Kidney International, Vol. 66 (2004), pp.

12661271

Vector length as a proxy for the adequacy of ultrafiltration

in hemodialysis
LUANA PILLON, ANTONIO PICCOLI, EDMUND G. LOWRIE, J. MICHAEL LAZARUS,
and GLENN M. CHERTOW
Division of Nephrology, Department of Medicine, New York University School of Medicine, New York, New York; Division of
Nephrology, Department of Medical and Surgical Sciences, University of Padua, Padua, Italy; Fresenius Medical Care North
America, Lexington, Massachusetts; and Division of Nephrology, Department of Medicine, University of California San Francisco,
San Francisco, California

Vector length as a proxy for the adequacy of ultrafiltration in

hemodialysis.
Background. Evaluation of dialysis adequacy has focused on
parameters of solute (principally urea) clearance. Relatively little attention has been paid to the adequacy of ultrafiltration. At
a given phase angle, the bioimpedance vector length reflects
the degree of tissue hydration, as the vector lengthens with
ultrafiltration.
Methods. We determined the relative risk of death associated
with different bioimpedance vector lengths in a 3009 patient
hemodialysis cohort using proportional hazards regression.
Results. The mean phase angle was 4.8 , and the mean vector length 300 70 ohm/m (range 140 to 630 ohm/m). Vector
length was much longer in women than men (mean 340 vs.
270 ohm/m) and significantly longer in African Americans and
patients without diabetes. Adjusted for the effects of age, gender, race, diabetes, vintage, weight, albumin, prealbumin, creatinine, hemoglobin, ferritin, and dialysis dose, the relative risk
(RR) of death was 0.75 (95% CI 0.57 to 0.88) per 100 ohm/m
decrease in vector length. The effect of vector length on RR was
somewhat more pronounced among men (vector length gender interaction, P = 0.07). Considering vector length of 300 to
350 ohm/m as the referent category, the RRs of death were 1.54
(95% CI 1.08 to 2.21) and 2.83 (95% CI 1.55 to 5.14) for patients
with vector length 200 to 250 and <200 ohm/m, respectively.
Conclusion. Shorter predialysis bioimpedance vectors, indicating greater soft tissue hydration, were associated with diminished survival in hemodialysis patients. These findings validate
clinical observations linking longevity to maintenance of dry
body weight.

The evaluation of hemodialysis adequacy has focused

on parameters of solute (principally urea) clearance.

Key words: ultrafiltration, hemodialysis, adequacy of dialysis, body fluid

volume, bioimpedance, diabetes, phase angle.
Received for publication December 20, 2003
and in revised form February 10, 2004
Accepted for publication April 21, 2004


C

2004 by the International Society of Nephrology

Lower volume normalized clearance time products

(Kt/V) have generally been associated with higher mortality rates in observational studies of maintenance
hemodialysis patients [1], although the recently completed HEMO Study suggested that higher rates of urea
clearance might not be of benefit [2]. In contrast to solute
kinetics, relatively little attention has been paid to evaluation of the adequacy of ultrafiltration. Fluid removal is
integral to dialysis since over- and underhydration have
been linked with intradialytic morbidity and long-term
cardiovascular complications. To date, the optimal postdialysis (dry) weight is determined clinically, generally
as the lowest weight a patient can tolerate without intradialytic symptoms or hypotension. This rather crude
trial-and-error method does not account for changes in
lean body mass, fat mass, or inflammatory or nutritional
status over time.
Bioelectrical impedance analysis (BIA) is a practical, portable, relatively inexpensive and reliable bedside
tool to aid the evaluation of body composition. Briefly,
impedance is a measurable property of electrical ionic
conduction of soft tissue, particularly lean tissue, since
fat and bone are poor conductors [3, 4]. An impedance
vector (Z) is the sum of resistance (R) and reactance (Xc).
Where R is the opposition to the flow of an alternating
current through intra- and extracellular ionic solutions,
Xc is the capacitance produced by interfaces, including
cell membranes and organelles, across tissues [5]. Alternatively, the impedance
can be expressed as a vector with
a magnitude |Z| = (R2 + Xc2 ) and phase angle (arc tan
Xc/R). In principle, at low frequencies (<5 kHz) current
would pass through the extracellular fluid while at higher
frequencies (>100 kHz) it would penetrate all fluid compartments. In practice, a variable amount of current can
cross muscle cells even at very low frequencies, particularly when the currents path is parallel to the fiber [5].
Estimation of total body water (TBW) with the conventional BIA approach assumes the body to be an isotropic
1266

Pillon et al: Vector length as a proxy for the adequacy of ultrafiltration

cylinder of a given length [height (H)] and constant

cross-sectional area, and uses regression equations based
on H2 /R and other variables, including age, gender, body
weight, and reactance. BIA estimation of masses and
volumes of body compartments are accurate in healthy
adults, although with 95% prediction intervals >3 kg
[69]. In order to increase accuracy of TBW and body cell
mass estimation from BIA, specific prediction equations
have been developed for hemodialysis patients. Unfortunately these equations result in a bias in the order of 4 to
5 kg with respect to reference methods [810].
Some of the important limitations of conventional BIA
can be overcome by direct evaluation of impedance vector readings. A graphic approach for monitoring relative
changes in TBW based on single-frequency impedance
analysis has been termed the resistance-reactance graph
(RXc graph) [11]. This approach makes no assumptions
about body geometry, hydration status, or the electric
model of cell membranes. Rather, an individuals heightadjusted resistance (R/H) and reactance (Xc/H) are
employed, yielding a phase angle and vector magnitude
unaffected by the obligate errors of regression adjustment. The individual impedance vector plotted as a point
on the R-Xc plane can be compared with normative data
derived from a representative sample of healthy individuals of the same gender and race or ethnicity to evaluate
the relative hydration status of the individual and can be
followed over time. Referent, bivariate tolerance intervals were derived that included 50%, 75%, and 95% of
healthy subjects from different races [12]. Gender, race,
or ethnicity, body mass index (BMI), and age, in decreasing order, influenced the vector distribution pattern [13].
Following clinical validation studies, normal hydration
was defined as the range of impedance vectors falling
within the reference 75% tolerance interval.
In hemodialysis patients, the thrice weekly wet-dry
weight cycling parallels the cyclic backward-forward displacement of the impedance vector over definite elliptical
areas on the R-Xc plane with vector displacement parallel to the major axis of the reference tolerance ellipses
[14]. About 50% of patients undergoing maintenance
hemodialysis cycle within the third quartile of impedance
vector distribution of the healthy population, which allowed the identification of patients with full versus intermittent versus no restoration of normal tissue electrical
conductivity.
In a previous study, we demonstrated that phase angle
was directly correlated with several parameters of nutritional status and independently associated with survival
in hemodialysis patients [15]. However, two persons at a
given phase angle may be widely disparate in body composition and volume status, based on the vector magnitude. Given that ultrafiltration lengthens the impedance
vector, we hypothesized that shorter vector length (i.e.,
inadequate ultrafiltration) would be associated with in-

1267

creased mortality, even after accounting for phase angle

and other confounding variables. This observation would
be consistent with our clinical impression that hemodialysis patients who achieve postdialysis weights at or near
target experience more favorable outcomes.
METHODS
Study subjects were 3009 prevalent adult hemodialysis patients from 101 free-standing Fresenius Medical
Care North America (FMCNA) dialysis units across the
United States. Inclusion criteria included age 18 years
and thrice weekly in-center hemodialysis for 3 months.
Patients with an amputation above the transmetatarsal
site were excluded from participation. BIA Quantum
(RJL Systems, Inc., Clinton Twp., MI, USA) was performed before a midweek dialysis session during the first
6 months of 1995. Briefly, an inner electrode was attached
to the dorsal surface of the wrist on the arm without
an arteriovenous fistula or graft. An outer electrode was
placed on the dorsal surface of the third metacarpal bone.
A second pair of electrodes was positioned on the anterior surface of the ipsilateral ankle and the dorsal surface
of the third metatarsal bone [6]. A single frequency lowamplitude imperceptible current (800 lA at 50 kHz) was
introduced via the electrodes on the hand and foot. The
electrodes at the wrist and ankle detected the voltage
decrease. The resistance and reactance in ohms was obtained directly from the device and normalized by the
subjects height in meters and expressed as R/H and
Xc/H in ohm/m according to the vector BIA (RXc graph)
methodology [11]. Phase angle (the arc tangent of the
reactance to resistance ratio) was calculated in radians,
and multiplied by 180/
(3.14159265) to covert radians to degrees. Reactance, resistance, phase angle, and
the derived estimates of total body water were merged
with the Patient Statistical Profile, a database with selected demographic, historic, and laboratory information
on patients cared for at FMCNA-affiliated dialysis facilities. Laboratory values were means of the three months
proceeding BIA testing. All patients were followed for
at least 1 year from the time of BIA testing or the point
of censoring. Duration of follow-up ranged from 2 days
to 18 months. Patients whose survival time was unknown
or uninterpretable (N = 19) (0.6%) were excluded from
the analysis.
Statistical analysis (general)
Continuous variables were described as mean standard deviation or median and interquartile range, and
compared using Student t test, analysis of variance or nonparametric tests, where appropriate. Linear regression
was used to determine independent correlates of vector
length. Categorical variables were described as proportions. Height normalized impedance vector magnitude

Pillon et al: Vector length as a proxy for the adequacy of ultrafiltration

Statistical analysis (bioimpedance vectors)

The parameters (slope and length of semiaxes) of the
bivariate confidence and tolerance ellipses were calculated following methods described elsewhere. In the statistical analysis of mean vectors, separate 95% confidence
ellipses indicate a statistically significant difference (P <
0.05) between mean vector positions on the R-Xc plane
(equivalent to a significant Hotellings T2 test) [18].

ND
D
ND
D

20

ac

20

Women

30

Bl
ac
k

Men

30

te

Bl

|Z/H| = [(R/H)2 + (Xc/H)2 ] in ohm/m was evaluated

as a continuous variable in the primary analysis. The association between vector length and patient survival was
explored using proportional hazards (Cox) regression,
with vector length expressed as a continuous variable
[16]. In this case, the relative risk (RR) represents the
expected change in risk per 100 ohm/m change in vector
length. Companion analyses were performed using vector length categories (six predefined categories, namely
<200, 200 to 250, 250 to 300, 300 to 350, 350 to 400, and
400 ohm/m) obviating the linearity assumption. Analyses were adjusted for case mix (age, gender, race, diabetes,
and vintage (time since initiation of dialysis), and for case
mix plus nutrition- and inflammation-related variables
significantly associated with survival on univariate analysis. Unadjusted, case mix-adjusted, and multivariableadjusted hazard ratios with 95% CI were calculated based
on model parameter coefficients and standard errors, respectively. Factors not included in multivariable regression models were re-entered individually to evaluate for
residual confounding (>5% change in vector length or
phase angle parameter estimate). Multiplicative interaction terms with vector length were tested to explore for
effect modification by other model covariates. Plots of
log (-log [survival rate]) against log (survival time) were
performed to establish the validity of the proportionality
assumption [17]. Subjects who underwent kidney transplantation (N = 82) (2.7%), recovered renal function
(N = 18) (0.6%), transferred dialysis facilities (N =
287) (9.7%), withdrew from dialysis (N = 42) (1.3%),
or were lost to follow-up for unknown reasons (N = 8)
(0.3%) were censored. Two-tailed P values less than 0.05
were considered statistically significant. Statistical analyses were conducted using SAS version 6.08 (SAS Institute, Cary, NC, USA).

Xc/H, ohm/m

1268

10

hi

10

ND
ND
D

te

hi

0
0

100 200 300

R/H, ohm/m

100 200 300 400

R/H, ohm/m

Fig. 1. Mean impedance vectors, standardized by height [resistance/height (R/H), reactance/height (Xc/H) (ohm/m) with 95%
confidence ellipses, by gender, race, or ethnicity, and diabetes status.
Separate 95% confidence ellipses indicate a significant (P < 0.05) vector displacement on the R-Xc plane. Abbreviations are: D, diabetic;
ND, nondiabetic.

phase angle across the population has been previously

reported [19].
Correlates of vector length
The mean phase angle was inversely correlated with
age (r = 0.33, P < 0.0001) and vintage (r = 0.15, P <
0.0001). The mean phase angle was significantly wider in
men than women (5.0 vs. 4.4 , P < 0.0001), blacks (5.3 )
than whites (4.3 ), Hispanics (4.7 ), or persons of other
races or ethnicities (4.6 ) (P < 0.0001), and in patients
without diabetes (4.9 vs. 4.5 in patients with diabetes,
P < 0.0001) (Fig. 1). In contrast, age was directly correlated with vector length (r = 0.19, P < 0.0001) and vector
length was substantially longer in women than men (337
vs. 270 ohm/m, P < 0.0001) (Fig. 1). Mean vector length
was slightly shorter among patients with diabetes (Fig. 1).
Linear regression showed significant associations among
vector length and age, gender, race, or ethnicity, diabetes,
and the diabetes gender interaction, with these factors
explaining 28% of the variation in vector length. Vector length was unrelated to vintage. Figure 2 shows the
predialysis distribution by gender and race (95%, 75%,
and 50% tolerance ellipses) for the nondiabetic patients
classified as white or black, for comparison with other
published data. These distributions can be utilized for
comparison and for bedside identification of shorter and
more down sloping vectors.
Vector length and mortality

RESULTS
The mean age was 60.5 15.4 years, 47.2% were
women, 46.9% black, 45.4% white, 6.5% Hispanic, and
1.2% of other races or ethnicities. Thirty-seven percent of
patients had diabetes mellitus. The median dialysis vintage was 2.6 years (interquartile range 1.4 to 4.9 years).
A detailed description of the distributions of R, Xc, and

The overall 1-year survival rate was 88%. Mortality was

significantly associated with age [RR 1.04 (95% CI 1.03
to 1.05) per year increase], and race or ethnicity [RR 0.59
(95% CI 0.48 to 0.74) in blacks and RR 0.39 (95% CI 0.22
to 0.65) in Hispanics, compared with whites]. The proxies
of nutritional status, serum albumin [RR 0.34 (95% CI
0.26 to 0.44) per g/dL increase] and creatinine [RR 0.87
(95% CI 0.84 to 0.90) per mg/dL increase] were inversely

1269

Pillon et al: Vector length as a proxy for the adequacy of ultrafiltration

3
Men

50

30

95%
75%
50%

40
30

10

0
0 100 200 300 400 500
R/H, ohm/m

B
Men

0
0

40

m/

10

oh

10

50%
75%
95%
30

20

H,

20

.80

R/

30

.93

4.8

95%
75%
50%

20

30

40

Women

1.54
.98

40
Xc/H, ohm/m

50
95%
75%
50%

/H

0
10

50

0 0

0 100 200 300 400

R/H, ohm/m

Xc

m/

40

10

h
,o

20

20

10

20

2.83
2
50

95%
75%
50%

Women

30

Xc/H, ohm/m

40

Relative risk

50

0
0 100 200 300 400 500
R/H, ohm/m

Fig. 2. Gender and race specific, bivariate distribution (95%, 75%,

and 50% tolerance ellipses) for an individual impedance vector from
white or black, nondiabetic, hemodialysis population, before a dialysis
session.

related to mortality. The relation between survival and

unadjusted RR was curvilinear (reverse J-shaped). The
unadjusted risk of death was increased among patients
with diabetes, but not significantly so in multivariable
analyses [RR 1.22 (95% CI 0.98 to 1.50)]. In this cohort,
mortality was not significantly associated with gender or
dialysis vintage.
There was a significant association between vector
length and mortality [multivariable RR 0.75 (95% CI
0.57 to 0.88 per 100 ohm/m change in vector length)].
To test the linearity assumption, we fit companion models in which vector length was divided into six pre-defined
categories (<200, 200 to 250, 250 to 300, 300 to 350, 350
to 400, and 400 ohm/m). Using 300 to 350 ohm/m as the
referent vector length, the RR of death was 1.54 (95%
CI 1.08 to 2.21) for vector length 200 to 250 ohm/m and
2.83 (95% CI 1.55 to 5.15) for vector length <200 ohm/m.
The increase in RR with shorter vector length was independent of age, gender, race, diabetes, vintage, albumin,
creatinine, hemoglobin, ferritin, and perhaps most important, phase angle. There was no interaction between
phase angle and vector length. In other words, shorter
vector lengths and narrower phase angles were indepen-

100 200 300 400

R/H, ohm/m

50

Fig. 3. The bivariate distribution of impedance vectors from 2990

hemodialysis patients is drawn using a set (95%, 75%, and 50%) of
tolerance ellipses. The multivariable relative risk of death is depicted
as a vertical bar on the midpoint of the six vector magnitude intervals
(arrows from a to f, categories of impedance vector |Z/H| <200, 200 to
250, 250 to 300, 300 to 350, 350 to 400, and 400 ohm/m, respectively).
The mean impedance vector is represented with a larger arrow (phase
angle 4.8 ). The interval |Z/H| between 300 and 350 ohm/m was considered the referent category for the relative risk calculation (vector
d). Statistically significant (P < 0.05) relative risk values are indicated
with a solid circle. Open circles indicate nonsignificant relative risks of
death.

dently associated with mortality. The association of vector length with mortality tended to be more pronounced
among men (vector length gender interaction, P =
0.07). Figure 3 shows the multivariable adjusted RR of
death within the six predefined vector length categories
plotted on the 50%, 75%, and 95% tolerance ellipses.

DISCUSSION
Inadequate fluid removal during hemodialysis can lead
to hypertension, heart failure, and stroke, among the
leading causes of death in dialysis patients. In contrast,
excessive fluid removal can lead to hypotension, arrhythmias, muscle cramping, nausea, vomiting, and other
adverse effects. Commonly, the rate of ultrafiltration exceeds the rate of vascular refill (especially with shorter
dialysis sessions), leading to inadequate intradialytic fluid

1270

Pillon et al: Vector length as a proxy for the adequacy of ultrafiltration

removal, hypervolemia, hypertension, the provision of

antihypertensive therapy, and ultimately a vicious cycle, yielding a steady-state well above dry weight.
Chronic hypervolemia can contribute to cardiomyopathy, pulmonary congestion, systemic and pulmonary
hypertension, and potentially, an increased risk of ischemic coronary artery disease, stroke, and sudden death.
Quantitative measures other than clinical symptoms and
more specific parameters than changes in overall body
weight would be valuable to guide prescription of dry
weight and potentially, avoid intra- and interdialysis complications of cardiovascular disease.
BIA allows a specific assessment of the electric properties of tissues that depend on both tissue hydration
and structure. With bioimpedance vector analysis (RXc
graph), different patterns of the impedance vector distribution have been found to reflect different hydration and
nutritional states independent of body weight [1114].
Phase angle has been associated with mortality in patients
with acquired immunodeficiency syndrome (AIDS) [20]
and end-stage renal disease (ESRD) [15, 21] and has been
directly correlated with other proxies of nutritional status in several diverse cohorts [22, 23]. Few studies have
jointly considered phase angle and vector magnitude.
The findings by gender and race are consistent with
reports in healthy adults from the Europe [12] and the
United States [14]. An interesting parallel can be made
with findings from a large Italian hemodialysis population
in whom impedance measurements were performed preand postdialysis with the same device and method. In the
Italian study, thrice weekly wet-dry weight cycling was
linked to a cyclic backward-forward displacement of the
impedance vector with displacement parallel to the major
axis of the reference, gender-specific, tolerance ellipses
from the healthy population [13]. Gender-specific tolerance ellipses of the Italian hemodialysis population were
of the same shape as those of the healthy Italian population. The predialysis vector position in Italian hemodialysis patients (of whom the vast majority were Caucasian)
was similar to that of the nondiabetic, white patients studied here.
In this study, we documented a significant association
between shorter vector length and mortality. We could
not demonstrate an association between longer vector
lengths and decreased mortality. Whether this reflects a
nonlinear relation between vector length and mortality
or insufficient power to detect such an association is unknown. Interestingly, the relation between total body water and vector magnitude is asymmetric (a hyperbolic
function), with a shortening of vectors outside of the
lower poles of the 75% tolerance ellipses when tissue
hydration increases.
We have previously shown an increase in the RR of
death for hemodialysis patients with phase angles below
4 . Shorter vectors tend to be seen with narrower phase

angles, due to the geometry of the bivariate distribution of

vectors. The geometric link between phase angle and vector length could lead to errors in determining a patients
nutrition and prognosis if one were to focus only on phase
angle while neglecting vector magnitude. For example,
wider phase angles with longer vectors are observed in
states of dehydration (e.g., postdialysis, cholera, prerenal azotemia), and narrower phase angles with shorter
vectors are observed with fluid overload (e.g., predialysis, edema) independent of either nutritional status or
prognosis [24].
There are several limitations to the analyses presented
here. BIA was performed predialysis. Since interdialytic
weight gain varies within and among individuals, there
was more noise in the estimates of resistance and reactance than there might have been had BIA been obtained
postdialysis. Therefore, it is even more noteworthy that a
significant association between vector length (as a proxy
for volume status) and mortality was demonstrated. We
did not record direct physical examination findings (e.g.,
edema, pulmonary rates). These might have substantiated alterations in impedance parameters. Aside from
height and weight, no other measures of body composition were obtained. Finally, the sample was restricted
to maintenance hemodialysis patients; results may not be
generalizable to peritoneal dialysis patients, persons with
acute renal failure on hemodialysis, or persons with less
severe degrees of chronic kidney disease.

CONCLUSION
In a large, nationally representative sample of
hemodialysis patients, we demonstrated a significant association between vector length (a known proxy for
volume status and ultrafiltration) and mortality. These
findings were independent of case mix, confounding laboratory values, and phase angle. These findings validate
the clinical observation linking longevity to maintenance
of dry body weight. Since hemodialysis population norms
of vector distribution, including length and phase angle, have been developed, bioelectrical impedance vector
analysis may be used to evaluate the adequacy of ultrafiltration in hemodialysis. Prospective, serial determinations of phase angle and vector length will be required to
determine the utility of bioelectrical impedance vector
analysis in clinical practice.
Reprint requests to Luana Pillon, M.D., New York University School
of Medicine, Old Bellevue 677, 550 First Avenue, New York, NY 10016.
E-mail: luana.pillon@med.nyu.edu

REFERENCES
1. DENKER BM, CHERTOW GM, OWEN WF, JR.: Hemodialysis, in The
Kidney, edited by Brenner BM, Philadelphia: Saunders, 2000, pp
23732453

Pillon et al: Vector length as a proxy for the adequacy of ultrafiltration

2. EKNOYAN G, BECK GJ, CHEUNG AK, et al: Effect of dialysis dose and
membrane flux in maintenance dialysis. N Engl J Med 347:2010
2019, 2002
3. FOSTER KF, LUKASKI HC: Whole-body impedanceWhat does it
measure? Am J Clin Nutr 64:388S396S, 1996
4. LUKASKI HC: Biological indexes considered in the derivation of
the bioelectrical impedance analysis. Am J Clin Nutr 64:397S404S,
1996
5. GRIMNES S, MARTINSEN G: Bioimpedance and Bioelectricity Basics,
London, Academic Press, 2000
6. KUSHNER RF: Bioelectrical impedance analysis: A review of principles and applications. J Am Coll Nutr 11:199209, 1992
7. HOUTKOOPER LB, LOHMAN TG, GOING SB, HOWELL WH: Why bioelectrical impedance analysis should be used for estimating adiposity. Am J Clin Nutr 64 (Suppl):43648S, 1966
8. KUSHNER RF, DE VRIES PM, GUDIVAKA R: Use of bioelectrical
impedance analysis measurements in the clinical management of
patients undergoing hemodialysis. Am J Clin Nutr 64:503S509S,
1996
9. CHERTOW GM, LOWRIE EG, WILMORE DW, et al: Nutritional assessment with bioimpedance analysis in maintenance hemodialysis
patients. J Am Soc Nephrol 6:7581, 1995
10. DAUGIRDAS JT, GREENE T, DEPNER TA, et al: Anthropometrically
estimated total body water volumes are larger than modeled urea
volume in chronic hemodialysis patients: Effects of age, race, and
gender. Kidney Int 64:11081119, 2003
11. PICCOLI A, ROSSI B, PILLON L, BUCCIANTE G: A new method for
monitoring body fluid variation by bioimpedance analysis: The RXc
graph. Kidney Int 46:534539, 1994
12. PICCOLI A, NIGRELLI S, CABERLOTTO A, et al: Bivariate normal values
of the bioelectrical impedance vector in adult and elderly populations. Am J Clin Nutr 61:269270, 1995

1271

13. PICCOLI A, PILLON L, DUMLER F: Impedance vector distribution by

sex, race, body mass index, and age in the United States: Standard
reference intervals as bivariate Z scores. Nutrition 18:153167, 2002
14. PICCOLI A, FOR THE ITALIAN HD-BIA STUDY GROUP: Identification
of operational clues to dry weight prescription in hemodialysis using
bioimpedance vector analysis. Kidney Int 53:10361043, 1998
15. CHERTOW GM, JACOBS DO, LAZARUS JM, et al: Phase angle predicts
survival in hemodialysis patients. J Renal Nutr 7:204207, 1997
16. COX DR: Regression models and life tables. J Royal Stat Soc [B]
74:187220, 1972
17. COLLETT D: Modelling Survival Data in Medical Research, London,
Chapman and Hall, 1994, pp 149197
18. HARRIS EK, BOYD JC: Statistical Bases of Reference Values in Laboratory Medicine, New York, Marcel Dekker Inc., 1995
19. CHERTOW GM, LAZARUS JM, LEW NL, et al: Bioimpedance norms
for the hemodialysis population. Kidney Int 52:16171621, 1997
20. OTT M, FISCHER H, POLAT H, et al: Bioelectrical impedance analysis
as a predictor of survival in patients with human immunodeficiency
virus infection. J Acquir Immune Defic Syndr Hum Retrovirol 9:20
25, 1995
21. MAGGIORE Q, NIGRELLI S, CICCARELLI C, et al: Nutritional and prognostic correlates of bioimpedance indexes in hemodialysis patients.
Kidney Int 50:21032108, 1996
22. IKIZLER TA, WINGARD RL, HARVELL J, et al: Association of morbidity
with markers of nutrition and inflammation in chronic hemodialysis
patients: A prospective study. Kidney Int 55:19451951 1999
23. TOSO S, PICCOLI A, GUSELLA M, et al: Altered tissue electric properties in lung cancer patients as detected by bioelectric impedance
vector analysis. Nutrition 16:120124, 2000
24. PICCOLI A, PILLON L, TABB` MG: Major confounders for reactance
as a marker of malnutrition in hemodialysis patients. Kidney Int
56:23112312, 1999