Dialyzer membrane permeability and survival in hemodialysis patients

●

vha2tbwadA7acpw©

I

Dimttcroadow0phitbTrbflt(fg

A

Dialyzer Membrane Permeability and Survival inHemodialysis Patients

Philippe Chauveau, MD, Hiep Nguyen, MD, Christian Combe, MD, PhD,Geneviève Chêne, MD, PhD, Raymond Azar, MD, Noël Cano, MD, Bernard Canaud, MD,

Denis Fouque, MD, PhD, Maurice Laville, MD, Xavier Leverve, MD, PhD, Hubert Roth, Eng,Michel Aparicio, MD, and the French Study Group for Nutrition in Dialysis

Background: We previously showed that nutritional protein concentrations were predictive of outcome, whereasariables reflecting body composition and dialysis dose were not, in a 30-month prospective follow-up of 1,610emodialysis patients. Information on dialysis membrane and erythropoietin use had to be evaluated in andditional follow-up. Methods: A subset of 650 patients from the initial cohort of 1,610 was analyzed for survival in a-year extension of follow-up. Detailed data were collected: demographics; cause of renal failure; time on dialysisherapy; type of membrane; erythropoietin treatment; body mass index (BMI); predialysis albumin, prealbumin, andicarbonate levels; and outcome. Normalized protein catabolic rate (nPCR), dialysis adequacy, and lean body massere computed from predialysis and postdialysis urea and creatinine values. Results: Patient characteristics werege of 61 � 16 years, 58% men, BMI of 22.7 � 4.4 kg/m2, time on dialysis therapy of 102 � 73 months, and 8.8% hadiabetes. Dialysis parameters were duration of 247 � 31 minutes, Kt/V of 1.4 � 0.3, and nPCR of 1.2 � 0.3 g/kg/d.lbumin level was 3.73 � 0.53 g/dL (37.3 � 5.3 g/L), and prealbumin level was 31 � 8 mg/dL. The survival rate was8.7% after 2 years. Survival was influenced by age, presence of diabetes, use of high-flux membrane, and serumlbumin level, but not other variables, including Kt/V and prealbumin level. Two-year variations in values for urea,reatinine, and weight were predictive of survival in univariate, but not multivariate, analyses. Conclusion: Inatients on dialysis therapy for a long period, better survival was observed when high-flux dialysis membranesere used. Am J Kidney Dis 45:565-571.2005 by the National Kidney Foundation, Inc.

NDEX WORDS: Hemodialysis (HD); nutrition; survival; albumin; dialysis membrane.

sm

tSvBRtttG

b

o

UsPKV

pd

ESPITE MANY ADVANCES in medicalmanagement and dialysis delivery, mortal-

ty remains unacceptably high in patients onaintenance hemodialysis therapy compared with

he general population. Among the different fac-ors that can contribute to this high mortality,omorbidities and protein-energy malnutritionepresent the leading causes of death. In a previ-us prospective study of 1,610 prevalent hemodi-lysis patients, we reported that age, presence ofiabetes, and nutritional protein levels were thenly variables predictive of mortality.1 In thisell-dialyzed population (mean Kt/V, 1.34 �.34), dialysis dose had no influence on survival,resumably because of the combined effects of aigh dialysis dose and narrow range of variationn Kt/V.1 Obviously, it is particularly importanto assess the effects of variables that are likely toe corrected and thus improve patient outcome.his is especially the case for such treatment-

elated factors as erythropoietin use and mem-rane permeability. Beneficial effects of high-ux membranes on survival were not shown in

he first published results of the HemodialysisHEMO) Study, except for relative risk for deathrom cardiac cause.2 Secondary analyses sug-

merican Journal of Kidney Diseases, Vol 45, No 3 (March), 2005

is on survival in patients treated by dialysis forore than 3.7 years.3

From the Département de Néphrologie, Centre Hospi-alier Universitaire and Université Bordeaux 2; Institut deanté Publique d’épidémiologie et de Développement, Uni-ersité Bordeaux 2, Bordeaux; Service de Médecine Interne, Centre Hospitalier Général, Dunkerque; Clinique Laésidence du Parc, Marseille; Service de Néphrologie, Hôpi-

al Lapeyronie, Montpellier; Service de Néphrologie, Hôpi-al Edouard Herriot, Lyon; and Laboratoire de Bioénergé-ique Fondamentale et Appliquée, Université Joseph Fourier,renoble, France.Received June 29, 2004; accepted in revised form Novem-

er 5, 2004.Originally published online as doi:10.1053/j.ajkd.2004.11.014

n January 25, 2005.The following physicians participated in this study: Pablo

reña, Aubervilliers; Valérie de Précigout, Catherine Las-eur, Bordeaux; Jean-Noël Ottavioli, La Roche-sur-Yon;aul Stroumza, Marseille; Catherine Delcroix, Nantes; Anneolko, Lorelei Boudier, Paris; Nathalie Pertuiset, Poissy;incent Lemaître, Valenciennes.Address reprint requests to Philippe Chauveau, MD, Dé-

artement de Néphrologie, Hôpital Pellegrin, 33076 Bor-eaux, France. E-mail: [email protected]© 2005 by the National Kidney Foundation, Inc.0272-6386/05/4503-0014$30.00/0

: pp 565-571 565

oppsah

S

pttFifdpn

Jedd(ci

fm1baleaHcaam

1t

C

2dBpcwidtf

noi

S

rttgcocnw

wwtcwh(palSuI

P

hiaenasgs

istdcSetp

m

CHAUVEAU ET AL566

In the present study, we followed up a cohortf 650 prevalent patients from the previouslyublished cohort for more than 4 years.1 Werospectively studied the effects on mortality ofome patient-related factors (age, comorbidities,nd nutritional status) and practice patterns, whichad not been assessed in our previous reports.

METHODS

tudy PopulationFrom January 1996 to July 1998, 20 dialysis centers

articipated in a prospective study of the influence of nutri-ion on survival in hemodialysis patients.1 Permeability ofhe membrane used during these 2 years was not known.rom this cohort of 1,610 patients, 843 patients (all patients

n the first cohort alive in July 1998) from 11 centers wereollowed up for 2 more years. We obtained a complete set ofata, including membrane permeability, for 650 of theseatients, which allowed us to study the relation betweenutritional factors and membrane permeability and survival.Study methods were reported previously.1,4 Briefly, in

anuary 1996, a standardized questionnaire was filled out forach patient by physicians of the participating centers. At theate of the last record, patients were categorized as still onialysis therapy, lost to follow-up, or not on dialysis therapytransplantation, change of treatment, or death). Underlyingauses of death were classified into 5 categories: cancer,nfection, cardiac, vascular, and other.

Predialysis and postdialysis blood samples were collectedrom a single midweek dialysis session according to recom-ended procedures for dialysis quantification in January

996 and July 1998. Predialysis bicarbonate, albumin, preal-umin, and cholesterol and predialysis and postdialysis ureand creatinine concentrations were determined by localaboratories using conventional autoanalyzers. In July 1998,rythropoietin treatment was recorded (as a binary variable),s well as prescription of low- or high-flux membrane.igh-flux membrane is defined as an in vitro ultrafiltration

oefficient greater than 20 mL/min. �2-Microglobulin clear-nce was not determined. Residual renal function was char-cterized by the persistence of diuresis greater than 500L/d.Differences in nutritional variables between 1998 and

996 also were considered new variables to be included inhe survival analysis.

alculationsData collection started July 1, 1998, and ended June 30,

000. Duration of follow-up in days was calculated as theifference between date of the last record and July 1, 1998.ody mass index (BMI) was obtained from height andostdialysis body weight, whereas ideal body weight wasalculated using the Lorentz formula.5 Dialysis adequacyas estimated by using a single-pool Kt/V formula accord-

ng to Garred et al.6 Normalized protein catabolic rate forry body mass (nPCR) was calculated from the urea genera-ion rate.7 Estimated lean body mass (LBM) was calculated

ine generation rate, described previously.8,9 The ratio ofbserved to expected LBM (obs/exp LBM) was used as anndex of LBM.1,4

tatistical AnalysisPatients were grouped according to their outcome at last

ecord: still on hemodialysis therapy, dead, modification ofreatment type (transplantation or peritoneal dialysis), or losto follow-up. Results are given as mean � SD, except forraphs, for which 95% confidence intervals are used. Eachontinuous variable was analyzed according to categorizedutcome by using a 1-way analysis of variance. Groupomparisons were performed using Fisher’s exact test. Sig-ificance is set at P less than 0.01. Categorical variablesere compared using chi-square tests.For survival analysis, data concerning patients who under-

ent transplantation, had a modification of treatment, orere lost to follow-up were censored at the date of modifica-

ion or last available information. The relationship betweenharacteristics measured at the date of inclusion and survivalas analyzed first by means of a univariate Cox proportionalazards model. Significant variables in univariate analysiswith P � 0.25) were introduced into a multivariate Coxroportional hazards model. A final model was derived frombackwards stepwise strategy using the Wald test, with P

ess than 0.05 as a criterion to retain variables at each step.AS, version 8.0, software (SAS Institute, Cary, NC) wassed for statistical analysis. Statistica, version 6.1 (Statsoftnc, Tulsa, OK), was used for graphical analysis.

RESULTS

atient Characteristics

Six hundred fifty patients who were treated byemodialysis in July 1998 in 11 centers werencluded in the present analysis. Distribution ofge, sex, underlying nephropathy, and most rel-vant clinical and biological data (Table 1) wereot different from the French dialysis populations described in the previously published cross-ectional study of 7,123 patients reported by ourroup.4 Diabetes was reported in 9.4% of theample in 1996 and 8.8% in 1998.

Seventy-three percent of patients were treatedn in-center facilities, and 23%, in home orelf-care dialysis. This distribution is identical tohat previously reported4 and the distribution ofialysis facilities in France. Dose of dialysis wasonsistent with current recommendations.10

ixty-seven percent of patients were treated withrythropoietin. Forty-six percent of patients werereated with a high-flux membrane. Only 7% ofatients had residual diuresis.Patient characteristics according to type ofembrane used are listed in Table 2. Patients

ob

PD

TtT1awdW

sdgtwPb

daswmsbmt

S

yiat0s(b�u0awefrwmd

b

ATKnBAP

b

CLUDST

c

NUTRITION, MEMBRANE PERMEABILITY, AND SURVIVAL IN DIALYSIS 567

lder and had lower serum prealbumin levels,ut higher albumin levels.

atient Outcome: Influence of Nutrition andialysis Adequacy

Outcomes of patients in July 2000 are listed inable 3, and clinical and biological characteris-

ics of alive or dead patients are listed in Table 4.he same observations as in the previous 1996 to998 study1 were made. Nutritional parameters,lbumin level, prealbumin level, LBM, and nPCRere significantly lower in patients who dieduring follow-up compared with the other groups.eekly dialysis time, but not Kt/V, remained

Table 1. Main Clinical Characteristics of 650Hemodialyzed Patients in July 1998

Age (y) 61 � 16Men (%) 58Time on dialysis (mo) 102 � 73Kt/V urea 1.4 � 0.3nPCR (g/kg/d) 1.2 � 0.3BMI (kg/m2) 22.7 � 4.4obs/exp LBM 0.96 � 0.24Albumin (g/dL) 3.73 � 0.53Prealbumin (mg/dL) 31 � 8Primary renal disease (%)

Chronic glomerulonephritis 31Interstitial nephritis 18Diabetes 9Nephrosclerosis 16Polycystic kidney disease 10Others and unknown 16

High-flux membrane (%) 46Treatment by erythropoietin (%) 67Residual diuresis (%) 7

NOTE. To convert serum albumin in g/dL to g/L, multiplyy 10; prealbumin in mg/dL to mg/L, multiply by 10.

Table 2. Main Characteristics of Patients Accordingto Membrane Permeability

Low-Flux High-Flux P *

ge (y) 63 � 16 58 � 16 �0.001ime on dialysis (mo) 96 � 75 104 � 75 0.3t/V urea 1.4 � 0.3 1.4 � 0.3 0.8PCR (g/kg/d) 1.15 � 0.33 1.17 � 0.31 0.5MI (kg/m2) 22.4 � 4.4 23.1 � 4.4 0.07lbumin (g/dL) 3.74 � 0.51 3.63 � 0.53 0.006realbumin (mg/dL) 31 � 8 32 � 8 0.03

NOTE. To convert serum albumin in g/dL to g/L, multiplyy 10; prealbumin in mg/dL to mg/L, multiply by 10.

l*Analysis of variance between groups.

ignificantly lower in patients who subsequentlyied. Weight and BMI were not different. Hemo-lobin levels were different between groups, buthe percentage of erythropoietin-treated patientsas not significantly different (58% versus 66%;� 0.06). Sex distribution was not different

etween groups.Patients who died were more likely to be

ialyzed in in-center facilities (86%; P � 0.008),finding consistent with patient selection for

elf-dialysis or home dialysis. Patients who diedere more likely to be dialyzed using a low-fluxembrane (66% versus 55%; P � 0.0006). Re-

idual renal function was not related to mortality,ut this concerned a limited number of patients,ainly because we studied a prevalent popula-

ion with a high dialysis vintage.

urvival Analysis

Overall survival rates were 86.4% after 1ear and 79.0% after 2 years (Fig 1). In univar-ate analysis, survival was significantly associ-ted with age (P � 0.0001), presence of diabe-es (P � 0.009), type of dialysis facility (P �.002), obs/exp LBM (P � 0.0001), predialy-is urea level (P � 0.002), creatinine levelP � 0.001), albumin level (P � 0.001), preal-umin level (P � 0.001), hemoglobin level (P

0.012), dialysis session time (�240 min-tes; P � 0.01), and membrane type (P �.0007; Fig 2). A specific analysis of the inter-ction between membrane influence and centeras performed because some facilities prefer-

ntially used low-flux membranes. Facility ef-ect was not significant, but membrane typeemained significant (P � 0.0007). Survivalas not related to sex, erythropoietin treat-ent, residual renal function, BMI, weekly

ialysis time, duration of dialysis therapy, cho-

Table 3. Patient Outcome at the End of Follow-Up,July 2000

hange of facility 7 (1.1)ost to follow-up 12 (1.9)nderwent transplantation during follow-up 22 (3.4)ied 138 (21.2)till on dialysis therapy 471 (72.4)otal 650

NOTE. Values expressed as number of patients (per-ent).

esterol level, nPCR, urea removal rate, Kt/V,

pltdl�v

m

bf

hvA

b

CHAUVEAU ET AL568

redialysis bicarbonate level, or hemoglobinevel. When we studied the influence of varia-ion in parameters between 1996 and 1998,ecreases in weight (P � 0.03), serum ureaevel (P � 0.05), and serum creatinine level (P

0.036) were associated negatively with sur-ival.In the multivariate Cox proportional hazardodel, age, diabetes, albumin level, and mem-

Table 4. Clinical and Biological Characteristics

Sex ratio (M/F)Age (y) 7Time on dialysis (mo) 9Weekly hemodialysis duration (h) 11.Predialysis creatinine (mg/dL) 10.Albumin (g/dL) 3.Prealbumin (mg/dL) 2Hemoglobin (g/dL) 10.Kt/V 1.nPCR (g/kg/d) 1.0BMI (kg/m2) 22.obs/exp LBM 0.8Facility (center/self-hemodialysis) 1Diuresis (yes/no)Membrane (high-flux/low-flux)Erythropoietin treatment (yes/no) 1

NOTE. To convert creatinine in mg/dL to �mol/L, multiply 10; prealbumin in mg/dL to mg/L, multiply by 10.*Analysis of variance between groups.

rane permeability remained the only significantactors associated with survival (Table 5).

DISCUSSION

In this prospective study of 650 prevalentemodialyzed patients, we observed overall sur-ival rates of 86.4% at 1 year and 79% at 2 years.fter multivariate adjustment, age, presence of

ents Who Died or Were Alive During Follow-Up

Alive P *

292/220 0.458 � 16 0.0001

101 � 73 0.312.4 � 1.6 0.000811.2 � 3.4 0.00013.7 � 0.4 0.000132 � 1 0.001

10.8 � 1.4 0.031.4 � 0.3 0.1

2 1.18 � 0.32 0.000622.8 � 4.5 0.7

1 0.99 � 0.24 0.0001370/107 0.0237/475 0.8

255/257 0.0001334/178 0.02

.4; serum albumin and hemoglobin in g/dL to g/L, multiply

Fig 1. Kaplan-Meier plot of

of Pati

Dead

84/541 � 131 � 849 � 1.62 � 2.95 � 0.49 � 15 � 1.44 � 0.37 � 0.36 � 4.06 � 0.211/1411/12746/9204/34

y by 88

overall survival in 650 pa-tients.

dibptrdfmsdc

vo

solD(Fbapr1

1fcahswalanetva

aals6

ADAH

b

NUTRITION, MEMBRANE PERMEABILITY, AND SURVIVAL IN DIALYSIS 569

iabetes, and serum albumin level significantlynfluenced patient outcomes, as well as mem-rane permeability. Because the influence of suchatient-related factors as age, diabetes, and nutri-ional status have been discussed in previouseports from our group,1,11,12 we focus the presentiscussion on the influence of treatment-relatedactors on outcome. Contrary to membrane per-eability, the other treatment-related factors as-

essed in this study, erythropoietin treatment andialysis dose, had no influence on patient out-ome.

Although many studies documented the fa-orable impact of increased hemoglobin levelsn survival of patients treated by hemodialy-

Fig 2. Cumulative survivalccording to membrane char-cteristics: high-flux versusow-flux (log-rank test, chi-quare � 13.3; P < 0.001) in50 patients.

Table 5. Significant Variables in the Multivariate CoxProportional Hazard Model

Relative Riskper Unit

95% ConfidenceInterval P

ge (/5 y) 1.26 1.17-1.36 �0.0001iabetes (v no) 1.71 1.05-2.81 0.03lbumin (/0.1 g/dL) 0.95 0.91-0.98 0.008igh-fluxmembrane(v no)

0.62 0.43-0.91 0.01

NOTE. To convert serum albumin in g/dL to g/L, multiply

is,13-15 hemoglobin level was not predictivef patient outcome in this cohort. Hemoglobinevels were similar to those reported in theialysis Outcomes and Practice Patterns Study

DOPPS) in Europe16 (ie, relatively lower inrance than in other European countries). Weelieve that the lack of influence of degree ofnemia correction on the outcome of theseatients may be the consequence of a narrowange of variation in hemoglobin values (10.8 �.4 g/dL [108 � 14 g/L]).In the present study, single-pool Kt/V was

.4 � 0.3, similar to values previously reportedor patients treated in France.1,17 The DOPPSlearly showed that high dialysis doses in Francere common and linked to dialysis times of 4ours and more, rather than the use of high-urface dialyzers.17 As in our previous reports,e found no relationship between dialysis dose

nd patient survival. Although it is well estab-ished that a reduction in dialysis dose is associ-ted with a progressive increase in mortality, thisegative effect has been observed when deliv-red Kt/V is less than 1.2, which is considered ahreshold for adequate dialysis.1,10,18,19 Con-ersely, greater dialysis dose does not seem to bessociated with significant advantages in terms

borr

hs4cimgHgwulatmfc

flltttrwbetpmtttemc

riuesflcsd

4tHidclmraadspddthrccalbididy

ddeowmlfltslm

trirlsm

CHAUVEAU ET AL570

elieve that the lack of influence of dialysis dosebserved in this cohort mainly is caused by aelatively high delivered Kt/V and a narrowange of variation (1.4 � 0.3).

Conversely, our data clearly show that use ofigh-flux membranes is associated with betterurvival. High-flux membranes were used in6% of patients in this cohort, which should beompared with 60% of French patients includedn the DOPPS cohort.17 We define high-fluxembrane as an in vitro ultrafiltration coefficient

reater than 20 mL/h/1 mm Hg (14 mL/h in theEMO Study). We did not evaluate �2-micro-lobulin clearance, as in the HEMO study,hereas synthetic or cellulosic membranes weresed in this study. It must be stressed that dia-yzer reuse is forbidden by law in France, as forny other single-use medical device. Therefore,he survival advantage observed with high-fluxembranes results from their intrinsic effect, not

rom any other maneuver, such as technical pro-edures needed by reuse.

Compared with low-flux dialyzers, use of high-ux membranes results in better clearance of

arger molecules that can be considered putativeoxins acting as enzymatic and metabolic inhibi-ors. Lower mortality risk has been reported withhe use of high-flux membrane dialyzers, whereaseuse with bleach was independently associatedith better survival, 2 arguments in favor of aeneficial effect of better clearance of large mol-cules.23 It has been reported that in addition toheir acknowledged effects on the reduction inrevalence of carpal tunnel syndrome, high-fluxembranes could produce significant benefits in

erms of lipoprotein profiles, left ventricular sys-olic function, inflammatory response, and nutri-ional status.24 However, despite these positiveffects, the influence of the use of high-fluxembranes on patient survival is not unequivo-

al and remains debated.Numerous retrospective observational studies

eported significant improvement (19% to 76%)n survival rates when high-flux membranes weresed compared with low-flux dialysis.25,26 How-ver, Locatelli et al,27 in a 2-year prospectivetudy of 380 patients randomly assigned to low-ux hemodialysis with biocompatible or bioin-ompatible membranes and high-flux hemodialy-is or hemodiafiltration, did not find significant

treatment groups. A potential explanation forhese different results may be provided by theEMO Study, which was designed to determine,

n 1,846 prevalent patients, whether increase inialysis dose or use of a high-flux membraneould improve survival or morbidity in hemodia-yzed patients. A primary analysis did not show aajor benefit, but a reduction in risk for cardiac-

elated events was observed in patients randomlyssigned to high-flux membranes.2 In secondarynalyses devoted to the effects of high-flux hemo-ialysis on clinical outcomes,28 patients weretratified into 2 groups to study the influence ofrior dialysis duration on effects of high-fluxialysis on clinical outcomes. In the long-uration group (8.6 � 4.9 years on dialysisherapy before randomization), patients using aigh-flux membrane had a significantly lowerisk for all-cause mortality (relative risk, 0.68),ardiac death (relative risk, 0.63), and first de-line of more than 15% from baseline in serumlbumin level (relative risk, 0.74) compared withow-flux dialysis. No significant effect of mem-rane permeability on these events was observedn the short-duration group (1.5 � 1.0 years onialysis therapy before randomization), and thenvestigators concluded “a benefit of high-fluxialysis is strongest for patients with severalears of prior dialysis.”Our findings closely match those of the long-

uration group of the HEMO Study.3 Time onialysis therapy was similar, and a beneficialffect of high-flux hemodialysis on survival wasbserved in both groups. Serum albumin levelas preserved better in patients using high-fluxembranes, but in our study, serum albumin

evel and high-flux membrane independently in-uenced patient outcome. It must be stressed that

here was no reuse of dialyzers in this study; theurvival advantage of high-flux dialyzers is re-ated to their intrinsic properties, not to any otheraneuvers, such as reuse.In conclusion, it can be proposed that during

he first years of dialysis treatment, the survivalate depends mostly on age, comorbidities, andnitial nutritional status,1,29,30 whereas treatment-elated factors have an important role in theong-term survival of hemodialyzed patients. Re-ults of our study support the use of high-fluxembranes in selected patients to improve sur-

ival.

nvS

ds

hH

st

NS

Wb

K2

kmN

itT

tA

i71

hi

ota

Aa1

tS

ht

aT

HcOD

Jtd

h1

Ds

h5

Lu1

tms2

Ip1

cct

hN

otC1

dtN

cL

i

NUTRITION, MEMBRANE PERMEABILITY, AND SURVIVAL IN DIALYSIS 571

REFERENCES1. Combe C, Chauveau P, Laville M, et al: Influence of

utritional factors and hemodialysis adequacy on the sur-ival of 1,610 French patients. Am J Kidney Dis 37:S81-88, 2001 (suppl 2)2. Eknoyan G, Beck GJ, Cheung AK, et al: Effect of

ialysis dose and membrane flux in maintenance hemodialy-is. N Engl J Med 347:2010-2019, 2002

3. Cheung AK, Levin NW, Greene T, et al: Effects ofigh-flux hemodialysis on clinical outcomes: Results of theEMO Study. J Am Soc Nephrol 14:3251-3263, 20034. Aparicio M, Cano N, Chauveau P, et al: Nutritional

tatus of haemodialysis patients: A French national coopera-ive study. Nephrol Dial Transplant 14:1679-1686, 1999

5. Leverve X, Cosnes J, Erny P, Hasselmann M: Traité deutrition Artificielle de l’Adulte (ed 2). Paris, France,pringer Verlag, 20016. Garred LJ, Barichello DL, DiGiuseppe B, McCreadyG, Canaud BC: Simple Kt/V formulas based on urea mass

alance theory. ASAIO J 40:997-1004, 19947. Daugirdas JT: Simplified equations for monitoring

t/V, PCRn, eKt/V, and ePCRn. Adv Ren Replace Ther:295-304, 19958. Canaud B, Garred LJ, Argiles A, et al: Creatinine

inetic modelling: A simple and reliable tool for the assess-ent of protein nutritional status in haemodialysis patients.ephrol Dial Transplant 10:1405-1410, 19959. Desmeules S, Levesque R, Jaussent I, et al: Creatinine

ndex and lean body mass are excellent predictors of long-erm survival in haemodiafiltration patients. Nephrol Dialransplant 19:1182-1189, 200410. National Kidney Foundation: K/DOQI Clinical Prac-

ice Guidelines for Hemodialysis Adequacy: Update 2000.m J Kidney Dis 37:S7-S64, 2001 (suppl 1)11. Chauveau P, Combe C, Laville M, et al: Factors

nfluencing survival in hemodialysis patients aged older than5 years: 2.5-Year outcome study. Am J Kidney Dis 37:997-003, 200112. Cano NJ, Roth H, Aparicio M, et al: Malnutrition in

emodialysis diabetic patients: Evaluation and prognosticnfluence. Kidney Int 62:593-601, 2002

13. Besarab A, Bolton WK, Browne JK, et al: The effectsf normal as compared with low hematocrit values in pa-ients with cardiac disease who are receiving hemodialysisnd epoetin. N Engl J Med 339:584-590, 1998

14. Xue JL, St Peter WL, Ebben JP, Everson SE, CollinsJ: Anemia treatment in the pre-ESRD period and associ-

ted mortality in elderly patients. Am J Kidney Dis 40:1153-161, 2002

15. Murphy ST, Parfrey PS: The impact of anemia correc-ion on cardiovascular disease in end-stage renal disease.emin Nephrol 20:350-355, 200016. Locatelli F, Pisoni RL, Combe C, et al: Anaemia in

aemodialysis patients of five European countries: Associa-

nd Practice Patterns Study (DOPPS). Nephrol Dialransplant 19:121-132, 200417. Hecking E, Bragg-Gresham JL, Rayner HC, et al:

aemodialysis prescription, adherence and nutritional indi-ators in five European countries: Results from the Dialysisutcomes and Practice Patterns Study (DOPPS). Nephrolial Transplant 19:100-107, 200418. Owen WF Jr, Lew NL, Liu Y, Lowrie EG, Lazarus

M: The urea reduction ratio and serum albumin concentra-ion as predictors of mortality in patients undergoing hemo-ialysis. N Engl J Med 329:1001-1006, 199319. Held PJ, Port FK, Wolfe RA, et al: The dose of

emodialysis and patient mortality. Kidney Int 50:550-556,99620. Owen WF Jr, Chertow GM, Lazarus JM, Lowrie EG:

ose of hemodialysis and survival: Differences by race andex. JAMA 280:1764-1768, 1998

21. Szczech LA, Lowrie EG, Li Z, et al: Changingemodialysis thresholds for optimal survival. Kidney Int9:738-745, 200122. Chertow GM, Owen WF, Lazarus JM, Lew NL,

owrie EG: Exploring the reverse J-shaped curve betweenrea reduction ratio and mortality. Kidney Int 56:1872-1878,99923. Port FK, Wolfe RA, Hulbert-Shearon TE, et al: Mor-

ality risk by hemodialyzer reuse practice and dialyzerembrane characteristics: Results from the USRDS Dialy-

is Morbidity and Mortality Study. Am J Kidney Dis 37:276-86, 200124. de Précigout V, Higueret D, Larroumet N, et al:

mprovement in lipid profiles and triglyceride removal inatients on polyamide membrane hemodialysis. Blood Purif4:170-176, 199625. Koda Y, Nishi S, Miyazaki S, et al: Switch from

onventional to high-flux membrane reduces the risk ofarpal tunnel syndrome and mortality of hemodialysis pa-ients. Kidney Int 52:1096-1101, 1997

26. Woods HF, Nandakumar M: Improved outcome foraemodialysis patients treated with high-flux membranes.ephrol Dial Transplant 15:S36-S42, 2000 (suppl 1)27. Locatelli F, Mastrangelo F, Redaelli B, et al: Effects

f different membranes and dialysis technologies on patientreatment tolerance and nutritional parameters. The Italianooperative Dialysis Study Group. Kidney Int 50:1293-302, 199628. Allon M, Depner TA, Radeva M, et al: Impact of

ialysis dose and membrane on infection-related hospitaliza-ion and death: Results of the HEMO Study. J Am Socephrol 14:1863-1870, 200329. Khan IH, Catto GR, Edward N, et al: Influence of

oexisting disease on survival on renal-replacement therapy.ancet 341:415-418, 199330. Hakim RM, Fearon DT, Lazarus JM: Biocompatibil-

ty of dialysis membranes: Effects of chronic complement