Identification of operational clues to dry weight prescription in hemodialysis using bioimpedance vector analysis

Abstract

Identification of operational clues to dry weight prescription in hemodialysis using bioimpedance vector analysis. In patients undergoing hemodialysis (HD) cyclic body fluid changes are estimated by body weight variations, which may be misleading. Conventional bioelectrical impedance analysis (BIA) produces biased estimates of fluids in HD due to the assumption of constant tissue hydration. We used an assumption-free assessment of hydration based on direct measurements of the impedance vector. The impedance vector (standard BIA at 50 kHz frequency) was measured in 1367 HD patients, ages 16 to 89 years with BMI 17 to 31 kg/m2, 1116 asymptomatic (680 M and 436 F), and 251 with recurrent HD hypotension (118 M and 133 F) before and after two HD sessions (thrice weekly bicarbonate dialysis, 210 to 240 min) removing 2.7 kg fluid. The vector distribution of HD patients was compared to 726 healthy subjects with the same age and BMI range. Individual vector measurements (resistance and reactance components) were plotted on the gender specific 50th, 75th and 95th percentiles of the vector distribution in the healthy population (reference tolerance ellipses) as a resistance-reactance graph (RXc graph). The wet-dry weight cycling of HD patients was represented on the resistance-reactance plane with a definite, cyclical, backward-forward displacement of the impedance vector. The vectors of patients with HD hypotension were less steep and more often shifted to the right, out of the reference 75% tolerance ellipse, than asymptomatic patients. A wet-dry weight prescription, based on BIA indications, would bring the vectors of patients back into the 75% reference ellipse, where tissue electrical conductivity is restored. Whether HD patients with vector cycling within the normal third quartile ellipse have better outcome awaits confirmation by longitudinal evaluation. Identification of operational clues to dry weight prescription in hemodialysis using bioimpedance vector analysis. In patients undergoing hemodialysis (HD) cyclic body fluid changes are estimated by body weight variations, which may be misleading. Conventional bioelectrical impedance analysis (BIA) produces biased estimates of fluids in HD due to the assumption of constant tissue hydration. We used an assumption-free assessment of hydration based on direct measurements of the impedance vector. The impedance vector (standard BIA at 50 kHz frequency) was measured in 1367 HD patients, ages 16 to 89 years with BMI 17 to 31 kg/m2, 1116 asymptomatic (680 M and 436 F), and 251 with recurrent HD hypotension (118 M and 133 F) before and after two HD sessions (thrice weekly bicarbonate dialysis, 210 to 240 min) removing 2.7 kg fluid. The vector distribution of HD patients was compared to 726 healthy subjects with the same age and BMI range. Individual vector measurements (resistance and reactance components) were plotted on the gender specific 50th, 75th and 95th percentiles of the vector distribution in the healthy population (reference tolerance ellipses) as a resistance-reactance graph (RXc graph). The wet-dry weight cycling of HD patients was represented on the resistance-reactance plane with a definite, cyclical, backward-forward displacement of the impedance vector. The vectors of patients with HD hypotension were less steep and more often shifted to the right, out of the reference 75% tolerance ellipse, than asymptomatic patients. A wet-dry weight prescription, based on BIA indications, would bring the vectors of patients back into the 75% reference ellipse, where tissue electrical conductivity is restored. Whether HD patients with vector cycling within the normal third quartile ellipse have better outcome awaits confirmation by longitudinal evaluation. bioelectrical impedance analysis dual energy X-ray absorptiometry height hemoglobin hemodialysis resistance total body water reactance In uremic patients undergoing hemodialysis (HD), estimating body fluid changes by body weight variations often results in bringing the patient to a point of reduced cardiovascular tolerance to fluid removal. All body composition methods quantify unknown components using three measurable quantities (that is, properties, components, and both) and two mathematical functions. Methods using bioelectrical impedance analysis (BIA) are property-based methods1.Heymsfield S.B. Wang Z.M. Visser M. Gallagher D. Pierson Jr., R.N. Techniques used in the measurement of body composition: An overview with emphasis on bioelectrical impedance analysis.Am J Clin Nutr. 1996; 64: 478S-484SPubMed Google Scholar. Impedance is a measurable property of electrical ionic conduction of soft tissues, as fat and bone are poor conductors2.Foster K.F. Lukaski H.C. Whole-body impedance – What does it measure?.Am J Clin Nutr. 1996; 64Suppl: 388S-396SGoogle Scholar, 3.Kushner R.F. Bioelectrical impedance analysis: A review of principles and applications.J Am Coll Nutr. 1992; 11: 199-209PubMed Google Scholar, 4.Barber D.C. Electrical impedance tomography.The Biomedical Engineering Handbook. edited by BRONZINO JD. CRC Press, Boca Raton1995: 1151-1164Google Scholar. Whole-body impedance, a complex number represented in the real-imaginary plane by the Z vector2.Foster K.F. Lukaski H.C. Whole-body impedance – What does it measure?.Am J Clin Nutr. 1996; 64Suppl: 388S-396SGoogle Scholar, 3.Kushner R.F. Bioelectrical impedance analysis: A review of principles and applications.J Am Coll Nutr. 1992; 11: 199-209PubMed Google Scholar, 4.Barber D.C. Electrical impedance tomography.The Biomedical Engineering Handbook. edited by BRONZINO JD. CRC Press, Boca Raton1995: 1151-1164Google Scholar, is a combination of resistance (R) (that is, the opposition to flow of an alternating current through intra- and extracellular ionic solutions, representing the real part of Z) and reactance (Xc) (the capacitance produced by tissue interfaces and cell membranes, representing the imaginary part of Z) across tissues. The arc tangent of Xc/R is called the phase angle (that is, the phase difference between voltage and current, determined by the reactive component of R). For a constant signal frequency, the electrical impedance of a conductor (Ohm) is proportional to the specific impedivity (Ohm × m) multiplied by the length (m) and divided by the cross sectional area (m2) of the conductor. In simple biological conductors without cells (such as saline, urine, and ultrafiltrate) no Xc value can be measured, and therefore specific impedivity becomes specific resistivity (the real part of impedivity)4.Barber D.C. Electrical impedance tomography.The Biomedical Engineering Handbook. edited by BRONZINO JD. CRC Press, Boca Raton1995: 1151-1164Google Scholar. In BIA studies, the height (H) is used as a measure of the human conductor length2.Foster K.F. Lukaski H.C. Whole-body impedance – What does it measure?.Am J Clin Nutr. 1996; 64Suppl: 388S-396SGoogle Scholar,3.Kushner R.F. Bioelectrical impedance analysis: A review of principles and applications.J Am Coll Nutr. 1992; 11: 199-209PubMed Google Scholar. Many conventional BIA predictions of total body water (TBW) from H2/R, and also of fat-free mass from TBW (fat-free mass = TBW/0.73), while requiring model assumptions, are nevertheless accurate in healthy adults1.Heymsfield S.B. Wang Z.M. Visser M. Gallagher D. Pierson Jr., R.N. Techniques used in the measurement of body composition: An overview with emphasis on bioelectrical impedance analysis.Am J Clin Nutr. 1996; 64: 478S-484SPubMed Google Scholar, 2.Foster K.F. Lukaski H.C. Whole-body impedance – What does it measure?.Am J Clin Nutr. 1996; 64Suppl: 388S-396SGoogle Scholar, 3.Kushner R.F. Bioelectrical impedance analysis: A review of principles and applications.J Am Coll Nutr. 1992; 11: 199-209PubMed Google Scholar. In HD patients, not only conventional BIA but also reference methods like dual energy X-ray absorptiometry (DXA) can be biased since for all methods based on nonchemical subordinate relationships, violation of the assumption of fixed hydration (73%) propagates errors in body compartment prediction1.Heymsfield S.B. Wang Z.M. Visser M. Gallagher D. Pierson Jr., R.N. Techniques used in the measurement of body composition: An overview with emphasis on bioelectrical impedance analysis.Am J Clin Nutr. 1996; 64: 478S-484SPubMed Google Scholar, 2.Foster K.F. Lukaski H.C. Whole-body impedance – What does it measure?.Am J Clin Nutr. 1996; 64Suppl: 388S-396SGoogle Scholar, 3.Kushner R.F. Bioelectrical impedance analysis: A review of principles and applications.J Am Coll Nutr. 1992; 11: 199-209PubMed Google Scholar, 5.Roubenoff R. Kehayas J.J. Dawson-Hughes B. Heymsfield S.B. Use of dual-energy X-ray absorptiometry in body composition studies: Not yet a “gold standard.”.Am J Clin Nutr. 1993; 58: 589-591PubMed Google Scholar, 6.Woodrow G. Oldroyd B. Turney J.H. Davies P.S.W. Day J.M.E. Smith M.A. Measurements of total body water by bioelectrical impedance in chronic renal failure.Eur J Clin Nutr. 1996; 50: 676-681PubMed Google Scholar. For instance, in lean subjects, skinfold thickness was reduced by 5% and DXA bone mineral density by 0.6% after hemodialysis7.Abrahamsen B. Hansen T.B. Hosberg I.M. Pedersen F.B. Beck-Nielsen H. Impact of hemodialysis on dual X-ray absorptiometry, bioelectrical impedance measurements, and anthropometry.Am J Clin Nutr. 1996; 63: 80-86PubMed Google Scholar. We recently overcame the need of assumptions for BIA, using direct measurements from the analyzer with the RXc graph method8.Piccoli A. Rossi B. Pillon L. Bucciante G. A new method for monitoring body fluid variation by bioimpedance analysis: The RXc graph.Kidney Int. 1994; 46: 534-539Abstract Full Text PDF PubMed Scopus (510) Google Scholar, which can evaluate tissue properties in any clinical condition. We used BIA on the standard 50 kHz frequency since measurements of R and Xc at multiple frequencies, despite their promising theory, did not provide any clinically significant improvement over measurements at only 50 kHz for the estimation of extracellular water and TBW2.Foster K.F. Lukaski H.C. Whole-body impedance – What does it measure?.Am J Clin Nutr. 1996; 64Suppl: 388S-396SGoogle Scholar, 9.Hannan W.J. Cowen S.J. Fearon K.C.H. Plester C.E. Falconer J.S. Richardson R.A. Evaluation of multi-frequency bio-impedance analysis for the assessment of extracellular and total body water in surgical patients.Clin Sci. 1994; 86: 479-485Crossref PubMed Scopus (99) Google Scholar, 10.Patel R.V. Peterson E.L. Silverman N. Zarowitz B.J. Estimation of total body and extracellular water in post-coronary artery bypass graft surgical patients using single and multiple frequency bioimpedance.Crit Care Med. 1996; 24: 1824-1828Crossref PubMed Scopus (46) Google Scholar, 11.Sinning W.E. De Oreo P.B. Morgan A.L. Clark Brister E. Monitoring hemodialysis changes with bioimpedance. What do we really measure?.ASAIO J. 1993; 39: M584-M589Crossref Google Scholar. In fact, in muscle the current passes through the intracellular space even at low frequencies since the current is mainly directed along the muscle fibers, according to an electrical behavior more complex than in suspended cells, where the extracellular or intracellular current path predominates at low or high frequencies, respectively2.Foster K.F. Lukaski H.C. Whole-body impedance – What does it measure?.Am J Clin Nutr. 1996; 64Suppl: 388S-396SGoogle Scholar,4.Barber D.C. Electrical impedance tomography.The Biomedical Engineering Handbook. edited by BRONZINO JD. CRC Press, Boca Raton1995: 1151-1164Google Scholar. In this study, we evaluated the performance of this new BIA approach for interpreting the cyclic body weight variation and determining optimal dry weight prescription in the HD patient. In 1996 a multicenter study was designed to establish first whether the cyclic fluid repletion and removal was associated with a definite BIA pattern that could be utilized in dialysis prescription (cross-sectional study phase), and second whether different BIA patterns were associated with different death risk levels over a two-year follow-up period (1996 to 1998; the longitudinal phase). Results concerning the cross-sectional evaluation of patients are reported. In April to September 1996 we studied 1489 patients (age range 16 to 89 years) undergoing chronic hemodialysis (HD) in Italy, selected from 39 dialysis units participating in the Italian HD-BIA Study Group. All patients had BIA measurements taken immediately before and after each of two HD sessions conducted within a week, making a total of 5956 BIA measurements. A random order was assigned to the sequence of HD sessions for each patient, one session after a short interdialysis period (48 hr) and the other after the long interdialysis period (72 hr). As reference normal group for BIA measurements we considered 726 healthy Italian subjects, 354 men and 372 women [age 15 to 85 years, and body mass index (BMI) 17 to 31 kg/m2], who also participated in a previous study12.Piccoli A. Nigrelli S. Caberlotto A. Bottazzo S. Rossi B. Pillon L. Maggiore Q. Bivariate normal values of the bioelectrical impedance vector in adult and elderly populations.Am J Clin Nutr. 1995; 61: 269-270PubMed Google Scholar. Uremic patients, adult Caucasian of both genders, undergoing ambulatory, standard chronic hemodialysis (consisting of bicarbonate dialysis, 210 to 240 min thrice weekly) from six months or more were eligible for the study. A sampling grid was provided to centers in order to balance the sample size for age within the gender (half above and half below 60 years). All subjects gave their informed consent to the study. A patient who is symptom-free and achieves the prescribed target dry weight at the end of treatment was defined as stable if no hemodynamic instability (such as hypotension and hypertension access) was observed over the prior three months. A patient with either symptomatic hypotensive episodes or a systolic blood pressure nadir below 90 mm Hg in 30% or more of sessions over the last three months (10 of 36 sessions or more), was defined as unstable. The HD prescription was not modified during the study, no matter what BIA results were obtained. Patients with cardiac (New York Heart Association class III or more), pulmonary, or hepatic failure, with cancer, or with previous kidney graft failure were not considered. As for the reference healthy population, we excluded from statistical analysis those HD patients with BMI values out of the range 17 to 31 kg/m2. Therefore, a total of 1367 patients (5468 BIA measurements), namely 1116 stable and 251 unstable patients, were considered in this study. Age, dialytic age and height (H) were recorded the day of the first BIA determination. Body weight immediately before and after the HD sessions was recorded from a bed-balance scale. The average BMI, obtained from the mean of pre- and post-dialysis weight, was considered in patient classification for body mass. Hemoglobin (Hb), albumin, and total plasma protein concentration were determined by routine methods in the week of BIA measurements, before the HD session. The plasma oncotic pressure was calculated from albumin and total protein concentration13.Miller P.L. Meyer T.W. Plasma protein concentration and colloid osmotic pressure in nephrotic rats.Kidney Int. 1988; 34: 220-223Abstract Full Text PDF PubMed Scopus (9) Google Scholar. In every center, BIA measurements were conducted by the same operator using an impedance plethysmograph that emitted 800 μA and 50 kHz alternating sinusoidal current (BIA-101; RJL/Akern Systems, Clinton Twp, MI, USA) and was connected to surface electrodes (standard, tetrapolar placement on the hand and foot) strictly following the method reported elsewhere3.Kushner R.F. Bioelectrical impedance analysis: A review of principles and applications.J Am Coll Nutr. 1992; 11: 199-209PubMed Google Scholar. Electrodes were placed on the side free from vascular access and were not removed during the HD session. The mean coefficient of variation was 1.5% for paired, pre- and post-HD intraindividual repeated measurements. According to the RXc graph method (see below), we standardized BIA measurements by the H of the subjects, thus expressing both R/H and Xc/H in Ohm/m. The programs of the statistical package BMDP14.Dixon W.J. Brown M.B. Engelman L. Jennrich R.I. BMDP Statistical Software Manual. UCLA, Berkeley1992Google Scholar were used for standard calculations, including the Student's t-test, the Hotelling's T2 test for vector analysis (Program 3D), the linear correlation coefficient r (Program 6D), the Χ2 test (Program 4F), and the analysis of covariance (Program 1V). Using the bivariate normal distribution15.Morrison D.F. Multivariate Statistical Methods. McGraw-Hill, New York1967Google Scholar,16.Lentner C. Introduction to statistics. Statistical tables. Mathematical formulae.Geigy Scientific Tables. 1982; vol 2 (8th ed, Ciba-Geigy Limited, Basle): 215-218Google Scholar of R/H and Xc/H, we calculated the bivariate 95% confidence limits for mean impedance vectors of the different classification groups (the ellipse containing both the magnitude and the phase angle of the mean vectors with 95% probability). We called the “RXc mean graph” the average of R/H and Xc/H plotted as arrowhead line segment with the 95% confidence ellipse Figure 1. We calculated the bivariate 50%, 75% and 95% tolerance limits16.Lentner C. Introduction to statistics. Statistical tables. Mathematical formulae.Geigy Scientific Tables. 1982; vol 2 (8th ed, Ciba-Geigy Limited, Basle): 215-218Google Scholar of the impedance vector in the reference healthy population (that is, the ellipses within which the vector of the individual subject falls with a probability of 50%, 75% and 95%, respectively; Figures 2 and 3). One point vector from an individual patient, either pre-HD or post-HD vector, was the average value of the two BIA measurements taken from the patient over the two HD sessions (one after 48 hr interdialysis period and the other after a 72 hr interdialysis period).Figure 3Dotted ellipses represent the distribution of 50% impedance vectors from unstable individual hemodialysis patients before (lower ellipse) and after fluid removal (upper ellipse). Symbols and reference tolerance ellipses were drawn as in Figure 2.View Large Image Figure ViewerDownload (PPT) The results considering gender, study population, and HD treatment effect on the impedance vector of 1367 patients (1116 stable and 251 unstable), are reported in Table 1. The two components of the impedance vector were significantly linearly correlated each other in all groups considered, although with r values lower in HD patients than healthy subjects of either gender Table 1. This correlation accounted for the elliptical shape of confidence and tolerance intervals of vectors in all groups Figures 1,2,3.Table 1Mean values with standard deviation (SD) of protocol variables The correlation coefficients between either vector components and age, dialytic age, BMI, hemoglobin, plasma albumin, and oncotic pressure, are reported in Table 2. The square of the highest r value indicated that less than 13% of either vector component variability was explained by any protocol variables.Table 2Linear correlation coefficient (r) between either component of the impedance vector and protocol variables Significantly shorter and down-sloping impedance vectors were observed in either HD group compared with healthy population (non overlapping 95% confidence ellipses, Figure 1). The position of group vectors from unstable patients significantly differed from stable HD patients of either gender (T2= 49.4 in men, and 31.8 in women, P < 0.001), since vectors of unstable patients were longer (Z/H = 303 vs. 294 Ohm/m in men, and 373 vs. 355 Ohm/m in women) and with a smaller phase angle (4.5° vs. 5.1° in men, and 4.2° vs. 4.7° in women). A significant gender effect was observed in all groups, with longer and less steep impedance vectors in women Figure 1. Mean values of either vector component, adjusted for age, were compared by the analysis of covariance (age as covariate). Adjusted R/H mean values significantly (P < 0.001) differed between healthy and stable HD groups in men, and among all three groups in women. Adjusted Xc/H mean values significantly (P < 0.001) differed among the three groups in either gender. Therefore, the pattern of vector positions before HD treatment was not significantly influenced by different age of groups. The vector displacement caused by fluid removal (that is, the dZ = post-HD Z minus pre-HD Z), which was comparable in men and women, differed between the HD groups, being significantly shorter and less steep in unstable patients (T2= 28.4, P < 0.001 in men, and T2= 7.3, P = 0.03 in women). Post-HD vectors were significantly longer than normal in either HD groups Figure 1. The amount of fluid removal did not significantly differ (Student's t-test) between stable and unstable HD groups, considering either the average of the two sessions Table 1 or each HD sessions, namely after the short interdialytic interval (-2.6 vs. -2.6 kg in men, and -2.4 vs. -2.2 kg in women) and after the long interdialytic interval (-2.6 vs. -2.6 kg in men, and -2.7 vs. -2.6 kg in women). Moreover, in all groups, the correlation coefficients between changes in either vector components and the amount of fluid removal were statistically significant (P < 0.05) and ranging in the interval -0.52 < r < -0.24. To derive bedside application of BIA patterns associated with the wet-dry weight cycling of HD patients, we plotted the vectors recorded from single patients both pre- and post-HD on the reference 50%, 75%, and 95% tolerance ellipses of the healthy population. In Figures 2 and 3, two dotted ellipses depict the distribution of 50% of individual vectors before and after HD, from stable and unstable patients, respectively. A greater number of vectors of the unstable HD group fell out of the reference 75% tolerance ellipse at the start of HD (43% vs. 27% in unstable vs. stable HD men, and 41% vs. 31%, respectively, in women). However, the distribution of the vectors falling out of the target boundary was asymmetric Figures 2 and 3, since 15% more vectors from the unstable HD group fell out of the right half boundary as compared to the left half. Indeed, the vectors from 83% unstable versus 68% stable HD male patients (X2= 5.2, P = 0.02), and 79% versus 63% stable HD female (X2= 5.4, P = 0.02), fell out of the right half ellipse (particularly out of the lower right quadrant). Therefore, a patient beginning a HD session with a vector falling out of the right boundary of the normal reference 75% tolerance ellipse, had a greater probability of symptomatic hypotension during treatment. A greater number of vectors of the unstable HD group also fell out of the normal reference 75% tolerance ellipse at the end of HD (59% vs. 44% in unstable vs. stable HD men, and 54% vs. 46%, respectively, in women). The asymmetry of the frequency distribution between unstable versus stable patients was even greater at the end of dialysis Figures 2 and 3, in the order of 25%, as the vectors from 71% versus 45% men (Χ2= 15.7, P < 0.001), and 67% versus 42% women (Χ2= 14.9, P < 0.001), fell out of the right half ellipse (particularly out of the upper right quadrant). There was no stable, definite correlation between either vector components and plasma albumin, oncotic pressure, and Hb in any group Table 2. On the other hand, hemoglobin weakly correlated with either plasma albumin (0.10 < r < 0.25) or oncotic pressure (0.10 < r < 0.29) in separate and pooled groups. Furthermore, both albumin and Hb were significantly lower in unstable HD patients (37 g/liter albumin and 96 g/liter Hb) as compared to stable HD patients (40 g/liter albumin and 101 g/liter Hb; T2= 60.9, P < 0.001) with severe anemia (Hb < 80 g/liter) and hypoalbuminemia (albumin < 30 g/liter) more frequent in unstable as compared to stable patients (16% vs. 8% anemia, and 12% vs. 5% hypoalbuminemia) (Χ2= 26.7 and Χ2= 32.8, P < 0.001, respectively). Dry weight prescription based on symptoms induced by inappropriate fluid removal is a daily clinical dilemma in the absence of objective criteria other than body wt17.Kushner R.F. De Vries P.M.J.M. Gudivaka R. Use of bioelectrical impedance analysis measurements in the clinical management of patients undergoing dialysis.Am J Clin Nutr. 1996; 64: 503S-509SPubMed Google Scholar. From this study, conducted in a large HD population, we obtained important clues for the routine monitoring of hydration independent of body wt. By using direct impedance measurements we demonstrated that: (1) The thrice weekly wet-dry weight cycling was bound to a cyclic backward-forward displacement of the impedance vector over definite elliptical areas on the R-Xc plane. The line of action of vector displacement paralleled the major axis of the reference tolerance ellipses from the healthy population. (2) About half of uremic patients undergoing maintenance HD were cycling within the third quartile of impedance vector distribution of the healthy population. This allowed the identification of patients with full versus either intermittent or no restoration of normal tissue electrical conductivity. (3) About 20% of patients with recurrent hypotension were characterized by longer, less steep vectors than the stable patients either at the beginning or at the end of the HD session. These vectors more frequently overshot the right boundary of the reference 75% tolerance ellipse. (4) The different vector position of unstable versus stable HD patients was associated with lower albumin and Hb concentrations, binding fluid status, hemodynamic instability and malnutrition. These results support a new operative definition for the optimal wet-dry weight cycling. The definition of optimal includes pre- and post-HD impedance vectors cycling within the third quartile tolerance ellipse of the reference healthy population. However, only the longitudinal phase of the study will establish whether patients from either group with vectors cycling within the target ellipse have better tolerance to HD treatment and a longer survival. To evaluate whether BIA information could be successfully used to modify dialysis prescription and improve outcomes, a further intervention study should be performed in which 50% of patients were provided impedance vector data and 50% were not given the data. Patients in the current study were classified according to their response to a dry weight target into an unstable group if hypotension occurred with fluid removal and into stable group if they remained asymptomatic. This might include a percentage of longer vectors from asymptomatic patients (still tolerating inappropriate fluid removal) and a percentage from unstable patients with symptomatic hypotension (possibly improving after less fluid removal). Therefore, the routine bedside use of the RXc graph in monitoring hydration of patients is expected to be of benefit to most patients whose vectors fall out of the reference 75% tolerance ellipse. Furthermore, the lower albumin and Hb concentration in unstable patients indicate that metabolic and nutritional factors are likely involved in the dialytic tolerance. The low phase angle in the unstable HD group was close to that found by others in HD and in AIDS patients with the highest mortality rate over two years18.Maggiore Q. Nigrelli S. Ciccarelli C. Grimaldi C. Rossi G.A. Michelassi C. Nutritional and prognostic correlates of bioimpedance indexes in hemodialysis patients.Kidney Int. 1996; 50: 2103-2108Abstract Full Text PDF PubMed Scopus (175) Google Scholar, 19.Chertow G.M. Lowrie E.G. Lew N.L. Lazarus J.M. Bioelectrical impedance analysis predicts survival in hemodialysis patients.J Am Soc Nephrol. 1996; 7 (abstract): 1442Google Scholar, 20.Ott M. Fischer H. Polat H. Helm E.B. Frenz M. Caspary W. Lembcke B. Bioelectrical impedance analysis as a predictor of survival in patients with human immunodeficiency virus infection.J Acquir Immune Defic Syndr. 1995; 9: 20-25Google Scholar. Interestingly, the slope of vector displacement after comparable fluid removal was also less steep in unstable versus stable HD patients Figure 1. However, contrary to literature indications18.Maggiore Q. Nigrelli S. Ciccarelli C. Grimaldi C. Rossi G.A. Michelassi C. Nutritional and prognostic correlates of bioimpedance indexes in hemodialysis patients.Kidney Int. 1996; 50: 2103-2108Abstract Full Text PDF PubMed Scopus (175) Google Scholar, 19.Chertow G.M. Lowrie E.G. Lew N.L. Lazarus J.M. Bioelectrical impedance analysis predicts survival in hemodialysis patients.J Am Soc Nephrol. 1996; 7 (abstract): 1442Google Scholar, 20.Ott M. Fischer H. Polat H. Helm E.B. Frenz M. Caspary W. Lembcke B. Bioelectrical impedance analysis as a predictor of survival in patients with human immunodeficiency virus infection.J Acquir Immune Defic Syndr. 1995; 9: 20-25Google Scholar, the interpretation of phase angle values should not be made independently of vector magnitude (that is, Xc independent of R values) since R and Xc are correlated in living anisotropic tissues and both impedance components are dependent on tissue structure4.Barber D.C. Electrical impedance tomography.The Biomedical Engineering Handbook. edited by BRONZINO JD. CRC Press, Boca Raton1995: 1151-1164Google Scholar. Thus, vector displacements parallel to the major axis of tolerance ellipses would indicate changes in tissue hydration with preserved tissue structure, and vector displacements following different trajectories would indicate combined changes in tissue structure and hydration8.Piccoli A. Rossi B. Pillon L. Bucciante