198 Volume 7 . Number 2 . 1996

Adequacy of Dialysis and Nutrition in ContinuousPeritoneal Dialysis: Association with Clinical Outcomes1CANADA-USA (CANUSA) Peritoneal Dialysis Study Group2’3

Principal Investigators: David N. Churchill, D. WayneTaylor, Prakash R. Keshaviah

Coordinating Center Investigators: Kevin E. Thorpe,Mary Louise Beecroft

Clinical Center Investigators: Kailash K. Jindal. Stanley

S.A. Fenton, Joanne M. Bargman, Dimitrios G.Oreopoulos. David N. Churchill, George G. Wu, SusanD. Lavoie, Adrian Fine, Ellen Burgess. James C.Brandes, Karl D. Nolph, Barbara F. Prowant. DenisPage, Francis X. McCusker, Brendan P. Teehan, MrinalK. Dasgupta, Kelvin Bettcher, Ralph Caruana

Baxter Healthcare Investigators: George deVeber,

Lee W. Henderson

(J, Am. Soc. Nephrol. 1996; 7:198-207)

ABSTRACTThe objective of the study presented here was toevaluate the relationship of adequacy of dialysis andnutritional status to mortality, technique failure, andmorbidity. This was a prospective cohort study ofconsecutive patients commencing continuous perito-neal dialysis in 14 centers in Canada and the UnitedStates. Between September 1 , 1990 and December31,1992, 680 patients were enrolled. Follow-up was termi-

nated December 31 , 1993. There were 90 deaths, 137transplants, and 1 18 technique failures. Fifteen with-drew from dialysis. Analysis of the patient and tech-nique survival used the Cox proportional hazardsmodel with adequacy of dialysis and nutritional statusas time-dependent covariates. The relative risk (PR) ofdeath increased with increased age, insulin-depen-dent diabetes mellitus, cardiovascular disease, de-

creased serum albumin concentration and worsenednutritional status (subjective global assessment andpercentage lean body mass). A decrease of 0. 1 unitKt/V per week was associated with a 5% increase inthe RR of death; a decrease of 5 L/1 .73 m2 creatinineclearance (CCr) per week was associated with a 7%Increase In the RR of death. The RR oftechnique failure

1 Received May 17, 1995. Accepted November 21, 1995.

2 � Appendix for participating invesligators and affiliated organizations.

3 correspondence to Dr. D. churchill, McMaster University, St. Joseph’s Hospital,50 charlton Avenue East, Hamilton, Ontario, canada L8N 4A6.

104&6673/0702.0198$03.00/0Journal of the American Society of Nephrologycopyright © 1996 by the American Society of Nephrology

was increased with decreased albumin concentra-tion and decreased CCr. Hospitalization was in-creased with decreased serum albumin concentra-tion, worsened nutrition according to subjectiveglobal assessment and decreased CCr. A weeklyKt/V of 2. 1 and a weekly CCr of 70 L/ 1.73 m2 wereeach associated with an expected 2-yr survival of78%.

Key Words: Clinical trial, urea, creatinine, multivariate, pa-

tient survival

C ontinuous ambulatory peritoneal dialysis(CAPD) ( 1 ,2) is the treatment used for approxi-

mately 1 4% of the world’s dialysis population (3). Withadjustment for comorbidity, Burton and Walls (4) andthe Canadian Organ Replacement Register (5) re-ported no difference between CAPD and hemodialysiswith respect to patient survival. However, Maiorca et

al. (6) found better survival for patients aged 53.5 yr

and greater treated with CAPD while Held et a!. (7)

reported an increased risk of death for older diabeticpatients treated with CAPD.

For patients treated with hemodialysis. longer dial-ysis times and better nutritional status are associated

with better patient survival (8,9). Increased removal ofsmall molecular weight sobutes is associated withdecreased morbidity ( 10). For patients treated withCAPD, some data suggest an association betweenimproved survival and better nutritional status( 1 1 , 1 2). There are conflicting reports regarding therelationship between estimates of adequacy of dialysisand clinical outcomes ( 13-18). The methodobogic lim-itations of these studies include small sample size

with inadequate statistical power, insensitive clinicaloutcomes, and use of univariate rather than multiva-nate statistical analysis (19).

In this article, we report the results of a multicenterprospective cohort study of 680 patients commencingcontinuous peritoneal dialysis in 14 centers in Can-ada and the United States between September 1 . 1990and December 3 1 , 1992, with follow-up until Decem-ber 3 1 , 1993. The objectives of the study were: (1 ) toevaluate the association of adequacy of dialysis withmortality. technique failure and hospitalization; and(2) to evaluate the association of nutritional statuswith those same variables.

METHODS

The clinical centers participating in the study, listed inorder of number of patients entered, were Victoria General

Churchill et al

Journal of the American Society of Nephrology 199

Hospital (Halifax, Nova Scotia); Toronto Western and Toronto

General Divisions ofThe Toronto Hospital (Toronto, Ontario);St. Joseph’s Hospital (Hamilton. Ontario): Credit Valley Hos-pital (Mississaugua, Ontario); Ottawa Civic Hospital (Ottawa.Ontario); St. Boniface Hospital (Winnipeg. Manitoba): Foot-hills Hospital (Calgary, Alberta); Medical College of Wiscon-sin (Milwaukee, Wisconsin); University of Missouri MedicalCenter (Columbia, Missouri); Ottawa General Hospital (Otta-wa, Ontario): Lankenau Hospital (Philadelphia, Pennsylva-nia): University of Alberta Hospital (Edmonton. Alberta):Medical College of Georgia (Augusta, Georgia). The Coordi-nating Center was the Department of Clinical Epidemiologyand Biostatistics, McMaster University, Hamilton, Ontario.

All patients commencing continuous peritoneal dialysis forthe first time between September 1 , 1990 and December 31,1992 were eligible for the study. Exclusion criteria were: (1)

being unlikely to survive for at least 6 months; (2) electiveliving donor kidney transplant within 6 months; (3) plannedmove from the study center within 6 months; (4) positivehepatitis B or HIV serology; (5) active systemic inflammatory

disease; or (6) failure to sign the informed consent form. Thedialysis prescription was that prescribed by the individualpatient’s physician. Changes were made for clinical indica-tions and without reference to the data recorded at theCoordinating Center.

The demographic data recorded at enrollment includedage. sex, race, functional status according to the Karnofskyscore (20), underlying renal disease, insulin-dependent dia-betes mellitus (IDDM). and history of cardiovascular disease(CVD). CVD was defined as a history of previous myocardialinfarction, angina, amputation for vascular disease, or ClassIII through IV congestive heart failure.

Estimates of nutritional status were obtained at enroll-ment and every 6 months thereafter. If there was an acutemedical problem. this estimate was obtained 1 month afterresolution ofthe problem. Serum albumin concentration wasdetermined by the bromcresol green method. The subjectiveglobal assessment (SGA) of nutritional status was deter-mined by using a modification of the method described byBaker et a!. and Detsky et al. (2 1 .22). This technique isreliable and valid for ESRD patients treated with CAPD (23).

Four items were scored on a seven-point Likard-type scale,with lower values representing worse nutritional status.These values were weight loss during the past 6 months,anorexia, subcutaneous fat, and muscle mass. These fouritems were given subjective weights to produce a globalassessment. Scores of 1 to 2 represented severe malnutri-tion; 3 to 5, moderate to mild malnutrition; and 6 to 7,normal nutrition. The protein catabolic rate (PCR) was deter-mined according to Randerson et at. (24) and normalized tostandard body weight (total body water/0.58). Total bodywater (V) was determined from the formula of Watson et at.(25). Percentage of lean body mass 1% LBM] was determinedfrom creatinine kinetics (26).

Adequacy of dialysis was estimated by measurement oftotal weekly Kt/V for urea, total weekly creatinine clearance(CCr), and serum beta-2-microglobulin (f32M]. Peritoneal Ktwas estimated from 24-h dialysate urea excretion and theserum urea concentration. Renal Kt was estimated from theconcurrent 24-h urine urea excretion. Peritoneal creatinineclearance was estimated from the 24-h dialysate creatinineexcretion and the serum creatinine concentration at thecompletion of the collection. Dialysate creatinine excretionwas corrected for glucose Interference by using a validatedcorrection factor (27) determined for each clinical chemistrylaboratory. The renal contribution to creatinine clearance

was estimated as the average of renal creatinine and ureaclearance (CCr + Curea)/2 (28). The serum f32M estimationswere performed by a solid-phase time-resolved fluoroimmu-noassay (DELFIA; Wallace Canada. Vaudreuil, Quebec) intwo reference laboratories.

The clinical outcomes were mortality, technique failure,and hospitalization. Technique failure was defined as trans-fer to hemodialysis or to conventional intermittent peritonealdialysis. Hospitalizations for vascular access surgery. renalallograft nephrectomy, elective pretransplant assessment,and renal transplantation were excluded from the analysis ofthe effect of adequacy of dialysis on hospitalization. Theduration of follow-up and total number of days hospitalizedwere used to calculate days hospitalized per month of follow-up.

Statistical analysis of patient mortality and techniquefailure used Andersen and Gill’s (29) extension to the Coxproportional hazards model (30) with estimates of nutritionalstatus and adequacy of dialysis as time-dependent covari-ates (31,32). Events (e.g., death or technique failure) wereattributed to the level of nutrition or adequacy of dialysisrecorded at the 6-month evaluation preceding the event.Transplantation, recovery of renal function, technique fail-ure and loss to follow-up were censored observations for thepatient survival analysis. Death, transplantation. recovery ofrenal function, and loss to follow-up were censored observa-tions for the technique survival analysis. All baseline demo-graphic and clinical variables and baseline serum albuminconcentrations were entered. Backward elimination removednonsignificant variables. The three estimates of adequacy ofdialysis (Kt/V. CCr, and (32M) were each added separately tothis model. Three estimates of nutritional status (NPCR.SGA, and % LBM) were each added separately to a modelcontaining the significant baseline demographic variables,serum albumin concentration, and one estimate of adequacyof dialysis. The likelihood ratio test (3 1 .33) was used todetermine whether or not the addition of a variable to a modeladded significantly to that model. The validity of the propor-tional hazards assumption was considered for all variablesremaining in the final models by examining the Schoenfeldresiduals (34).

To further explore the effect of Kt/V and CCr on patientsurvival, the expected survival time was determined forseveral theoretical Kt/V and CCr values. Survival curveswere constructed from the fitted models (3 1 ) for weekly Kt/Vvalues of 2.3, 2.1, 1.9, 1.7, and 1.5, and for weekly CCrvalues of 95, 80, 70, 55, and 40 L/wk per 1 .73 m2. Thesesurvival curves assume no change in the Kt/V and CCrvalues over follow-up time. Peritoneal and renal clearancewere considered equivalent with decreases in renal functioncompensated by Increased peritoneal clearance.

A log-linear model was used to analyze the hospitalizationdata (35). The dependent variable was the number of dayshospitalized per month of follow-up. The baseline explana-tory variables were: country (Canada/United States); age atenrollment; sex: race (Caucasian / non-Caucasian); Karnof-sky score (<80 or �80): renal disease: IDDM; and CVD. Forexplanatory variables estimated at 6-month intervals duringthe study. the weighted average was used. These variableswere serum albumin concentration, SGA. NPCR. % LBM,Kt/V, CCr, and serum f32M.

The sample-size calculation was made on the basis of thepatient survival analysis. For the multivariate analysis. ap-proximately ten events (death) per independent variable arerequired to produce a model of reasonable accuracy (36).Given the ten independent variables to be evaluated, the

Adequacy and Nutrition in CAPD

200 Volume 7 . Number 2 ‘ 1996

study population had to be large enough to experience about100 deaths during the follow-up period. The patient enroll-ment target was 700 patients, with follow-up extended untilthe requisite number of events was accumulated.

RESULTS

There were 680 patients enrolled in the study. Therewere 97.9% treated with CAPD and 2. 1% with contin-uous cycling peritoneal dialysis (CCPD). The meanand median daily prescribed instilled volumes were7.7 and 8.0 L, respectively. The mean age was 54.3 yr.with a range of 18 to 82 yr. The distribution ofdemographic and clinical factors is shown In Table 1.The mean height was 1 .66 m, mean weight was 67.8kg, and mean body mass index was 24.6 kg per m2.The baseline values for adequacy of dialysis and nu-tritional status are shown in Table 2. By SGA assess-ment, there were 30 patients (4.2%) with severe mal-

nutrition, 364 (5 1 .2%) with mild to moderate

malnutrition, and 3 1 7 (44.6%) considered well flour-

ished.

TABLE 1 . Demographic and clinical factorsa

Factor N (%)

RaceCaucasian 558 82.1African American 57 8.4Asian 38 5.6Aboriginal 20 2.9Other 7 1.0

Education<High school 315 46.3High school 187 27.5Technical 79 11.6College/university 99 14.6

Karnofsky Score80-100 399 58.750-70 268 39.4<50 13 1.9

GenderMale 394 57.9Female 286 42.1

Renal DiseaseDiabetes 202 29.7Glomerulonephritis 160 23.5Nephrosclerosis 95 14.0Renal vascular disease 42 6.2Polycystic kidney 47 6.9diseaseOther 134 19.7

ComorbidityIDDM 151 22.2CHF(111 orlV) 156 22.9Ml 99 14.6Angina 159 23.4Amputation 19 2.8

0 IDDM, insulin-dependent diabetes m ellitus; CHF. congestive heartfailure; Ml, myocardlal infarction.

TABLE 2. Adequacy and nutritional status(baseline)a

Mean.

Median1st

Quartile3rd

.Quartile

Kt/V (weekly) 2.38 2.28 1.96 2.68Kt/V renal 0.71 0.64 0.30 0.99Kt/V peritoneal 1.67 1.61 1.37 1.89

CCr (1/week per 83.0 78.8 63.8 98.01.73 m2)

CCr renal 38.8 34.6 18.2 53.2CCr peritoneal 44.2 43.6 38.5 50.1

Serum �32M 22.5 21.0 16.0 27.0(mg/I)

Serum albumin 34.9 35.0 31.0 39.0

(g/L)NPCR (g/kg) 1.04 1.01 0.87 1.19SGA (1-7) 5.19 5.00 4.00 6.00% IBM 62.6 60.9 53.3 71.0

a Kt/V. for urea; OCr. creatinine clearance rate; NPCR, normalizedprotein catabolic rate; SGA, subjective global assessment; LBM, leanbody mass.

TABLE 3. Weekly mean Kt/V, CCr, and estimates ofnutritional status

Follow-Up (Months) 0 6 12 18 24

N 680 525 321 166 78

Kt/V 2.38 2.25 2.10 2.02 1.99aPeritoneal 1.67 1.67 1 .68 1.66 1.70Renal 0.71 0.58 0.41 0.39 0.28#{176}

CCr (1.73 m2) 83.0 74.7 68.3 65.7 61.6#{176}Peritoneal 44.2 44.6 46.4 45.3 473b

Renal 38.8 30.1 21.9 20.4 14.3#{176}

Nutrition

Serum albumin 34.9 35.1 35.1 35.1 35.2(g/L)NPCR (g/Kg/day) 1.04 1.03 0.97 0.96 0.99

SGA (0-7) 5.19 5.96 6.00 6.00 601b

% IBM 62.6 63.3 63.5 63.1 65.3

a p < o.ooooi . baseline versus 24 months (paired t test).b p < 0.005. baseline versus 24 months (paired t test).

The changes In adequacy of dialysis and nutritionalstatus are shown In Table 3. The decrease in patientnumbers is partly because of death, transfer to hemo-dialysis, and transplantation. but also reflects the

staggered entry of patients over the study period. For

total weekly Kt/V, there is a progressive decrease from2.38 to 1 .99. This decrease is entirely the result of lossof the renal component. which decreases from 0.71 to0.28 whereas the peritoneal component remains sta-

ble at 1 .67 to 1 .70. For total weekly CCr, there is aprogressive decrease from 83.0 to 6 1 .6 L/ 1 .73 m2. As

for Kt/V, this is entirely the result of loss of the renalcomponent, which decreases from 38.8 to 14.3whereas the peritoneal CCr increases slightly from44.2 to 47.3 L/wk per 1 .73 m2. The serum j32M

0 6 12 18 24

Time in Months

Churchill et al

Journal of the American Society of Nephrology 201

Increased from 22.5 to 24.3 mg/L. There is a signifi-

cant increase in SGA, but not for serum albuminconcentration, NPCR, or % LBM.

During the follow-up period of 1 0, 1 38 patient-months, there were 90 deaths, of which 68 were

classified as cardiovascular, five as infection, four as

malignancy, and 1 3 as various causes. There were

137 transplants and 1 18 transfers ( 1 1 1 to hemodial-

ysis and seven to IPD). Among the remaining 335patients, 264 were alive on continuous peritonealdialysis on December 3 1 , 1 993 and were administra-tiveby censored at that time. Follow-up had been dis-

continued on 7 1 other patients for the following rea-sons: noncompliance with dialysate collection (22

patients); elective withdrawal from dialysis ( 15 pa-tients); move to a non-study dialysis center ( 13 pa-tients); loss to follow-up ( 13 patients); and recovery ofrenal function (8 patients). The Kaplan-Meier esti-

mates of patient survival, technique survival, and theprobability of being alive and on CAPD/CCPD areshown In Figure 1 . The probability of patient survivalto 24 months is 78%; of technique survival, 75%; andof being alive and receiving CAPD treatment, 58%.

Mortality

The final Cox proportional hazards model resultsare summarized in Tables 4 and 5. In Table 4, Kt/V isused as the estimate of adequacy, whereas CCr isused in Table 5. The Schoenfeld residuals were plottedagainst time for each variable and showed no serious

violation of the proportional hazards assumption.

Furthermore, Harrell’s z test detected no violations ofthe assumption (37,38).

The relative risk (RR) of death for each variable wasslightly different for the two models (Kt/V and CCr).

Accordingly, where different, both values are pro-vided. The SGA has been used as the estimate ofnutritional status rather than NPCR or % LBM be-cause it was considered more credible than estimateswith statistical coupling with adequacy. The PR of

0)C

>

C/)C

C.)

a.

Figure 1 . Probabilities of patient and technique survival.

TABLE 4. Cox proportional hazards model#{176}

VariableRelativeMortality

Risk

95%Confidence

Interval

Age(peryear) 1.03 1.01-1.05IDDM 1.45 0.89-2.36CVDb 2.09 1.33-3.28

Country (USA) 1.93 1.14-3.28Serum albumin ( I 1 g/L)Kt/V ( I 0.1 units/wk)SGA ( f 1 unit)

0.940.940.75

0.90- 0.970.90-0.990.66-0.85

a Kt/V as estimate of adequacy of dialysis and SGA as the estimate ofnutritional status.b CVD, cardiovascular disease.

TABLE 5. Cox proportional hazards model#{176}

VariableRelativeMortality

Risk

95%Confidence

Interval

Age(peryear) 1.03 1.01-1.05IDDM 1.49 0.92-2.42CVD 2.12 1.35-3.34Country(USA) 1.95 1.14-3.31Serum albumin ( f 1 g/1) 0.94 0.90-0.97CCr ( f 5 L/wk/1 .73 m2) 0.93 0.88- 0.98SGA ( I 1 unIt) 0.75 0.66-0.85

a OCr as the estimate of adequacy of dialysis and SGA as theestimate of nutritional status.

TABLE 6. Expected 2-Year patient survival accordingto sustained weekly Kt/V and CCr (1/1.73 m2)

Kt/V Survival % CCr Survival %

2.3 81 95 862.1 78 80 811.9 74 70 781.7 71 55 721.5 66 40 65

death is 1 .03 greater for a 1 -yr difference in age (I.e.,

for two patients identical for all other risk factors.

being 1 yr older increases the PR ofdeath by 3%). TheRR for those with IDDM was 1 .45 to 1 .49 whereasthose with a history ofCVD had an RR of2.09 to 2.12.Patients in the United States had a RR ofdeath 1 .93 to1 .95 greater than those in Canada. A 1 g/L-lowerserum albumin concentration was associated with an6% increase in the RR of death. A 1 -U lower SGA scorewas associated with a 25% increase In the RR of death.

For two patients who differed only in weekly Kt/V, a0. 1 lower value was associated with a 6% increase inthe PR of death. For a 5 L/wk per 1 .73 m2 lower CCr,there was a 7% increase In the RR of death. Inseparate models, 1%-bower % LBM and a 1 mg/L-greater serum �2M were associated with 3% and 2%increases in the RR of death, respectively.

8

C

CO

C

a.

8

0N.

0 6 12 18 24

Time in Months

0)C

>

Cl)

C

8a.

0 6 12 18

Time in Months

Figure 2. PredIcted probability of patient survival by Kt/V.

Adequacy and Nutrition in CAPD

202 Volume 7 . Number 2 . 1996

The expected patient survival was estimated forseveral sustained weekly Kt/V and CCr values. Forweekly Kt/Vvalues of2.3, 2.1, 1.9, 1.7, and 1.5, the

expected 2-yr survivals were 81%, 78%, 74%, 71%,

and 66%. For weekly CCr values of95, 80, 70, 55, and

40 L/ 1 732 the expected 2-yr survivals were 86%,8 1 % . 78% , 72% , and 65% . These data are shown inTable 6 and Figures 2 and 3.

Technique Failure

There were 1 18 transfers to hemodialysis or IPD.There were 36 transfers because of peritonitis or

exit-site infection, and five because of catheter mal-function. There were 19 transfers attributed to mad-equate dialysis ( 1 2 transfers because of inadequate

clearance and 7 because of failure of ultrafIltration).

Patient and partner fatigue accounted for two trans-

fers each; social causes were responsible for eight

failures. The remaining 46 were considered to be theresults of “other medical” causes. Among the baselinedemographic and clinical variables, a decreased se-rum albumin concentration was associated with anincreased risk of technique failure. The RR associated

with a 1 g/L increase was 0.95 (95% CI, 0.92 to 0.98).

CCr was the only estimate of adequacy which addedsignificantly to this model. The RR associated with a 5

L/wk per 1 .73 m2 increase was 0.95 (95% CI, 0.9 1 to0.99). None of the other estimates of nutrition was

associated with technique failure. The PR of techniquefailure increased by 5% with a 1 g/L decrease Inserum albumin concentration and with a 5 L/wk

decrease in CCr.

Hospitalization

There were 1 239 hospitalizations; the causes werecardiovascular disease ( 189, 15.3%); peritonitis (285,23.0%); gastrointestinal disease ( 120, 9.7%); other

infections (78, 6.3%). No other category accounted formore than 5%. There are more days hospitalizedassociated with increased age, female sex, worse func-

Figure 3. PredIcted probability of patient survival by CCr.

tional status. CVD, IDDM, and bower serum albuminconcentration. Malnutrition, as estimated by the SGA,had a strong association with hospitalization. Amongthe estimates of adequacy of dialysis, increased CCr

and decreased (32M were associated with fewer hospi-

tal days, whereas Kt/V was less strongly associated(Table 7).

About 50% of the patients were not hospitalized.The data were highly skewed and Inappropriate forusual linear regression methods. Log-linear modelingwas used. Among the three estimates of adequacy of

dialysis. CCr and �2M were significant; Kt/V was not.

Among the three estimates of nutritional status, SGAand % LBM were significant; NPCR was not. The final

model, using SGA as the nutrition estimate and CCr

as the adequacy estimate, is shown in Table 8. All

variables were statistically significant (P < 0.05). The

exponential coefficient is an expression of the relative

length of time hospitalized compared with the alterna-tive within the variable. For country, the value 1.23indicates that patients in centers in the United States

were hospitalized on average 1 .23 times as bong as acomparable patient in Canada. Each additional yearof age translates into 1 .02 times the time spent inhospital. Those with Karnofsky scores <80 spent 1.63times the days hospitalized compared with those withscores �80. Patients with IDDM and CVD spent 1.47and 1 .22 as many days hospitalized as those withoutIDDM and CVD, respectively. For SGA, a 1 -U increase

was associated with 0.82 times as many days hospi-talized as those with a score 1 U bower. For serum

/32M, a value 1 mg/L greater was associated with 1.02

times as many days hospitalized. A 5 L per weekgreater CCr was associated with a 0.99 times as manydays hospitalized.

DISCUSSION24 In a prospective cohort study of patients treated

with continuous peritoneal dialysis, the association ofnutrition and adequacy of dialysis with clinical out-

TABLE 7. Days hospitalized per year

AbsentPresent

a The e�0& Is the multiplicative factor Indicating the relative timehospitalized compared to the alternative within the variable (e.g.. forUSA. the alternative is Canada). For continuous variables, the alter-natives are 1 year younger for age. 1 gIL less for serum albuminconcentration, 1 unit less for SGA. and 5 L/wk per 1 .73 m2 less for CCr.b For ESRD, the alternative is diabetes.

18.6

21.7

10.118.7

29.0

22.716.7

11.3

30.2

15.226.2

15.531.7

7.312.723.635.6

13.3

24.871.0

14.417.9

27.8

18.118.7

25.8

26.219.714.6

8

0

0)C

3Cl)

CG) 0

a.

0

0

- canada----. USA

Number At RiskCanada 491 343 189USA 81 59 26 10

0 18 24

Churchill et al

Journal of the American Society of Nephrology 203

Factor Days

CountryCanadaUSA

Age (yr)<4545-64

>64

SexFemalesMales

Karnofsky�80

<80

CVD

IDDMAbsentPresent

Serum albumin (g/L)>4035-4030-34<30

Malnutrition (SGA)NoneMild-moderateSevere

CCr (L/wk/ 1.73 m2)>8060-80<60

Kt/V (wk)>2.11.7-2.1<1.7

f32M (mg/L)>3020-30<20

comes has been evaluated by using multivariate sta-

tistical techniques to control for baseline demographic

and clinical variables. Better nutrition was associatedwith better patient survival and fewer days hospital-ized. A higher “dose” of dialysis. inclusive of residualrenal function, was associated with better patient

survival, better technique survival, and fewer dayshospitalized.

The increased RR of death associated with increasedage, IDDM, and history of cardiovascular disease wasexpected. The increased RR of death among patients

in the centers in the United States (RR 1 .93 to 1.95)was unexpected. The Kapban-Meier patient survival

curves (Figure 4) show a 2-yr survival probability of

TABLE 8. Multiplicative factor on days hospitalizedper month

.Variable COEF#{176}e

Country (USA) 1.23Age (1 year) 1.02Karnofsky (<80) 1.63ESRDb

(glomerulonephritis) 0.58(hypertension-RVD) 0.80(other) 0.73

IDDM (present) 1.47CVD (present) 1.22Serum albumin (1 g/L) 0.95SGA (1 unit) 0.82CCr (5 L/wk per 1.73 m2) 0.99

6 12

Time in Months

Figure 4. Predicted probability of patient survival by country.

80% for those in the Canadian centers compared with

63% in the U.S. centers. The mean ages were 54.4 and53.4 yr in Canada and the United States, respectively.The serum albumin concentration was 34.9 g/L inboth countries. Demographic and clinical character-istics of the patients are compared In Table 9. Statis-

tically significant differences were present for race,

body surface area, Kt/V (total and peritoneal), andCCr (peritoneaJ). However, all of these variables werecontrolled in the multivariate analysis. The survivaldata observed in this study are consistent with datafrom Canada (4) and the United States (7). The differ-ent survival rates are not explained by the variablesentered in this analysis.

The degree to which serum albumin concentrationis an estimate of nutrition is controversial (39). Fine

and Cox (40) reported excellent patient survivalamong patients with persistently low serum albumin

Adequacy and Nutrition in CAPD

204 Volume 7 . Number 2 . 1996

TABLE 9. Canada-USA Comparison

Factor Canada USA

GenderMale 59% 54%Female 41% 46%

Race#{176}Caucasian 85% 66%African-American 4% 30%

IDDM 22% 25%

CVD 34% 42%

BSA (m2)#{176} 1.74 1.80

Kt/V (wk)Totala 2.4 2.2Peritoneal#{176} 1.7 1.5Renal 0.7 0.7

CCr (1 .73 m2/wk)Total 83.8 78.4Peritoneala 44.7 41.2Renal 39.1 37.2

op< 0.05.

values. Struijk et al. (4 1 ) found no association be-tween serum albumin concentration and patient sur-

vival. The PR for death was 0.89 for a 1 g/L-greaterserum albumin concentration (95% CI, 0.76 to 1.03)

compared with a PR of 0.94 (95% CI, 0.90 to 0.97) inthis study. The back of statistical significance in thestudy from the Netherlands may be related to lowstatistical power (36,42). Teehan et al. (43) found thatan average serum albumin concentration less than 35g/L was associated with very poor survival. We foundthe serum albumin concentration to be strongly asso-dated with mortality. technique failure and hospital-ization. A difference of 1 g/L was associated with a 6%change in the PR of death, a 5% change in the RR oftechnique survival, and a 5% change in days hospi-

tabized.

The SGA is a valid clinical estimate of nutrition for

CAPD patients (23). Young and colleagues used SGAIn a study of prevalent CAPD patients and reportedthat 8% were severely malnourished, 32% had mild tomoderate malnutrition, and 60% showed no evidence

for malnutrition (44). We found severe and mild tomoderate malnutrition in 4.2% and 5 1 .2% respec-

tiveby of incident CAPD patients. A 1 -U change in this7-point scale is associated with a 25% difference in theRR of death.

Another estimate of nutrition is % LBM, which can

be estimated by bioebectrical impedance. near infra-red, anthropometrics, or creatinine kinetics (26,45,46). Keshaviah et at. (26) reported an excellent corre-

lation between bioelectrical impedance and nearinfrared, whereas estimates from creatinine kinetics

were systematically lower. We found that a 1% differ-ence in % LBM was associated with a 3% change in the

RR of death.NPCR was not associated with mortality. Harty et at.

(47) suggested that the normalized protein catabolic

rate is a flawed marker of nutrition in CAPD patients.They reported negative correlations between PCR nor-malized by dry weight and other estimates of nutri-

tion. Normalization using standard body weight (V/0.58) dId not discriminate between well and poorlynourished patients whereas use of ideal body weightfrom National Health and Nutrition Examination Sur-vey (NHANES) tables (48) did.

Each of the three estimates of adequacy of dialysis

were independently associated with patient survival.

A 0. 1 difference in weekly Kt/V was associated with a6% difference in the RR of death; a 5-L/week per 1.73

m2 difference in CCr was associated with a 7%difference.

When controlling for baseline demographic andclinical variables, there were associations between

improved patient survival and increased serum albu-mm concentration, greater % LBM, higher SGAscores, lower serum �2M concentration, greater Kt/V,and greater CCr.

Several sources of bias were considered. We hadcensored the 15 withdrawals from dialysis. Analysiswith the withdrawals treated as patient death did notchange the association of estimates of nutrition andadequacy of dialysis with patient survival. Events inany 6-month period had been attributed to the esti-mate of adequacy at the beginning of that period. For

events occurring later in a period, the Kt/V and CCrvalues would have been systematically bower than thevalue at the beginning of that period. A randomizedclinical trial with constant dialysis dose would providea more accurate evaluation of the association betweenadequacy and clinical outcomes. Estimates of ade-quacy and estimates of nutrition were treated astime-dependent covariates. If nutrition were signifi-cantly affected by adequacy of dialysis, the risk esti-

mate of mortality for adequacy would be incorrect.Analysis with estimates of nutrition entered only as

baseline variables did not change the risk estimates ofmortality for adequacy from those determined by us-

ing time-dependent covariates. If there were a rela-tionship between the estimates of adequacy and thequality of care provided, the latter could be a con-

founder, but, in a cohort study, cannot be evaluated.The expected 2-yr survival probabilities (Table 6) are

based on the assumption that increases in peritonealclearance will compensate for the loss of residualrenal function and sustain Kt/V and CCr values at agiven level. However, the survival probabilities weregenerated by deaths that occurred during loss ofresidual renal function without a compensatory in-

crease in peritoneal clearance. The assumption thatan increase in peritoneal clearance can clinically com-

pensate for the loss of residual renal function is

unproven.Theoretical constructs suggest that the target

weekly Kt/V should be 2.0 to 2.25 (1,2,49,50,51).

Several studies have addressed the relationship be-tween weekly Kt/V and patient survival by using

Churchill et al

Journal of the American Society of Nephrology 205

univariate analyses (13,15,52,53). Blake et a!. (13)found no effect of Kt/V on patient survival among 76patients. In a study of 102 CAPD patients in Spainover 1 2 months, the mean weekly Kt/V value for the92 patients who survived was 2.0, compared with 1.7

for those who died ( 15). Teehan et al. reported a 5-yrsurvival of >90% for those with a weekly Kt/V >1.89,

compared with <50% for those with less than 1.89(52). Lamelre et al. (53) reported the changes in Kt/V

for 16 patients who had survived for 5 yr on CAPD.There was progressive loss of renal function over 5 yr.The peritoneal contribution remained > 1 .9 per wk.

These clinical studies suggest a target weekly Kt/V

value of 1 .9 to 2. 1 . Unlike Kt/V, there are no theoret-

ical constructs on which to base adequate dialysisaccording to CCr. The recommendation of a total CCr40 to 50 L per 1 .73 m2 per week is based on clinicalexperience (54).

Chen et al. have shown an increase in serum �2Mwith longer duration of CAPD treatment (56). We

found that the serum �32M increased from 22.5 to 24.3mg/L over 24 months. Chen et a!. (56) also reported a

weekly peritoneal dialytic �32M removal of 1 19 mg;

Lysaght et at. reported 22 1 mg removed per wk (57).Given a daily production of 1000 to 1500 mg per wk(58), a decrease in residual renal function should be

associated with an increase in serum �2M.The only independent variables associated with

technique failure were serum albumin concentration

and CCr. The former could reflect malnutrition, over-

hydration, or both, whereas the latter is an estimate of

dose of dialysis. Among the factors associated with

more days hospitalized were increased age, worse

functional status, IDDM, and CVD. Malnutrition, es-timated by the SGA, had a very strong associationwith an increase in days hospitalized. A similar asso-ciation with decreased serum albumin concentrationmay be partly a reflection of nutritional status. In-

creased CCr was associated with fewer days hospital-ized; there was no association with Kt/V.

In this study, the expected 2-yr survival associated

with a sustained weekly Kt/V value of 2. 1 was 78%.This same 2-yr survival was associated with a weekly

CCr of 70 L/wk per 1 .73 m2. Because ofthe significant

level of residual renal function in the study popula-tion, these Kt/V and CCr doses were achieved simul-taneously. With progressive boss of residual renal

function, achievement of the Kt/V target may not beassociated with simultaneous achievement of the CCrtarget. Further study is required to determine whichestimate of adequacy should be used clinically.

The study population had a mean age of 54 yr; 30%had diabetes mebbitus as the cause for ESRD. The 2-yrsurvival of 78% exceeds the 73% for a cohort of 45 to

64-yr-old patients commencing dialysis in Canadabetween 1989 and 1993 (5) and a 66% 2-year survivalreported in the United States (64). This difference inpatient survival times between Canada and the UnitedStates is consistent with that observed in this study.

Adequate dialysis may be defined as the dose asso-

dated with an acceptable level of morbidity and mor-tality; optimal dialysis may be defined as the dosebeyond which the incremental clinical benefit does notjustify the added cost or patient burden. Within thedialysis dose range in this study, higher Kt/V and

higher CCr values were associated with better clinicaloutcomes. The predicted 2-yr survival of 78% wasassociated with a weekly Kt/V value of 2. 1 . This valueis within the 2.0 to 2.25 suggested by theoretical

constructs ( 1 ,2,45,47) and sligl’itly higher than the 1.9to 2.0 suggested by clinical studies ( 15,48,49). A 2-yrsurvival of 78% is better than that currently observedIn North America (5,59). Pending further studies todefine optimal dialysis. it would appear reasonable toprovide as high a dialysis dose as is feasible with a

Kt/V of 2. 1 or a CCr of 70 L/week per 1 .73 m2considered a reasonable target.

Strategies to optimize dialysis dose include pre-

scription of exchange volumes based on body surface

area (60) and the addition of a nocturnal exchangewith simple automated devices. The optimal fill vob-

ume for a patient with a body surface area of 1 . 73 m2is 2.5 L. This increases to 3.0 to 3.5 for body surface

area >2.0 m2 (60). For CCPD patients, one can in-

crease clearance by adding daytime exchanges. Ifincreased volumes and frequency of exchanges usedto compensate for declining renal function are unac-ceptable to the patient, or if therapy cost is excessive,transfer to hemodialysis should be considered. Ac-cording to the peak urea concentration hypothesis. aweekly continuous peritoneal dialysis Kt/V of 2. 1 isequivalent to a thrice weekly hemodialysis Kt/V of 1.5

(5 1). A transfer to hemodialysis on the basis of Inabil-ity to achieve a weekly Kt/V of 2. 1 would require that

a hemodialysis Kt/V of 1 .5 be provided.

The determination of optimal peritoneal dialysisdose will require further study. In order to achieve it ina cost-effective manner that is consistent with a highquality of life, health-care providers and Industrymust work together to optimize the dialysis prescrip-tion.

APPENDIX

CANUSA Peritoneal Dialysis Study GroupAffiliations

D.N. Churchill, D.W. Taylor, K.E. Thorpe. M.L. Bee-

croft, St. Joseph’s Hospital. McMaster University,

Hamilton, Ontario.

P.R. Keshaviah. G. deVeber, L.W. Henderson, BaxterHealthcare, Round Lake, Illinois.

K.K. Jindal, Victoria General Hospital, Halifax, Nova

Scotia.S.S.A. Fenton, J.M. Bargman, D.G. Oreopoubos, The

Toronto Hospital, University of Toronto. Ontario.G.G. Wu, Credit Valley Hospital, Mississauga. On-

tario.

S.D. Lavoie, Ottawa Civic Hospital, University of Ot-tawa, Ottawa, Ontario.

Adequacy and Nutrition in CAPD

206 Volume 7 . Number 2 - 1996

A. Fine, St. Boniface Hospital. University of Manitoba,

Winnipeg. Manitoba.

E. Burgess, Foothills Hospital, University of Calgary,

Calgary, Alberta.

J.C. Brandes, Medical College of Wisconsin. Milwau-

kee, Wisconsin.

K.D. Nolph, B.F. Prowant, University ofMissouri Med-ical Center, Columbia, Missouri.

D. Page, Ottawa General Hospital, University of Ot-

tawa, Ottawa, Ontario.

F.X. McCusker, B.P. Teehan, Lankenau Hospital, Tho-mas Jefferson Medical College, Philadelphia, Penn-

sylvania.

M.K. Dasgupta. K. Bettcher, University of Alberta

Hospital. University of Alberta, Edmonton, Alberta.

R. Caruana, Medical College of Georgia. Augusta,

Georgia.

ACKNOWLEDGMENTS

This study was funded by the Kidney Foundation ofCanada (Premiere

Clinical Research Award) and by Baxter Healthcare in Canada and in

the United States of America. Funding was also provided by the

Father Sean O’Sullivan Research Centre (St. Joseph’s Hospital. Ham-

ilton, Ontario); the Kidney Foundation of Manitoba (St. Boniface

Hospital. Winnipeg. Manitoba): the Foothills Hospital Foundation,

Calgary. Alberta: the Nephrolog,y Research Fund ofVictoria Hospital.

Halifax. Nova Scotia; and the Dialysis Research Fund of the Toronto

General Division of the Toronto Hospital. Toronto. Ontario. We thank

Ms. Barbara Leavitt. Mr. Walker Woodworth, and Dr. John Moran of

Baxter Healthcare for their support in this study. We also thank Dr.

Edward Vonesh for statistical consultation and Ms. Deirdre Hobeck

for her secretarial skills. Finally. we express our appreciation to the

research assistants in each of the clinical centers for their invaluable

work.

REFERENCES

1 . Popovich PP. MoncriefJW, Decherd JB. Bomar JB, PyleWK: The definition of a novel portable/wearable equilib-rium peritoneal dialysis technique [Abstract]. ASAIO (AmSoc Artif Intern Organs) Trans 1976;5:64A.

2. Popovich RP, Moncrief JW: Kinetic modeling of perito-neal transport. Contrib Nephrol 1979; 17:59-72.

3. Nolph KD: What’s new in peritoneal dialysis-An over-view. Kidney Int 1992:42ESuppl 381:5148-5152.

4. Burton PR, Walls J: Selection-adjusted comparison oflife expectancy of patients on continuous ambulatoryperitoneal dialysis. hemodialysis and transplantation.Lancet 1987: 1: 1 1 15-1 1 19.

5. Canadian Institute for Health Information: CanadianOrgan Replacement Register, 1993 Annual Report. On-tario: Don Mills: 1995.

6. MaJorca R, Vonesh EF, Cavalbi P. et at: A multicenter,selection adjusted comparison of patient and techniquesurvivals on CAPD and hemodialysis. Pent Dial mt i � i:11:118-127.

7. Held PJ, Port FK, Turenne MN, et at.: Continuousambulatory peritoneal dialysis and hemodialysis: Com-parison of patient mortality with adjustment for comor-bid conditions. Kidney mt 1994:45:1163-1169.

8. Lowrie EG, Lew NL: Death risk in hemodialysis patients:The predictive value of commonly measured variablesand an evaluation of death rate differences betweenfacilities. Am J Kidney Dis 1990:15:458-482.

9. Held PJ, Levin NW, Bovbjerg RR, Pauly MV, DiamondLH: Mortality and duration of hemodialysis treatment.JAMA 199 1:265:871-875.

10. Gotch F, Sargent JA: A mechanistic analysis of theNational Cooperative Dialysis Study (NCDS). Kidney Int1985:28:526-534.

1 1 . Fenton SSA, Johnston N, Delmore T, et at.: Nutritional

assessment of continuous ambulatory peritoneal dialy-sis patients. ASAIO (Am Soc Artif Intern Organs) Trans1987:33:650-653.

12. Teehan BP, Schleifer CR, Brown JM, Sigler MH, Rai-mondo J: Urea kinetic analysis and clinical outcome onCAPD. A five year longitudinal study. Adv Pent Dial1990;6: 18 1-185.

13. Blake PG. Sombolos K, Abraham G, et at.: Lack ofcorrelation between urea kinetic indices and clinicaloutcomes in CAPD patients. Kidney mt 199 1:39:700-706.

14. Keshaviah PR, Nolph 1W, Prowant B, et at: Definingadequacy of CAPD with urea kinetics. Adv Pent Dial1990:6:173-177.

15. DeAlvaro F, Bajo MA. Alvarez-Ude F, et at.: Adequacy ofperitoneal dialysis: Does KT/V have the same predictivevalue as in HD? A multicenter study. Adv Pent Dial1992:8:93-97.

16. Brandes JC, Piering WF. Beres JA, Blumenthal 55,Fritsche C: Clinical outcome of continuous peritonealdialysis predicted by urea and creatinine kinetics. J AmSoc Nephrol 1992:2:1430-1435.

1 7. Arkouche W, Debawari E, My H, et at.: Quantification ofadequacy of peritoneal dialysis. Pent Dial Int 1993:l3ISuppl 21:S215-S218.

18. Tzamaloukas AR, Murata GH, Sena P: Assessing theadequacy of peritoneal dialysis. Pent Dial mt 1993:13[Suppl 21:5236-5239.

19. Churchill DN: Adequacy of peritoneal dialysis. Howmuch dialysis do we need? Kidney Int 1994:46[Suppl48]:S2-S6.

20. Karnofsky DA. Burchenal JH: The clinical evaluation ofchemotherapeutic agents in cancer. In: McLeod CM, Ed.Evaluation of Chemotherapeutic Agents. New York: Co-lumbia University Press: 1949.

2 1 . Baker JP, Detsky AS, Wesson DE, et al.: Nutritionalassessment. A comparison of clinical judgement andobjective measurements. N Engl J Med 1982:306:969-972.

22. Detsky AS, McLaughlin JR. Baker JP, et at.: What issubjective global assessment? J Parenter Enteral Nutr1987; 11:8-13.

23. Enia G. Sicuso C, Alati G, Zoccali C: Subjective globalassessment ofnutrition in dialysis patients. Nephrol DialTransplant 1993;8: 1094-1098.

24. Randerson DH, Chapman GV, Farrell PC: Amino acidand dietary status in CAPD patients. In: Atkins RC,Farrell PC. Thomson N, Eds. Peritoneal Dialysis. Edin-burgh: Churchill-Livingstone: 1981:180-191.

25. Watson PE, Watson ID, Batt RD: Total body water vol-umes for adult males and females estimated from simpleanthropometric measurements. Am J Clin Nutr 1980:33:27-39.

26. Keshaviah PR, Nolph KD, Moore IlL, et at.: Lean bodymass estimation by creatinine kinetics. J Am Soc Neph-rol 1994:4:1475-1485.

27. Farrell SC, Bailey MP: Measurement of creatinine inperitoneal dialysis fluid. Ann Clin Biochem 1991:28:624-625.

28. Nobph KD: Update on peritoneal dialysis worldwide. PentDial Int 1993: 1 3ESuppl 21:�15-�19.

29. Andersen PK, Gill 1W: Cox’s regression model for count-ing processes: A large sample study. Ann Stat 1982; 10:1100-1120.

30. Cox D: Regression models and life-tables. J R Stat Soc1972:34:187-201.

3 1 . Cox DR, Oakes D. Analysis of Survival Data. London:Chapman and Hall: 1984.

32. Kalbfleisch JD, Prentice RL. The Statistical Analysis ofFailure Time Data. New York: Wiley; 1980.

33. Cox DR, Hinkley DV. Theoretical Statistics. London:Chapman and Hall; 1974.

34. Schoenfeld D: Partial residuals for the proportional haz-ards regression model. Biometnika 1982:69:239-241.

35. McCullagh P. NelderJA. Generalized Linear Models. 2ndEd. London: Chapman and Hall; 1989.

36. Harrell FE Jr. Lee KL, Matchar DB, Reichent TA: Regres-sion models for prognostic prediction: Advantages, prob-

Churchill et al

Journal of the American Society of Nephrology 207

lems and suggested solutions. Cancer Treat Rep 1985;69: 107 1-1077.

37. Statistical Sciences. S-PLUS Guide to Statistical andMathematical Analysis. Version 3.2. Seattle: StatSci, ADivision of Mathsoft Inc.; 1993.

38. Harrell F. The PHGLM procedure. SAS SupplementalLibrary User’s Guide. Version 5 Edition. Cary, NC: SASInstitute, Inc.: 1986.

39. Heimburger 0. Bergstrom J, Lindholm B: Is serumalbumin an index of nutritional status in continuousambulatory peritoneal dialysis patients? Pent Dial mt1994; 14:108-114.

40. Fine A, Cox D: Modest reduction of serum albumin incontinuous peritoneal dialysis patients is common andof no apparent clinical consequence. Am J Kidney Dis1992;20:50-54.

41 . Struijk DG, Krediet RT, Koomen GCM, Boeschoten EW,Arisz L: The effect of serum albumin at the start ofcontinuous ambulatory peritoneal dialysis treatment onpatient survival. Pent Dial Int 1994; 14:121-126.

42. Concato J, Feinstein AR, Holford TR: The risk of deter-mining risk with multivariable models. Ann Intern Med1993:1 18:201-2 10.

43. Teehan BP, Schleifer CR, Brown J: Urea kinetic model-ing is an appropriate assessment of adequacy. SeminDialysis 1992:5:189-192.

44. Young GA, Kopple JD, Lindholm B, et at.: Nutritionalassessment of CAPD patients: An international study.Am J Kidney Dis 1991;17:462-471.

45. Bergia R, Bellini ME, Vabenti M, et al.: Longitudinalassessment of body composition in continuous ambula-tory peritoneal dialysis using bioelectrical impedanceand anthropometric measurements. Pent Dial bnt 1993;l3lSuppl 21:�512-�514.

46. Schmidt R, Dumler F, Cruz C, Lubkowski T, Kilates C:Improved nutritional follow-up of peritoneal dialysis pa-tients with bioelectrical Impedance. Adv Pent Dial 1992:8:157-159.

47. Hasty JC, Boulton H, Curwell J, et at.: The normalizedprotein catabolic rate is a flawed marker of nutrition inCAPD patients. Kidney Int 1994:45:103-109.

48. Frisancho AR: New standards of weight and body com-position by frame size and height for assessment of

nutritional state of adults and the elderly. Am J ClinNutr 1984:40:808-819.

49. Teehan BP, Schleifer CR, Sigler MH, Gilgore GS: Aquantitative approach to the CAPD prescription. PentDial Bull 1985:5:152-156.

50. Keshaviah PR, Nolph KD, Van Stone JC: The peakconcentration hypothesis: A urea kinetic approach tocomparing the adequacy of continuous penitoneal dialy-sis (CAPD) and hemodialysis. Pent Dial mt 1989:9:257-260.

5 1 . Keshaviah PR: Adequacy of CAPD: A quantitative ap-proach. Kidney Int 1992;42[Suppl 38l:S 1 60-S 164.

52. Teehan BP, Schleifer CR, Brown J: Urea kinetic model-ing is an appropriate assessment of adequacy. SeminDialysis 1992:5:189-192.

53. Lameire NH, Vanholder R, Veyt D, Lambert M-C, Ringoir5: A longitudinal five year survey of urea kinetic param-eters in CAPD patients. Kidney Int 1992;42:426-432.

54. Twardowski Z, Nolph K: Penitoneal dialysis: How muchis enough? Semin Dial 1988:1:75-76.

55. Blake PG. Izatt 5, Sombolos K, Oreopoulos DG: Totalweekly creatinine clearance in CAPD-How much dopatients get and does it correlate with outcomes? lAb-stract] Pent Dial Int 1992; l2lSuppl 1 l:99A.

56. Chen H-C, Guh J-Y, Laf Y-H, Tsai J-H: Serial changes inserum f3-2-microglobulin in CAPD. Pent Dial mt 1993;13:238-239.

57. Lysaght MJ, Pollock CA, Moran JE, Ibels LS, Farrell PC:Beta-2 microglobulin removal during continuous ambu-latory penitoneal dialysis (CAPD). Pent Dial mt 1989:9:29-35.

58. Karlson FA, Wibell L, Ervin PE: Beta-2 microglobulmn inclinical medicine. Scand J Lab Clin Invest 1980;4OESuppll:293-299.

59. US Renal Data System. USRDS 1994 Annual Report.Bethesda, MD: The National Institute of Health, NationalInstitute of Diabetes and Digestive and Kidney Diseases;1994.

60. Keshaviah P, Emerson PF, Vonesh EF, Brandes JC:Relationship between body size, fill volume, and masstransfer area coefficient in penitoneal dialysis. J Am SocNephrol 1994:4:1820-1826.