 9, 2001; accepted in revised form July 27, 2001. Supported in part by a grant from Tanita Corporation of America, Arlington Heights, IL; and Mr. and Mrs. Everett P. Weaver (D.M.R.). Address reprint requests to Robert F. Kushner, MD, Wellness Institute, Northwestern Memorial Hospital, 150 East Huron, Suite 1100, Chicago, IL 60611. E-mail: rkushner@nmh.org 2002 by the National Kidney Foundation, Inc. 0272-6386/02/3901-0020$35.00/0 doi:10.1053/ajkd.2002.29907 therapy were recruited for the study. All patients were medically stable and dialyzed on a thrice-weekly schedule at Northwestern Memorial Hospital (Chicago, IL). None of those studied had a pacemaker or implanted defibrillator. Measurements (described next) were performed during four consecutive hemodialysis sessions. Eighteen healthy adult subjects (8 men, 10 women) served as controls. The study was approved by the Northwestern University Institutional Review Board, and all participants signed an informed consent form. Protocol BP-BIA. Immediately before hemodialysis, patients stood barefoot on the BP-BIA device (TBF-105; Tanita Corp of America, Arlington Heights, IL). In this system, a leg-to-leg conduction pathway is used for impedance measurements. Two subdivided steel foot-pad electrodes are mounted on a platform scale. Each foot is divided in half so that the anterior and posterior portions form two separate electrodes. Current is applied through the anterior portion of the footpad electrode, and the voltage decrease across the posterior (heel) electrode is then measured. Impedance of the lower extremities and body weight is measured simultaneously while the subject stands on the scale. Measurement of impedance uses a standard 50 khz to 0.8 ma sine wave constant current. 2 Measured impedance (ohms) and estimated V using the manufacturer s predictive equations were recorded. These equations incorporate variable impedance, height, weight, age, and sex for the prediction of V. Height was measured by a wall-mounted statiometer to the nearest 0.1 cm. Watson equations. Anthropometrically derived sexspecific urea distribution volumes were calculated using the Watson equations, 3 as cited in the DOQI guidelines. Equations were generated from 458 adult men and 265 adult women obtained from dilution studies. Subjects were selected representing a wide range of height, weight, percentage of body fat, and total body water (TBW). Most individuals were healthy, and none had clinical evidence of edema or conditions that would affect normal water metabolism. These equations are commonly used in clinical practice for pharmacokinetic estimates of V (W-V). SDs and total variances (r 2 ) for equations for men and women are 3.76 l and 70.4% and 3.60 l and 73.6%, respectively. W-V was calculated before each dialysis session as follows: 154 American Journal of Kidney Diseases, Vol 39, No 1 (January), 2002: pp 154-158

BIPEDAL BIOELECTRICAL IMPEDANCE AND DIALYSIS 155 (Men) V 2.447 0.09516 (age) 0.1074 (height) 0.3362 (weight) (Women) V 2.097 0.1069 (height) 0.2466 (weight) where V is in liters, age is in years, height is in centimeters, and weight is in kilograms. UK modeling. Formal UK modeling requires accurate measures of predialysis and postdialysis blood urea nitrogen levels for the first dialysis treatment of the week, predialysis blood urea nitrogen value for the second dialysis session of the week; predialysis and postdialysis weights, actual hemodialysis treatment time, and effective dialyzer clearance. These values are entered into a computer software program that computes V by a mathematical iteration of two formulas. UK modeling is calculated monthly in the Northwestern Memorial Hospital dialysis unit. Statistical Analysis Between-day coefficients of variation (CVs) for BP- BIA-V and body weight were calculated as SD/mean 100%. Regression analysis and paired t-test were performed to examine the correlation and statistical significance of the mean difference observed in BP-BIA-V with W-V and UK-V. To estimate the suitability of replacing one method (BP-BIA-V) with another (W-V), limits of agreement were determined using the method of Bland and Altman. 4 RESULTS Physical characteristics of study patients and controls are listed in Table 1. Hemodialysis patients were generally older than controls. Mean CV of predialysis body weights over the four sessions was 0.83% 0.42%, with individual variation ranging from 0.36% to 1.6%, suggesting a stable hemodialysis population. This was not significantly different from the CV of 0.50% 0.28% for repeated measures of W-V (individual variation range, 0.17% to 1.08%) and 2.1% 1.28% for repeated measures of BP- BIA-V (individual variation range, 0.84% to 4.45%). In comparison, the CV for UK-V values for the same patients over the preceding 8 months was 13.9% (range, 5.9% to 31.6%). The correlation (r) between BP-BIA-V and W-V was 0.97 for control subjects and 0.93 for hemodialysis patients (Fig 1). Conversely, the correlation between BP-BIA-V and UK-V was poorer (r 0.58) for hemodialysis patients (data not shown). The closeness of agreement between BP-BIA-V and W-V is shown in Bland-Altman plots (Fig 2). For control subjects, the two methods were not significantly different, with BP- BIA-V predicting a mean overestimation (bias) of 1.2 2.3 L compared with W-V. Conversely, BP-BIA consistently overestimated V compared with W-V among hemodialysis patients. The mean bias overestimation was 5.1 3.1 L for women and 5.9 2.2 L for men. DISCUSSION Kinetic modeling is performed to obtain a number that is believed to indicate the dose of dialysis a patient receives. The number is derived from three components: dialyzer clearance, treatment duration, and patient s V. The more precisely we measure clearance, time, and V, the more dependable the number we obtain to indicate the dose of dialysis. Multiple factors have been shown to affect the calculation of UK-V; a single patient may have values that vary from 44% to 123% of body weight in a 7-month period. 5 Therefore, it would be useful if we could predict volume with more certainty, especially if V is being used to help calculate dialysis time. BIA is a potentially useful tool for the measurement of V in the hemodialysis population. BIA has traditionally been performed using the distal tetrapolar electrode placement technique, which requires placement of four gel-impregnated surface electrodes on the foot and ankle and hand and wrist of the right side of the body. Using either single-frequency or multifrequency BIA instruments, V is estimated from the measured impedance along with several anthropometric Table 1. Physical Characteristics of Patients and Controls Patient Males Patient Females Control Males Control Females Age (y) 55 (28-77) 59 (47-73) 42 (32-58) 45 (26-60) Height (cm) 176.2 (165-186) 153.9 (148-159) 176.3 (170-186) 159.3 (146-170) Weight (kg) 76.4 (65.2-88.9) 63.1 (43.2-74.0) 90.9 (62.1-137.5) 65.6 (53.7-90.3) Weight gain (kg) 3.3 (1.3-4.5) 3.0 (1.2-4.2) NOTE. Values are the mean with the range in parentheses.

156 KUSHNER AND ROXE Fig 1. Correlation between prediction of W-V and BP-BIA-V for (A) control subjects (r 0.97; P < 0.001) and (B) hemodialysis patients (r 0.93; P 0.02). ( ), Men; (}), women. variables, including height, weight, and sex. Estimation of V and body fat by BIA has been validated in multiple patient populations. 6 However, practical application of the tetrapolar technique poses several problems for the hemodialysis population. These include positional changes in impedance with the change from standing to lying, presence of a fistula or shunt in the arm, and crowded environment of the dialysis unit. 7 BP-BIA offers a new application of BIA technology that may be better suited to the hemodialysis patient population. Impedance and body weight are measured simultaneously, with little need for additional nursing time, training, or expertise. It has the special virtue that a printout of BP-BIA-V is generated with every dialysis session; therefore, 12 values are generated in a typical month. One can average these results in an approximate time-averaged value for V; thus, any unusual circumstance at any given dialysis session will not unduly affect calculations. For estimation of V, BP-BIA avoids consideration of when and how samples for urea reduction ratio are drawn. Another potential advantage of BP- BIA is that it can be performed at home with simultaneous measurement of body weight and body water. Nunez et al 2 have shown that the leg-to-leg system has overall performance char-

BIPEDAL BIOELECTRICAL IMPEDANCE AND DIALYSIS 157 Fig 2. Graphical plot of the Bland-Altman representation of the difference between the W-V and BP-BIA-V methods for (A) control subjects and (B) hemodialysis patients. ( ), Men; (}), women. For all controls, mean difference was 1.25 2.3 L; for male patients, 5.9 2.2 L; and for female patients, 5.1 3.1 L. acteristics for impedance measurement similar to conventional arm-to-leg gel electrode BIA. Although BP-BIA-V showed a high correlation with W-V in hemodialysis patients using standard regression analysis (r 0.93), we observed a tendency for BP-BIA to overestimate V compared with W-V in both men and women (Fig 2). This bias may result from bioelectrical changes that occur in renal failure, including alterations in impedance and specific resistance of body fluids and tissues that are dependent on electrolyte concentrations. 7,8 The other likely source of error is the gravity dependence of intradialytic fluid in the lower extremities. The leg-to-leg conduction pathway of BP-BIA thus would overestimate V. The consequence of this error would be to provide more dialysis than might be necessary to achieve a target value for Kt/V. Such increased dialysis may affect postdialysis potassium and other solute values. Because it is not clear whether adequate dialysis is optimal dialysis, this overestimation does not appear to be a practical impediment to the use of BP-BIA-V to estimate V. Although the predictive equation overestimated TBW in hemodialysis patients, there was remarkable intraindi-

158 KUSHNER AND ROXE vidual consistency in measurements among dialysis sessions. This reproducibility is an important factor for the dialysis population. In our study, we did not use gold standard methods to measure V, such as deuterium or oxygen. 18 Rather, we compared the level of agreement between BP-BIA-V and W-V. This formula was chosen as a valid comparison because it often is used to predict V in dialysis units and has been shown to provide a reliable estimation of Kt/V compared with the oxygen 18 reference method. 9 The mean CV for predicting W-V (0.50% 0.28%) was very similar to the CV in body weights between dialysis sessions (0.83% 0.42%). This is not surprising because W-V is directly dependent on body weight as the primary predictive variable. If we are to achieve more accurate kinetic modeling, improvement in the methods used to obtain values for clearance, time, or volume is needed. Our data suggest that BP-BIA reproducibly estimates V with a lower CV than UK and avoids the technical problems, costs, and blood draws associated with that method. These advantages facilitate calculations of Kt/V and dialysis time; thus, BP-BIA becomes a preferable method over UK for estimation of V in hemodialysis patients. Nonetheless, we cannot conclude whether BP-BIA-V is superior to W-V because we did not use a gold standard technique and consequently cannot recommend its selected use at this time. Additional studies are needed to explore the causes for overestimation of V compared with W-V. BP-BIA should not be used in patients with pacemakers, defibrillators, or implanted internal or electronic devices until safety concerns are resolved. REFERENCES 1. National Kidney Foundation: DOQI Clinical Practice Guidelines for Hemodialysis Adequacy. New York, NY, National Kidney Foundation, 1997, p 25 2. Nunez C, Gallegher D, Visser M, Pi-Sunyer FX, Wang Z, Heymfield SB: Bioimpedance analysis: Evaluation of leg-to-leg system based on pressure contact foot-pad electrodes. Med Sci Sports Exerc 29:524-531, 1997 3. Watson PE, Wastson ID, Batt RD: Total body water volumes for adult males and females from simple anthropometric measurements. Am J Clin Nutr 33:27-39, 1980 4. Bland JM, Altman DG: Statistical method for assessing agreement between two methods of clinical measurement. Lancet 1:307-310, 1986 5. Depner TA: Assessing adequacy of hemodialysis: Urea modeling. Kidney Int 45:1522-1535, 1994 6. Yanorski SZ, Hubbard VS, Heymsfield SB, Lukaski HC (eds): Bioelctrical Impedance Analysis. Proceedings of a National Institutes of Health technology assessment conference. December 12-14, 1994. Am J Clin Nutr 64:S387-S532, 1996 (suppl) 7. Kushner RF, de Vries PMJM, Gudivaka R: Use of bioelectrical impedance analysis measurements in the clinical management of patients undergoing dialysis. Am J Clin Nutr 64:S503-S509, 1996 (suppl) 8. Jaeger JQ, Mehta R: Assessment of dry weight in hemodialysis: An overview. J Am Soc Nephrol 10:392-403, 1999 9. Arkouche W, Fouque D, Pachiaudi C, Normand S, Laville M, Delawari E, Riou JP, Traeger J, Laville M: Total body water and body composition in chronic peritoneal dialysis patients. J Am Soc Nephrol 8:1906-1914, 1997