Is it time to revisit residual renal function in haemodialysis? (Review Article)


Associate Professor Carmel Hawley, Department of Renal Medicine, Level 2, Ambulatory Renal and Transplant Services Building, Princess Alexandra Hospital, Ipswich Road, Woolloongabba, Brisbane, Qld 4102, Australia. Email:


SUMMARY:  Residual renal function (RRF) is not currently emphasized for patients undergoing haemodialysis (HD). The role of RRF is well recognized in the peritoneal dialysis population as studies have clearly demonstrated a survival benefit with preservation of RRF. There is however, data to suggest that RRF is important in HD patients as well. Contemporary HD therapies using high flux biocompatible synthetic dialysers, bicarbonate buffered ultrapure dialysis fluids with ultrafiltration control appear to allow better preservation of RRF. The long held belief that peritoneal dialysis is better at preserving RRF than HD may no longer be true. More robust studies are required to determine the relative importance of RRF in HD and strategies to best preserve this vital asset.

The importance of residual renal function (RRF) in dialysis patients has begun to re-emerge over the last few years, as outcome studies in peritoneal dialysis (PD) patients have clearly demonstrated a significant survival benefit with preservation of RRF. However, most nephrologists do not appear to place the same emphasis on monitoring and preservation of RRF in haemodialysis (HD) patients.

The role of RRF in HD patients is often neglected as there is a common notion that RRF is lost rapidly after initiation of haemodialysis and hence inconsequential in terms of patient outcome. There has also been the concern that PD is minimally adequate; thus, RRF may play a more essential role in PD rather than in HD where it was thought that the loss of RRF can be compensated by the concomitant increase in small solute clearance provided by HD therapies. It does not help that the measurement of RRF in HD is more complicated than that in PD, as HD patients are not in a steady state because of the intermittent nature of the dialysis therapies.


Residual renal function is a significant determinant of morbidity and mortality in dialysis patients. Several studies have reported that RRF, but not the dose of PD predicted mortality.1–4 The original CANUSA study where total (peritoneal and renal) small solute clearance significantly predicted mortality, resulted in the assumption that peritoneal small solute clearance must be important.5 However, a re-analysis of the CANUSA data by Bargman et al. showed that the predictive power of lower mortality lay exclusively in the RRF, and not in the small solute clearance component.6 Each increment of 5 L/week per 1.73 m2 in residual kidney glomerular filtration rate (GFR) was associated with a 12% reduction in the relative risk (RR) of death while there was no association with peritoneal creatinine clearance. Every 250 mL of daily urine output conferred a 36% reduction in mortality. What is not commonly recognized in the nephrology community is that there were data to suggest that RRF is important in the HD population as well.


Shemin et al. reported that the presence of RRF, even at a low level, was associated with a lower mortality risk in HD patients in a prospective cross-sectional observational study involving 114 prevalent HD patients in a single centre followed up over 2 years.7 The presence of any RRF (measured as urea clearance or creatinine clearance) was protective against mortality (odds ratio for death 0.44, 95% confidence interval 0.24–0.81, P = 0.008), when compared with those without RRF, even after adjustment for duration of dialysis treatment and other baseline variables.

Termorshuizen et al. analysed 740 HD patients who were part of the Netherlands Cooperative Study on the Adequacy of Dialysis (prospective, multicentre, observational cohort study of incident dialysis patients), and showed that RRF and delivered Kt/V were both positively associated with better survival: each increase of 1/week in renal Kt/V was associated with a RR of death of 0.44 (95% confidence interval 0.30–0.65, P < 0.0001) and each increase in delivered Kt/V was associated with a RR of death of 0.76 (95% confidence interval 0.64–0.92, P = 0.0035).8 However, the effect of delivered Kt/V on mortality was strongly dependent on the presence of RRF: low values for delivered Kt/V were associated with a significantly higher mortality in anuric patients only. In statistical terms, the effect of RRF appeared to be stronger than the effect of delivered Kt/V – in agreement with the findings in PD studies that dialysis dose as measured by the removal of low molecular weight solutes, has a much weaker effect on the survival of dialysis patients than the magnitude of RRF.

Unfortunately, patients with significant RRF (>1.5 mL/min per 35 L of urea) were excluded from the HEMO study; hence no conclusions could be drawn from this large HD study with regards to RRF and survival.9


There are a number of biologically plausible reasons why preservation of RRF might be an important strategy in our attempt to improve survival of HD patients. Much of the information, however, is extracted from studies conducted in PD patients, as very few studies have examined RRF in HD patients. The potential benefits in the PD population which are described in the following paragraph include better clearance of middle and larger molecular weight toxins, better volume and blood pressure control, decreased inflammation, improved appetite and nutritional status, relative preservation of renal endocrine functions, improved phosphate control and quality of life.

Patients with RRF have better clearance of middle molecules and larger molecular weight toxins. Presence of RRF is associated with lower β2 microglobulin10–17 and p-cresol levels.10 The contribution of RRF to the elimination of these middle and larger molecular weight toxins and possibly other uraemic toxins, may serve to improve survival in a manner not measurable by conventional small solute adequacy studies. RRF improves fluid control in PD patients, as the capacity for sodium18,19 and water removal is enhanced in view of the ‘continuous’ nature of RRF. It reduces the need for strict fluid restriction, and plays a crucial role in the prevention of fluid overload in PD patients. This may also help to optimize blood pressure control,20 and prevent or alleviate left ventricular hypertrophy (LVH) which is commonly seen in dialysis patients, especially in fluid overloaded states.21 Loss of RRF in PD patients was associated with worsening blood pressure control in a retrospective study of 207 patients by Menon et al.20 and Wang et al. demonstrated an independent effect of the loss of RRF with more severe LVH in a retrospective cross-sectional study of 158 PD patients.21 Given the fact that cardiovascular mortality is the leading cause of death in dialysis patients, it is perhaps not surprising that better control of volume status and blood pressure leads to improved survival.22 Loss of RRF is associated with increased inflammation in PD patients and inflammation has been identified as a predictor of all-cause and cardiac mortality.22–25 RRF also helps to maintain optimal nutritional status. It has been associated with preservation of appetite, increased dietary protein and caloric intake, increased lean body mass, and improved nutritional parameters including normalized protein catabolic rate, serum albumin and transferrin.13,26–30 Haemoglobin levels are higher in patients with better preserved RRF,13,21 possibly related to higher levels of endogenous erythropoietin and consequently a decreased requirement for exogenous erythropoietin administration in these patients.21 Higher levels of RRF may also preserve the endocrine functions of the kidneys better, including possibly higher levels of active Vitamin D metabolites. Other benefits of preserving RRF include better phosphate balance11,13,31 and improved quality of life.2,32,33

Less information is available in the published literature on this issue in HD patients. Studies have reported lower β2 microglobulin levels, improved nutritional markers, an increase in true dry weight, higher haemoglobin and endogenous erythropoietin levels, lower potassium and uric acid levels, better phosphate balance and improved quality of life in HD patients with better preserved RRF.12,32–40

Residual renal function is also an important component of the overall treatment efficacy delivered by either PD or HD, and contributes to the success of the concept of incremental dialysis whereby patients with significant RRF can be initiated on smaller doses of dialysis at the beginning, hence according such patients better quality of life and less restrictions from the full dose of dialysis right from the onset.


On the negative side, maintenance of RRF may prolong protein malnutrition and worsen hypoalbuminemia in highly proteinuric patients.41 There has also been an intriguing concern as to whether preservation of RRF in dialysis patients is actually at the expense of chronic volume overload. Gunal et al. in a prospective crossover study of 19 incident HD patients, subjected these patients to 3 months of antihypertensive therapy initially, followed by 3 months of strict volume control with salt restriction and aggressive ultrafiltration.42 There was a 6% reduction in the left ventricular mass index measured by echocardiography after antihypertensive therapy, compared with a 36% reduction after strict volume control. A dramatic drop in daily urine output after achieving euvolaemia with strict volume control was noted. The authors concluded that volume control had more significant effects when compared with blood pressure control alone with antihypertensives in inducing regression of LVH and that loss of RRF is the price to be paid for effective volume control, as persistence of RRF was due to a state of volume overload. This study was in direct contrast to the retrospective study by Wang et al. where loss of RRF was associated with more severe LVH.21 Important limitations of the study by Gunal et al. should be kept in mind, including a very small study population, better blood pressure reduction achieved after strict volume control as opposed to antihypertensive therapy, short follow-up period, and lack of adjustment for the longitudinal decline of RRF with time as the patients received antihypertensive therapy before commencing strict volume control.


Peritoneal dialysis therapies have long been considered to preserve RRF better than HD therapies. This was first reported by Rottembourg et al. in 1982 when they showed that creatinine clearance remained significantly higher in PD patients than in matched patients on HD.43 Their results were subsequently duplicated by various authors (Table 1).44–49 These studies all reported a significantly higher rate of decline of RRF in HD compared with PD patients. This was postulated to be due to improved cardiovascular stability with decreased hypotensive episodes in PD patients as a result of the continuous nature of the dialysis therapy, and/or the fact that the peritoneal membrane is more biocompatible than the membranes used in haemodialysers. Other factors which have been implicated include removal of GFR stimulatory factors and the osmotic drive during HD. An important point of note is that many of these studies were conducted using old HD technologies and practices of bioincompatible low flux cellulose membranes, acetate buffered fluids (with higher incidence of hypotensive events),51,52 and no volumetric ultrafiltration control. Comparison of these studies is difficult because of differences in study design and statistical methods, and flaws in methodology (including retrospective design, small study population, short follow-up period, Neyman bias, patient selection bias and lack of adjustment for baseline variables) in some of the studies. Moreover, there was also no standard measure of RRF in these studies, and methods of measuring residual GFR ranged from simple urine volume to creatinine clearance to arithmetic mean of urea and creatinine clearances.

Table 1.  Studies comparing the rate of decline of RRF in HD versus PD patients
Author YearStudy designHD/PD nMean follow up monthsMeasure of RRFBaseline GFR HD/PD mL/min per 1.73 m2GFR during follow up HD/PD mL/min per 1.73 m2Rate of decline of RRF HD/PD %/monthDifference in rate of decline %Comments
  • Rates were not given in the original article but were calculated by Jansen et al.49

  • Calculated by Jansen et al.49 CCr, creatinine clearance; CUr, urea clearance; HD, haemodialysis; GFR, glomerular filtration rate; PD, peritoneal dialysis; RRF, residual renal function.

More rapid decline of RRF in HD vs. PD43–49
Rottembourg et al. 198243Prospective25/2518CCr
Urine volume
4.3/4.42.1/3.8 (12 months)
1.3/4.0 (18 months)
Cancarini et al. 198644Retrospective75/8626.4 (HD)
25.2 (PD)
CCr Urine volume5.8/6.4 (mL/min)1.3/3.9 (end of study)  Possible selection bias
Lysaght et al. 199145Retrospective7/5521.5 (HD)
33.6 (PD)
CCr5.0/4.5 5.8/2.950 
Moist et al. 200046Retrospective811/103212Urine volume7.3/7.569% of HD patients vs 38% of PD patients lost RRF (defined as urine volume less than 200 mL daily) at 12 months  Urine volume was used alone as a surrogate measure of RRF; Data on baseline urine volume were not collected
Misra et al. 200147Retrospective40/103 ½ (CCr + CUr)4.2/5.1 7.0/2.269Adjusted for informativecensoring
Lang et al. 200148Prospective31/1524CCr7.5/7.43.8/6.0 (12 months)
1.8/3.1 (24 months)
Jansen et al. 200249Prospective279/24312½ (CCr + CUr)5.1/5.81.4/2.2 (12 months)10.7/8.124Possible selection bias with regards to HD patients; adjusted for informative censoring
Identical decline of RRF in HD vs. PD50
McKane et al. 200250Retrospective300/17548 (maximum)CUr   No significant differenceAdjusted for informative censoring


This issue has been intensely debated. Most, but not all recent studies (Table 2) have reported that the use of biocompatible membranes such as polysulphone is associated with a slower rate of decline of RRF when compared with traditional bioincompatible cellulose membranes.40,46,48,49,53–56 The inflammatory milieu generated by the use of bioincompatible cellulose membranes is generally thought to be associated with a more rapid decline in RRF.57,58 One limitation of these studies is the possible confounding factor of flux characteristics as the biocompatible synthetic membranes were largely high flux, while the bioincompatible membranes were low flux in nature. In a prospective cohort study of 522 incident dialysis patients by Jansen et al. the increased rate of decline of RRF in HD as compared with PD patients was only modest, albeit still significant, when compared with earlier studies (Table 1). This may be explained in part by the fact that more biocompatible cellulose derivatives and synthetic dialyser membranes were used in the study, whereas unmodified cellulose membranes were used in the earlier studies.49 Moist et al. failed to demonstrate a significant difference in the rate of decline of RRF between synthetic and cellulose dialyser membranes. However, they used urine volume alone as a surrogate measure of RRF at follow up, and only 18% of the patients were using the cellulose membranes.46 This study had other limitations: follow-up data on RRF were not available for a large proportion of patients because of mortality before data collection and baseline data for urine volume were not collected.

Table 2.  Biocompatibility of dialyser membranes and rate of decline of RRF
Author YearStudy designnFollow up MonthsMeasure of RRFPredictorOutcomeComments
  1. CA, cellulose acetate; CCr, creatinine clearance; CUr, urea clearance; CP, cuprophan; GFR, glomerular filtration rate; MC, modified cellulose; PAN, polyacrylonitrile; PMMA, polymethyl methacrylate; PS, polysulphone; RRF, residual renal function; UC, unmodified cellulose.

Caramelo et al. 199453Prospective Randomized229CCr
Urine volume
CP vs. PAN/PSNo difference in decline of RRF between PAN/PS and CP membranesShort follow up; increased incidence of hypertension; excluded patients with less than 500 mL of daily urine volume
Van Stone 199554Retrospective33460 (maximum)½ (CCr + CCr)CP vs. PS/PMMA/CARRF declined at a faster rate with CP compared with more biocompatible membranes 
Hakim et al. 199655Prospective Randomized15918CUrUC vs. PMMATrend towards slower decline of RRF in PMMA group (not statistically significant)Large number of dropouts in the PMMA group
Hartmann et al. 199740Prospective Randomized2012CCr
Urine volume
CA vs. PSRRF declined at a faster rate with CA compared with PS membranesIncreased haematocrit and (renal + delivered) Kt/V in PS group
McCarthy et al. 199756Retrospective100PS: 16.1
CA: 12.3 (mean)
CUrCA vs. PSRRF declined at a faster rate in non-diabetic patients with renal parenchymal disease using CA compared with PS membranesDifferences in reprocessing of dialysers; increased delivered Kt/V in PS group
Moist et al. 200046Retrospective81112 (mean)Urine volumeUC vs. MC/syntheticNo statistically significant difference in RRF loss between MC/synthetic and UC membranesOnly 18.3% of patients were using UC membranes; urine volume alone was used as surrogate measure of RRF
Lang et al. 200148Prospective Randomized3024CCrCP vs. PSRRF declined at a faster rate with single use CP compared with PS membranes 
Jansen et al. 200249Prospective27912½ (CCr + CCr)MC vs. syntheticNo effect of dialyser membrane on residual GFR at 3 monthsNo analysis for effect of dialyser membrane on residual GFR beyond 3 months


Recent studies utilizing contemporary HD technology have shown that RRF is better preserved than in the past. These studies, reflecting contemporary conditions of HD therapy using high flux biocompatible synthetic membranes and ultrapure, bicarbonate buffered dialysis fluids with ultrafiltration control (reducing the risk of intradialytic hypotensive events) have shown that the decline in RRF has been significantly retarded compared with previous studies.50,59

In a prospective randomized study involving 30 incident HD patients, Schiffl et al. demonstrated that the rate of decline of RRF (measured by creatinine clearance and urine volume) with single use biocompatible high flux synthetic membranes, bicarbonate buffer and ultrafiltration control, was slower for patients using ultrapure dialysis fluid (bacterial growth of 0 CFU/mL, endotoxins < 0.03 EU/mL) compared with conventional dialysis fluid (bacterial growth ranging from 0 to 230 CFU/mL, endotoxins ranging from <0.03 to >0.25 EU/mL in 1 patient) over a 24 month follow-up period.59 This was associated with lower serum C-reactive protein and interleukin-6 levels in the group using ultrapure dialysis fluid. In a retrospective study involving 475 incident dialysis patients, McKane et al. reported that the rate of decline of RRF (measured by urea clearance) was identical in HD patients using biocompatible high flux synthetic (91.3% polysulphone, 3.6% polyacrylonitrile) dialyser membranes with bicarbonate buffer and ultrapure water (0 CFU/mL, <0.015 EU/mL), when compared with CAPD patients, over a 48-month follow-up period.50 This is the only, and most recent comparative (albeit retrospective) study to date which did not demonstrate a more rapid decline of RRF in HD compared with PD.

These studies have rekindled interest in the HD community, necessitating a re-look into the issue of RRF in HD patients, given the benefits of RRF and better preservation of RRF with contemporary HD practices (Table 3).

Table 3.  Possible factors accounting for better preservation of RRF in contemporary HD
1.Biocompatible synthetic dialyser membranes
2.Ultrapure water
3.Improved cardiovascular stability (through the use of bicarbonate buffered fluids and ultrafiltration control)
4.High flux dialysers


Unfortunately there does not appear to be a consensus for the measurement of RRF in the HD population. Unlike the PD population where there is a consistent method of measuring RRF60 this is not the case for HD patients. It is well accepted that serum creatinine alone is not a good measure of RRF. GFR equations including the Cockcroft Gault and Modification of Diet in Renal Disease formulas have not been validated in the dialysis population. Van Olden et al. reported that cimetidine did not improve the accuracy of urinary creatinine clearance in HD patients, and that creatinine clearance in the last 24 h of dialysis interval appeared to give a better approximation of GFR because of minimal tubular secretion of creatinine in that period.61 One might assume that urea clearance will provide a more accurate estimate of GFR in HD patients given the substantial tubular secretion of creatinine at low GFR. Milutinovic et al., however, reported that creatinine clearance, rather than urea clearance, correlated better with inulin clearance, at values between 1 and 5 mL/min of inulin clearances.62 They also reported that the mean of urea and creatinine clearances as a measure of GFR had a similar correlation with inulin clearance as that of creatinine clearance alone (correlation coefficient: 0.92 vs 0.90) when inulin clearances were between 1 and 5 mL/min. At inulin clearances of less than 1 mL/min, there was not much separating the use of creatinine clearance, urea clearance or the mean of the two. Use of urine volume as a measure of RRF also has its advocates and objectors. The use of radionuclide methods for measuring RRF is not practical in a day-to-day clinical setting.

The updated KDOQI clinical practice guidelines in 2006 recommended using urea clearance (corrected for body surface area) as a measure of RRF in both HD and PD patients, as residual urea clearance is lower than that of residual kidney GFR, and hence protects the patient from under-dialysis.63 Inclusion of native kidney urea clearances in kinetic modelling programs also allows accurate calculation of G (urea generation rate) and normalized protein catabolic rate as an aid to nutritional assessment. At the same time, it provides an opportunity to add RRF to dialyser clearance. The guidelines mentioned that residual urea clearance is best measured from a timed 24 h urine collection but did not specify if the urine collection needs to be timed to the dialysis cycle.

The European BEST practice guidelines published in 2002 recommended using the mean of urea and creatinine clearances (corrected for body surface area), as a measure of the residual kidney GFR in both HD and PD patients.64 However in HD, unlike PD, the blood urea and creatinine concentrations as well as the residual kidney GFR vary during the dialysis cycle; hence urine should be collected over a complete dialysis cycle, beginning at the initiation of one dialysis and ending at the start of the next. To correct for the fluctuations in blood urea and creatinine concentrations, blood should be sampled immediately after the end of dialysis and before the start of the next dialysis, for calculation of mean blood urea and creatinine levels.


Apart from the complexities involved in the measurement of GFR in HD patients, the total clearance achieved is not a simple arithmetic addition of residual GFR and dialyser clearance because of the intermittent nature of HD therapies. The continuous nature of RRF contrasts with the intermittent schedule of HD and evidence suggests that continuous clearance is more efficient than intermittent clearance, and that the contribution of RRF to overall kidney and dialyser function is greater than the simple addition of time-averaged clearances would suggest. One can reduce the discontinuous clearance of HD to a continuous equivalent (standard Kt/V) allowing simple addition, or convert the continuous native kidney clearance to an intermittent equivalent before addition.65,66

This consideration is particularly important when patients undergo a less intensive HD regime when they have significant RRF or what is commonly known as the incremental approach to HD dosing. RRF should be measured regularly in such instances to prevent under-dialysis from progressive loss of RRF.


Factors affecting RRF have been extensively studied in the PD population,46,49,67–74 but information in the HD population has been sparse. Demographic factors (including age, gender, race), aetiology of renal failure, blood pressure control, proteinuria, nephrotoxic agents, medications (including angiotensin converting enzyme inhibitors {ACE-I}, angiotensin II receptor antagonists {ARB}, β-hydroxy-β-methylglutaryl Co-A reductase inhibitors, calcium channel blockers), comorbidities, acute illnesses and chronic inflammation have all been implicated in one way or other, although the data are conflicting at the present moment, especially in HD patients.45,49,53,59,75

Dialysis-related factors including biocompatibility of dialyser membranes, intradialytic hypotension, use of bicarbonate-buffered dialysis fluid and ultrapure water have been alluded to earlier. The role of membrane flux has not been established as no studies have directly compared the decline of RRF between low and high flux synthetic membranes.

Novel therapies of HD, including short daily HD and nocturnal HD are gaining recognition in the dialysis community. Although data from well-conducted, adequately powered randomized controlled trials have not been published yet, accrued worldwide experiences with these modalities have been encouraging. RRF in this group of patients have not been studied although it is reasonable to postulate that these patients might have better preservation of RRF. The ability to decrease ultrafiltration rates (as a result of increased frequency and/or duration of dialysis) has markedly reduced the incidence of intradialytic hypotension, leading possibly to better preservation of RRF. It is possible that improved preservation of RRF could have contributed in part to the clinical benefits reported with daily/nocturnal dialysis. Further studies are required to validate this hypothesis.


It is counterintuitive to advocate aggressive measures to preserve RRF prior to initiation of dialysis, only to ignore it once dialysis has been initiated. Very frequently, HD patients are subjected to nephrotoxic insults including intravenous contrast, antibiotics (e.g. aminoglycosides) and intradialytic hypotension with little consideration of RRF which is often thought to be non-existent or not important once dialysis has been initiated.

From the recent studies available, it does appear that biocompatible, preferably high flux synthetic membranes with bicarbonate buffered dialysis fluids, and/or ultrapure water can retard the loss of RRF in HD patients.40,50,56,59 Intradialytic hypotension should be avoided and overzealous ultrafiltration to achieve rapid euvolaemia at the expense of hypotension should be discouraged. Attention to maintenance of hydration is important in relation to general anaesthesia, major surgery and acute illnesses. It also makes sense to avoid nephrotoxic agents including intravenous radiocontrast, aminoglycosides, and non-steroidal anti-inflammatory drugs where possible, until further data are available on the use of such agents in the HD population with significant RRF. Diuretics appear to have a role in maximizing urine output and minimizing the need for aggressive ultrafiltration but did not influence the decline of RRF as measured by small solute clearances in PD patients.76,77 There have also been studies demonstrating the benefit of Ramipril and Valsartan in retarding the loss of RRF in PD patients.78,79 Whether ACE-I and/or ARB's play a similar role in HD patients is not well defined although Moist et al. failed to demonstrate a statistically significant benefit of the use of ACE-I in slowing the loss of RRF in the HD population (limitations of this study have been described earlier).46


Although RRF is important in HD there seem to be more questions than answers currently. Clinical trials are urgently needed. We need to define the best method of measuring RRF in HD patients, establish the rate of decline of RRF under contemporary HD conditions and the predictors of decline of RRF in this population, to allow formulation of strategies to best preserve RRF. Does early initiation of HD affect the rate of decline of RRF? Is RRF important in determining the survival of patients on HD? With increasing numbers of patients being placed on daily/nocturnal HD regimes as well as newer techniques of haemofiltration/haemodiafiltration, we also need to know if these practices affect the rate of loss of RRF.


Residual renal function appears to play an important role even in HD patients, as we are unable to replace the loss of RRF with enhanced small solute clearances, regardless of the technological advancements in HD therapies. Emphasis should be placed on the routine measurement of RRF and continued preservation of RRF in this population. Well-conducted, adequately powered prospective studies are required to determine the magnitude of this benefit and how best to achieve this in the HD population.