Effect of Hemodiafiltration on Quality of Life over Time

Summary

Background and objectives

It is unclear if hemodiafiltration leads to a better quality of life compared with hemodialysis. It was, therefore, the aim of this study to assess the effect of hemodiafiltration on quality of life compared with hemodialysis in patients with ESRD.

Design, setting, participants, & measurements

This study analyzed the data of 714 patients with a median follow-up of 2 years from the Convective Transport Study. The patients were enrolled between June of 2004 and December of 2009. The Convective Transport Study is a randomized controlled trial on the effect of online hemodiafiltration versus low-flux hemodialysis on all-cause mortality. Quality of life was assessed with the Kidney Disease Quality of Life—Short Form. This questionnaire provides data for a physical and mental composite score and describes kidney disease-specific quality of life in 12 domains. The domains have scales from 0 to 100.

Results

There were no significant differences in changes in health-related quality of life over time between patients treated with hemodialysis (n=358) or hemodiafiltration (n=356). The quality of life domain patient satisfaction declined over time in both dialysis modalities (hemodialysis: −2.5/yr, −3.4 to −1.5, P<0.001; hemodiafiltration: −1.4/yr, −2.4 to −0.5, P=0.004).

Conclusions

Compared with hemodialysis, hemodiafiltration had no significant effect on quality of life over time.

Introduction

Health-related quality of life (HRQOL) is a multidimensional concept. It comprises a patients’ perspective on physical, mental, and social domains of health (1). In a chronic condition like ESRD, improving patients’ HRQOL is considered a relevant treatment goal (2). The HRQOL of patients with ESRD has been described as worse than the HRQOL of patients with congestive heart failure, chronic lung disease, or cancer (3). Besides an important outcome in itself, reduced HRQOL has been related to hospitalization and mortality (4,5).

Of the various renal replacement therapies, observational studies have suggested that kidney transplantation offers the most preferable health status in general (6). However, because the availability of donor kidneys is limited, the effect of contemporary renal replacement therapies on HRQOL is an important issue. Hemodiafiltration (HDF) is a relatively new therapy (7). It combines diffusion with convection to clear middle molecular weight substances more effectively compared with hemodialysis (HD) (7,8). Recently, it was shown by our group that treatment with HDF did not result in reduced mortality or cardiovascular events compared with HD (9). The effect of an improved uremic environment on HRQOL, however, remains unclear: earlier studies on the effect of HDF on HRQOL were inconclusive and based on findings from small (n=44–111) or nonrandomized studies (10–15). Therefore, the aim of this study was to assess the effect of HDF compared with HD on HRQOL in a large randomized clinical trial.

Materials and Methods

Patients and Study Design

This randomized analysis is based on data obtained from 714 HD patients who participated in the Convective Transport Study (CONTRAST). Between June of 2004 and December of 2009, they were recruited from dialysis centers in The Netherlands (n=26), Canada (n=2), and Norway (n=1). As described previously, CONTRAST is a randomized, controlled trial (ISRCTN38365125) that compares the effects of low-flux HD and online HDF on all-cause mortality and cardiovascular events (8). The analysis of HRQOL was prespecified as a secondary outcome measure. Online HDF was performed in the postdilution mode with a target convection volume of 6 L/h. This amount was based on a targeted filtration rate between 25% and 33% of an extracorporeal blood flow rate between 300 and 400 ml/min (so-called filtration fraction), which seemed achievable in clinical practice. Blood flow rates could be increased in the HDF arm to improve convection volumes. For HDF, synthetic high-flux dialyzers were used (FX80, 27%; FX100, 11%; Optiflux F200NR, 8%; Fresenius Medical Care, Bad Homburg, Germany; Polyflux 170H, 25%; Polyflux 210H, 27%; Gambro AB, Stockholm, Sweden; other dialyzers, 2%). HD patients were treated with synthetic low-flux dialyzers (F6HPS, 5%; F8HPS, 45%; Optiflux 18NR, 9%; Fresenius; Polyflux 14L, 2%; Polyflux 17L, 30%; Gambro; other, 9%).

Patients were randomized centrally into a 1:1 ratio and stratified per participating center. Patients were eligible for inclusion if they were treated two or three times per week with HD for at least 2 months, with a minimum dialysis urea Kt/V (i.e., urea clearance divided by volume of distribution) of ≥1.2. Exclusion criteria were age below 18 years, treatment with HD, HDF, or high-flux HD in the 6 months before randomization, a life expectancy less than 3 months because of nonrenal disease, participation in another clinical intervention trial evaluating cardiovascular outcomes, and severe incompliance regarding frequency and duration of dialysis treatment (8). The study was conducted in accordance with the Declaration of Helsinki and approved by the medical ethics review boards of all participating hospitals. Written informed consent was obtained from all patients before enrolment.

Measurements

At baseline, standardized forms were used to collect demographic, clinical, and laboratory data. Demographic data included age, sex, race, educational level, and employment status. Clinical characteristics recorded were medical history, including the cause of kidney disease, presence of diabetes mellitus, previous cardiovascular disease, HD dose (single-pool Kt/V urea), time on dialysis, treatment time in hours, and estimated residual GFR. Laboratory values were measured in the different participating hospitals using standard techniques. Study visits were performed at 3-month intervals. Interdialytic urinary samples were collected in patients with a urinary production of 100 ml/d or more. Estimated residual GFR was calculated as the mean of creatinine and urea clearances and adjusted for body surface area (milliliters per minute per 1.73 meter²) (16). Additional information regarding the measured characteristics has been reported previously (8). Because of the nature of the intervention, it was not possible to blind the patients, the local study nurses, or the investigators for the treatment assignment.

Kidney Disease Quality of Life—Short Form

HRQOL was assessed at baseline and every subsequent year of follow-up with the validated Kidney Disease Quality of Life—Short Form (KDQOL-SF) version 1.3 (http://gim.med.ucla.edu/kdqol/downloads/-download.html) (17,18). This questionnaire covers different domains to face the multidimensional nature of HRQOL. It was distributed by local study nurses who helped the patients to fill out the forms, if necessary. The KDQOL-SF can be split up in a generic part and a disease-specific part. First, the generic part is formed by the Short Form with 36 questions (SF-36) version 1 (19). The eight domains of the SF-36 can be summarized in two summary scores, one for physical functioning (the physical composite score [PCS]) and one for mental functioning (the mental composite score [MCS]). These summaries are constructed in a way that a score of 50 represents the mean of the general US population with an SD of 10 (20). Because the PCS and MCS may be easier to understand than the eight individual domains of the SF-36, we focused on these composite scores with regard to generic HRQOL. Second, the disease-specific part of the KDQOL-SF consists of 44 kidney disease-targeted questions. The responses to these items are condensed in 12 domains. The HRQOL domains have a score from 0 to 100, with higher scores indicating the absence of problems. A difference of five points has been proposed to be clinically relevant with regard to individual domains, and a difference of three points has been proposed to be clinically relevant with regard to the composite scores (19,20).

Statistical Analyses

Data were analyzed according to an intention-to-treat principle (i.e., according to assigned instead of received treatment). All variables were reported as proportions, means with SD, or medians with interquartile range when appropriate. Differences in changes in HRQOL domains over time were assessed with multilevel linear regression. Because HRQOL scores from the same patient are likely to be dependent, there is a two-level hierarchy with HRQOL scores over time (level 1) nested within patients (level 2). For every HRQOL domain, we evaluated which model had the best fit. We expanded the model step by step and assessed for every step if this expansion led to a significantly lower log-likelihood using the chi-squared likelihood ratio test. We started off with fixed coefficients and then a random intercept for HRQOL and a random slope for HRQOL, and we evaluated different covariance structures and growth curves. In the end, the vast majority of models had a random intercept and slope for HRQOL over time, an unstructured covariance structure, and a first-order polynomial. The residuals of all models had a normal distribution. Differences in subgroups were evaluated using interaction terms. Eight subgroups were evaluated: age, sex, residual renal function (diuresis≥100 m/24 h), dialysis vintage, diabetes, albumin, previous cardiovascular disease, and convection volume. The Benjamini and Hochberg False Discovery Rate was used to correct for multiple comparisons (21). Missing data were handled with multilevel models (22). Results were considered statistically significant when P<0.05 (two-tailed comparison). Statistical analyses were performed with SPSS 18 (SPSS Inc., Chicago, IL).

Results

Patient Characteristics

The baseline characteristics of the patients on HD (n=356) and HDF (n=358) are summarized in Table 1. The mean age was 64 years in both groups (SD HD±13, SD HDF±14); 65% of patients on HD were male, and 60% of patients on HDF were male. Figure 1 shows the patient flow over time. Of 714 patients enrolled, 14 patients did not fill out a single KDQOL-SF (11 patients assigned to HD and 3 patients assigned to HDF). Over time, 93% of the KDQOL-SFs were filled out. The HRQOL domain sexual function was not analyzed, because most patients (88%) did not fill out its questions. The median follow-up of these patients was 2.0 years (interquartile range=1.1–3.0). The average convection volume over time in patients on HDF was 19±6 (SD) L/session.

Table 1.

Patient characteristics at baseline

	HD (n=356)	HDF (n=358)
Demographic
Age (years)	64±13	64±14
Sex (% male)	65	60
Caucasian (%)	83	85
Socioeconomic status
Employed (%)	10	11
High educational statusa (%)	21	20
Clinical parameters
Cause of kidney failure (%)
Vascular	27	29
Diabetes mellitus	17	21
Nephritis	24	20
Cystic kidney disease	7	7
Other/unknown	25	23
Dialysis vintage (yr)	2.2 (1.0–4.0)	1.8 (1.0–3.8)
Dialysis frequency (% 3×/wk)	95	93
Session time (h)	4.0 (3.5–4.0)	4.0 (3.5–4.0)
spKt/V urea	1.4±0.2	1.4±0.2
eGFR (ml/min per 1.73 m2)b	3.2 (1.2–5.4)	3.1 (1.7–6.2)
BMI (kg/m2)	26±5	25±5
Diabetes (%)	23	26
History of cardiovascular disease (%)	46	42
Hemoglobin (g/dl)	12±1	12±1
Albumin (g/dl)	3.7±0.5	3.7±0.5
Phosphorus (mg/dl)	5.1±1.5	5.1±1.6

Mean ± SD or median (interquartile range). To convert hemoglobin in grams per deciliter to millimoles per liter, multiply by 0.62. To convert albumin in grams per deciliter to grams per liter, multiply by 10. To convert phosphorus in milligrams per deciliter to millimoles per liter, multiply by 0.323. HD, hemodialysis; HDF, hemodiafiltration; spKt/V, urea clearance divided by volume of distribution, single pool; eGFR: estimated residual GFR; BMI, body mass index.

^aHigh educational status is college or university level.

^bIn 377 patients with diuresis≥100 ml/24 h (53%).

Figure 1.

Patient flow diagram. Because patients were included at different points in time, potential follow-up time varies per patient. The median follow-up of these patients was 2.0 years (interquartile range=1.1–3.0). The line between consecutive years of follow-up is dashed to point out that a measurement was sometimes skipped (e.g., 17 patients did not fill out the KDQOL-SF at baseline but did fill it out after 1 year of follow-up). HD, hemodialysis; HDF, hemodiafiltration; HRQOL, health-related quality of life; KDQOL-SF, Kidney Disease Quality of Life—Short Form; PD, peritoneal dialysis.

Quality of Life at Baseline and over Time

Table 2 shows the HRQOL at baseline. The mean PCS was 40±10 in HD patients and 39±11 in HDF patients. The mean MCS was 50±12 in both groups.

Table 2.

Health-related quality of life at baseline

	HD (n=356)	HDF (n=358)
Generic domains (SF-36)
Physical summary (PCS)	40±10	39±11
Mental summary (MCS)	50±12	50±12
Kidney disease-specific domains
Symptom/problem list	80±12	79±14
Effects of kidney disease on daily life	72±19	71±19
Burden of kidney disease	46±26	45±25
Work status	0 (0−50)	0 (0−50)
Cognitive function	78±20	80±18
Quality of social interaction	83±16	81±17
Sleep	62±21	62±20
Social support	83 (67−100)	83 (67−100)
Dialysis staff encouragement	75 (63−100)	75 (63−100)
Overall health	55±24	54±21
Patient satisfaction	70±23	70±24

Mean ± SD or median (interquartile range). The domains have a range from 0 to 100, with higher scores indicating a preferable health status or a relative absence of problems. HD, hemodialysis; HDF, hemodiafiltration; SF-36, Short Form with 36 questions version 1; PCS, the physical composite score; MCS, the mental composite score.

In patients on both HD and HDF, multiple HRQOL domain scores declined significantly over time (Table 3). The HRQOL domain patient satisfaction declined in both dialysis modalities (HD: −2.5 points/yr, 95% confidence interval=−3.4 to −1.5, P<0.001; HDF: −1.4 points/yr, 95% confidence interval=−2.4 to −0.5, P=0.004). One HRQOL domain improved, namely overall health, in patients on HDF (+0.9, 0.1–1.7, P=0.03). Over time, there were no significant differences in HRQOL between patients on HDF and HD, although a trend was observed to a worse MCS (−0.5, −1.1 to 0.0, P=0.06) and an improved effects of kidney disease on daily life (+1.1, −0.1 to 2.4, P=0.06) in patients on HDF.

Table 3.

Health-related quality of life over time in patients treated with hemodialysis and hemodiafiltration

	HD (n=356)		HDF (n=358)		HDF versus HD
	Δ	P Value	Δ	P Value	Δ	P Value
Generic domains (SF-36)
Physical summary (PCS)	−0.8 (−1.1 to −0.5)	<0.001a	−0.2 (−0.6 to 0.1)	0.13	0.4 (−0.3 to 1.1)	0.29
Mental summary (MCS)	0.1 (−0.3 to 0.5)	0.78	−0.5 (−0.8 to −0.1)	0.01a	−0.5 (−1.1 to 0.0)	0.06
Kidney disease-specific domains
Symptom/problem list	−1.1 (−1.8 to −0.4)	0.003a	−0.6 (−1.2 to 0.0)	0.07	0.4 (−0.5 to 1.4)	0.33
Effects of kidney disease on daily life	−1.3 (−2.3 to −0.4)	0.01a	−0.3 (−1.1 to 0.4)	0.38	1.1 (−0.1 to 2.4)	0.06
Burden of kidney disease	−0.2 (−1.0 to 0.6)	0.69	−0.9 (−1.7 to −0.1)	0.02a	−0.7 (−1.7 to 2.4)	0.21
Work status	−0.4 (−1.4 to 0.6)	0.42	−1.1 (−2.0 to −0.1)	0.03a	−0.6 (−2.0 to 0.8)	0.41
Cognitive function	−0.4 (−0.9 to 0.2)	0.22	−1.1 (−1.7 to −0.4)	0.001a	−0.7 (−2.0 to 0.7)	0.35
Quality of social interaction	−0.1 (−0.9 to 0.6)	0.71	−0.4 (−1.2 to 0.4)	0.36	−0.2 (−1.3 to 0.9)	0.68
Sleep	0.3 (−0.3 to 0.9)	0.35	0.4 (−0.2 to 1.1)	0.17	0.1 (−0.8 to 1.0)	0.76
Social support	−0.2 (−1.1 to 0.7)	0.65	−0.4 (−1.2 to 1.1)	0.37	−0.2 (−1.4 to 1.1)	0.81
Dialysis staff encouragement	−1.0 (−1.8 to −0.1)	0.02	0.0 (−0.7 to 0.8)	0.96	1.0 (−0.1 to 2.1)	0.09
Overall health	0.2 (−0.8 to 1.2)	0.69	0.9 (0.1 to 1.7)	0.03a	0.7 (−0.5 to 2.0)	0.25
Patient satisfaction	−2.5 (−3.4 to −1.5)	<0.001a	−1.4 (−2.4 to −0.5)	0.004a	0.0 (0.0 to 0.0)	0.29

The median follow-up was 2.0 years (interquartile range=1.1–3.0). HD, hemodialysis; HDF, hemodiafiltration; Δ, change in health-related quality of life over time per year; SF-36, Short Form with 36 questions version 1; PCS, the physical composite score; MCS, the mental composite score.

^aStatistically significant P value.

Subgroup analyses did not identify patient groups in which HDF may have had a clinical relevant benefit on HRQOL.

Discussion

This study showed that HDF had no significant effect on patients’ HRQOL compared with HD. Although there were changes over time in HRQOL for patients on HDF and HD, these changes did not result in significant differences over time between both dialysis modalities. There was a trend to a worse MCS and an improved effects of kidney disease on daily life in patients on HDF versus HD. Because their effect sizes were small and opposite, they implicate no clear benefit for either therapy.

To our knowledge, this study is the first large randomized study to analyze the effect of HDF on HRQOL. The results of earlier studies were inconclusive (Table 4): two studies showed an increase in physical HRQOL in patients on HDF (13,15), and three studies found no significant differences between dialysis modalities (11,12,14). Furthermore, no clinical meaningful differences in HRQOL have been identified in patients on low- versus high-flux HD (23) or hemofiltration versus HD (24,25). It should be noted that these studies used different questionnaires to assess HRQOL. Perhaps, this difference might explain why two studies did find a positive effect of HDF on HRQOL, whereas others did not. However, although the work by Lin et al. (13) used a self-developed form, the work by Schiffl (15) applied the Kidney Disease Questionnaire, which was also used in the work by Ward et al. (12). The works by Lin et al. (13) and Schiffl (15) found a positive effect of HDF on physical HRQOL, but the work by Ward et al. (12) did not.

Table 4.

Previous studies on hemodiafiltration and health-related quality of life

Ref	Design	Intervention	Number of Patients	Follow-Up (mo)	HRQOL Instrument	Effect on HRQOL
11	Cross-sectional	HDF ⇔ HD ⇔ PD	1013 (71 on HDF)	N/A	KS and SIP	No difference
12	RCT	HDF ⇔ high-flux HD	44 (24 on HDF)	±12	KDQ	No difference
13	RCT	HDF ⇔ high-flux HD	111a	Unclear	Self-developed	Better physical wellbeing in HDF (32%)
15	Crossover	HDF ⇔ high-flux HD	76	2×24	KDQ	Better perception of physical symptoms in HDF (26%)
14	Observational	HDF ⇔ high-flux HD ⇔ low-flux HD	2165 (253 on HDF)	±21 (HDF)	SF-36	No difference

HRQOL, health-related quality of life; HDF, hemodiafiltration; HD, hemodialysis; PD, peritoneal dialysis; KS, Karnofsky Performance Scale (34); SIP, sickness impact profile (35); RCT, randomized clinical trial; KDQ, Kidney Disease Questionnaire (36); SF-36, Short Form 36 (19).

^aRandomization into four groups: 3×/wk HD, 3×/wk HDF, and two intermediate versions with a 2×/wk versus 1×/wk distribution of HD or HDF.

When we put the baseline HRQOL of our population in the context of other large studies, the previous finding that dialysis patients have a severely impaired physical functioning but mental functioning comparable with the general population normal scores was supported (present study: PCS=39±11, MCS=50±12; Hemodialysis [HEMO] Study: PCS=36±10, MCS=50±11 [26]; Dialysis Outcomes and Practice Patterns Study: PCS=35±11, MCS=45±12 [27]).

A difference in HRQOL of five points or more has been proposed as being clinically relevant (19). Such a notion would mean that, in our study, only the decline in patient satisfaction is likely to have clinical relevance over several years of time. This decline was shown in both dialysis modalities. It is in agreement with previous findings that showed a negative relation with dialysis vintage (28,29). Patient satisfaction has been related to various factors (28–31), including a positive association with compliance (30) and a negative association with comorbidity (28). An increase in age seems less relevant, because age has been both positively (32) and negatively (29) related to patient satisfaction.

Between dialysis modalities, there was no clinical relevant or statistical significant difference in HRQOL. Apparently, the effect of HDF on the removal of middle molecules did not impact patients’ health status. A limitation of our study is the favorable patient selection within the context of a randomized clinical trial. This limitation might affect changes in HRQOL over time within groups, but it is not likely to affect differences in HRQOL between HDF and HD. The external validity of our study is supported by the fact that the variables age, sex, and primary kidney disease of our patients are similar to the variables in all Dutch patients registered in the RENINE database (www.renine.nl). Also, it was not registered if patients needed assistance to fill out the KDQOL-SF (e.g., from a nurse). Differences in the number of patients needing assistance in completing the questionnaire could have introduced bias. However, there is no reason to assume differences between HD and HDF populations in this respect. A third limitation is that the variability in HRQOL might be underestimated because of missing data (7%). One could, furthermore, speculate that HRQOL differences might arise during a prolonged follow-up period. However, because there was no statistically significant difference between HDF and HD in multiple HRQOL domains, it is unlikely that a more pronounced, clinically relevant difference would arise over time, especially when taking the high mortality rate of dialysis patients into account (33).

In conclusion, the results from this first large randomized, controlled trial comparing the effect of HDF with HD on HRQOL show that HDF had no effect on this important outcome. Apparently, alterations in the dialysis prescription, which can be made within the context of three dialysis sessions (4 h/wk) as studied herein and in the HEMO study, are unlikely to significantly impact HRQOL.

Disclosures

A.H.A.M, G.A.d.W., M.P.C.G., E.L.P., N.C.v.d.W., C.H.d.H., R.L., M.A.v.d.D., and M.L.B. have no disclosures. M.J.N. received funding for research from Baxter and Fresenius. P.M.t.W. received funding for research from Abbott, Baxter, Gambro, Fresenius, and Roche and honoraria for lectures from Amgen, Roche, Genzyme, and Fresenius. P.J.B. received funding for research from Fresenius and Gambro and honoraria for lectures from Fresenius, Gambro, Solvay, and Novartis.