Peritoneal Dialysis International, Vol. 26, pp. 458–465
Printed in Canada. All rights reserved.
0896-8608/06 $3.00 + .00
Copyright © 2006 International Society for Peritoneal Dialysis
WHAT IS THE LINK BETWEEN POOR ULTRAFILTRATION AND INCREASED
MORTALITY IN ANURIC PATIENTS ON AUTOMATED PERITONEAL
DIALYSIS? ANALYSIS OF DATA FROM EAPOS
Simon J. Davies,1 Edwina A. Brown,2 Werner Reigel,3 Elaine Clutterbuck,4 Olof Heimbürger,5 Nicanor
Vega Diaz,6 George J. Mellote,7 Javier Perez–Contreras,8 Renzo Scanziani,9 Christian D’Auzac,10
Dirk Kuypers,11 and Jose C. Divino Filho,12 on behalf of the EAPOS Groupa
♦♦Background: Primary analysis of the European Automated
Peritoneal Dialysis Outcomes Study (EAPOS) found that patients with daily ultrafiltration (UF) below a predefined
target of 750 mL at baseline experienced increased mortality and continuing low UF over 2 years.
♦♦Setting: Multicenter, prospective observational study of
prevalent, functionally anuric patients on automated peritoneal dialysis (APD) treated to predefined standards.
♦♦Methods: Secondary data analysis to determine clinical
covariates that might support a link between poor UF and
outcome, including pattern of comorbidity, prescription,
nutrition as determined by Subjective Global Assessment
(SGA), membrane function, and blood pressure (BP). Ultrafiltration was treated as a categorical (comparing patients above and below target at baseline) and continuous
dependent variable in univariate and multivariate regression. The relationship between BP and survival was also
explored.
♦♦Results: Of 177 patients recruited from 28 centers across
Europe, 43 were below the UF target at baseline. Compared
to those above target, there were no differences in the
spread of comorbidity, type of APD prescription, SGA, BP,
hemoglobin, HCO3, or parathyroid hormone, at baseline or
at any later time. At baseline, plasma calcium and, at
12 months, plasma phosphate were lower in the low UF
group. There was a weak positive correlation between baseline systolic or diastolic BP and UF, which remained on multivariate analysis but accounted for just 9% of the
variability in BP. There was no clear relationship between
baseline BP and survival, although, if anything, low BP was
associated with earlier death. Poor UF was associated with
lower mean dialysate glucose concentration during the first
4 months and with consistently worse membrane function.
a
EAPOS Group: see Appendix.
458
♦♦Conclusions: The increased mortality associated with poor
UF is likely multifactorial and not easily explained by clear
differences in comorbidity, nutritional state, or other indices of treatment at baseline. The lower plasma phosphate
suggests a subsequent fall in appetite. Poor BP control is
unlikely to be the explanation, and a link between lower
BP, reduced UF, and earlier death is suggested. Failure to
achieve adequate UF due to worse membrane function remains an important and potentially reversible or preventable cause.
Perit Dial Int 2006; 26:458–465
www.PDIConnect.com
KEY WORDS: Ultrafiltration; solute transport; blood
pressure; fluid status; survival; observational cohort
study.
T
he European Automated Peritoneal Dialysis Outcomes
Study (EAPOS) was undertaken to establish the efficacy of automated peritoneal dialysis (APD) in functionally anuric patients treated according to previously
agreed upon targets for anemia, several biochemical
variables, small solute clearance, and daily ultrafiltration (UF) (1–3). The study showed that good clinical outcomes are possible and that the most important
predictors of survival are age, comorbidity (especially
diabetes), and severe malnutrition. Membrane permeability and small solute clearance, as measured by creatCorrespondence to: S.J. Davies, Department of Nephrology,
University Hospital of North Staffordshire, Princes Road,
Hartshill, Stoke-on-Trent ST4 7LN, United Kingdom.
[email protected]
Received 8 September 2005; accepted 22 December 2005.
Downloaded from http://www.pdiconnect.com/ by guest on September 30, 2014
University Hospital of North Staffordshire,1 Stoke on Trent; Charing Cross Hospital,2 London, UK;
Klinikum Darmstadt,3 Darmstadt, Germany; Hammersmith Hospital,4 London, UK; Karolinska
University Hospital,5 Huddinge, Sweden; Hospital Universitario Dr. Negrin,6 Las Palmas, Spain;
Adelaide & Meath Hospital,7 Dublin, Ireland; Hospital General Universitario,8 Alicante, Spain; Ospedale
Provinciale Desio,9 Desio, Italy; Hôpital Européen Georges Pompidou,10 Paris, France;
U.Z. Gasthuisberg,11 Leuven; Baxter Renal Division Europe,12 Brussels, Belgium
PDI
JULY 2006 – VOL. 26, NO. 4
EAPOS: ULTRAFILTRATION AND MORTALITY
inine clearance, did not predict survival. Strikingly, patients who were below the predefined UF target at the
start of the study (>750 mL/day) proved difficult to get
above target throughout the 2-year study period, and
had a significantly increased mortality. They also had a
relatively poor UF capacity at the start of the study compared to those individuals achieving the UF target. There
was no difference in solute transport between the two
groups. This finding is supported by the recent analysis
of survival of anuric peritoneal dialysis (PD) patients by
the Netherlands Cooperative Study on the Adequacy of
Dialysis (NECOSAD) group (4).
A number of possible explanations linking poor UF to
reduced survival might be put forward. These could include one or more of the following:
14 European countries. Enrollment included 177 of
204 screened patients who were followed for 2 years or
until they stopped PD. Clinicians were asked to optimize
the dialysis prescription to achieve predefined treatment
standards during the first 6 months: these were creatinine clearance ≥60 L/week/1.73 m 2 and daily UF
≥750 mL. Standard pH 5.5, 40-mmol lactate-buffered
glucose dialysate was used throughout the study and clinicians used icodextrin at their own discretion to achieve
UF targets. In addition, targets were set for blood pressure (BP) and biochemical and hematological parameters
as published by the UK Renal Association (5).
1. Increased mortality by association between poor
appetite and reduced fluid intake, leading to low UF
requirements;
2. Inadequate UF leading to poor control of fluid status due to reluctance to prescribe higher glucose
concentrations, or underuse of icodextrin, or inadequate tools for assessing fluid status. This mechanism might increase mortality via hypertension due
to increased intravascular volume or expansion of
the extracellular interstitial space, a well-documented phenomenon in PD patients.
3. Inability of clinicians to increase UF due to symptomatic hypotension and thus increased mortality
by association with worse cardiac function; and
4. An unmeasured difference (e.g., inflammation) in
the mix of comorbidity not revealed by the initial
analysis.
All patients had a peritoneal equilibration test (PET)
at baseline and at 12 and 24 months, from which their
solute transport status and UF capacity were determined
as described previously (6). Weekly creatinine clearance
was calculated from creatinine concentration in plasma,
urine, and dialysate, 24-hour urine volume, and 24-hour
dialysate volume, using PD Adequest (Baxter Healthcare,
Brussels, Belgium). Daily UF was calculated as the difference between the volume of total dialysate infused
(including both night and daytime fluid) and the volume drained over 24 hours for the same period as the
solute collection. Comorbidities were counted for each
patient to enable calculation of the Stoke comorbidity
score (7,8). Nutrition was assessed at baseline and at
yearly visits using Subjective Global Assessment (SGA)
(9). Data management was centralized; peritoneal,
urine, and dialysate concentrations of creatinine and all
other blood tests were measured in local laboratories.
The results of these and all other demographic data were
collated in the central EAPOS office (Baxter Healthcare),
where the PET results and creatinine clearances were
calculated.
METHODS
STUDY DESIGN
The design and analysis of primary end points (patient and technique survival) of EAPOS are described in
detail elsewhere (1–3). In brief, it was a prospective
study of functionally anuric patients (urine volume
<100 mL and/or residual renal function <1 mL/minute/
1.73 m2) treated with APD, undertaken in 28 centers in
STATISTICAL DESIGN AND ANALYSIS
Previous analysis of the EAPOS data set established a
relationship between baseline UF and outcome in two
ways. First, there was a categorical analysis according
to whether patients were above or below the predefined
threshold of daily net UF (750 mL) at the start of the
study. Patients below this target, in whom there was an
intention to increase UF, had a higher mortality. To reflect this in the present analysis, all available clinical
variables were compared according to this baseline categorization, with comparisons between groups being
made at different time points using unpaired tests, and
459
Downloaded from http://www.pdiconnect.com/ by guest on September 30, 2014
The purpose of this secondary analysis of the EAPOS database was to see if any of the above associations could
be demonstrated and to see how they interacted with
the only factors identified in the primary analysis to be
associated with less UF, namely, membrane UF capacity
and underutilization of hypertonic glucose in the early
stages of the study.
PERITONEAL MEMBRANE FUNCTION, SOLUTE CLEARANCES,
COMORBIDITY SCORING, NUTRITION, AND DATA MANAGEMENT
DAVIES et al.
JULY 2006 – VOL. 26, NO. 4
within group longitudinally using paired tests. In the
second approach used in the primary analysis, UF was
treated as a continuous variable, both at baseline and
longitudinally as a time-related variable; in this analysis, increased UF was associated with improved survival
independent of age, gender, body surface area (BSA),
time on treatment, comorbidity, and nutritional state.
To reflect this in the present analysis, UF was treated as
a continuous variable when correlated with other clinical factors, using univariate or multivariate regression.
To explore the possibility that BP was linked to survival,
both a categorical analysis (Kaplan–Meier plot of tertiles
combined with between-group comparison using the log
rank test) and Cox regression, treating BP as a continuous variable, were used. Data are generally expressed as
mean ± SD unless otherwise stated.
COMPARISON OF BASELINE CHARACTERISTICS ACCORDING TO
BASELINE UF CATEGORY
Baseline characteristics are summarized in Table 1. It
can be seen that the demographic characteristics were
broadly similar in both UF categories. If anything, the
prevalence of peripheral vascular disease was greater in
the patients with higher daily UF, although the overall
spread of comorbidity was very similar. Transplant his-
tory, use of icodextrin, and the various approaches to
prescribing APD were also the same in both groups.
COMPARISON OF LONGITUDINAL VARIABLES ACCORDING TO
BASELINE UF CATEGORY
Hemoglobin, bicarbonate, parathyroid hormone, BP,
and SGA showed no between-group differences at any
time point; nor was within-group longitudinal change
significant (Table 2). There was a small but significant
increase in BSA in the higher UF group (from 1.74 to
1.77 m2), implying that these patients gained weight
relatively with the loss of larger patients from PD, resulting in no change in average BSA at each time-point.
Plasma calcium was significantly lower at baseline in the
low UF category (p = 0.015), but this difference disappeared by 6 months. Patients in this category also tended
to have lower phosphate levels, especially at 12 (p < 0.05)
and 18 months (NS). At the start of the study, patients
below the UF target had dialysis prescriptions with a
lower average glucose concentration. Over the next
6 months, this was gradually increased so that it became
identical between the two groups for the rest of the study
(see Figure 1). Membrane function differed significantly
at baseline, such that patients below the UF target had
lower UF capacity (p = 0.022) but similar solute transport. Longitudinal membrane function (see Figure 2)
showed a significant increase in solute transport and a
TABLE 1
Demographic Characteristics of Patients Below and Above Ultrafiltration (UF) Target at Baseline
N
Cancer
Ischemic heart disease
Peripheral vascular disease
Left ventricular dysfunction
Diabetes mellitus
Systemic collagen vascular disease
Comorbidity: other life-threatening condition
Distribution of comorbidity score (0:1:2:3:4:5 comorbidities)
Distribution of grade of comorbidity (None:1–2:>2)
Gender (male)
Age (years)
Number of previous transplants (0:1:2:3)
Body surface area at baseline (m2)
Use icodextrin
Distribution of prescription type
Dry day
Wet day
Extra day exchange
Tidal
460
Group 1 (UF<750 mL)
Group 2 (UF>750 mL)
p Value
43
2%
23%
9%
14%
21%
0%
9%
21:13:7:1:1:0
49%:47%:4%
58%
53.7±17.3
28:13:2:0
1.72
47%
131
4%
21%
27%
11%
13%
2%
8%
61:42:18:8:1:1
47%:46%:7%
56.5%
51.4±15.4
91:33:6:1
1.77
46%
0.64
0.71
0.017
0.66
0.20
0.42
0.73
0.83
0.79
0.85
0.72
0.73
0.67
0.99
4.6%
14%
63%
18.4%
6.8%
18.3%
61%
12.2%
0.67
Downloaded from http://www.pdiconnect.com/ by guest on September 30, 2014
RESULTS
PDI
PDI
JULY 2006 – VOL. 26, NO. 4
EAPOS: ULTRAFILTRATION AND MORTALITY
TABLE 2
Longitudinal Changes in Clinical Variables According to Whether Patients Were
Below or Above the Ultrafiltration (UF) Target at Baseline
Remaining in study (n)
BSA (m2)
Hemoglobin (g/dL)
Phosphate (mmol/L)
Calcium (mmol/L)
PTH (pg/L)
Systolic BP (mmHg)
Diastolic BP (mmHg)
SGA (score)
0
6
Months into study
12
18
24
<750 mL
>750 mL
<750 mL
>750 mL
<750 mL
>750 mL
<750 mL
>750 mL
<750 mL
>750 mL
<750 mL
>750 mL
<750 mL
>750 mL
<750 mL
>750 mL
<750 mL
>750 mL
<750 mL
>750 mL
43
131
1.72
1.77
11.4±1.7
11.0±1.8
1.60±0.48
1.71±0.49
2.35±0.28a
2.46±0.23
27.9±4.1
27.0±3.8
208±76
44±221
133±23
136±25
79±14
82±15
1.63±0.59
1.54±0.57
26
98
1.72
1.77
11.8±1.6
11.5±2.1
1.57±0.51
1.60±0.50
2.36±0.29
2.46±0.23
28.0±4.0
26.5±3.5
118±252
38±299
125±22
131±27
76±16
80±13
—
—
18
77
1.74
1.77
11.9±1.5
11.2±1.7
1.38±0.44b
1.62±0.50
2.37±0.29
2.41±0.24
26.5±4.0
26.9±4.2
19±368
51±223
127±29
129±27
76±16
78±16
1.59±0.51
1.55±0.55
11
53
1.78
1.76
11.3±1.7
11.2±1.9
1.24±0.47
1.54±0.50
2.36±0.27
2.38±0.19
26.5±4.0
26.5±3.7
55±279
31±364
123±28
132±29
73±21
78±17
—
—
7
48
1.68
1.77c
10.4±1.6
11.2±1.7
1.49±0.48
1.47±0.40
2.30±0.14
2.43±0.22
27.2±6.4
26.8±4.8
85±125
87±310
133±19
130±24
74±11
78±14
1.71±0.76
1.6±0.56
Average % glucose concentration
BSA = body surface area; PTH = parathyroid hormone; BP = blood pressure; SGA = Subjective Global Assessment.
a p = 0.015, between groups.
b p = 0.05, between groups.
c p < 0.01, paired test, longitudinal increase from baseline.
Figure 1 — Longitudinal average glucose concentration according to whether patients were above (open squares) or below
(closed squares) the 750 mL ultrafiltration target at baseline.
During the first few months of the study, patients below baseline were prescribed significantly lower glucose concentrations. *p < 0.05.
reduction in UF capacity in the patients with UF capacity
above target. Membrane function always tended to be
worse with respect to UF capacity in those below target,
and this was statistically significant on between-group
analysis at baseline.
Figure 2 — Mean (SE) longitudinal membrane function (at 0,
12, and 24 months) according to whether patients were above
(open squares) or below (closed squares) the 750 mL ultrafiltration (UF) target at baseline. *At baseline, the betweengroup difference in UF capacity is significant (p = 0.022). †In
those patients >750 mL UF at baseline (open squares), there
was a significant longitudinal increase in solute transport and
decline in UF by 24 months (p < 0.01).
461
Downloaded from http://www.pdiconnect.com/ by guest on September 30, 2014
Bicarbonate (mmol/L)
UF group
DAVIES et al.
JULY 2006 – VOL. 26, NO. 4
PDI
CROSS-SECTIONAL ANALYSIS OF BASELINE CORRELATES,
NUTRITIONAL STATUS, AND DIALYSIS PRESCRIPTION
Figure 3 —Patient survival according to baseline systolic blood
pressure (BP) divided into tertiles: BP < 126 mmHg (solid light
line), 126 – 150 mmHg (solid heavy line), and > 150 mmHg
(broken line). There is no significant difference in survival (log
rank test, p = 0.23), although the lowest-BP group appears to
have a disproportionate early death rate.
TABLE 3
Multivariate Regression of Correlates with
Systolic Blood Pressure at Baseline
ANALYSIS OF BP
The relationship between baseline BP and survival was
also explored. There was no significant relationship if
systolic BP was treated as a continuous variable, with a
relative risk of 0.99 per 1 mmHg of increasing pressure
(p = 0.188). Blood pressure, which was normally distributed at baseline, was also treated as a categorical variable by tertiles (low <126, medium 126 – 150, and high
>150 mmHg) and comparison of survival was made (see
Kaplan–Meier plot, Figure 3). There was no significant
difference in survival by category (p = 0.23), although it
can be seen that survival is not proportional, with a tendency toward earlier death in the lowest BP tertile.
Multivariate analysis of clinical covariates associated
with baseline systolic BP is summarized in Table 3. In
this analysis, the only significant correlation was with
net daily fluid removal, and the relationship is positive
such that a higher UF was associated with a higher BP.
While the overall model is significant (ANOVA p = 0.04),
it must be emphasized that it accounts for only 9% of
the variability in systolic BP. Adding grade of comorbidity or types of comorbidity had no effect on the overall
model. A similar analysis was performed using diastolic
BP as the dependent variable (data not shown). This result was almost identical to the systolic BP model, except that, in addition, there was a negative correlation
462
Covariate
t
Constant
6.09
Age
0.045
Gender (male=1)
–1.09
Ultrafiltration at baseline (mL)
2.47
Urine volume at baseline
0.09
SGA at baseline (A=1, B=2, C=3)
–1.634
Average glucose concentration of dialysate –0.73
Solute transport a baseline
1.47
p Value
<0.001
0.96
0.27
0.015
0.93
0.104
0.47
0.145
Dependent variable: systolic blood pressure.
between age and diastolic pressure, with the model accounting for 11% of the variability.
DISCUSSION
Having shown in the primary analysis of the EAPOS
patient cohort that, in those patients who were below the
predefined daily UF target of 750 mL, death was more
likely, we were anxious to examine the clinical data gathered for any potential explanation of this association. In
interpreting these analyses, two important factors must
be remembered. First, while EAPOS can be described as
an intervention study with predefined minimum targets,
Downloaded from http://www.pdiconnect.com/ by guest on September 30, 2014
Univariate correlations observed at baseline for the
whole study population included weak but significant
relationships between age and comorbidity score (r =
0.24, p < 0.001) and plasma phosphate (r = –0.34, p <
0.001), but not between any of these covariates and the
SGA score. There was correlation between SGA and baseline plasma calcium concentration (r = –0.23, p = 0.003).
Average concentration of glucose prescription at baseline correlated weakly with age (r = –0.2, p = 0.012), UF
(r = 0.16, p = 0.02), SGA (r = –0.26, p = 0.001), and BSA
(r = –0.2, p = 0.011). Analysis of dialysis prescription
showed that a dry day (n = 11) was associated with a significantly lower daily UF at baseline: 1006 ± 320 mL compared to 1251 ± 768 mL with all other regimes (p = 0.045).
This effect disappeared at 6 months because of prescription change. Survival was not influenced by prescription
approach, for example, tidal versus dry/wet day, or wet
day with additional exchange. Inequality of size meant
that analysis for a center effect was limited to country
of origin. No effect of nationality on achieved UF at baseline could be identified (ANOVA p = 0.28).
PDI
JULY 2006 – VOL. 26, NO. 4
no significant relationship between BP and survival in
this population, whether treated as a continuous or a
categorical variable on either multivariate or univariate
analysis (data not shown), although a tendency for better survival was seen with higher BP. It is conceivable
that clinicians were reluctant to press for higher daily
UF in patients with relatively low BP in this study, leading to an associated increase in mortality. This cannot
have been the main mechanism, however, as betweengroup comparisons of BP were very similar.
Another possible explanation for the association between poor outcome and reduced UF in this study would
be inflammation (15). One of the main shortcomings of
EAPOS is the lack of measurement of either plasma albumin or a marker, such as C-reactive protein. The reasons
for this were logistic in that central laboratory storage
and measurement of blood samples were beyond the
funding arrangements. One possible clue to this link,
inflammation, may be inferred from the weak but significant negative correlation between plasma calcium
and comorbidity, and the significantly lower plasma calcium observed in the poor UF group at baseline. Clinicians were asked to report corrected calcium levels, but
it is possible that this correction was insufficient or omitted in some cases, such that reduced calcium concentration reflected lower albumin levels. If so, this suggests
that these patients did have increased inflammation at
the start of the study despite similar age and distribution of comorbidity. It seems unlikely, however, that this
has a large enough effect alone, as the main clinical correlates with inflammation described in other studies are
age, comorbidity, and residual renal function.
Following this more detailed analysis, relatively low
use of hypertonic glucose and membrane function, specifically UF capacity, remain the clinical variables that
are most clearly related to lower achieved UF. In EAPOS,
clinicians were allowed to deliver APD according to their
clinical preference. This resulted in a number of identifiable patterns of prescription, including dry days, that
were associated with slightly less UF at baseline. This
effect had disappeared by 6 months and was not used
disproportionately in patients below target. It is apparent, however, that care must be taken in using dry days
in the prescription given to anuric APD patients with respect to fluid removal, let alone concerns over adequate
middle molecule clearance.
In conclusion, no clear explanation of the cause of
increased mortality associated with reduced UF in this
study is apparent from this analysis, although it is likely
to be multifactorial. Clinicians need to give special attention to anuric patients who are consistently achieving UF volumes of less that 750 mL per day, especially if
463
Downloaded from http://www.pdiconnect.com/ by guest on September 30, 2014
there was no randomization, so any attempts to draw
cause-and-effect conclusions must be considered with
extreme care. Second, because there was a different death
rate between the two groups, and thus reason for leaving
the study, any conclusions drawn from the longitudinal
analyses must consider informative censoring, especially
as a reason for lack of difference between groups. With
these limitations in mind, we were especially interested
to see if differences in BP, nutrition, glucose prescription,
membrane function, and approach to dialysis prescription could be implicated in this excess mortality.
The SGA is a useful tool in assessing clinical nutritional
state, which is well validated in PD patients, in whom
SGA is also able to detect longitudinal change and predict survival (10,11). Indeed, it was specifically designed
to pick up changes in appetite, gastrointestinal symptoms, and muscle wasting. Baseline SGA, specifically,
severe malnutrition (SGA = C), predicted poor survival
independently of fluid removal in the EAPOS patients.
There is no evidence, however, from this analysis that
poor nutrition at baseline was the explanation for reduced survival in those patients whose UF was below target. Furthermore, there was no longitudinal change in
BSA, and thus weight, of the low UF patients who remained in the study, in contrast to the significant trend
for weight gain in those that were above the UF target
and who remained in the study (see Table 2). It can be
inferred from the relatively low plasma phosphate, in
particular at 12 months, that appetite was reduced in
patients with less UF, especially as the mean value at this
point of the study is well below any treatment target.
The fact that SGA score remained similar between the
groups may reflect the fact that patients with worsening
SGA had already died or left the study. It could be, therefore, that patients in this group developed worsening
nutrition during the study. This might have resulted in a
lack of need for greater fluid removal, as perceived by
patients or their clinicians, or because poor fluid control itself is associated with worse appetite. It is clear
that further studies are required to evaluate the relationship between achieved UF, fluid status, and nutrition. Cross-sectional data would suggest that there is a
link, but longitudinal intervention studies are required
(12,13).
This analysis did not find any evidence to suggest that
the excess mortality associated with poor UF was attributable to poor BP control. There was a weak association
between increasing BP and UF volume at baseline, but
no longitudinal or between-group differences were seen.
The relationship between BP and survival in observational studies of dialysis patients can be complex due to
reverse causality in patients with low BP (14). There was
EAPOS: ULTRAFILTRATION AND MORTALITY
DAVIES et al.
this is in the context of poor membrane function or declining appetite. While it is not clear why clinicians were
reluctant or unable to increase the prescription of glucose and icodextrin in this study, it is certain that, due
to differences in membrane function, it would have been
necessary in order to gain equivalent fluid removal. It
would, therefore, be unsafe to conclude that the association between UF and survival is not amenable to therapeutic intervention. Further studies, preferably involving
randomization of treatment strategy, are required in
order to understand the relationship between reduced
fluid removal, fluid status, and increased mortality, and
optimal management.
APPENDIX
464
PDI
ACKNOWLEDGMENT
Baxter Europe (Brussels, Belgium) supported the study by providing data management and statistical support and by supporting investigator meetings.
REFERENCES
1. Brown EA, Davies SJ, Heimburger O, Meeus F, Mellotte G,
Rosman J, et al. Adequacy targets can be met in anuric
patients by automated peritoneal dialysis: baseline data
from EAPOS. Perit Dial Int 2001; 21(Suppl 3):S133–7.
2. Brown EA, Davies SJ, Rutherford P, Meeus F, Borras M,
Riegel W, et al. Survival of functionally anuric patients on
automated peritoneal dialysis: the European APD Outcome
Study. J Am Soc Nephrol 2003; 14:2948–57.
3. Davies SJ, Brown EA, Frandsen NE, Rodrigues AS,
Rodriguez-Carmona A, Vychytil A, et al. Longitudinal membrane function in functionally anuric patients treated with
APD: data from EAPOS on the effects of glucose and icodextrin prescription. Kidney Int 2005; 67:1609–15.
4. Jansen MA, Termorshuizen F, Korevaar JC, Dekker FW,
Boeschoten E, Krediet RT. Predictors of survival in anuric
peritoneal dialysis patients. Kidney Int 2005; 68:
1199–205.
5. Treatment of adult patients with renal failure: recommended standards and audit measures. The Standards and
Audit Subcommittee of the Renal Association on behalf
of the Renal Association and the Royal College of Physicians. J R Coll Physicians Lond 1995; 29:190–1.
6. Twardowski ZJ, Nolph KD, Khanna R, Prowant BF, Ryan LP,
Moore HL, et al. Peritoneal equilibration test. Perit Dial
Bull 1987; 7:138–47.
7. Davies SJ, Phillips L, Naish PF, Russell G. Quantifying comorbidity in peritoneal dialysis patients and its relationship to other predictors of survival. Nephrol Dial Transplant
2002; 17:1085–92.
8. Van Manen JG, Korevaar JC, Dekker FW, Boeschoten EW,
Bossuyt PM, Krediet RT. Adjustment for comorbidity in
studies on health status in ESRD patients: which comorbidity index to use? J Am Soc Nephrol 2003; 14:478–85.
9. Enia G, Sicuso C, Alati G, Zoccali C. Subjective global assessment of nutrition in dialysis patients. Nephrol Dial
Transplant 1993; 8:1094–8.
10. Churchill DN, Taylor DW, Keshaviah PR. Adequacy of dialysis and nutrition in continuous peritoneal dialysis: association with clinical outcome. J Am Soc Nephrol 1996;
7:198–207.
11. Davies SJ, Phillips L, Griffiths AM, Naish PF, Russell GI.
Analysis of the effects of increasing delivered dialysis
treatment to malnourished peritoneal dialysis patients.
Kidney Int 2000; 57:1743–54.
12. Asghar RB, Green S, Engel B, Davies SJ. Relationship of
demographic, dietary, and clinical factors to the hydration status of patients on peritoneal dialysis. Perit Dial
Int 2004; 24:231–9.
Downloaded from http://www.pdiconnect.com/ by guest on September 30, 2014
Members of EAPOS Group: M. Borras, Hospital Amau
de Vilanova, Lerida, Spain; E. Brown, Charing Cross Hospital, London, UK; A. Caillette-Beaudoin, Association
Calydial, Irigny, France; E. Clutterbuck, Hammersmith
Hospital, London, UK; C. D’Auzac, Hôpital Européen
Georges Pompidou, Paris, France; S.J. Davies, University
Hospital of North Staffordshire, Stoke-on-Trent, UK; A.
Ekstrand, Helsinki University Hospital, Helsinki, Finland;
N.E. Frandsen, Central Hospital Esbjerg, Esbjerg, Denmark; P. Freida, Centre Hospitalier Louis Pasteur,
Cherbourg, France; C. Friedrichsohn,* Universitätskliniken des Saarlandes, Homburg-Saar, Germany;
O. Heimbürger, CAPD-Enheten, Stockholm, Sweden;
D. Kuypers, U.Z. Gasthuisberg, Leuven, Belgium; E.
MacNamara, Centre Hospitalier Germon et Gauthier,
Bethune, France; R. Mactier, Stobhill Hospital NHS Trust,
Glasgow, Scotland, UK; G. Malmsten, Orebro Medical Center Hospital, Orebro, Sweden; F. Mastrangelo, Ospedale
Multizonale “Vito Fazzi,” Lecce, Italy; F. Meeus, Centre
Hospitalier Louise Michel Evry, Evry, France; G.J.
Mellotte, Adelaide & Meath Hospital, Dublin, Ireland;
J. Perez-Contreras, Hospital General Universitario,
Alicante, Spain; W. Riegel,† Universitätskliniken des
Saarlandes, Homburg-Saar, Germany; A.S. Rodrigues,
Hospital Geral de Santo António, Porto, Portugal; A.
Rodriguez-Carmona, Hospital Juan Canalejo, Coruna,
Spain; J. Rosman, Westeinde Hospital, Den Haag, The
Netherlands; P. Rutherford, Maelor General Hospital,
Clydd, Wales, UK; R. Scanziani, Ospedale Provinciale
Desio, Desio, Italy; N. Vega Diaz, Hospital Nuestra
Senora del Pino, Las Palmas, Spain; A. Vychytil, Univ.
Klinik für Innere Medizin III, Wien, Austria; T. Weinreich,
Dialyse Institüt Villingen-Schwenningen, Germany.
*Present address: Dialyse Institüt Villingen-Schwenningen, Germany; †Present address: Klinikum Darmstadt,
Darmstadt, Germany.
JULY 2006 – VOL. 26, NO. 4
PDI
JULY 2006 – VOL. 26, NO. 4
13. Jones CH, Wells L, Stoves J, Farquhar F, Woodrow G. Can a
reduction in extracellular fluid volume result in increased
serum albumin in peritoneal dialysis patients? Am J Kidney Dis 2002; 39:872–5.
14. Goldfarb-Rumyantzev AS, Baird BC, Leypoldt JK, Cheung
AK. The association between BP and mortality in patients
EAPOS: ULTRAFILTRATION AND MORTALITY
on chronic peritoneal dialysis. Nephrol Dial Transplant
2005; 20:1693–701.
15. Stenvinkel P, Heimburger O, Paultre F, Diczfalusy U, Wang
T, Berglund L, et al. Strong association between malnutrition, inflammation, and atherosclerosis in chronic renal
failure. Kidney Int 1999; 55:1899–911.
Downloaded from http://www.pdiconnect.com/ by guest on September 30, 2014
465