Show Summary Details
Page of



Ken Farrington

and Roger Greenwood



Timing of initiation of dialysis—the IDEAL (Initiating Dialysis Early And Late) study revealed that planned early initiation was not associated with an improvement in survival or clinical outcomes.

Frequency of dialysis—the randomized Frequent Hemodialysis Network Trial showed the benefits of 6 times weekly versus conventional 3 times weekly sessions.

Management of heparin-induced thrombocytopenia—recommendations for the use of heparinoid or argatroban.

Management of hypertension in dialysis patients—discussion of the general move for guidelines to become less prescriptive, recognizing risks of overtreatment as well as of undertreatment.

Management of hyperlipidaemia in dialysis patients—discussion of the findings of the SHARP trial, which showed that lowering LDL cholesterol with simvastatin plus ezetimibe reduced major atherosclerotic events, but did not have significant effects on cardiovascular or total mortality.

Updated on 25 May 2011. The previous version of this content can be found here.
Page of

PRINTED FROM OXFORD MEDICINE ONLINE ( © Oxford University Press, 2015. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in Oxford Medicine Online for personal use (for details see Privacy Policy).

Subscriber: null; date: 20 March 2018


Over the past four decades, maintenance haemodialysis has proved to be a highly successful treatment for patients with endstage renal disease. In the developed world, the haemodialysis population continues to increase and is becoming more elderly and dependent. However, despite considerable advances in haemodialysis technology and other significant improvements, such as those in renal anaemia management, the long-term clinical outcomes for patients remain much less good than those of other people with comparable characteristics but without renal failure.

Dialysis adequacy—working definitions are based on small-solute—typically urea—removal, which occurs mainly by diffusion. Current guidelines recommend targeting a normalized urea clearance (eKt/V) in excess of 1.2 per session for thrice-weekly treatment. Higher doses of dialysis delivered thrice weekly do not produce significant improvement in outcomes, but there is evidence that frequent treatment (6 times weekly) can do so, and there is also a need to incorporate more holistic approaches into the concept of adequacy.

Technical aspects of haemodialysis—high-flux membranes are needed to achieve significant removal of middle molecules, of which β‎2-microglobulin is the prime example. Use of such membranes can slow the progression of dialysis-related amyloidosis, a complication of long-term haemodialysis related to β‎2-microglobulin retention, but firm evidence of an effect on survival is lacking. The technique of haemodiafiltration adds a significant convective component to high-flux haemodialysis, providing improved β‎2-microglobulin clearance, with observational studies suggesting a survival benefit.

Vascular access—the need to secure and maintain reliable vascular access is fundamental to achieving adequate dialysis and maintaining health. An arteriovenous fistula is the preferred option, with fewer complications and longer survival than other access options. The current overdependence on tunnelled lines contributes to morbidity and excess mortality, mainly from line-related sepsis, and represents a failure in access provision.

Complications of haemodialysis—the main acute complication of haemodialysis is intradialytic hypotension, resulting from an imbalance between the ultrafiltration rate and the rate of vascular refill. Underlying cardiovascular disease, antihypertensive drugs, autonomic dysfunction, shortened dialysis times, large interdialytic fluid gains, and inaccurate dry-weight assessment all predispose. Postdialysis headache is relatively common, as is prolonged bleeding from the fistula after needle removal. Dialyser reactions are less common. More serious bleeding problems with anticoagulation are rare. True heparin-induced thrombocytopenia (HIT-II) is very rare, but is potentially life threatening.

Comorbidities in patients on dialysis—the high mortality in the dialysis population is mainly due to accelerated cardiovascular disease. Traditional risk factors for atheroma such as hypertension and dyslipidaemia are common, but difficult to interpret: there seems to be a paradoxical relationship between blood pressure and mortality in patients on haemodialysis (‘reverse epidemiology’), and there are questions about the appropriateness of blood pressure targets in the absence of randomized interventional trials of the effect of blood pressure reduction in this population. Similarly, while therapy to lower LDL cholesterol may reduce the incidence of major atherosclerotic events in dialysis patients, it does not produce a significant reduction in cardiovascular or overall mortality. In addition, there are a myriad of less traditional risk factors operating in dialysis patients, including anaemia, disturbances of calcium and phosphate metabolism, hyperhomocysteinaemia, increased oxidative stress, and elevated inflammatory markers. It may be that the combined effect of all these risk factors outweighs the benefits of a single intervention, and more global approaches are required. It remains to be seen whether the improved control of the uraemic milieu possible with more frequent dialysis schedules will impact on survival.

Is haemodialysis in the best interest of the patient?—haemodialysis, although universally applicable, has its limitations. Dialysis initiation may not improve quality of life or survival for some very high-risk dependent patients, for whom supportive therapy, and not dialysis, may be the best treatment option.


The availability of effective renal replacement therapy (RRT) has transformed the outlook for patients with endstage renal failure (ESRF) over the past 40 years, replacing certain and imminent death with the potential for long-term survival. The modalities of RRT comprise haemodialysis, peritoneal dialysis and renal transplantation. These therapies are appropriately regarded as complementary but, because of its very few contraindications, haemodialysis can be regarded as the default modality. Growing numbers of patients worldwide are now dependent on regular haemodialysis to sustain life, the escalating proportion of older and frailer patients being direct testimony to the flexibility of the treatment. Inevitably, this success needs qualification. (1) The functions of the kidney are many and diverse and dialysis provides only partial replacement for some, notably the excretion of nitrogenous waste products, and the control of water, electrolyte, and acid–base balance. Additional measures are required to manage problems such as anaemia and abnormal mineral metabolism (see Chapter 21.6). (2) Although there is agreement about the general aims of dialysis treatment—to maintain life and prevent or reduce morbidity from uraemia—there is no consensus on the best way to achieve these aims. There is still debate about what constitutes adequate dialysis and even about which parameters we should be trying to control. (3) Although dialysis undoubtedly prolongs life in patients with ESRF, mortality still far exceeds that in the general population. This is mainly due to cardiovascular disease, which is endemic and runs an accelerated course in dialysis patients, whatever the modality employed. This chapter outlines the scope of current practice in haemodialysis, stressing those aspects most relevant to clinical management.

Development of haemodialysis

The pioneers

The process of dialysis was discovered in the mid-19th century as a means of separating dissolved elements by diffusion through a semipermeable membrane (Graham, 1861). Subsequent evolution of the technique from the first use of an artificial kidney in experimental studies in dogs (by Abel and colleagues, 1913) to the first successful dialysis treatment of a patient with acute renal failure (by Kolff, 1943) followed technical advances in the development of the semipermeable membrane cellophane, which was originally introduced as sausage skin, and of anticoagulants—first hirudin, then heparin. The extension of the technique to the maintenance treatment of patients with ESRF followed the development in the 1960s of the Scribner silastic shunt and the Brescia–Cimino arteriovenous fistula, which allowed safe and reliable long-term venous access.

Expanding services

Haemodialysis treatment was originally reserved for young and otherwise fit patients. Rigid selection criteria were applied, but ethical concerns, together with rising patient expectations, eventually led to more liberal policies. There were marked geographical differences, largely economically driven, in subsequent rates of expansion of dialysis programmes, in the modalities employed, and in the patterns of service provision. In the United States of America and most of Europe, haemodialysis was rapidly decentralized from the pioneering units into small but numerous hospital-based units and freestanding facilities in cities, towns, and rural areas (Fig. By contrast, there was little decentralization in the United Kingdom from the 50 or so large metropolitan centres, and self-supervised home haemodialysis was the preferred means of expansion. Selection criteria remained tight.

Fig. Prevalent patient counts, by modality, for all patients receiving renal replacement therapy in the United States of America between 1980 and 2004.

Prevalent patient counts, by modality, for all patients receiving renal replacement therapy in the United States of America between 1980 and 2004.

(Data from the United States Renal Data Systems (USRDS).)

The advent of continuous ambulatory peritoneal dialysis (CAPD) in the early 1980s was seized on as a means to liberalize access to treatment in the United Kingdom. CAPD became the dominant dialysis mode, engulfing the home haemodialysis programme and relegating centre-based haemodialysis to the status of rescue mode for those in whom CAPD was precluded or had failed. However, in the last 15 years the haemodialysis base in the United Kingdom has expanded, mainly through the building of satellite units, owing to the combined effects of an increasingly aged and frail dialysis population, many of whom are unable to cope with the requirements of CAPD or other peritoneal dialysis techniques, recognition that with loss of residual renal function peritoneal dialysis may not provide adequate treatment (see below), and also because of the flow of capital from commercial dialysis companies. This has led to a convergence towards the American/European pattern of provision (Fig. Haemodialysis has become the dominant dialysis modality in the United Kingdom, with peritoneal dialysis falling to less than 20% of prevalent patients.

Fig. Modality of renal replacement therapy in the United Kingdom between 1997 and 2005. APD, automated peritoneal dialysis; CAPD, continuous ambulatory peritoneal dialysis; HD, haemodialysis.

Modality of renal replacement therapy in the United Kingdom between 1997 and 2005. APD, automated peritoneal dialysis; CAPD, continuous ambulatory peritoneal dialysis; HD, haemodialysis.

Impact of adequacy

Although patients continued to dialyse thrice weekly, there was a general trend to reduce dialysis times in response to increasing funding constraints and patient preference. ‘Short dialysis’ (<4 h/session) was held responsible for the excess mortality in American haemodialysis patients in the 1980s, which focused attention on the concept of dialysis adequacy. Most units now prescribe and monitor dialysis dose using the urea kinetic methods established at that time. The same concepts were also applied to CAPD, leading to recognition that CAPD adequacy is critically dependent on residual renal function, and that loss of residual renal function compromises adequacy, sometimes to the point of technique nonviability, especially in larger patients. This was a significant factor in the steady relative decline in the United Kingdom CAPD population during the 1990s (Fig., a decrease that has been more than offset by increased centre-based haemodialysis provision. The automated peritoneal dialysis (APD) programme has also grown, but it remains small. In the last decade or so, dialysis practice in the United Kingdom has tended to converge to the American and European model.

Changing demographics

Demographic changes have placed increased demands on nephrological and other specialist resources. The median age of new patients commencing on dialysis in the United Kingdom has increased considerably over the last two decades and is now over 64 years, with around one-quarter over 75 years. Older people account for most of the increase in incidence and prevalence rates, which now exceed 100 and 600 patients per million population, respectively. The proportion of patients commencing dialysis with extrarenal comorbidities, particularly cardiovascular disease, has increased dramatically, and multiple pathologies are common. Around 20% of United Kingdom dialysis patients are diabetic, and in the United States of America this figure is around 50% (see Chapter 21.6). Many such patients have widespread micro- and macrovascular complications at the time of dialysis initiation.

Impact of transplantation

The haemodialysis population has continued to grow despite active transplant programmes. Two main factors account for this. First, the major growth in the dialysis population is among older people, who tend to have a high comorbid load such that renal transplantation is contraindicated. Second, the donor pool is limited, despite initiatives to increase this, such as encouragement of living donation and use of asystolic donors (see Chapter 21.7.3).

Impact of conservative (nondialytic) management

There is increasing recognition that dialysis treatment may be of limited benefit to some elderly, dependent patients, many of whom have multiple comorbidities. In such patients, dialysis may not enhance quality of life or life expectancy and some may choose not to dialyse, opting instead for a conservative approach. Increased awareness of and provision of conservative management pathways and services may produce a small reduction in the rate of growth of the dialysis population. Where conservative management programmes have been established, up to 15% of patients reaching ESRF are supported by this modality.

Technical aspects of haemodialysis

Principles of dialysis

Dialysis is a physicochemical process allowing separation of the components of a complex solution by solute exchange across a semipermeable membrane. Such membranes act as molecular size-selective filters, the size threshold depending on the nature of the membrane. In haemodialysis, the membrane is interposed between the patient’s bloodstream and a rinsing solution (dialysis fluid). Diffusive and convective mass transfer takes place across the membrane, allowing changes in the composition of body fluid compartments.

The rate of diffusive transport of a solute is dependent on its molecular weight, electrical charge, the blood–dialysis fluid concentration gradient, blood and dialysis flow rates, and membrane characteristics (membrane area and membrane mass transfer-area coefficient, KoA). With current designs of dialysers, blood is perfused through a bundle of narrow capillaries with a total surface area between 1 and 1.5 m2. Small molecules such as urea (60 Da) are cleared well, but larger molecules such as albumin (60 000 Da) cannot pass through the membrane. The clearance of middle molecules such as β‎2-microglobulin (11 800 Da) can be improved somewhat by using high-flux membranes, which have pores of sufficient size to allow their passage, although fluid layering still presents a substantial barrier to slowly diffusing large molecules.

Convection involves the bulk movement of solvent and solute across the membrane. The driving force is transmembrane hydrostatic pressure, which can be adjusted by application of variable negative pressure to the dialysate side of the membrane. Solute transport (by solvent drag) is independent of diffusion. In general, convection contributes little to the clearance of rapidly diffusible small solutes such as urea, but can add significantly to the diffusive clearance of middle molecules by high-flux membranes. Convective movement of water from blood across the membrane is known as ultrafiltration, the rate of which depends on the hydrostatic pressure difference across the membrane and on its permeability to water (ultrafiltration coefficient, Kuf).

Membranes and dialysers

The original haemodialysis membranes were fashioned from regenerated cellulose, but technology has since proliferated and there are now three major classes of membrane (Table In dialysers, these semipermeable membranes are arranged to form separate adjacent paths for blood and dialysis fluid, which flow on opposite sides of the membrane in opposite directions to maximise diffusion gradients.

Table Haemodialysis membranes

Membrane class


Hydraulic permeability

β‎2-Microglobulin clearance


Regenerated cellulose




Modified cellulose

  • Cellulose acetate

  • Cellulose diacetate

  • Cellulose triacetate

Low/high flux




  • Polymethylmethcrylate

  • Polyacrylonitrile

  • Polysulphone

  • Polyamide

  • Polycarbonate

High/low flux



Dialysers are classified by design type (modern dialysers are usually of hollow-fibre design), membrane composition, surface area, and membrane permeability characteristics. Membrane permeability is defined in terms of dialyser clearance Kd for a range of solutes, and ultrafiltration coefficient Kuf, which is the water flux per unit of transmembrane pressure. A range of Kd values is normally quoted for a particular solute, these being derived from the dialyser’s mass transfer-area coefficient KoA for that solute at a variety of appropriate blood and dialysis fluid flow rates. In contrast to cuprophane, high-flux synthetic membranes are highly permeable (high Kuf and high Kd for middle molecules), remove β‎2-microglobulin and other potentially toxic middle molecules, and tend to be more biocompatible, meaning that they cause less activation of inflammatory cells, the complement cascade, and contact pathways, and less cytokine production. However, in spite of numerous experimental demonstrations of such potentially beneficial effects, it is still unclear whether clinical use of biocompatible membranes translates into improved outcomes.

High-flux membranes permit high-volume ultrafiltration, which can be ulitised to combine convective clearance with the diffusive clearance offered by haemodialysis in the technique of haemodiafiltration (see later). As a consequence of the high flow rates employed and the countercurrent flow, there is an unavoidable movement of dialysis fluid into blood. This so called ‘obligatory back-filtration’ (Fig. provides some useful convective clearance (internal haemodiafiltration), but also underscores the need to use ultrapure water in the preparation of dialysis fluid. Most would argue that this is highly desirable anyway. The observed absence of clinical sequelae from back-filtration may reflect the effectiveness of the membrane as a bacterial filter, but the absorptive capacity (for endotoxins) of the ‘thick’ matrix supporting the diffusing surface of the membrane may also be important.

Fig. ‘Obligatory back-filtration’ across a high-flux dialysis membrane. At the ‘arterial’ end of the dialyser (left side) there is convective movement of ultrafiltrate from blood to dialysis fluid, but at the ‘venous end’ (right side) the situation is reversed and some dialysis fluid moves across the membrane and into the blood.

‘Obligatory back-filtration’ across a high-flux dialysis membrane. At the ‘arterial’ end of the dialyser (left side) there is convective movement of ultrafiltrate from blood to dialysis fluid, but at the ‘venous end’ (right side) the situation is reversed and some dialysis fluid moves across the membrane and into the blood.

Dialysis water and fluids

Patients on haemodialysis are intimately exposed to huge quantities of water. Normally people are exposed (by drinking and eating) to around 15 litres of water each week, from which they are protected by the gut. By contrast, haemodialysis patients are exposed to 300 to 600 litres/week and protected only by the dialysis membrane. The potential for poisoning by waterborne impurities is significant. Aluminium and chloramines are examples of proven toxins, which must be removed. Bacterial and endotoxin contamination can produce major acute problems, including intradialytic pyrexias and hypotension, and may also add to the inflammatory milieu, which is a feature of uraemia. Use of ultrapure water is important generally, but perhaps especially in high-flux modes in which dialysis fluid is infused, passively (back-filtration) or actively (online haemodiafiltration), directly into the patient’s bloodstream. Mains water supplied to the renal unit is purified by a combination of techniques which include softening and deionization, carbon adsorption, reverse osmosis, ultraviolet irradiation, and ultrafiltration. Regular monitoring is required to ensure that chemical and microbiological standards are maintained. Currently, the latter accept low levels of endotoxin and bacterial growth on culture; ultrapure water is defined by the absence of both.

Acid and bicarbonate concentrates are mixed with treated water in a single-patient proportionating system within the dialysis machine to produce dialysis fluid of the desired composition (Table Regulation of dialysis fluid composition is the main tool to maintain normal electrolyte and mineral concentrations and normal acid–base balance in body fluid compartments. The concentrations of substances in the dialysis fluid may be set to effect net diffusive loss (e.g. potassium), net diffusive gain (e.g. bicarbonate), or net zero balance (e.g. calcium). Sodium concentrations are generally set to achieve net zero diffusive balance with the required sodium removal being attained by ultrafiltration (Fig. Although there is great potential for individualization, a programme-standard composition is typically adopted, which may be varied in particular circumstances.

Table Bicarbonate dialysis: concentrate composition achieved in the dialysate


Acid (mmol/litre)

Base (mmol/litre)

























Fig. Diffusive transport within a dialyser.

Diffusive transport within a dialyser.

The extracorporeal circuit

As illustrated in Fig., blood is withdrawn from the patient via the ‘A’ needle inserted in the fistula (or the arterial limb of a dual-lumen dialysis catheter) by a peristaltic pump, circulated through the dialyser, and returned to the patient through the ‘V’ needle (or the venous limb of a dual-lumen dialysis catheter). The circuit is anticoagulated by heparin, which is infused downstream from the blood pump. The arterial pressure monitor (Pa) protects the fistula or central vein by detecting excessive negative pressure. The venous pressure monitor (Pv) protects against blood loss from the circuit to the environment, although it is important to understand that leaks arising from dislodgement of the ‘V’ needle from the fistula may not be detected, since the major resistance to flow at this point in the circuit arises from the needle itself and not the fistula. Obstruction downstream of the needle or catheter does raise venous pressure. The bubble trap level detector protects against air embolus: a falling level activates a venous clamp and stops the blood pump.

Fig. A standard extracorporeal circuit. Pa, arterial pressure detector; Pv, venous pressure detector; A needle, arterial needle; V needle, venous needle.

A standard extracorporeal circuit. Pa, arterial pressure detector; Pv, venous pressure detector; A needle, arterial needle; V needle, venous needle.

The dialysis machine

The dialysis machine supplies dialysis fluid at the prescribed flow rate, temperature, and chemical composition in a fail-safe manner. It also monitors the extracorporeal circuit and in fail-safe mode activates the venous clamp and switches off the blood pump. Ultrafiltration is also controlled, volumes being preset by the operator (see below). In addition to housing the blood pump and heparin pump, most modern machines also incorporate additional devices allowing single-needle dialysis and haemodiafiltration (see below), and they may also include other technical advances such as blood temperature monitors, allowing control of thermal balance during dialysis (which may improve haemodynamic stability) and measurement of access recirculation, blood volume monitors, which detect changes in haematocrit during dialysis and are potentially useful in predicting episodes of hypotension, and devices for measuring ionic dialysance, which can allow online monitoring of adequacy parameters. The clinical usefulness of these recent technical developments remains to be fully established.

Control of ultrafiltration

Modern dialysis machines use volumetric methods that permit precise control of ultrafiltration. A balancing system placed in the dialysis fluid pathway in the machine balances dialysis fluid flow rates to and from the dialyser (Fig. The ultrafiltration pump can be preset by the operator to remove fluid from the return limb from the dialyser, causing an equal volume to be drawn from the blood across the dialysis membrane into the dialysis fluid in order to maintain balance.

Fig. The dialysis fluid pathway and the volumetric control of ultrafiltration. F1, filter; A, arterial line; V, venous line.

The dialysis fluid pathway and the volumetric control of ultrafiltration. F1, filter; A, arterial line; V, venous line.


Unfractionated heparin (UFH) remains the standard anticoagulant. It is usually administered by initial intravenous bolus of 1000 to 2000 units and subsequent infusion at a similar hourly rate, stopping 30 min before the end of the session in patients with fistulas and grafts. Monitoring using whole-blood activated clotting times is possible, although many units reserve such measurements for problem situations. There is an increasing use of low-molecular-weight heparins (LMWH), which offer the advantage of single-bolus administration and appear to have a similar efficacy vs safety profile to UFH. However, measurement of anticoagulant effect is more difficult, requiring estimation of antifactor Xa inhibition, and—unlike with UFH—anticoagulation cannot be reversed with protamine. Heparin-free dialysis, employing regular saline flushes of the circuit, is the preferred strategy in patients at high risk of bleeding. In rare cases of heparin-induced thrombocytopenia (HIT) in which heparin-free dialysis is not possible, heparinoid, or argatroban, a direct thrombin inhibitor, are probably the agents of choice, depending on availability. Lepirudin, a recombinant hirudin, is an alternative, although it is exclusively renally excreted and so has a prolonged half-life in this setting. LMWH should not be used in this situation.

Haemodialysis/filtration techniques

Conventional haemodialysis

Conventional haemodialysis uses low-flux membranes in standard circuits, producing diffusive but little convective solute removal. Smaller molecules such as urea are cleared efficiently, but middle-molecule clearance is poor. Previously, the definition would also have included the use of acetate as buffer, but bicarbonate dialysis is now the norm.


Haemofiltration is a purely convective treatment, the inefficiency of which for small-molecule clearance limits its applicability in the intermittent treatments required in chronic renal failure, but in continuous mode the technique has become established as a treatment of acute renal failure in the critical care setting. Highly permeable membranes are used, permitting high-volume ultrafiltration of 20 to 50 litres per session. Substitution fluid is delivered from commercially prepared bags, either on the arterial side of the filter (predilution) or into the venous bubble-trap (postdilution). Middle-molecule clearance is excellent. There are a number of variants: continuous arteriovenous haemofiltration (CAVH), which required femoral artery cannulation, has given way to continuous venovenous haemofiltration (CVVH), which requires a pumped venous supply. Continuous venovenous haemodialysis (CVVHD) is a further variant. These continuous treatments are well tolerated haemodynamically and avoid the peaks and troughs in metabolic, electrolyte, acid–base, and volume control which are a feature of intermittent treatments. The inefficiency for small-molecule clearance may not be such a barrier for the use of haemofiltration in chronic renal failure if the current interest in more frequent (quotidian) dialysis continues (see below).

High-flux haemodialysis

Concerns about the biocompatibility of cuprophane, and its poor clearance of middle molecules, especially β‎2-microglobulin, have fuelled the increasing use of high-flux membranes. High-flux haemodialysis uses highly permeable, usually biocompatible membranes, which provide good diffusive clearance of small solutes combined with much better diffusive removal of middle molecules than conventional dialysis. This is augmented by a convective contribution to middle-molecule clearance resulting from a degree of obligate back-filtration within the dialyser.


Haemodiafiltration is the addition of a prescribed convective component to the technique of high-flux haemodialysis. Adding a greater convective component (haemofiltration) to the diffusive and convective clearances offered by high-flux haemodialysis allows the benefits of both modalities to be maximized. In each high-flux session, 12 to 25 litres of ultrafiltrate is removed and replaced by substitution fluid. The cheap online production by the dialysis machine of ultrapure substitution fluid from dialysis fluid has allowed haemodiafiltration to become established as a viable routine therapy of ESRF (Fig. Fluid is removed from the dialysis fluid pathway supplying the dialyser by a haemofiltration pump (HF), passed through an ultrafilter (F2), and infused in to the ‘V’ line (postdilution) or the ‘A’ line (predilution, not shown). To maintain balance in the fixed volume (volumetric) loop between the balance chamber and the dialyser, an equal volume of fluid is drawn from the blood across the dialysis membrane. The added convective component of haemodiafiltration produces a small improvement in small-solute clearance over that seen in high-flux dialysis, but a significant improvement in middle-molecule clearance (Fig. serum β‎2-microglobulin levels are reduced, and dialysis-related amyloidosis may be delayed; haemodynamic stability and survival may also be improved.

Fig. The dialysis fluid pathway for online haemodiafiltration. Fluid is removed from the dialysis fluid pathway supplying the dialyser by a haemofiltration pump (HF), passed through an ultrafilter (F2), and infused into the ‘V’ venous line (postdilution) or the ‘A’ arterial line (predilution, not shown).

The dialysis fluid pathway for online haemodiafiltration. Fluid is removed from the dialysis fluid pathway supplying the dialyser by a haemofiltration pump (HF), passed through an ultrafilter (F2), and infused into the ‘V’ venous line (postdilution) or the ‘A’ arterial line (predilution, not shown).

Fig. Comparison of solute removal by diffusion and convection according to molecular weight. Convection has better middle-molecule clearance but much poorer clearance of small-molecular-weight solutes than diffusion. Haemodiafiltration combines the strengths of both techniques to broaden the spectrum of solute removal.

Comparison of solute removal by diffusion and convection according to molecular weight. Convection has better middle-molecule clearance but much poorer clearance of small-molecular-weight solutes than diffusion. Haemodiafiltration combines the strengths of both techniques to broaden the spectrum of solute removal.

Adequacy of dialysis

The assessment of how much dialysis is required to maintain health is still controversial over 40  years after its introduction for endstage renal failure. We know that predialysis blood levels of urea and creatinine can be misleading as indicators of the adequacy of dialysis. Low levels are just as likely to indicate malnutrition and muscle wasting as good dialysis, and have been associated with increased mortality. Instead, dialysis dose is normally defined in relation to small-solute clearance, either by the urea reduction ratio (URR—the percentage reduction in blood urea during dialysis), or more commonly by urea kinetic modelling (UKM). UKM derives a normalized urea clearance, Kt/V, where K is the dialyser urea clearance, t is the duration of dialysis in minutes, and V is the urea distribution volume estimated as total body water from anthropomorphic data. This parameter, which derives from kinetic modelling of urea removal during dialysis (Fig., can be used as a measure of dialysis dose or as a tool to prescribe the dialysis.

Fig. Blood urea concentration during and after a haemodialysis session, showing an exponential decline during treatment with a postdialysis rebound due to two-pool effects.

Blood urea concentration during and after a haemodialysis session, showing an exponential decline during treatment with a postdialysis rebound due to two-pool effects.

Urea reduction ratio

The URR is the simplest measure of dialysis adequacy and is given by:(Equation

where C0 is the initial blood urea concentration and Cpost is the blood urea concentration in a blood sample taken immediately postdialysis. URR takes no account of urea generation, ultrafiltration, or residual renal function, but does correlate with outcome, testifying to its clinical utility. Most guidelines recommend a target URR of greater than 65%.

The urea kinetic modelling approach

Assuming urea is distributed in a single pool within the body and that the effects of urea generation and ultrafiltration during dialysis are small, the blood urea concentration (Ct) at any time (t) during the dialysis is given by:(Equation

Hence, the delivered dose of dialysis, Kt/V, can be calculated from the expressionEquation
The expression can be corrected for urea generation and ultrafiltration during dialysis according to a formula by Daugirdas:Equation
where R = Cpost/C0, T is the treatment time in hours, Δ‎W is the interdialytic weight gain, and W is the postdialysis weight.

The single-pool assumption (which is that all the urea in the body is in a pool that is equally accessible to removal by dialysis) is also an oversimplification. The rapid removal of urea (and other solutes) from the bloodstream during dialysis creates intercompartmental disequilibria: the intracellular concentration of urea exceeds the extracellular, and that in poorly perfused peripheral pools exceeds that in well-perfused body compartments. Urea exchange between these compartments continues after cessation of dialysis and causes a postdialysis rebound of blood urea concentration (Fig. This rebound can be substantial in high-efficiency treatments and can cause overestimation of Kt/V delivery by as much as 20%, making the single-pool assumption untenable in these circumstances. There are a number of ways of dealing with this problem: the most straightforward is to incorporate a postdialysis sample which is delayed until rebound is complete (30–60 min)—the equilibrated postdialysis sample (though in practice patients find this inconvenient). Much more complex is to model the system as two pools, requiring the assumption of a number of physiological parameters and iterative solution by computer. There are a number of less complex approximations to equilibrated Kt/V (eKt/V), which are usually preferred and have been shown to produce equivalent results (Equations and Equation

where sp Kt/V is the single pool Kt/V, and t is duration of dialysis session in minutes.

A further predialysis sample taken prior to the subsequent dialysis session allows an estimate of normalized protein catabolic rate (nPCR) to be calculated from the interdialytic rise in blood urea (see Box This can provide valuable information about nutritional status.

Defining target Kt/V

In the early 1980s, the National Cooperative Dialysis Study (NCDS), a randomized controlled trial, established a Kt/V of 0.9 per session as the minimum threshold dose for thrice-weekly dialysis, provided the patients were adequately nourished as defined by an nPCR greater than 0.8 g/kg body weight per day. Lower doses were associated with increased short-term morbidity and mortality. It was also inferred from the study (perhaps inappropriately) that the duration of the dialysis session did not influence outcome provided small-solute clearance was adequate. Subsequently, a wealth of observational data suggested that higher delivered Kt/V levels produced improved outcomes. This debate culminated in the HEMO study, another randomized controlled trial, which found no additional benefit of high compared with standard doses (mean delivered two-pool Kt/V 1.53 vs 1.16), except perhaps in women. Neither was there a benefit of high-flux over conventional haemodialysis, certainly during the first few years of treatment. The HEMO study has probably defined the adequacy limits of thrice-weekly treatment and in doing so highlighted some of the limitations of the Kt/V concept. The target recommended by K/DOQI and the United Kingdom Renal Association is an equilibrated value (eKt/V) of at least 1.2 for thrice-weekly treatment (approximating to a single-pool Kt/V of 1.4).

Incremental dialysis

This approach recognizes that the target total urea Kt/V has dialysis (Kdt/V) and residual renal (KRt/V) components. As residual renal function declines during the first few years of dialysis, the dialysis component is gradually increased to ensure the target continues to be achieved. This allows a gentler initiation to dialysis and maximizes the use of scarce resources, but does require regular estimates of residual renal function, and also assumes an equivalence of renal and dialyser clearance, which holds for urea but not necessarily for other solutes, or other renal functions. It is relevant that the viability of CAPD as a renal replacement modality also depends on this assumption, and in practice the dosing of CAPD and peritoneal dialysis techniques are similarly incremental.

Adequacy, duration, and frequency

The relationship between small-solute clearance, session duration, and session frequency is complex. The concentrations in the blood of small solutes decline in an exponential fashion during a dialysis session (Fig. Their rate of removal is thus maximal during the initial part of the session and increasing the duration of a dialysis session produces diminishing returns in terms of small-solute clearance, which is much greater if the same weekly duration (say 12 h) is configured as 6 × 2 h rather than 3 × 4 h. The standards discussed above for Kt/V are based on thrice-weekly dialysis. Comparing dialysis doses delivered by different frequency schedules and even different modalities is difficult and requires resort to other mathematical methods. The standard Kt/V (sKt/V) is such an approach, in which weekly doses delivered by any intermittent schedule can be compared after translation to an equivalent dose considered to have been delivered continuously over that week. Figure shows the relationship between Kt/V per session, session frequency, and weekly sKt/V, illustrating the major benefits of increased frequency.

Fig. Relationship between weekly standard Kt/V, sessional equilibrated Kt/V, and haemodialysis session frequency. CAPD, continuous ambulatory peritoneal dialysis. Used with the permission of the Nature publishing Group.

Relationship between weekly standard Kt/V, sessional equilibrated Kt/V, and haemodialysis session frequency. CAPD, continuous ambulatory peritoneal dialysis. Used with the permission of the Nature publishing Group.

Other approaches to adequacy

Small-solute clearance is just one measure of the adequacy of dialysis. Adequate dialysis, by this criterion, can be delivered in short sessions using high blood flows and large dialysers to augment clearance, but it is difficult to control other parameters such as phosphate and blood pressure, which also affect morbidity and mortality. In addition, larger molecules such as β‎2-microglobulin are certainly toxic in dialysed patients but do not figure in our currently accepted notions of adequacy. Broader concepts of adequacy are required which encompass these other factors impacting on the patient’s global wellbeing.

Quotidian dialysis

Quotidian dialysis refers to schedules in which dialysis is carried out every day, or at least six times weekly. The HEMO trial effectively defined the limits of standard thrice-weekly dialysis. The prediction is being borne out in practice that more frequent dialysis up to six times weekly, either overnight or short daily dialysis, not only enhances urea clearance to a degree which is not possible by extended thrice-weekly sessions but also provides excellent control of many other parameters. A number of uncontrolled studies have suggested substantial improvements in patient well-being. Recently, the Frequent Hemodialysis Network Trial demonstrated, in a randomized controlled setting, that short daily treatments–6 times weekly, as compared to conventional 3 times weekly sessions–was beneficial with respect to the composite outcomes of death, or change in left ventricular mass and death, or change in a physical-health composite score. In-centre haemodialysis was employed in both arms of the study, although the therapy may be more sustainable as a home treatment. These benefits, and a greater freedom from fluid and dietary restrictions, may more than compensate for the higher levels of commitment required. How many patients will choose (and be allowed by health care systems that need to restrain costs) to enhance their long-term health prospects by quotidian dialysis remains to be seen.

Management of haemodialysis patients

Predialysis care and initiation of dialysis

The predialysis period, as discussed in Chapter 21.6, is an important time for planning and preparation. Ideally, patients will be referred to the renal services in good time to allow this. Embarking on a career of renal replacement therapy is a major life event. Patients and their families need to have the relevant information presented in an assessible way. They need the time to consider this and to discuss their options with members of the multidisciplinary team. The facts should be presented as clearly and as honestly as possible, and the patients allowed to choose whether or not they wish to receive dialysis treatment. In some cases, there are medical imperatives, which preclude some or occasionally all replacement options. For some patients, especially those who are frail and dependent, the modality choice offered should include conservative (nondialytic) management, it being particularly important to emphasize that conservative management is not ‘no treatment’ (Box Patients opting for haemodialysis should be assessed early for fistula formation and the surgery planned 3 to 6 months before the putative initiation date. Vaccination against hepatitis B should also be carried out in this period: the earlier this occurs in the course of progressive renal failure, the greater the likelihood of a successful response. The prospects for pre-emptive (avoiding dialysis) transplantation should also be considered.

Unfortunately, there is still a very high incidence of late referral of patients with endstage renal disease to renal services, involving up to 30 to 40% of new patients in some areas, which precludes effective planning and increases morbidity and mortality.

The timing of initiation of dialysis is subjective. There is no absolute level of residual renal function that mandates initiation. The IDEAL (Initiating Dialysis Early And Late) study recently demonstrated, in a randomized controlled trial, that planned early initiation was not associated with an improvement in survival or clinical outcomes. These findings reinforce most current guidelines in recommending that the decision to start dialysis in those with CKD stage 5 should be individualized and based on careful consideration of many factors in discussion with the patient. In addition to the absolute level of kidney function, its rate of decline, and the patient’s symptoms and signs of renal failure, account should also be taken of the patient’s nutritional and functional status, and the physical, psychological, and social consequences of starting dialysis in that individual. Patients with comorbidities, such as diabetes and heart failure, tend to start at higher levels of estimated GFR, perhaps because of the earlier development of symptoms. Although there is no absolute threshold level of renal function below which dialysis is necessary, it is of note that most patients in the ‘late’ arm of the IDEAL study actually started dialysis at higher levels of Cockcroft–Gault creatinine clearance than the targeted 5 – 7 ml/min, because of the development of symptoms.

Prescribing dialysis

Target Kt/V is normally 1.2 to 1.3. Patients with intercurrent illness require a higher target. The value of K is obtained from data sheets from the dialyser manufacturer, taking into account membrane area and blood and dialyser flow rates. The value of V is obtained empirically from age, sex, weight, and height, usually by the Watson formula. The dose of dialysis prescribed can be adjusted to achieve the target Kt/V by changing the surface area of the membrane, blood flow rate, dialysis fluid flow rate (these influence urea clearance by the dialyser), and dialysis duration. This logic allows adequate dialysis to be delivered in a shorter time using larger dialysers and high flow rates (high-efficiency dialysis). If the target Kt/V includes a component for residual renal function, then residual urea clearance should be measured monthly and dialysis time readjusted accordingly. Membrane type and flux should be specified. Use of high-flux synthetic membranes is increasingly standard, rather than restricted to patients showing evidence of amyloid deposition. Setting the dry weight (see below) allows the ultrafiltration requirement to be specified for each dialysis. The prescription should also refer to the heparin loading dose and maintenance infusion rate. Whether dialysis prescription is individualized in this manner or whether a unit approach is taken with standardization of modality, dialyser, and session duration, it is important to have a means of monitoring the effectiveness of delivery of the prescribed dose.

Monitoring dialysis delivery

Dialysis delivery should be monitored monthly. Prescription and monitoring protocols are now often computer based. Pre- and postdialysis blood sampling allows calculation of the URR, and single-pool Kt/V and eKt/V can be calculated using Equation, and Failure to reach adequacy targets needs investigation. There are many possible causes of underdelivery, including poor blood flows, inadequate vascular access, recirculation within the vascular access, clotting problems within the dialyser, multiple alarms and poor compliance, which both reduce actual dialysis duration. A low nPCR should provoke an evaluation of nutritional intake. It is vital to ensure correct sampling technique, particularly for the postdialysis sample, to avoid serious potentially misleading artefacts.

Dry weight

Regulation of salt and water balance is one of the key functions of the kidney. Renal failure results in salt and water retention, which along with activation of the renin–angiotensin–aldosterone system, contributes to hypertension, left ventricular hypertrophy, and dilatation, which are potent causes of morbidity and mortality. The removal of the excess fluid accumulated between dialysis sessions is a vital function of dialysis, so we need ways of estimating the degree of this excess. Dry weight is an important concept dating back to the early days of maintenance haemodialysis. It assumes that body weight at any time consists of two components: the dry weight or target weight, at which the patient’s fluid compartments are normal in volume, and an excess weight consisting of surplus volume, which expands body fluid compartments and elevates blood pressure. The only way of defining dry weight is by trial and error. The protocol requires cessation of antihypertensive agents and weight reduction during successive dialyses during the first few weeks or months after initiation. The dry weight is the point at which the patient is oedema free and below which hypotension occurs on further fluid removal. The implicit assumption is that patients on dialysis have normal cardiovascular responses, which may have been reasonable in the highly selected dialysis population of 1970 but is much less tenable in the older, sicker patients on dialysis today. Applying such principles, most of the early patients on dialysis became normotensive without the need for antihypertensive agents. In most current patients, the target weight is likely to be the best achievable weight and hypertension is more likely. Shorter treatment times and less rigorous salt restriction have undoubtedly added to these difficulties. It is also vital to recognize that dry weight is not static: it changes over time, falling during periods of intercurrent illness and poor nutrition and increasing during recovery from such episodes, hence it requires regular clinical review. The potential adjunctive roles of methods such as natriuretic peptide measurement, ultrasonic inferior vena cava diameter estimation, blood volume monitoring, and bioimpedance spectroscopy in helping to define volume status in haemodialysis patients are still debated.


Many factors contribute to hypertension in patients on dialysis, including stimulation of the renin–angiotensin–aldosterone system and sympathetic overactivity, but the overriding factor is volume overload. In many units, 60% or more of patients receive antihypertensive agents. Insufficient emphasis on dietary sodium restriction and adequate ultrafiltration, coupled with a tendency for the early use of antihypertensive drugs, especially in patients with cardiac dysfunction, may render achievement of dry weight more difficult and further compromise blood pressure control. Hypertension is an important risk factor for cardiovascular disease mortality in the general population. However, unlike in the general population, many studies demonstrate a paradoxical relationship between blood pressure and mortality in haemodialysis populations—low to normal blood pressure being associated with poor outcome, and high pressure conferring a survival advantage (‘reverse epidemiology’). This has been attributed to an increased incidence of cardiac failure in patients with low-normal blood pressure. Alternatively, the deleterious effects of hypertension may be being swamped by the numerous other toxic factors operating in endstage renal failure. There may also be a problem with the definition of hypertension in this setting. Fluid status varies throughout the dialysis cycle: predialysis patients are generally fluid overloaded, hence predialysis blood pressure is a reflection of the cardiovascular response to fluid overload; likewise, postdialysis blood pressure reflects the cardiovascular response to fluid removal. There has been a general move for guidelines to become less prescriptive with respect to blood pressure targets in haemodialysis patients. For instance, current UK Renal Association guidelines confine themselves to suggesting that systolic blood pressure in the interdialytic period should not regularly exceed 160 mmHg, and that a predialysis systolic blood pressure regularly less than 120 mmHg should occasion concern. These reflect disquiet about the reliability of pre- and postdialysis blood pressure readings, about the dangers of overtreatment—in particular in relation to the phenomenon of intradialytic cardiac stunning—and generally about the sparsity of randomized interventional trials of the effect of blood pressure reduction in this population. There is some evidence that use of self-monitored home blood pressure readings can be of benefit to guide treatment, and greater use of this simple technique might help reduce both under- and overtreatment. Drug therapy is generally regarded as second-line treatment to be deployed after the achievement of optimal fluid status, but it is of note that recent meta-analyses have suggested a survival benefit for the use of antihypertensive medication in haemodialysis patients. This finding could relate to their cardioprotective actions as much as to their effect on blood pressure. No class of antihypertensive agent is contraindicated in dialysis patients, but their mode of excretion and dialysability should be considered, and care is often required with dosing schedules.

Vascular access

Securing and maintaining adequate, reliable, and robust vascular access is a major task, the importance of which cannot be overstated.

Temporary access

Beginning a career on maintenance haemodialysis with the need for temporary access should in most cases be regarded as a failure of adequate predialysis planning, although—as stated above—in a significant proportion of patients the initial presentation with endstage renal failure is as a uraemic emergency. If acute access is required, temporary, noncuffed, dual-lumen catheters can be inserted into femoral, internal jugular, or subclavian veins. The femoral route is simplest, safest, and preferred in very sick patients, but the infection risk is high if femoral catheters are left in situ for more than a few days. Use of the subclavian route risks stenosis of the vein and compromises future ipsilateral fistula formation.

Permanent access

Options for permanent access include the subcutaneous arteriovenous fistula (AVF), arteriovenous polytetrafluoroethene (PTFE) grafts, and cuffed, tunnelled, internal jugular venous catheters. AVF are preferred: they have the lowest complication rate and the longest life span, and are created by end-to-side or side-to-side anastamosis, preferably of the radial artery and cephalic vein (Brescia–Cimino fistula) in the nondominant forearm. Use of other sites such as brachial artery and cephalic vein is common. After anastamosis, venous distension and arterialisation occur, and needling is usually possible by about 6 weeks. Primary nonfunction is unfortunately common, and attention to the preservation of these vessels in predialysis patients is important. Distal ‘steal’ resulting in critical underperfusion of the forearm and hand may occur with large brachiocephalic fistulae. Stenosis and thrombosis may occur, which, in addition to predisposing to premature fistula loss, may lead to access recirculation (see below). Access monitoring may help to avoid such problems, facilitating early diagnosis and pre-emptive radiological or surgical intervention. Fistula flow assessment by ultrasound dilution is probably the monitoring mode of choice.

The many patients that still present late for dialysis require primary central venous access by default, and tunnelled catheters are also required when other access options have been exhausted. They carry a high risk of infection, are prone to clotting, and provide much lower blood flow rates than the AVF. Despite this they are in common usage: a recent United Kingdom national survey carried found that only 31% of patients started their haemodialysis career with a native AVF and that around one-third of hospital admissions in haemodialysis patients are due to vascular access problems. Limiting the use of lines to the 15% or so of haemodialysis patients with no other option is an important quality target, which would make a significant impact on haemodialysis morbidity and mortality.


If there is a stenosis in a fistula severe enough to limit fistula blood flow to a level less than that demanded by the blood pump in the extracorporeal circulation, then blood returning from the dialyser to the fistula can be drawn directly from the ‘V’ needle to the ‘A’ needle and dialysed again. This is known as access recirculation, an effect that can also be produced by misplacement of fistula needles with the ‘A’ needle downstream to the ‘V’ needle. Also during dialysis, a proportion of the blood returning through the ‘V’ needle will pass directly to the ‘A’ needle after passage through the heart and lungs without traversing a capillary bed to be ‘replenished’ with solute. This is known as cardiopulmonary recirculation and is an inevitable consequence of having a fistula as the access. The higher the blood pump speed, the greater the degree of recirculation in all of these circumstances. Access recirculation is a major cause of underdelivery of prescribed dialysis dose, and unexplained reductions of monitored Kt/V or urea reduction ratios demand further investigation to exclude this. Recirculation can be detected and quantified by techniques such as ultrasound dilution, thermal dilution, and ionic dialysance. Significant recirculation (>10%), requires further investigation, which may include Doppler ultrasonography or fistulography. Fistuloplasty or surgical reconstruction may be required if stenotic lesions are identified.

Diet and nutrition

Malnutrition is common and is often due to underdialysis. The early signs are subtle and often masked by fluid overload, which can compound the problem. There are no simple, foolproof laboratory tests. Hypoalbuminaemia indicates the presence of inflammation as much as it reflects malnutrition. Monitoring of nPCR may be helpful, but cannot replace regular dietetic review. When malnutrition is identified, a range of oral nutritional supplements may be deployed, and there is a limited role for intradialytic parenteral nutrition.

Protein requirements in patients on haemodialysis are not well characterized but have been estimated at 1.2 g/kg ideal body weight per day, which is considerably more than the nonuraemic requirement. Intakes greatly in excess of this may cause problems unless dialysis dose is correspondingly increased. There is no role for protein restriction. The energy requirement of a moderately active patient on haemodialysis is about 35 kcal/kg body weight per day, which is similar to that of normal subjects. Attempts should be made to limit interdialytic fluid gains to 1 to 2 litres, which is difficult when residual renal function has been lost. The value of limiting sodium intake (40–80 mmol/day) to control thirst is often understated. Potassium restriction (to about 60 mmol/day) is usually required when residual renal function is minimal. The recommended intake of elemental calcium is 1 to 1.5 g/day, but achieving this is seldom difficult given the extensive use of calcium salts as phosphate binders, although phosphate restriction to about 0.8 g/day of elemental phosphorus may reduce the requirement for these agents. There is no consensus on the need to supplement water-soluble vitamins (B and C), but the practice is widespread. Vitamin B12 supplements are recommended in high-flux treatments.


Low-density-lipoprotein (LDL) cholesterol levels are normal or near-normal in haemodialysis patients, but the overall lipid profile is highly atherogenic and characterized by marked accumulation of apo B-containing triglyceride-rich particles. There are elevated very-low-density lipoprotein (VLDL) and intermediate-density lipoprotein (IDL) levels, decreased high-density lipoprotein (HDL) levels, and a shift of LDL particle size toward a small dense apo-B-rich LDL predominance. In the general population, primary and secondary prevention trials have shown significantly improved cardiovascular outcomes with statins, but the same may not be true in the haemodialysis population. Two randomized trials (4D and AURORA) examined the effects of lowering LDL cholesterol with statin therapy in dialysis patients. In neither case was there a significant benefit on the primary outcome of cardiac/cardiovascular death, nonfatal MI, or stroke. The initial results from the SHARP study, however, suggest that LDL lowering with simvastatin plus ezetemibe was associated with a significant reduction in major atherosclerotic events, which was of a similar order in patients with moderate to severe CKD and those on dialysis. The effects on cardiovascular and total mortality were not significant. Whether this justifies routine LDL-lowering therapy in patients on dialysis is not yet clear.

Infection control

Strict adherence to universal precautions is necessary to minimize the risk of cross-infection by blood-borne viruses. Transmission from contaminated external surfaces, rather than through the dialyser membrane, is the major cross-infection threat. Screening of patients about to start dialysis for evidence of prior infection with hepatitis B and C and HIV is routine, and should be repeated at least 3-monthly thereafter. Patients negative for hepatitis B surface antigen should be vaccinated: patients who are positive should be segregated and use a dedicated machine, as should those positive for hepatitis C or HIV.

Other aspects of care

Cardiovascular risk factors such as smoking should be addressed, and exercise encouraged, both during dialysis sessions and in general. Low-dose aspirin may be beneficial. Folate and vitamin B supplements may reduce elevated homocysteine levels. There is a high incidence of sexual dysfunction, especially in males: some may benefit from androgen replacement, sildenafil may be effective if not contraindicated, and skilled counselling may be helpful.

Complications of haemodialysis

Acute complications


The severest forms of disequilibration occur shortly after dialysis initiation (dialysis disequilibrium syndrome). The main predisposing factors are late presentation with severe uraemia and aggressive dialysis initiation with lengthy dialyses and high solute clearance rates. Restlessness, headache, tremors, fits, and coma can result. Dialysis should not be initiated in this way. Cerebral oedema due to fluid shifts induced by intercompartmental differences in urea concentrations and paradoxical cerebrospinal fluid acidosis are among the suggested causes. Short initial treatments using dialysers of small surface area and low blood pump settings prevent the problem.

Postdialysis headache is a common symptom in patients undergoing regular haemodialysis and may be a minor manifestation of disequilibrium. Postdialysis fatigue is common in regular dialysis patients, lasting from a few minutes to many hours. It is reported that many of these symptoms are alleviated in quotidian dialysis.


Symptomatic hypotension occurs in up to 30% of dialysis sessions. Symptoms include nausea, vomiting, cramps, palpitations, dizziness, and syncope. The major cause is hypovolaemia, resulting from an imbalance between the rate of fluid removal from the circulation by ultrafiltration and the rate of vascular refilling from the interstitium. Underlying cardiovascular disease, the use of antihypertensive drugs, autonomic dysfunction, shortened dialysis times, excessive interdialytic fluid gains, and failure to reset dry weight after flesh weight gain all increase the likelihood. Accordingly, the mainstays of management are careful assessment and reassessment of target weight, limited use of antihypertensive agents, reduction of interdialytic weight gain by fluid and sodium restriction, and reduction of ultrafiltration rate. Newer dialysis machines have the capacity to monitor relative blood volume and to profile the sodium concentration of dialysis fluid throughout the dialysis session, techniques which may be useful in particular situations, but neither is yet used routinely. Convective therapies such as haemodiafiltration are associated with superior haemodynamic stability, possibly owing to blood cooling causing vasoconstriction. Episodes of hypotension may occasionally have more sinister causes such as primary myocardial events and heparin-induced bleeding. Cardiac arrhythmias are common, especially in patients with left ventricular hypertrophy and coronary artery disease, and are predisposed to by rapid intradialytic electrolyte fluxes, especially changes in serum potassium levels and in acid–base balance.

Dialyser reactions

Dialyser reactions are uncommon with the use of biocompatible membranes, and gamma- or steam-sterilised dialysers. Ethylene oxide, a device sterilant, has been implicated as a major cause of these reactions, along with other leachable materials. Type A reactions are anaphylactoid, rare, but life-threatening: symptoms usually begin in the first few minutes of dialysis but can be delayed for 20 min or more. Dialysis must be stopped, the lines clamped and discarded, and corticosteroids and antihistamines (adrenaline in very severe cases) administered. Type B reactions are less severe but more common: they occur 30 to 60 min after starting dialysis, which can be continued and the patient managed with supportive treatment. Reactions with reused dialysers are usually due to disinfectants, such as formaldehyde and peracetic acid, used in reprocessing. AN69 membranes may provoke anaphylactoid reactions, especially with concomitant angiotensin converting enzyme (ACE) inhibition, when bradykinin release by the negatively charged membrane and its reduced degradation due to ACE inhibition are responsible. A modified AN69 membrane is now available, coated with polyethyleneimine, which neutralises surface negative charge and decreases kinin generation.


Febrile reactions occur occasionally. Use of disposable dialysers and the increasing use of ultrapure water have dramatically reduced their incidence, and they are now much more likely to be due to infection related to the use of tunnelled lines for access than to contaminated dialysis fluid.

Coagulation problems

Circuit clotting may be due to underanticoagulation, and prolonged bleeding from the fistula after needle removal may be due to overanticoagulation. More serious bleeding problems with anticoagulation are rare. Thrombocytopenia is common in patients on heparin and usually mild and transient (HIT-I). True heparin-induced thrombocytopenia (HIT-II) is a rare (1–4%) but potentially life-threatening syndrome caused by platelet-activating antibodies to complexes of platelet factor 4 (PF4) and heparin. Characteristic features are thrombocytopenia, a systemic reaction within 30 min of intravenous UFH administration, and a hypercoagulable state with a high risk of thromboembolic complications. Severe thrombocytopenia and/or thrombosis in a patient on UFH should raise strong suspicions. The presence of antiheparin/PF4 antibody is confirmatory in these circumstances, in which case UFH and LMWH should be avoided. Danaparoid and argatroban are probably the best alternatives.

Other complications

Use of modern fail-safe dialysis machines and ultrapure water systems has fortunately rendered a number of previously well-described complications of haemodialysis exceedingly rare, including air embolism, severe hypercalcaemia due to dialysis against hard water, and acute haemolysis.

Chronic complications

Cardiovascular disease

Volume overload, hypertension, anaemia, hyperparathyroidism, excessive fistula flow rates, and uraemia itself all predispose to left ventricular hypertrophy, which is an independent risk factor for mortality. Correction of anaemia can favourably influence the natural history of left ventricular hypertrophy in patients on dialysis. Aortic pulse wave velocity predicts cardiovascular mortality, and is a major determinant of systolic hypertension. Haemodialysis patients also have a variety of other traditional risk factors, such as lipid abnormalities, and nontraditional risk factors, such as disturbances of calcium and phosphate metabolism, hyperhomocysteinaemia, increased oxidative stress, and elevated inflammatory markers.


Erythropoietin deficiency is the major cause of the anaemia of chronic kidney disease, but a number of additional causes of anaemia may arise from the haemodialysis process itself. Iron deficiency can result from the repeated loss of small amounts of blood in extracorporeal circuits. Intravenous iron preparations such as iron saccharate can be used in moderate doses (e.g. 100 mg on each of 10 successive dialyses) to correct iron deficiency, and as maintenance treatment (such as 50 mg weekly). Deficiencies of other haematinics can occur, particularly in high-flux treatments, when regular supplementation with vitamin B12 and folate is recommended. Mechanical and chloramine-induced haemolysis should not occur with modern techniques. There are other causes of erythropoietin resistance, the most potent being infection, in this context often arising from central venous lines. Some studies suggest that high-flux haemodialysis and haemodiafiltration may have beneficial effects on anaemia management, possibly related to removal of middle-molecule inhibitors of erythropoiesis. See Chapter 21.6 for more detailed discussion of the management of renal anaemia.

Bone disease

Haemodialysis patients have bone disease associated with chronic renal impairment, but may also develop skeletal complications of their treatment. Accumulation of β‎2-microglobulin may cause dialysis-related amyloidosis which may be associated with significant bone problems, including bone cysts and spondyloarthropathy (see below). Control of serum phosphate levels is crucial in controlling secondary hyperparathyroidism and preventing metastatic calcification: this requires a three-pronged approach of dietary restriction, phosphate binders, and adequate dialysis, the importance of the latter being highlighted by the superb control gained, in the absence of phosphate restriction and binder usage, in patients on daily nocturnal dialysis. The setting of the dialysis fluid calcium concentration can significantly affect the spectrum of bone disease, high levels (1.75 mmol/litre) can suppress parathyroid hormone levels but risk adynamic bone disease and metastatic calcification; by contrast, levels of 1.25 mmol/litre or lower may be useful treatments of adynamic bone and relative hypoparathyoidism. Dialysis fluid magnesium concentrations may also impact on bone metabolism, but the clinical relevance of this is not established. The toxic effects of acid accumulation on the skeleton and other systems can be avoided by adequate dialysis and adjustment of the dialysis fluid bicarbonate concentration. The incidence of osteoporosis is increasing in the haemodialysis population, related to increasing age, low levels of physical activity, previous steroid therapy, and relative hypogonadism. The effects of heparin on bone are well established, but evidence for a specific role of heparin in the development of osteoporosis in this setting is sparse. Aluminium-related bone disease, along with other manifestations of aluminium accumulation, including progressive dementia and anaemia, occurred in ‘epidemic’ form due to aluminium contamination of dialysis water. Use of modern water purification methods has seen the disappearance of this problem and new cases should no longer be seen. See Chapter 21.6 for more detailed discussion of the management of renal mineral and bone disorders.


Dialysis-related amyloidosis is a serious complication of chronic dialysis. Its incidence increases with duration of haemodialysis and symptomatic involvement is almost universal after 15 years. Older patients are more susceptible. The syndrome manifests mainly as carpal tunnel syndrome and destructive arthropathy associated with bone cysts and a destructive spondyloarthropathy, but other organs can be involved. Deposits of amyloid, mainly composed of β‎2-microglobulin fibrils, can be found at these and other sites. β‎2-Microglobulin is an 11 800-Da protein that is part of the human class 1 major histocompatibility complex. It is 95% eliminated by glomerular filtration, hence levels are elevated in renal failure, and these are not cleared by low-flux membranes. Elevated plasma levels are the major predisposing factor to amyloid deposition, but other factors may also be important, including modification of β‎2-microglobulin by advanced glycation endproducts and by oxidative and carbonyl stress, and it is also possible that β‎2-microglobulin or modified β‎2-microglobulin is directly toxic to tissues. Use of high-flux synthetic membranes, especially in haemodiafiltration mode, reduces plasma levels of β‎2-microglobulin, although the levels remain about 10-fold higher than in those with normal renal function. High-flux dialysis and especially haemodiafiltration may prevent or delay the onset of symptomatic disease, and the use of ultrapure water may also be protective. Treatment options are limited for established disease, but renal transplantation may enable slow resorption of deposits.

Outcomes of haemodialysis

Dialysis undoubtedly prolongs the life of patients with endstage chronic renal failure, but survival remains markedly inferior to that of age-matched peers with normal renal function. Cardiovascular disease is the main cause of death, followed by infection. Comparison of outcome in the different eras of dialysis is fraught with problems, largely because of the dramatic differences in case mix of patients entering programmes. Age, comorbidity, and functional status are independent predictors of morbidity (rate of admission to hospital) and mortality. Late presentation for dialysis has a profound effect on survival; and we have previously alluded to the effects of dialysis adequacy and nutrition on outcome. It is difficult to compare the outcome of patients treated with haemodialysis and peritoneal dialysis in any meaningful way. Data from single centres, multicentre studies, and analysis of registry data do not show consistent differences in survival between these modalities. There are a number of confounding factors. Patients initiated on CAPD are younger, have fewer coexisting nonrenal comorbidities, better functional status, and are less likely to have presented late. In addition, technique survival is poor in CAPD, and many patients require transfer to haemodialysis because of peritonitis, inadequate dialysis, or ultrafiltration failure. Quality-of-life assessments are similar in both groups, but both are inferior to those obtained in patients with successful transplants. It is probably safe to conclude that, in the early years of therapy at least, morbidity and mortality are similar on both modalities if risk-stratified groups are compared. There are few data to allow comparison of the outcome of conventional haemodialysis and more modern haemodialysis modes. The HEMO study demonstrated no overall survival benefit of high-flux over low-flux treatments, but there was a suggestion that patients whose dialysis vintage was greater than the mean 3.7 years fared better on high-flux treatment, and β‎2-microglobulin correlated with survival. Similarly, the MPO study, while finding no overall survival benefit of high-flux over low-flux membranes, did demonstrate a benefit in high-risk patients with low serum albumin levels. Observational data suggest improved survival with haemodiafiltation, although there are no controlled data. There is some evidence that high-flux modes and haemodiafiltration protect against the development of dialysis-associated amyloidosis. As alluded to earlier, short daily haemodialysis has been shown to impact on morbidity and mortality.

The future of haemodialysis

Haemodialysis is likely to remain centre based for the most patients. Technical advances will allow treatments to become more tailored to the specific requirements of the individual. Online dialysis quantification could guarantee the adequacy of each session. Online blood volume monitoring, possibly combined with bioimpedance spectroscopic monitoring of the extracellular fluid volume, coupled with algorithms to control ultrafiltration rate, dialysis fluid temperature, and sodium content on a minute-to-minute basis, could prevent intradialytic hypotension and allow patients to finish dialysis at their optimal achievable weight. Current concepts of the relationship between dialysis dose and sessional frequency, and the encouraging initial results with daily dialysis, suggest that the treatment may emerge as a self-supervised home modality, perhaps for a small proportion of younger, less dependent patients for whom transplantation or retransplantation is not an option. Vascular access is likely to remain the Achilles heel.

Further reading

Agarwal R, et al. (2009). Home blood pressure measurements for managing hypertension in hemodialysis patients. Am J Nephrol, 30, 126—34Find this resource:

Arieff AI (1994). Dialysis disequilibrium syndrome: current concepts on pathogenesis and prevention. Kidney Int, 45, 629–35.Find this resource:

Bowry SK (2002). Dialysis membranes today. Int J Artif Organs, 25, 447–60.Find this resource:

Canaud B, et al. (2006). Mortality risk for patients receiving hemodiafiltration versus hemodialysis: European results from the DOPPS. Kidney Int, 69, 2087–93.Find this resource:

Chandna SM, et al. (1999). Is there a rationale for rationing chronic dialysis? A hospital based cohort study of factors affecting survival and morbidity. BMJ, 318, 217–23.Find this resource:

Cooper BA, et al. (2010). IDEAL Study. A randomized, controlled trial of early versus late initiation of dialysis. N Engl J Med, 363, 609–19.Find this resource:

Daugirdas JT (1993). Second generation logarithmic estimates of single-pool variable volume Kt/V: an analysis of error. J Am Soc Nephrol, 4, 1205–13.Find this resource:

Daugirdas JT, Schneditz D (1995). Overestimation of hemodialysis dose depends on dialysis efficiency by regional blood flow but not by conventional two pool urea kinetic analysis. ASAIO J, 41, M719–24.Find this resource:

Davenport A (2006). Intradialytic complications during hemodialysis. Hemodial Int, 10, 162–7.Find this resource:

Dember LM, Jaber BL (2006). Dialysis-related amyloidosis: late finding or hidden epidemic? Semin Dial, 19, 105–9.Find this resource:

Department of Health (2002). Good practice guidelines for renal dialysis/transplantation units: prevention and control of blood-borne virus infection.

Eknoyan G, et al. (2002). Effect of dialysis dose and membrane flux in maintenance hemodialysis. N Engl J Med, 347, 2010–19.Find this resource:

Feest TG, et al. (2005). Trends in adult renal replacement therapy in the UK: 1982–2002. QJM, 98, 21–8.Find this resource:

FHN Trial Group, Chertow GM, Levin NW, et al. (2010). In-center hemodialysis six times per week versus three times per week. N Engl J Med, 363, 2287–300.Find this resource:

Fluck R (2006). Vascular access survey.

Gotch FA (2001). Evolution of the single-pool urea kinetic model. Semin Dial, 14, 252–6.Find this resource:

Gotch FA, Sargent JA (1985). A mechanistic analysis of the National Cooperative Dialysis Study (NCDS). Kidney Int, 28, 526–34.Find this resource:

Guerin AP, et al. (2006). Cardiovascular disease in the dialysis population: prognostic significance of arterial disorders. Curr Opin Nephrol Hypertens, 15, 105–10.Find this resource:

Heerspink HJ, et al. (2009). Effect of lowering blood pressure on cardiovascular events and mortality in patients on dialysis: a systematic review and meta-analysis of randomised controlled trials. Lancet, 373, 1009–15Find this resource:

Hoenich NA, Ronco C (2007). Haemodialysis fluid: composition and clinical importance. Blood Purif, 25, 62–8.Find this resource:

Jaeger JQ, Mehta RL (1999). Assessment of dry weight in hemodialysis: an overview. J Am Soc Nephrol, 10, 392–403.Find this resource:

Locatelli F, et al. (2009). Membrane Permeability Outcome (MPO) Study Group. Effect of membrane permeability on survival of hemodialysis patients. J Am Soc Nephrol, 20, 645–54.Find this resource:

Mitra S, Chandna SM, Farrington K (1999). What is hypertension in chronic haemodialysis? The role of interdialytic blood pressure monitoring. Nephrol Dial Transplant, 14, 2915–21.Find this resource:

National Kidney Foundation K/DOQI (2006). Clinical practice guidelines for hemodialysis adequacy.

Owen WF Jr, et al. (1993). The urea reduction ratio and serum albumin concentration as predictors of mortality in patients undergoing hemodialysis. N Engl J Med, 329, 1001–6.Find this resource:

Pickering TG (2006). Target blood pressure in patients with end-stage renal disease: evidence-based medicine or the emperor’s new clothes? J Clin Hypertens (Greenwich), 8, 369–75.Find this resource:

Pierratos A, et al. (2006). Daily hemodialysis 2006. State of the art. Minerva Urol Nefrol, 58, 99–115.Find this resource:

Smith C, et al. (2003). Choosing not to dialyse: evaluation of planned non-dialytic management in a cohort of patients with end-stage renal failure. Nephron Clin Pract, 95, c40–6.Find this resource:

Sonawane S, Kasbekar N, Berns JS (2006). The safety of heparins in end-stage renal disease. Semin Dial, 19, 305–10.Find this resource:

Tattersall JE, et al. (1996). The post-hemodialysis rebound: predicting and quantifying its effect on Kt/V. Kidney Int, 50, 2094–102.Find this resource:

UK Renal Asociation (2006). Clinical practice guidelines.

Vilar E, Fry AC, Wellsted D, Tattersall JE, Greenwood RN, Farrington K (2009). Long-term outcomes in online hemodiafiltration and high-flux hemodialysis: a comparative analysis. Clin J Am Soc Nephrol, 4, 1944–53.Find this resource:

Wanner C, et al. (2005). Atorvastatin in patients with type 2 diabetes mellitus undergoing hemodialysis. N Engl J Med, 353, 238–48.Find this resource: