CLINICAL RESEARCH www.jasn.

org

High-Efficiency Postdilution Online Hemodiafiltration

Reduces All-Cause Mortality in Hemodialysis Patients
Francisco Maduell,* Francesc Moreso,† Mercedes Pons,‡ Rosa Ramos,§ Josep Mora-Macià,|
Jordi Carreras,¶ Jordi Soler,** Ferran Torres,††‡‡ Josep M. Campistol,*
and Alberto Martinez-Castelao,§§ for the ESHOL Study Group
*Nephrology Department, Hospital Clinic, Barcelona, Spain; †Nephrology Department, Hospital Universitari Vall
d’Hebron, Barcelona, Spain; ‡CETIRSA, Barcelona, Spain; §Hospital San Antonio Abad, Vilanova i la Geltru, Spain;
|
Fresenius Medical Care, Granollers, Spain; ¶Diaverum Baix Llobregat, L’Hospitalet, Llobregat, Spain; **Fresenius
Medical Care, Reus, Spain; ††Biostatistics Unit, School of Medicine, Universitat Autònoma de Barcelona, Barcelona,
Spain; ‡‡Biostatistics and Data Management Platform, IDIBAPS, Hospital Clinic, Barcelona, Spain; and §§Nephrology
Department, Hospital Universitari Bellvitge, L’Hospitalet, Bellvitge, Spain

ABSTRACT
Retrospective studies suggest that online hemodiafiltration (OL-HDF) may reduce the risk of mortality
compared with standard hemodialysis in patients with ESRD. We conducted a multicenter, open-label,
randomized controlled trial in which we assigned 906 chronic hemodialysis patients either to continue
hemodialysis (n=450) or to switch to high-efficiency postdilution OL-HDF (n=456). The primary outcome
was all-cause mortality, and secondary outcomes included cardiovascular mortality, all-cause hospitaliza-
tion, treatment tolerability, and laboratory data. Compared with patients who continued on hemodialysis,
those assigned to OL-HDF had a 30% lower risk of all-cause mortality (hazard ratio [HR], 0.70; 95% con-
fidence interval [95% CI], 0.53–0.92; P=0.01), a 33% lower risk of cardiovascular mortality (HR, 0.67; 95%
CI, 0.44–1.02; P=0.06), and a 55% lower risk of infection-related mortality (HR, 0.45; 95% CI, 0.21–0.96;
P=0.03). The estimated number needed to treat suggested that switching eight patients from hemodial-
ysis to OL-HDF may prevent one annual death. The incidence rates of dialysis sessions complicated by
hypotension and of all-cause hospitalization were lower in patients assigned to OL-HDF. In conclusion,
high-efficiency postdilution OL-HDF reduces all-cause mortality compared with conventional hemodialysis.
J Am Soc Nephrol 24: 487–497, 2013. doi: 10.1681/ASN.2012080875

In the last decades, renal replacement therapy with
hemodialysis has become a standard of care for
patients with ESRD. Despite continuous improve-
ment, annual mortality among these patients ranges
between 15% and 25%.1,2 Hemodialysis techniques
are based on the ability of molecules to diffuse
across a semipermeable membrane, which allows
adequate clearance of low molecular weight parti-
cles. To increase the clearance of middle-to-large
molecules, synthetic membranes with high water
permeability (high-flux membranes) were intro-
duced years ago. The anticipated benefit of high-
flux over low-flux hemodialysis on patient survival
was not confirmed in the Hemodialysis (HEMO)
study. 3 However, the Membrane Permeability
Outcome (MPO) study,4 as well as a post hoc anal-
ysis in diabetic patients, showed that high-flux
hemodialysis improved long-term survival in pa-
tients with hypoalbuminemia.
Clearance of middle-to-large molecules can be
increased by combining diffusive and convective
transport through hemodiafiltration. The intro-
duction of online hemodiafiltration (OL-HDF)
using ultrapure dialysate as the source of the
Received August 30, 2012. Accepted November 22, 2012.
Published online ahead of print. Publication date available at
www.jasn.org.
Correspondence: Dr. Francisco Maduell, Department of Ne-
phrology and Renal Transplantation, Hospital Clinic, University of
Barcelona, Villarroel, 170 - 08036 Barcelona, Spain. Email:
fmaduell@clinic.ub.es
Copyright © 2013 by the American Society of Nephrology

J Am Soc Nephrol 24: 487–497, 2013 ISSN : 1046-6673/2403-487 487

 CLINICAL RESEARCH www.jasn.org
replacement fluid has allowed the convective volume to be
increased and has reduced the cost of the procedure.5 Ran-
domized studies with limited sample sizes and nonrandomized
studies have shown that OL-HDF improves hemodynamic
stability and response to erythropoietic-stimulating agents
(ESAs) and reduces the incidence of hemodialysis-associated
amyloidosis and chronic inflammation. 6–11 The effect of
OL-HDF on patient survival is derived from noncontrolled
studies. The European Dialysis Outcomes and Practice Pattern
Study was associated with a mortality risk reduction of 35%
in patients treated with high-efficiency hemodiafiltration com-
pared with those treated with conventional hemodialysis.12
These results were confirmed in other noncontrolled studies
conducted in several European countries, 13–15 although
two recent, randomized studies failed to show differences in
patient survival16,17
In 2007, the Catalonian Health Authorities approved a
specific additional reimbursement for OL-HDF to dialysis care
providers and the Catalonian Society of Nephrology promoted
this randomized study (On-Line Hemodiafiltration Survival
Study, or Estudio de Supervivencia de Hemodiafiltración On-
Line [ESHOL]) to compare the effect of OL-HDF over hemo-
dialysis on patient survival.18
RESULTS
Baseline Patient Characteristics
During the recruitment period, 939 patients were assessed for
eligibility in 27 Catalonian dialysis units. Thirty-three patients
were not included in the randomization, 18 because they did
not meet the inclusion criteria, 5 because they refused to
provide informed consent, and 10 for other reasons. Finally,
906 patients were included in the randomization (Figure 1).
Patient characteristics at enrollment in both groups are sum-
marized in Table 1.
Hemodialysis Treatment
Hemodialysis treatment parameters during follow-up are
summarized in Table 2. Length of dialysis time did not differ
between the two groups, whereas blood flow (Qb) and dialy-
sate flow (Qd) were higher in the OL-HDF group throughout
the study. Among patients randomized to hemodialysis, ap-
proximately 92% (range, 91.8%–93.6%) were treated with
high-flux membranes during the study. The median quarterly
replacement volume and convective volume during the study
in the OL-HDF group ranged from 20.8 to 21.8 L/session and
22.9–23.9 L/session, respectively.
Primary Outcome
All-Cause Mortality
The mean follow-up was 1.9161.10 years (median 2.08 years;
interquartile range [IQR], 0.86–3.00). During the observation
period, 355 patients prematurely finished the study (Figure 1)
because of kidney transplantation (n=180), change of dialysis
488 Journal of the American Society of Nephrology
unit (n=58), organizational changes (n=33), withdrawal of
consent (n=27), need for a temporary catheter (n=19), change
of treatment (n=15), or because of other, not predefined rea-
sons (n=23). All of these patients were censored at the time of
premature termination.
There were 207 deaths (22.8%) during the follow-up, with
3-year all-cause mortality rates of 18.6% and 27.1% in the
OL-HDF and the hemodialysis groups, respectively, implying a
30% risk reduction (hazard ratio [HR], 0.70; 95% confidence
interval [95% CI], 0.53–0.92; P=0.01) (Figure 2 and Table 3).
The main causes of death were cardiovascular diseases
(44.4%) and infectious diseases (15.5%).
Sensitivity analyses were performed on the basis of the
following variables, which were found to be independent
predictors for all-cause mortality: age, sex, diabetes, the
Charlson comorbidity index, and vascular access (Table 4).
First, these variables were included in four distinct multivar-
iate analyses to assess the covariate-adjusted risk estimates for
the intervention. Second, the treatment risk estimates were
also calculated in all subgroups arising from these variables,
using the original categories for nominal variables and tertiles
for continuous variables. All HRs were consistent for both
types of analysis and the statistical tests for interaction were
not significant except for the Charlson comorbidity index
(Figure 3).The estimated number needed to treat (NNT)
for all-cause mortality at 1, 2, and 3 years was 9.75 (5.03–
47.41), 7.67 (4.32–33.57), and 7.67 (4.51–31.83), respectively
(Table 3).
Secondary Outcomes
Cardiovascular Mortality
The analysis of the distinct causes of cardiovascular mortality
(Table 3) showed no significant differences between groups
in the number of deaths due to heart failure, ischemic heart
disease, mesenteric thrombosis, arrhythmia, or peripheral
arteriopathy. However, stroke mortality was significantly
lower in the OL-HDF group than in the hemodialysis group
(P=0.03 by the log-rank test). The Cox proportional hazards
model showed that OL-HDF caused a significant 61% risk
reduction in mortality from stroke (HR, 0.39; 95% CI,
0.16–0.93).
Other Causes of Mortality
Infection-related mortality was also significantly lower in the
OL-HDF group than in the hemodialysis group (P=0.03 by the
log-rank test; Table 3). The Cox proportional hazards model
showed that OL-HDF produced a significant 55% risk
reduction in mortality from infection (HR, 0.45; 95% CI,
0.21–0.96). There were no differences in sudden death, cancer,
cachexia, or other causes of mortality (Table 3).
Hospitalizations
The rate of all-cause hospital admissions showed a relative risk
reduction of 22% favorable to the OL-HDF group (rate ratio,
0.78; 95% CI, 0.67–0.90; P=0.001) (Table 5).
J Am Soc Nephrol 24: 487–497, 2013
 www.jasn.org CLINICAL RESEARCH

Figure 1. Flow chart of study populations, including the number of patients who were screened, underwent randomization, and
completed the study treatment or presented the primary variable. HD, hemodialysis.

Table 1. Demography characteristics and baseline parameters in randomized episodes per 100 patient-years in the OL-
patients HDF group versus 937.7 episodes per 100
All (n=906) Hemodialysis (n=450) OL-HDF (n=456) patient-years in the hemodialysis group (rate
ratio, 0.72; 95% CI, 0.68–0.77; P,0.001)
Age (yr) 65.4614.4 66.3614.3 64.5614.4
Male sex 606 (66.9) 289 (64.2) 317 (69.5)
(Table 4). The number of intradialysis epi-
Diabetes 226 (24.9) 122 (27.1) 104 (22.8) sodes of arrhythmia and thoracic pain
Charlson comorbidity index 7.0 (5.0–8.0) 7.0 (5.0–8.0) 6.0 (5.0–8.0) episodes did not differ between the groups
Time on dialysis (mo) 28.0 (12.0–59.0) 27.0 (12.0–58.0) 28.5 (12.0–60.0) (Table 5).
Vascular access
Fistula 779 (86.0) 372 (82.7) 407 (89.3) Dialysis Dose and b2-Microglobulin
Graft 34 (3.8) 19 (4.2) 15 (3.3) During the study period, the dialysis dose
Catheter 93 (10.3) 59 (13.1) 34 (7.5) significantly increased in both groups but
Data are presented as mean 6 SD, n (%), or median (IQR). was higher in patients randomized to OL-
HDF than in those randomized to hemodi-
alysis (Supplemental Table 1).
BP and Intradialysis Tolerance OL-HDF led to a lower accumulation of b2-microglobulin than
Neither predialysis nor postdialysis systolic and diastolic BP did hemodialysis. b2-Microglobulin increased from month 0 to
were significantly modified by treatment, although BP control month 36 in both groups but to a lesser extent in the OL-HDF
was improved in both treatment groups during the study group (Supplemental Table 1).
(Supplemental Table 1). The percentage of patients requiring
antihypertensive drugs did not differ between the groups Nutrition
(Supplemental Table 2). Dry body weight and albumin changes during follow-up did
The incidence of intradialysis symptoms was affected by not differ between treatment groups. However, albumin
treatment assignment. There were 679.2 intradialysis hypotension significantly decreased in both groups throughout the study.

J Am Soc Nephrol 24: 487–497, 2013 OL-HDF and Survival: ESHOL Study 489
490
Table 2. Dialysis parameters outcome
Length of Follow-Up
CLINICAL RESEARCH

P
Baseline 6 mo 12 mo 18 mo 24 mo 30 mo 36 mo
Dialysis time (min)
Hemodialysis 234.1 (232.2, 236.1) 235.5 (234.1, 236.9) 235.6 (234.1, 237.1) 235.7 (234.1, 237.4) 236.5 (234.7, 238.3) 238.3 (236.3, 240.2) 237.5 (235.3, 239.6) 0.07a
OL-HDF 235.8 (234.2, 237.3) 234.4 (233.0, 235.9) 234.2 (232.6, 235.7) 234.8 (233.3, 236.4) 235.7 (233.9, 237.5) 235.9 (234.0, 237.8) 236.6 (234.6, 238.7) 0.14b
www.jasn.org

Time 3 time — 0.30 0.18 0.43 0.55 0.09 0.58 0.80c

Qb (ml/min)
Hemodialysis 380 (374, 387) 367 (363, 371) 371 (366, 375) 375 (370, 380) 370 (365, 376) 378 (372, 384) 380 (374, 387) ,0.001a
OL-HDF 392 (387, 398) 384 (380, 388) 386 (381, 391) 384 (379, 389) 385 (379, 390) 385 (379, 391) 389 (383, 395) ,0.001b

Journal of the American Society of Nephrology

Time 3 time — ,0.001 ,0.001 0.01 0.001 0.09 0.04 0.54c
Qd (ml/min)
Hemodialysis 531 (520, 542) 560 (553, 568) 554 (546, 561) 550 (542, 558) 545 (536, 554) 548 (538, 557) 537 (526, 547) ,0.001a
OL-HDF 553 (541, 564) 571 (564, 579) 571 (563, 578) 575 (566, 583) 575 (566, 583) 580 (570, 590) 566 (556, 577) ,0.001b
Time 3 time — 0.04 0.002 0.001 0.001 0.001 0.001 0.14c
Replacement
volume (L/session)
Hemodialysis — — — — — — — ,0.001a
OL-HDF — 21.8 (21.6, 22.0) 21.5 (21.3, 21.8) 21.1 (20.8, 21.4) 20.8 (20.5, 21.1) 21.2 (20.9, 21.5) 21.7 (21.3, 22.0) 0.01b
Time 3 time 0.001 0.001 0.001 0.001 0.001 0.001 0.01c
Intradialysis weight
change (pre–post)
Hemodialysis 2.1 (2.0, 2.2) 2.1 (2.0, 2.2) 2.1 (2.0, 2.3) 2.1 (1.9, 2.2) 2.3 (2.1, 2.4) 2.0 (1.8, 2.1) 2.1 (1.9, 2.3) 0.91a
OL-HDF 2.1 (2.0, 2.2) 2.1 (1.9, 2.2) 2.3 (2.1, 2.4) 2.1 (2.0, 2.3) 2.1 (1.9, 2.2) 2.1 (2.0, 2.3) 2.1 (1.9, 2.2) 0.12b
Time 3 time 0.77 0.13 0.56 0.08 0.17 0.76 0.56c
Convection
volume (L/session)
Hemodialysis 2.1 (2.0, 2.2) 2.1 (1.8, 2.4) 2.2 (1.9, 2.4) 2.1 (1.7, 2.4) 2.3 (1.9, 2.6) 2.0 (1.6, 2.4) 2.1 (1.7, 2.5) ,0.001a
OL-HDF 2.1 (2.0, 2.2) 23.9 (23.6, 24.2) 23.8 (23.5, 24.1) 23.2 22.9, 23.5) 22.9 (22.5, 23.2) 23.3 (22.9, 23.7) 23.7 (23.3, 24.1) 0.003b
Time 3 time 0.001 0.001 0.001 0.001 0.001 0.001 0.01c
Data are shown for every 6 months but were collected every 3 months. Data are presented as mean (95% CI) at baseline, and least-squares mean (95% CI) at 6 months, 12 months, 18 months, 24 months, 30 months,
and 36 months. Analysis was conducted using a mixed model for repeated measures. Time 3 time indicates time 3 time comparisons, which are given as P values.
a
P value of treatment effect.
b
P value of time effect.
c
P value of treatment 3 time interaction.

J Am Soc Nephrol 24: 487–497, 2013

 www.jasn.org CLINICAL RESEARCH

The normalized protein catabolic rate was higher in patients
randomized to OL-HDF (Supplemental Table 1).
Anemia
Hemoglobin, the transferrin saturation index, and ferritin did
not differ between the groups during the study period (Sup-
plemental Table 1). Moreover, there were no differences in the
proportion of patients treated with distinct ESA. Intravenous
iron supplements and ESA doses did not differ between the
groups (Supplemental Table 2).
Phosphate and Other Biochemical Parameters
Serum phosphate (Supplemental Table 1) and the number of
phosphate binding tablets (Supplemental Table 2) did not
significantly change during the study and did not differ between
the treatment groups. There were no significant changes for
predialysis C-reactive protein, creatinine, sodium, potassium,
uric acid, calcium, and intact parathyroid hormone (Supple-
mental Table 3).
Influence of Convection Volume on All-Cause Mortality
In post hoc analyses, the association between convective vol-
ume per session in comparison with convective volume per
square meter of body surface area and convective volume per
body mass index was evaluated (Supplemental Table 4). The
convective volume per session is the variable that best repre-
sents the trend compared with the ratio of the convective vol-
ume per session both with the body mass index and with the
body surface area. In the group of patients
with the highest delivered convection vol-
ume, mortality in the intermediate tertile
(23.1–25.4 L) and upper tertile (.25.4 L)
was considered lower than that in patients
randomized to hemodialysis (HR, 0.60;
95% CI, 0.39–0.90; and HR, 0.55; 95%
CI, 0.34–0.84, respectively).

DISCUSSION

In this prospective, randomized clinical

trial, we found that high-efficiency OL-
HDF in patients with ESRD on hemodial-
ysis was associated with a 30% reduction in
all-cause mortality compared with conven-
tional high-flux hemodialysis. This mor-
tality reduction was related to a significant
Figure 2. Kaplan–Meier curves for 36-month survival in the intention-to-treat pop- risk reduction for stroke and infectious
ulation (P=0.01 by the log-rank test). HD, hemodialysis. mortality. The estimated NNT showed that

Table 3. Primary outcome: Mortality

Hemodialysis Group (n=450) OL-HDF Group (n=456)
(867.3 patient-years at risk) (863.1 patient-years at risk)
HR (95% CI) Pa
Events/100 Events/100
Events Events
Patient-Years Patient-Years
Death from any cause 122 14.1 85 9.8 0.70 (0.53–0.92) 0.01
Cardiovascular cause 55 6.3 37 4.3 0.67 (0.44–1.02) 0.06
Heart failure 10 1.2 7 0.8 0.69 (0.26–1.82) 0.46
Ischemic heart disease 15 1.7 14 1.6 0.93 (0.45–1.94) 0.86
Mesenteric thrombosis 6 0.7 5 0.6 0.84 (0.26–2.77) 0.78
Stroke 18 2.1 7 0.8 0.39 (0.16–0.93) 0.03
Dysrhythmia 5 0.6 3 0.3 0.59 (0.14–2.47) 0.46
Peripheral arteriopathy 1 0.0 1 0.0 0.97 (0.06–15.48) 0.98
Infection 22 2.5 10 1.2 0.45 (0.21–0.96) 0.03
Tumor 6 0.7 10 1.2 1.67 (0.61–4.59) 0.32
Sudden death 14 1.6 14 1.6 0.99 (0.47–2.08) 0.98
Cachexia 8 0.9 4 0.5 0.51 (0.15–1.70) 0.27
Death from other causes 17 2.0 10 1.2 0.59 (0.27–1.28) 0.18
Data are given as n or NNT (95% CI). NNTs (95% CIs) for all-cause mortality at 1, 2, and 3 years were 9.75 (5.03–47.41), 7.67 (4.32–33.57) and 7.67 (4.51–31.83), respectively.
a
P value by the log-rank test.

J Am Soc Nephrol 24: 487–497, 2013 OL-HDF and Survival: ESHOL Study 491
 CLINICAL RESEARCH www.jasn.org

Table 4. Univariate cox regression analysis for all-cause mortality again, the all-cause outcome was not affected by
HR (95% CI) P treatment allocation over 2 years of follow-up.17
OL-HDF (reference: hemodialysis group) 0.70 (0.53–0.92) 0.01 On the basis of the results of the CONTRAST
Age (per 1-yr increment) 1.05 (1.03–1.06) ,0.001 and Turkish studies, the convective volume seems
Sex (reference: female) 1.43 (1.05–1.94) 0.02 to be an important issue. In the post hoc analysis,
Diabetes (reference: absence of diabetes) 1.43 (1.07–1.91) 0.02 both studies showed a 39% and 46% mortality
Charlson comorbidity index (per 1-unit increment)a 1.37 (1.28–1.47) ,0.001 risk reduction in patients receiving high convec-
Vascular access (reference: fistula) 1.45 (1.20–1.76) ,0.001 tion volumes (.22 and .20 L/session, respec-
a
The Charlson comorbidity index was calculated excluding diabetes. tively). Similarly, the post hoc analysis in the
ESHOL study showed a 40% and 45% mortality
to prevent one annual death, eight patients would need to be risk reduction in patients receiving convection volumes between
switched from hemodialysis to OL-HDF. To our knowledge, 23–25 L/session and .25 L/session, respectively. These results
this is the first prospective, randomized clinical trial showing provide evidence of the need to deliver high convection volumes
that high-efficiency postdilution OL-HDF reduces all-cause to reduce all-cause mortality. To achieve this goal, high blood
mortality compared with conventional hemodialysis in the flow rates and long dialysis times are required. In this ESHOL
chronic prevalent population. The survival benefit observed in study, the mean blood flow rate was higher (387 ml/min) than in
this study has sufficient statistical robustness to support the the CONTRAST and Turkish studies (300 and 310 ml/min, re-
finding that OL-HDF was the major determinant of survival. spectively), whereas the mean length of dialysis was longer (236
Furthermore, the results were consistent in distinct subgroups of minutes) than in the CONTRAST study (226 minutes) and was
patients according to age, sex, diabetes mellitus, the Charlson similar to that of the Turkish study (236 minutes). These factors
comorbidity index, and vascular access. Although all patient led to a higher mean delivered convection volume in this study
groups benefitted from OL-HDF, the subgroups obtaining the (23.7 L/session) than in the CONTRAST (20.7 L) and Turkish
greatest benefit were older, had no diabetes, were dialyzed through (20.7 L) studies. These results indicate that the treatment modality
an arteriovenous fistula, and had a higher Charlson comorbidity could modify patient survival when a sufficient convective volume
index. is reached. Therefore, in future studies, the convective volume dose
The ESHOL study was designed to compare high-flux should be defined as a minimum postdilution volume, a minimum
hemodialysis versus postdilution OL-HDF. However, a small infusion flux, a convective volume normalized to body-size, or
proportion of patients on low-flux hemodialysis (6%) were other factors.
allowed to be included; therefore, this study represents routine The reduction in all-cause mortality associated with OL-
clinical practice in Catalonia. In this study, we did not record HDF treatment observed in this trial was focused on cardio-
the number of screened patients, which is a potential limitation vascular and infectious diseases. Cardiovascular disease is the
to the external validity of the study. However, the patients most common cause of mortality in chronic hemodialysis
included in this trial corresponded to .25% of all patients patients and the mortality rate is still 10 times higher than in the
maintained on hemodialysis in Catalonia and the mean age as general population.19 Cardiovascular disease contributed to
well as the proportion of patients with diabetes were similar to 44.4% of mortality in our study and there was a trend toward
data reported by the Catalonian Registry of Renal Patients.1 lower risk in the OL-HDF group. This result links with the
Importantly, as in other trials conducted in hemodialysis finding that convective techniques are associated with a higher
patients, a significant number of observations were censored, removal of several mediators involved in inflammation regu-
mainly due to renal transplantation. Indeed, in our trial, 38% lation, cytokine production, and accelerated atherosclerosis.20
of the patients were censored before reaching the end-point of Remarkably, stroke risk reduction largely contributed to the
the study or completing 3 years of follow-up, a figure close to reduction of cardiovascular mortality, even though the clinical
those reported in other hemodialysis studies (32% for the variables associated with stroke in epidemiologic studies such
HEMO study,3 37% for the MPO study,4 and 33% for the as BP control21 or ESA doses22 did not differ between the
Convective Transport [CONTRAST] study ). 16 groups. The significant reduction of intradialytic symptomatic
There issome controversy on the potential benefits of OL-HDF hypotension episodes with OL-HDF observed in this trial may
in mortality risk reduction. Retrospective, observational studies have contributed to reducing cardiovascular events and
suggested that OL-HDF could improve patient survival,12–15 especially stroke. This benefit of OL-HDF on intradialytic hy-
although two recent, prospective, randomized studies failed to potension episodes has previously been described.23,24 In ad-
demonstrate a survival advantage of OL-HDF over hemodi- dition, infection-associated mortality was also reduced by
alysis.16,17 In the CONTRAST study,16 also conducted in a prev- OL-HDF. ESRD patients have a significant risk of infectious
alent population, 714 patients were randomized to low-flux complications, which represent the first cause of hospitaliza-
hemodialysis or OL-HDF and no survival difference between tion and the second cause of death in hemodialysis patients.
the groups was observed at the end of the study with a mean Hemodiafiltration has been associated with a reduced in-
follow-up of 3 years. In the Turkish study, 782 prevalent patients flammatory state11 and improved granulocyte function.25
were randomized to high-flux hemodialysis or OL-HDF and, Several granulocyte-inhibitory proteins retained in uremic

492 Journal of the American Society of Nephrology J Am Soc Nephrol 24: 487–497, 2013
 www.jasn.org CLINICAL RESEARCH

Figure 3. Sensitivity analyses for the main outcome showing HRs (95% CIs) for the intervention based on relevant variables that were
found to be independent predictors for all-cause mortality. Multivariate I, age, sex, vascular access, diabetes, and the Charlson co-
morbidity index (excluding diabetes); multivariate II, age, sex, vascular access, and the Charlson comorbidity index (excluding diabetes);
multivariate III, age, sex, vascular access, and diabetes; multivariate IV, age, sex, vascular access, and the Charlson comorbidity index
(including diabetes); Y, yes; N, no; Cath, catheter; Fist, fistula; M, male; F, female; T1, T2, and T3, first, second, and third tertiles; HD,
hemodialysis.

J Am Soc Nephrol 24: 487–497, 2013 OL-HDF and Survival: ESHOL Study 493
 CLINICAL RESEARCH www.jasn.org

Table 5. Outcome data: Hospitalizations and intradialysis symptoms

Hemodialysis Group (n=450) OL-HDF Group (n=456)
(867.3 Patient-Years at Risk) (863.1 Patient-Years at Risk) Rate Ratio
Pa
No. of No. of Events/100 No. of No. of Events/ (95% CI)
Events Patient-Years Events 100 Patient-Years
All-cause hospitalizations 412 47.5 317 36.7 0.78 (0.67–0.90) 0.001
Infections 73 8.4 72 8.3
Vascular access 98 11.3 56 6.5
Heart failure 28 3.2 15 1.7
Ischemic heart disease 25 2.9 16 1.9
Respiratory disease 26 3.0 28 3.2
Gastrointestinal bleeding 10 1.2 4 0.5
Other reasons 152 17.5 126 14.6
Symptomatic hypotension episodesb 8133 937.7 5862 679.2 0.72 (0.68–0.77) ,0.001
Dysrythmiab 444 51.2 477 55.3 1.08 (0.86–1.35) 0.50
Thoracic painb 327 37.7 318 36.8 0.98 (0.75–1.28) 0.87
Data are the number of hospitalizations (hospital admissions per 100 patient-years) and number of complicated dialysis sessions (number of symptoms per 100
patient-years).
a
Penalized quasi-likelihood under restricted maximum likelihood models.
b
Pooled data (data were collected 1 month per each 3-month period).

patients—contributing to the high incidence of infections—
could be better removed with OL-HDF therapies.26 OL-HDF
was also associated with a significant reduction in hospitaliza-
tion rates, which could reasonably be linked to the decrease in
cardiovascular and infectious complications.
Clearance of middle-to-large molecules depends on the
type of dialysis membrane and the amount of convection
volume and may be increased with OL-HDF treatment.
However, the observed increase of b2-microglobulin serum lev-
els in both groups during this study was an unexpected finding.
Previous data27 showed that there is a positive correlation be-
tween infusion volume and the b2-microglobulin reduction ra-
tio but this does not mean that predialysis levels were reduced,
because b2-microglobulin has a low distribution volume.28
Moreover, several studies have shown that predialysis b2-
microglobulin levels were reduced after patients were switched
from high-flux hemodialysis to OL-HDF.7,29 Unfortunately, re-
sidual renal function, one of the major determinants of b2-
microglobulin serum levels, was not monitored during the study
period. These results suggest that the benefit of OL-HDF to
patient survival partly depends on clearance of molecules other
than b2-microglobulin. OL-HDF can remove other middle-
sized molecules or protein-bound uremic toxins more efficiently
than hemodialysis, which may influence endothelial function,
inflammatory status, or vascular calcification, providing cardi-
oprotective effects and/or improving the immunologic system.
The ESHOL study has some weaknesses but also major
strengths. A limitation is that a small proportion of patients
(6.3%) were treated with low-flux membranes and, addition-
ally, residual renal function was not monitored. Small differ-
ences might be suggested in diabetic patients, grafts, or
catheters between the two groups in our study. To address
this point, additional multivariate Cox regression sensitivity
analyses were conducted, adjusting by age, sex, diabetes, the
Charlson comorbidity index, and type of vascular access. The
results from the multivariate analyses were robust and the risk
estimates and 95% CIs were very much in line with the main
analysis conducted with any baseline adjustments. Because this
was a randomized study, the apparent small baseline differ-
ences were considered likely to be obtained by chance rather
than due to bias, and this statement holds true because of the
randomization procedure. The lower number of intradialysis
hypotension episodes observed in this study was not apparently
due to major differences in sodium removal among treatments,
as suggested by the similar pretreatment and post-treatment
plasma sodium concentrations, the similar ultrafiltration rate,
and the similar dry body weight behavior during the follow-up;
however, our measurement was a crude estimation of dialytic
sodium removal and therefore the possibility of a negative
sodium balance during OL-HDF could not be ruled out.23 The
strengths of this study include the randomized design, the
large sample size and long follow-up, as well as the require-
ment of high convection volumes. Hemodialysis units partici-
pating in this clinical trial were trained to utilize high blood
flow rates and long dialysis times in order to deliver a mean
convective volume close to 24 L/session.
In summary, the results of the ESHOL trial indicate that
high-efficiency postdilution OL-HDF reduces all-cause mor-
tality versus conventional hemodialysis in prevalent patients.
Furthermore, the main causes of mortality, cardiovascular and
infectious diseases, were significantly reduced by OL-HDF. In
view of these results, OL-HDF may become the first-line option
in hemodialysis patients.
CONCISE METHODS
Study Design
The study design was previously reported.30 The ESHOL study was
a prospective, randomized, open-label clinical trial in patients with

494 Journal of the American Society of Nephrology J Am Soc Nephrol 24: 487–497, 2013

ESRD under hemodialysis in Catalonia, Spain. All hemodialysis units
in Catalonia, both in-hospital and out-of-hospital units, were invited
to participate. All patients signed consent forms approved by the
Hospital Ethics Committee. The registered protocol number is
NCT 00694031.18
The primary objective was to assess the effect of postdilution OL-
HDF compared with hemodialysis, either low-flux or high-flux, on
all-cause mortality. The primary outcome variable was the time to
occurrence of death from any cause. Key secondary outcomes were
cardiovascular mortality, other causes of mortality, all-cause hospi-
talization, dialysis dose (Kt/V and urea reduction ratio), BP control
and intradialysis tolerance (symptomatic hypotension episodes,
arrhythmia, and thoracic pain), nutrition (dry body weight, normal-
ized protein catabolic rate, albumin), anemia, and phosphate and
b2-microglobulin serum levels.
Study Population
The study population was previously described.30 Essentially, the in-
clusion criteria were patients aged .18 years with ESRD receiving
thrice-weekly standard hemodialysis for .3 months. Exclusion cri-
teria were as follows: active systemic diseases, liver cirrhosis, malig-
nancy, immunosuppressive therapy, infradialysis dose (Kt/V,1.3),
single needle dialysis, and temporary nontunnelized catheter.
Randomization and Dialysis Treatment Parameters
Patients were randomized 1:1 to continue on thrice-weekly hemo-
dialysis or to start OL-HDF three times a week. A central computerized
random-generator was utilized to allocate patients to each study group
and randomization was stratified by center. The length of the recruit-
ment period was 16 months and the study was completed in order to
provide a follow-up of 3 years for all surviving patients.
Treatment Procedures
Synthetic high-flux dialyzers were used for OL-HDF (FX60: 59.7%
and FX80: 8.6% [Fresenius Medical Care, Bad Homburg, Germany];
Polyflux 170H: 7.9%, Polyflux 210H: 10.3%, and Arylane H9: 1.5%
[Gambro AB, Stockholm, Sweden]; or other dialyzers: 12.5%).
Patients randomized to hemodialysis were treated with synthetic
high-flux (FX60: 58.7%, FX80: 8.4%, Polyflux 170H: 10.4%, Polyflux
210H: 5.5%, Arylane H9: 0.9% and other dialyzers: 12.5%) or low-flux
dialyzers (8.1%).
Both OL-HDF and hemodialysis were performed with ultrapure
dialysis fluids, defined as ,0.1 CFU/ml and ,0.03 EU/ml. The length
of dialysis sessions in each treatment modality was not modified.
When OL-HDF could not be performed temporarily for technical
reasons, affected patients were treated with the same high-flux
membranes. For patients on postdilution OL-HDF, a minimum of
18 L/session of replacement volume was requested. Patients not
receiving the allocated treatment modality for .2 consecutive
months were withdrawn from the study.
The composition of dialysate was the same in both groups, as was
the reinfusate in OL-HDF (sodium 138–140 mmol/L, potassium 1.5–
2.0 mmol/L, calcium 1.25–1.75 mmol/L, magnesium 0.5 mmol/L,
chloride 106–109 mmol/L, bicarbonate 34–37 mmol/L, acetate 3–4
mmol/L, and glucose 1.0 g/L).
J Am Soc Nephrol 24: 487–497, 2013
www.jasn.org CLINICAL RESEARCH
Study Variables
Before randomization, the Charlson comorbidity index score was
calculated for each patient. The following parameters were recorded at
baseline and every 3 months: dialyzer characteristics, dialysis time, Qb,
Qd, vascular access, replacement volume, dry body weight, predialysis
and postdialysis body weight, convective volume, and predialysis and
postdialysis systolic and diastolic BP.
The following laboratory data were recorded at baseline and every
3 months: predialysis urea, creatinine, bicarbonate, sodium, potassium,
C-reactive protein, uric acid, albumin, calcium, phosphate, intact
parathyroid hormone, b2-microglobulin, hemoglobin, transferrin
saturation index, and ferritin. Using predialysis and postdialysis
BUN in a mid-week dialysis session, the urea reduction rate, dialysis
dose (Kt/V by Daugirdas’ second-generation single-pool variable vol-
ume formula) and normalized protein catabolic rate were calculated
by standard formulas. All laboratory determinations were performed
locally by standard procedures in certified laboratories. The doses of
ESAs, iron supplements, antihypertensive drugs, and phosphate
binders were also recorded at baseline and every 3 months.
Clinical monitoring included intradialysis symptoms (sympto-
matic hypotension, episodes of arrhythmia and thoracic pain),
hospital admissions for any reasons, and withdrawals from the study
and their causes. To calculate the number of dialysis sessions
complicated by intradialysis symptoms, the last 12 sessions preceding
the quarterly visit were considered.
Sample Size
According to the Catalonian Registry of Renal Patients, 3-year patient
survival in hemodialysis patients was 70%.1 Because several retro-
spective studies have suggested a 35% reduction of mortality in pa-
tients on OL-HDF,12 we expected a 3-year survival of 80% in patients
on OL-HDF. When the sample size in each group is 296, with a total
number of events required of 143, a 0.05 level two-sided log-rank test
will have 80% power to detect the difference between an OL-HDF
group proportion of 0.8 and a hemodialysis group proportion of 0.7
at time t, assuming a constant HR of 0.626. To compensate for drop-
outs, which were expected due to renal transplantation or loss to
follow-up, we planned to enroll at least 400 patients per group.
Statistical Analyses
All data analyses were carried out according to a pre-established statisti-
cal analysis plan. All-cause mortality, as well as cardiovascular death,
cachexia, infection, tumors, sudden death, and death from other causes
were described by means of the Kaplan–Meier method. The log-rank test
was used for hypothesis testing, and the HR and its 95% CI were esti-
mated from the unadjusted Cox model. Additional multivariate Cox
regression sensitivity analyses were conducted, adjusting by age, sex,
diabetes, the Charlson comorbidity index (the original scale and also
excluding diabetes), and the type of vascular access to assess the robust-
ness of the study results. The NNT estimated after the switch from
hemodialysis to OL-HDF was also calculated.
Time repeated measurements were analyzed using linear mixed
models including treatment, time, and the treatment by time in-
teraction term. Gaussian continuous variables were approached
through mixed models for repeated measurements, adjusting by
OL-HDF and Survival: ESHOL Study 495
 CLINICAL RESEARCH www.jasn.org
the baseline value and, for non-Gaussian data (binomial and Poisson
distribution variables), by penalized quasi-likelihood under restricted
maximum likelihood models. Treatment inferences, effect estimates,
and 95% CIs were taken from these models.
Two-sided significance tests were used throughout, and a P value of
,0.05 was considered significant. All statistical analyses were per-
formed using the SAS 9.2 statistical package.
ACKNOWLEDGMENTS
The following institutions and investigators participated in the ESHOL
study: M. Pons, B. Insensé, C. Perez, and T. Feliz (CETIRSA, Barcelona);
R. Ramos, M. Barbetta, and C. Soto (Hospital San Antonio Abad,
Vilanova i la Geltru); J. Mora, A. Juan, and O. Ibrik (Fresenius Medical
Care, Granollers); A. Foraster and J. Carreras (Diaverum Baix Llobregat,
Hospitalet); F. Moreso, J. Nin, and A. Fernández (Fresenius Medical
Care, Hospitalet); J. Soler, M. Arruche, C. Sánchez, and J. Vidiella
(Fresenius Medical Care, Reus); F. Barbosa, M. Chiné, and S. Hurtado
(Fresenius Medical Care Diagonal, Barcelona); J. Llibre, A. Ruiz,
M. Serra, M. Salvó, and T. Poyuelo (CETIRSA, Terrassa); F. Maduell,
M. Carrera, N. Fontseré, M. Arias, and Josep M. Campistol (Hospital
Clínic, Barcelona); A. Merín and L. Ribera (Fresenius Medical Care Julio
Verne, Barcelona); J.M. Galceran, J. Mòdol, E. Moliner, and A. Ramirez
(Fundació Althaia, Manresa); J. Aguilera and M. Alvarez (Hospital Santa
Tecla, Tarragona); B. de la Torre and M. Molera (Diaverum Bonanova,
Barcelona); J. Casellas and G. Martín (Diaverum IHB, Barcelona);
E. Andres and E. Coll (Fundació Puigvert, Barcelona); M. Valles and
C. Martínez (Hospital Josep Trueta, Girona); E. Castellote (Hospital
General, Vic); J.M. Casals, J. Gabàs, and M. Romero (Diaverum,
Mataró); A. Martinez-Castelao and X. Fulladosa (Hospital Universitari
Bellvitge, Hospitalet); M. Ramirez-Arellano and M. Fulquet (Hospital
de Terrassa); A. Pelegrí, M. el Manouari, and N. Ramos (Diaverum
Verge de Montserrat, Santa Coloma); J. Bartolomé (Centre Secretari
Coloma, Barcelona); R. Sans (Hospital de Figueres); E. Fernández and
F. Sarró (Hospital Arnau de Vilanova, Lleida); T. Compte (Hospital
Santa Creu, Tortosa); F. Marco and R. Mauri (Diaverum Nephros,
Barcelona); and J. Bronsoms (Clínica Girona). The clinical trials unit
comprised J.A. Arnaiz, H. Beleta, and A. Pejenaute (UASP Farmacología
Clínica, Hospital Clínic Barcelona). Statistical analyses were performed
by F. Torres, J. Ríos, and J. Lara (Biostatistics Unit, School of Medicine,
Universitat Autònoma de Barcelona; and Biostatistics and Data Man-
agement Platform, IDIBAPS, Hospital Clinic, Barcelona).
The Catalan Society of Nephrology has endorsed the ESHOL study.
This study was partly supported by grants from Fresenius Medical
Care and Gambro through the Catalan Society of Nephrology.
DISCLOSURES
None.
REFERENCES
1. Generalitat de Catalunya. Departament de Salut, Organització Catalana
de Trasplantaments: Registre malalts renals de Catalunya. Informe
496 Journal of the American Society of Nephrology
estadístic 2005-2006. Available at: http://biblio.idescat.cat/docs/pec/
paae2008/gi04652006.pdf. Accessed June 2012
2. Registro Español de Enfermos Renales: Dialysis and transplant report in
Spain, 2006 (in English). Nefrologia 29: 525–533, 2009
3. Eknoyan G, Beck GJ, Cheung AK, Daugirdas JT, Greene T, Kusek JW,
Allon M, Bailey J, Delmez JA, Depner TA, Dwyer JT, Levey AS, Levin
NW, Milford E, Ornt DB, Rocco MV, Schulman G, Schwab SJ, Teehan
BP, Toto R; Hemodialysis (HEMO) Study Group: Effect of dialysis dose
and membrane flux in maintenance hemodialysis. N Engl J Med 347:
2010–2019, 2002
4. Locatelli F, Martin-Malo A, Hannedouche T, Loureiro A, Papadimitriou
M, Wizemann V, Jacobson SH, Czekalski S, Ronco C, Vanholder R;
Membrane Permeability Outcome (MPO) Study Group: Effect of
membrane permeability on survival of hemodialysis patients. J Am Soc
Nephrol 20: 645–654, 2009
5. Canaud B, Flavier JL, Argilés A, Stec F, NGuyen QV, Bouloux C, Garred
LJ, Mion C: Hemodiafiltration with on-line production of substitution
fluid: Long-term safety and quantitative assessment of efficacy. Contrib
Nephrol 108: 12–22, 1994
6. Locatelli F, Marcelli D, Conte F, Limido A, Malberti F, Spotti D: Com-
parison of mortality in ESRD patients on convective and diffusive ex-
tracorporeal treatments. The Registro Lombardo Dialisi E Trapianto.
Kidney Int 55: 286–293, 1999
7. Maduell F, del Pozo C, Garcia H, Sanchez L, Hdez-Jaras J, Albero MD,
Calvo C, Torregrosa I, Navarro V: Change from conventional haemo-
diafiltration to on-line haemodiafiltration. Nephrol Dial Transplant
14: 1202–1207, 1999
8. Lornoy W, Becaus I, Billiouw JM, Sierens L, Van Malderen P, D’Haenens
P: On-line haemodiafiltration. Remarkable removal of b2-microglobulin.
Long-term clinical observations. Nephrol Dial Transplant 15[Suppl 1]:
49–54, 2000
9. Ward RA, Schmidt B, Hullin J, Hillebrand GF, Samtleben W: A com-
parison of on-line hemodiafiltration and high-flux hemodialysis: A
prospective clinical study. J Am Soc Nephrol 11: 2344–2350, 2000
10. Wizemann V, Lotz C, Techert F, Uthoff S: On-line haemodiafiltration
versus low-flux hemodialysis. A prospective randomized study. Neph-
rol Dial Transplant 15[Suppl 1]: 43–48, 2000
11. Carracedo J, Merino A, Nogueras S, Carretero D, Berdud I, Ramírez R,
Tetta C, Rodríguez M, Martín-Malo A, Aljama P: On-line hemodiafil-
tration reduces the proinflammatory CD14+CD16+ monocyte-derived
dendritic cells: A prospective, crossover study. J Am Soc Nephrol
17: 2315–2321, 2006
12. Canaud B, Bragg-Gresham JL, Marshall MR, Desmeules S, Gillespie
BW, Depner T, Klassen P, Port FK: Mortality risk for patients receiving
hemodiafiltration versus hemodialysis: European results from the
DOPPS. Kidney Int 69: 2087–2093, 2006
13. Jirka T, Cesare S, Di Benedetto A, Perera Chang M, Ponce P, Richards N,
Tetta C, Vaslaky L: Mortality risk for patients receiving hemodiafiltration
versus hemodialysis. Kidney Int 70: 1524, author reply 1524–1525, 2006
14. Panichi V, Rizza GM, Paoletti S, Bigazzi R, Aloisi M, Barsotti G, Rindi P,
Donati G, Antonelli A, Panicucci E, Tripepi G, Tetta C, Palla R; RISCAVID
Study Group: Chronic inflammation and mortality in haemodialysis:
Effect of different renal replacement therapies. Results from the RIS-
CAVID study. Nephrol Dial Transplant 23: 2337–2343, 2008
15. Vilar E, Fry AC, Wellsted D, Tattersall JE, Greenwood RN, Farrington K:
Long-term outcomes in online hemodiafiltration and high-flux hemodi-
alysis: A comparative analysis. Clin J Am Soc Nephrol 4: 1944–1953, 2009
16. Grooteman MPC, van den Dorpel MA, Bots ML, Penne EL, van der
Weerd NC, Mazairac AH, den Hoedt CH, van der Tweel I, Lévesque R,
Nubé MJ, ter Wee PM, Blankestijn PJ; CONTRAST Investigators: Effect
of online hemodiafiltration on all-cause mortality and cardiovascular
outcomes. J Am Soc Nephrol 23: 1087–1096, 2012
17. Ok E, Asci G, Toz H, Ok ES, Kircelli F, Yilmaz M, Hur E, Demirci MS,
Demirci C, Duman S, Basci A, Adam SM, Isik IO, Zengin M, Suleymanlar
G, Yilmaz ME, Ozkahya M; On behalf of the ‘Turkish Online
J Am Soc Nephrol 24: 487–497, 2013

Haemodiafiltration Study’: Mortality and cardiovascular events in on-
line haemodiafiltration (OL-HDF) compared with high-flux dialysis:
Results from the Turkish OL-HDF Study. Nephrol Dial Transplant 28:
192–202, 2013
18. Survival Study in Patients Undergoing On-line Hemodiafiltration
(ESHOL): A randomized study to evaluate survival in patients un-
dergoing on-line HDF. Available at: http://clinicaltrials.gov/ct2/show/
NCT00694031. Accessed
19. Foley RN, Parfrey PS, Sarnak MJ: Epidemiology of cardiovascular disease
in chronic renal disease. J Am Soc Nephrol 9[Suppl]: S16–S23, 1998
20. Kim ST, Yamamoto C, Asabe H, Sato T, Takamiya T: Online haemo-
diafiltration: Effective removal of high molecular weight toxins and
improvement in clinical manifestations of chronic haemodialysis pa-
tients. Nephrology (Carlton) 2[Supp 1]: S183–S186, 1996
21. Iseki K, Fukiyama K; Okawa Dialysis Study (OKIDS) Group: Clinical
demographics and long-term prognosis after stroke in patients on
chronic haemodialysis. Nephrol Dial Transplant 15: 1808–1813, 2000
22. McMurray JJ, Uno H, Jarolim P, Desai AS, de Zeeuw D, Eckardt KU,
Ivanovich P, Levey AS, Lewis EF, McGill JB, Parfrey P, Parving HH, Toto
RM, Solomon SD, Pfeffer MA: Predictors of fatal and nonfatal cardio-
vascular events in patients with type 2 diabetes mellitus, chronic kidney
disease, and anemia: An analysis of the Trial to Reduce Cardiovascular
Events with Aranesp (darbepoetin-alfa) Therapy (TREAT). Am Heart J
162: 748–755, e3, 2011
23. Locatelli F, Altieri P, Andrulli S, Bolasco P, Sau G, Pedrini LA, Basile C,
David S, Feriani M, Montagna G, Di Iorio BR, Memoli B, Cravero R,
Battaglia G, Zoccali C: Hemofiltration and hemodiafiltration reduce
intradialytic hypotension in ESRD. J Am Soc Nephrol 21: 1798–1807,
2010
24. Donauer J, Schweiger C, Rumberger B, Krumme B, Böhler J: Reduction
of hypotensive side effects during online-haemodiafiltration and low
J Am Soc Nephrol 24: 487–497, 2013
www.jasn.org CLINICAL RESEARCH
temperature haemodialysis. Nephrol Dial Transplant 18: 1616–1622,
2003
25. Kaiser JP, Oppermann M, Götze O, Deppisch R, Göhl H, Asmus G,
Röhrich B, von Herrath D, Schaefer K: Significant reduction of factor D
and immunosuppressive complement fragment Ba by hemofiltration.
Blood Purif 13: 314–321, 1995
26. Haag-Weber M, Cohen G, Hörl WH: Clinical significance of granulocyte-
inhibiting proteins. Nephrol Dial Transplant 15[Suppl 1]: 15–16,
2000
27. Lornoy W, Becaus I, Billiouw JM, Sierens L, van Malderen P: Remarkable
removal of beta-2-microglobulin by on-line hemodiafiltration. Am
J Nephrol 18: 105–108, 1998
28. Odell RA, Slowiaczek P, Moran JE, Schindhelm K: b 2-microglobulin
kinetics in end-stage renal failure. Kidney Int 39: 909–919, 1991
29. Tiranathanagul K, Praditpornsilpa K, Katavetin P, Srisawat N,
Townamchai N, Susantitaphong P, Tungsanga K, Eiam-Ong S: On-line
hemodiafiltration in Southeast Asia: A three-year prospective study of a
single center. Ther Apher Dial 13: 56–62, 2009
30. Maduell F, Moreso F, Pons M, Ramos R, Mora-Macià J, Foraster A, Soler
J, Galceran JM, Martinez-Castelao A; Online Hemodiafiltration Study
Group from the Catalonian Society of Nephrology: Design and patient
characteristics of ESHOL study, a Catalonian prospective randomized
study. J Nephrol 24: 196–202, 2011
See related editorial, “Has the Time Now Come to More Widely Accept He-
modiafiltration in the United States?,” on pages 332–334.
This article contains supplemental material online at http://jasn.asnjournals.
org/lookup/suppl/doi:10.1681/ASN.2012080875/-/DCSupplemental.
OL-HDF and Survival: ESHOL Study 497
Supplemental Tables 1-4

Table 1. Secondary efficacy variables: Laboratory data, dry body weight, and BP
Length of Follow-Up P
Baseline 6 mo 12 mo 18 mo 24 mo 30 mo 36 mo
Kt/V
Hemodialysis 1.66 1.68 1.73 1.75 1.78 1.76 1.80 <0.001a
(1.62, 1.69) (1.65, 1.70) (1.70, 1.77) (1.71, 1.78) (1.74, 1.82) (1.72, 1.80) (1.75, 1.84)
OL-HDF 1.67 1.80 1.83 1.84 1.84 1.85 1.88 <0.001b
(1.64, 1.70) (1.77, 1.83) (1.80, 1.86) (1.81, 1.88) (1.80,1.88) (1.81, 1.89) (1.83, 1.92)
Time  time — 0.001 0.001 0.001 0.019 0.001 0.010 0.81c
URR (%)
Hemodialysis 74.1 75.4 76.4 76.8 77.2 77.1 77.5 <0.001a
(72.8, 75.5) (74.9, 76.0) (75.8, 77.0) (76.2, 77.5) (76.5, 77.9) (76.4, 77.8) (76.7, 78.3)
OL-HDF 74.3 77.7 78.1 78.4 78.3 78.7 79.4 <0.001b
(72.4, 76.3) (77.2, 78.3) (77.6, 78.7) (77.8, 79.0) (77.6, 79.0) (77.9, 79.4) (78.6, 80.1)
Time  time — 0.001 0.001 0.001 0.021 0.003 0.002 0.76c
nPCR (g/kg per
day)
Hemodialysis 1.09 1.10 1.10 1.12 1.11 1.09 1.12 0.01a
(1.07, 1.11) (1.08, 1.11) (1.08, 1.12) (1.10, 1.14) (1.09, 1.14) (1.07, 1.12) (1.09, 1.15)
OL-HDF 1.10 1.12 1.12 1.14 1.13 1.15 1.12 0.09b
(1.08, 1.12) (1.10, 1.14) (1.10, 1.14) (1.12, 1.16) (1.10, 1.15) (1.12, 1.17) (1.09, 1.15)
Time  time — 0.08 0.19 0.32 0.34 0.004 0.77 0.20c
Albumin (g/dl)
Hemodialysis 4.06 4.08 4.09 4.07 3.97 3.88 3.88 0.08a
(4.02, 4.11) (4.05, 4.12) (4.05, 4.13) (4.03, 4.11) (3.92, 4.01) (3.83, 3.93) (3.82, 3.94)
OL-HDF 4.11 4.06 4.04 4.04 3.95 3.83 3.85 <0.001b
(4.07, 4.15) (4.02, 4.10) (4.00, 4.08) (4.00, 4.08) (3.90, 4.00) (3.78, 3.88) (3.79, 3.90)
Time  time — 0.40 0.08 0.35 0.65 0.21 0.42 0.19c
Dry body
weight (kg)
Hemodialysis 66.8 67.6 67.4 66.9 66.7 66.6 66.3 0.51a
(65.6, 68.1) (67.3, 68.0) (67.0, 67.8) (66.5, 67.3) (66.2, 67.1) (66.2, 67.1) (65.8, 66.8)
OL-HDF 67.9 67.5 67.3 67.4 66.9 66.9 66.8 <0.001b
(66.6, 69.2) (67.1, 67.8) (66.9, 67.6) (66.9, 67.8) (66.4, 67.3) (66.4, 67.3) (66.3, 67.3)
Time  time — 0.52 0.68 0.12 0.60 0.54 0.20 0.33c
Hemoglobin
(g/dl)
Hemodialysis 12.0 12.0 11.8 11.9 11.7 11.6 11.6 0.33a
(11.8, 12.1) (11.8, 12.1) (11.7, 11.9) (11.7, 12.0) (11.6, 11.9) (11.4, 11.8) (11.4, 11.8)
OL-HDF 12.0 11.8 11.7 11.9 11.8 11.6 11.7 0.002b
(11.9, 12.1) (11.7, 12.0) (11.6, 11.9) (11.8, 12.1) (11.6, 12.0) (11.4, 11.8) (11.5, 11.8)
Time  time — 0.21 0.62 0.67 0.49 0.73 0.81 0.79c
TSI (%)
Hemodialysis 30.5 31.7 30.6 29.4 28.5 28.0 29.4 0.04a
(28.9, 32.0) (30.4, 33.0) (29.2, 32.0) (27.8, 31.0) (26.7, 30.3) (26.1, 29.8) (27.3, 31.4)
OL-HDF 27.9 28.9 28.4 29.8 28.3 28.1 29.4 0.20b
(26.6, 29.2) (27.6, 30.3) (27.0, 29.9) (28.2, 31.4) (26.5, 30.0) (26.3, 30.0) (27.4, 31.4)
Time  time — 0.004 0.04 0.72 0.85 0.91 1.00 0.31c
Ferritin (ng/ml)
Hemodialysis 409 396 383 386 390 417 392 0.002a
(382, 436) (372, 420) (357, 409) (358, 415) (359, 422) (383, 451) (354, 430)
OL-HDF 372 353 315 347 371 380 421 0.004b
(347, 396) (329, 378) (288, 341) (319, 376) (339, 402) (346, 414) (385, 458)
Time  time — 0.02 0.001 0.06 0.39 0.14 0.28 0.14c
Phosphorus
(mg/dl)
Hemodialysis 4.58 4.69 4.52 4.58 4.43 4.44 4.44 0.62a
(4.45, 4.71) (4.57, 4.81) (4.39, 4.65) (4.43, 4.73) (4.27, 4.59) (4.26, 4.61) (4.25, 4.64)
OL-HDF 4.73 4.49 4.53 4.56 4.58 4.62 4.52 0.87b
(4.60, 4.87) (4.37, 4.62) (4.40, 4.67) (4.42, 4.71) (4.42, 4.74) (4.45, 4.79) (4.34, 4.70)
Time  time — 0.03 0.90 0.88 0.19 0.15 0.57 0.08c
2-
microglobulin

Page 1 of 4
(mg/L)
Hemodialysis 24.8 24.4 24.1 29.0 31.0 30.6 29.8 <0.001a
(23.6, 26.1) (23.0, 25.8) (22.6, 25.6) (27.3, 30.7) (29.1, 33.0) (28.5, 32.7) (27.4, 32.2)
OL-HDF 23.9 24.2 25.8 25.9 29.6 30.2 29.2 <0.001b
(22.7, 25.0) (22.7, 25.7) (24.2, 27.5) (24.1, 27.7) (27.5, 31.6) (28.0, 32.4) (26.8, 31.7)
Time  time — 0.89 0.13 0.02 0.30 0.78 0.73 <0.001c
SBP predialysis
(mmHg)
Hemodialysis 136 (134, 139) 138 (135, 140) 138 (135, 140) 136 (134, 139) 136 (133, 139) 133 (130, 136) 136 (132, 139) 0.36a
OL-HDF 136 (134, 139) 139 (137, 141) 138 (136, 141) 138 (135, 140) 137 (135, 140) 137 (134, 140) 133 (130, 136) <0.001b
Time  time — 0.33 0.80 0.47 0.43 0.09 0.27 0.53c
DBP predialysis
(mmHg)
Hemodialysis 71 (70, 73) 72 (70, 73) 71 (69, 73) 69 (67, 70) 70 (68, 72) 68 (67, 70) 71 (69, 73) 0.54a
OL-HDF 73 (71, 74) 71 (69, 72) 71 (69, 72) 70 (68, 71) 69 (67, 71) 69 (67, 70) 70 (68, 72) <0.001b
Time  time — 0.29 0.76 0.49 0.53 0.98 0.55 0.96c
Data are shown for every 6 months but were collected every 3 months. Analysis was conducted using a
mixed model for repeated measures. Data are presented as the mean (95% CI) at baseline and least-
squares mean (95% CI) at 6 months, 12 months, 18 months, 24 months, 30 months, and 36 months. Time
 time indicates time  time comparisons, which are given as P values. URR, urea reduction ratio; nPCR,
normalized protein catabolic rate; TSI, transferrin saturation index; SBP, systolic BP; DBP, diastolic BP.
a
P value for treatment effect.
b
P value for time effect.
c
P value for treatment  time interaction.

Table 2. Anemia, antihypertensive, and phosphate binding medication

Length of Follow-Up P
Baseline 6 mo 12 mo 18 mo 24 mo 30 mo 36 mo
Iron dosage (mg/wk)
Hemodialysis 38.7 35.2 37.2 34.1 37.8 40.4 41.0 0.001a
(35.1, 42.2) (31.8, 38.6) (33.5, 40.9) (30.0, 38.2) (33.3, 42.3) (35.6, 45.2) (35.7, 46.3)
OL-HDF 40.8 38.1 42.8 46.4 43.0 42.6 39.9 0.14b
(37.3, 44.3) (34.7, 41.6) (39.0, 46.5) (42.3, 50.4) (38.6, 47.4) (37.8, 47.3) (34.8, 45.0)
Time  time — 0.24 0.04 0.001 0.11 0.53 0.77 0.06c
ESA
EPO
Hemodialysis 213 (47.4) 166 (41.2) 101 (30.0) 74 (26.8) 45 (19.5) 39 (19.4) 31 (18.7) 0.55a
OL-HDF 203 (44.6) 148 (39.0) 104 (31.7) 80 (27.9) 50 (21.0) 44 (21.3) 37 (20.3) <0.001b
Time  time — 0.33 0.75 0.66 0.35 0.48 0.47 0.73c
Darbe
Hemodialysis 200 (44.5) 161 (40.0) 134 (39.8) 108 (39.0) 89 (38.5) 71 (35.3) 57 (34.3) 0.85a
OL-HDF 201 (44.1) 157 (41.3) 138 (42.1) 107 (37.3) 87 (36.6) 75 (36.2) 67 (36.8) <0.001b
Time  time — 0.66 0.39 0.82 0.47 0.53 0.40 0.60c
Cera
Hemodialysis 5 (1.1) 38 (9.4) 64 (19.0) 62 (22.4) 62 (26.8) 57 (28.4) 53 (31.9) 0.96a
OL-HDF 6 (1.3) 32 (8.4) 54 (16.5) 58 (20.2) 67 (28.2) 57 (27.5) 49 (26.9) <0.001b
Time  time — 0.58 0.51 0.77 0.28 0.86 0.50 0.68c
EPO dose (IU/kg
per week)
Hemodialysis 133.8 133.7 139.5 164.1 156.4 173.9 156.6 0.15a
(119.9, 147.7) (118.5, 148.9) (121.0, 157.9) (142.1, 186.0) (129.2, 183.6) (143.4, 204.4) (121.5, 191.6)
OL-HDF 129.3 135.3 157.3 154.2 126.7 132.7 131.2 0.31b
(115.4, 143.3) (119.2, 151.3) (138.7, 175.8) (132.6, 175.8) (100.6, 152.8) (103.0, 162.4) (97.3, 165.1)
Time  time — 0.89 0.18 0.53 0.12 0.06 0.31 0.02c
Darbe dose (g/kg
per week)
Hemodialysis 0.58 0.61 0.60 0.69 0.60 0.65 0.67 0.64a
(0.52, 0.65) (0.53, 0.68) (0.51, 0.68) (0.60, 0.78) (0.50, 0.70) (0.54, 0.76) (0.54, 0.80)

Page 2 of 4
 OL-HDF 0.63 0.64 0.64 0.70 0.60 0.66 0.58 0.14b
(0.56, 0.69) (0.57, 0.72) (0.56, 0.72) (0.61, 0.80) (0.50, 0.70) (0.54, 0.77) (0.46, 0.70)
Time  time — 0.50 0.46 0.84 0.97 0.93 0.33 0.68c
Cera dose (g/kg
per week)
Hemodialysis 0.56 0.54 0.61 0.59 0.56 0.57 0.64 0.64a
(0.35, 0.78) (0.39, 0.69) (0.48, 0.73) (0.46, 0.72) (0.42, 0.69) (0.43, 0.71) (0.49, 0.79)
OL-HDF 0.79 0.63 0.55 0.70 0.57 0.64 0.64 0.41b
(0.26, 1.33) (0.47, 0.80) (0.41, 0.69) (0.56, 0.83) (0.44, 0.70) (0.51, 0.78) (0.49, 0.79)
Time  time — 0.41 0.56 0.27 0.88 0.48 0.98 0.11c
Antihypertensive
therapy
Hemodialysis 260 (57.8) 212 (52.0) 179 (51.9) 151 (52.6) 128 (53.3) 110 (53.1) 87 (50.6) 0.99a
OL-HDF 270 (59.2) 213 (55.2) 179 (53.6) 149 (50.9) 127 (52.3) 106 (51.0) 91 (49.5) <0.001b
Time  time — 0.53 0.78 0.44 0.80 0.82 0.51 0.93c
Data are shown for every 6 months but were collected every 3 months. Analyses were conducted using a
mixed model for repeated measures for continuos variables and a penalized quasi-likelihood under REML
models for categorical variables. For continuous variables, data are presented as mean (95% CI) at
baseline and as least-squares mean (95% CI) at 6 months, 12 months, 18 months, 24 months, 30 months,
and 36 months. For categorical variables, data are presented as n (%). Time  time indicates time  time
comparisons, which are given as P values. EPO, erythropoietin ; Darbe, darbepoetin alfa; Cera,
methoxy-polyethylene glycol epoetin .
a
P value for treatment effect.
b
P value for time effect.
c
P value for treatment  time interaction.

Table 3. Laboratory data: Other biochemical parameters

Length of Follow-Up P
Baseline 6 mo 12 mo 18 mo 24 mo 30 mo 36 mo
CRP (mg/L)
Hemodialysis 11.3 14.1 13.3 17.1 14.0 12.0 11.9 0.67a
(9.5, 13.2) (11.0, 17.2) (9.9, 16.7) (13.3, 20.9) (9.8, 18.2) (7.4, 16.6) (6.3, 17.4)
OL-HDF 11.5 13.8 14.0 14.6 16.7 13.2 14.3 0.16b
(10.0, 13.0) (10.5, 17.2) (10.4, 17.5) (10.7, 18.5) (12.3, 21.0) (8.5, 18.0) (8.8, 19.8)
Time  time — 0.91 0.79 0.37 0.38 0.72 0.55 0.75c
Creatinine (mg/dl)
Hemodialysis 8.0 (7.8, 8.2) 8.0 (7.9, 8.2) 8.1 (7.9, 8.2) 8.0 (7.9, 8.2) 8.1 (7.9, 8.3) 7.9 (7.6, 8.1) 7.9 (7.7, 8.1) 0.90a
OL-HDF 8.1 (7.8, 8.3) 7.9 (7.7, 8.0) 8.0 (7.8, 8.1) 8.1 (8.0, 8.3) 8.1 (7.9, 8.3) 7.9 (7.7, 8.1) 8.0 (7.8, 8.2) 0.002b
Time  time — 0.14 0.32 0.40 0.87 0.82 0.62 0.77c
Sodium (mmol/L)
Hemodialysis 138.7 138.4 138.4 138.4 138.7 138.7 138.3 0.07a
(138.4, 139.0) (138.1, 138.7) (138.1, 138.7) (138.1, 138.8) (138.3, 139.1) (138.3, 139.1) (137.8, 138.8)
OL-HDF 138.8 138.9 138.5 138.8 138.9 138.6 138.3 0.002b
(138.5, 139.1) (138.6, 139.2) (138.2, 138.8) (138.4, 139.1) (138.5, 139.2) (138.2, 139.0) (137.8, 138.8)
Time  time — 0.02 0.72 0.14 0.62 0.72 0.96 0.56c
Potassium (mmol/L)
Hemodialysis 5.3 (5.2, 5.4) 5.3 (5.2, 5.4) 5.3 (5.2, 5.3) 5.2 (5.1, 5.3) 5.2 (5.2, 5.3) 5.2 (5.1, 5.3) 5.1 (5.0, 5.3) 0.20a
OL-HDF 5.2 (5.2, 5.3) 5.2 (5.1, 5.2) 5.2 (5.1, 5.2) 5.2 (5.1, 5.3) 5.3 (5.2, 5.4) 5.2 (5.1, 5.3) 5.1 (5.0, 5.2) 0.51b
Time  time — 0.03 0.14 0.54 0.74 0.46 0.94 0.40c
Uric acid (mg/dl)
Hemodialysis 5.5 (5.4, 5.6) 5.8 (5.6, 5.9) 5.8 (5.6, 5.9) 5.6 (5.4, 5.7) 5.8 (5.6, 5.9) 5.6 (5.4, 5.8) 5.8 (5.5, 6.0) 0.95a
OL-HDF 5.6 (5.5, 5.7) 5.6 (5.4, 5.7) 5.6 (5.4, 5.7) 5.6 (5.5, 5.8) 5.8 (5.6, 5.9) 5.7 (5.5, 5.9) 5.8 (5.6, 6.1) 0.07b
Time  time — 0.03 0.04 0.71 0.81 0.57 0.67 0.24c
Calcium (mg/dl)
Hemodialysis 9.1 (9.0, 9.2) 9.1 (9.0, 9.1) 9.1 (9.0, 9.1) 9.0 (9.0, 9.1) 9.0 (8.9, 9.0) 9.0 (8.9, 9.1) 9.0 (8.9, 9.1) 0.42a
OL-HDF 9.1 (9.0, 9.1) 9.1 (9.0, 9.1) 9.1 (9.0, 9.2) 9.1 (9.0, 9.2) 9.0 (8.9, 9.0) 9.0 (8.9, 9.1) 9.1 (9.0, 9.2) 0.01b
Time  time — 0.87 0.54 0.17 0.95 0.38 0.07 0.58c

Page 3 of 4
iPTH (pg/ml)
Hemodialysis 280 285 280 264 288 272 288 0.84a
(256, 305) (256, 304) (256, 304) (237, 290) (259, 317) (240, 303) (253, 322)
OL-HDF 261 253 268 261 286 281 313 0.004b
(240, 281) (230, 292) (244, 292) (235, 287) (257, 315) (250, 312) (281, 346)
Time  time — 0.05 0.50 0.89 0.94 0.67 0.29 0.36c
Data are shown for every 6 months but were collected every 3 months. Analysis was conducted using a
mixed model for repeated measures. Data are presented as the mean (95% CI) at baseline and least-
squares mean (95% CI) at 6 months, 12 months, 18 months, 24 months, 30 months, and 36 months. Time
 time indicates time  time comparisons, which are given as P values. CRP, C-reactive protein; iPTH,
intact parathyroid hormone.
a
P value for treatment effect.
b
P value for time effect.

Table 4. Risk of all-cause mortality results by achieved convective volume, convective

volume/BMI, and convective volume/BSA
Hemodialysis OL-HDF Pa Pb
Tertile 1 Tertile 2 Tertile 3
Convective volume (L/session) <23.1 23.1–25.4 >25.4
n (%) 124 (26.4) 33 (22.9) 27 (18.2) 23 (16.0) 0.03 0.003
HR (95% CI) Ref. 0.90 (0.61–1.31) 0.60 (0.39–0.9) 0.55 (0.34–0.84) 0.01 0.001
Convective volume/BMI (L/kg per m2) <0.9 0.9–1.1 >1.1
n (%) 124 (26.4) 25 (17.2) 32 (21.9) 26 (17.9) 0.05 0.03
HR (95% CI) Ref. 0.62 (0.4–0.94) 0.74 (0.49–1.08) 0.66 (0.42–0.99) 0.04 0.02
2
Convective volume/BSA (L/m ) <13.2 13.2–14.8 >14.8
n (%) 124 (26.4) 29 (20.0) 29 (19.9) 25 (17.2) 0.06 0.01
HR (95% CI) Ref. 0.75 (0.49–1.11) 0.66 (0.43–0.98) 0.62 (0.39–0.93) 0.04 0.01
n (%) indicates the number of exitus (% of exitus from total of group participants). BMI, body mass
index; BSA, body surface area.
a
P value for the statistical significance test (Fisher's exact test; log-rank test).
b
P value for the trend test (Cochran-Armitage trend test; log-rank test).

Page 4 of 4