Last updated: Lesson of the Month - August 2014…
on 04 Aug 2014

Haemodialysis (HD)


Adam Kirk and James Tattersall

Haemodialysis is a method of removing excess fluid, salt and wastes from the blood, effectively replacing the excretion functions of failed kidneys.

Haemodialysis is used in hospitalised patients, particularly during critical illness causing acute kidney injury (AKI). In this case, the treatment may be delivered continuously while the patient is in bed. Haemodialysis is also used in otherwise healthy patients with End Stage Renal Disease (ESRD), who are living relatively normal lives. In this case, the treatment is delivered over a total of about 12-24 hours in 3-7 sessions per week (usually three sessions of four hours each per week).

These sessions are delivered while the patient is sitting in a chair, or overnight while the patient is asleep. The procedure is not painful or uncomfortable and does not require an anaesthetic. Haemodialysis requires trained operators, specialised equipment and supplies. It can be performed in a patient’s home. Transportable equipment is available.

The haemodialysis blood circuit

A dialysis machine pumps blood from the patient, through disposable tubing, through a dialyser, or artificial kidney, and back into the patient. Waste solute, salt and excess fluid is removed from the blood as it passes through the dialyser.

The dialysis machine also pumps a special fluid, the dialysis fluid, through a separate compartment in the dialyser. The blood and dialysis fluid are separated by a thin membrane, so they do not mix. Wastes pass through the membrane from blood into the dialysis fluid. Certain salts, required for health, may pass in the opposite direction, from dialysis fluid to blood. The ‘used’ dialysis fluid, carrying the wastes eventually flows into the drain.





The dialysis machine is controlled by an integral computer. In addition to pumping the blood, it prepares the dialysis fluid, monitors the system to ensure that the dialysate is continuously at the correct pressure, temperature and composition; so that blood flows freely at the correct pressures and that no air has entered the blood. The dialysis machine also controls and monitors the removal of fluid by filtration (actually ‘ultrafiltration’ as explained later).

Haemodialysis requires up to 100 litres of dialysis fluid per treatment session (or up to 50 litres per day for continuous treatments). The dialysis fluid is sometimes provided in pre-prepared sterile bags. This allow the machine to be more compact (e.g. for portable systems or for bedside use in intensive care units). More commonly, the dialysis machine generates the dialysis fluid as required during treatment from connected supplies of purified water and concentrated salt and sugar solution.

To prevent transfer of blood-borne viruses between patients, and to simplify cleaning, the entire blood pathway (consisting of blood tubing, dialyser and any needles) is sterile, discarded after a single treatment. The dialysis machine itself can be used for multiple patients if cleaned between patients. The mechanical parts of the blood pump do not contact directly with the blood; they propel the blood along the tubing by squeezing the tube from outside using rollers. Similarly, the sensors which measure pressure in the blood at various points along the blood pathway are separated from the blood by multiple membranes; and, in some cases, an air gap, to prevent direct contact between blood and machine.

Objectives of Haemodialysis

Haemodialysis is the default treatment for patients with ESRD. Short-term objectives are to:

  • Correct electrolyte balance
  • Correct metabolic acidosis
  • Correct fluid state
  • Remove toxins

Longer-term objectives are to:

  • Optimise the patients functional status
  • Control BP
  • Prevent uraemia and its complications
  • Improve survival


Key point: currently, over 2 million people worldwide receive haemodialysis. As in the rest of the world, the number of patients receiving haemodialysis in the UK is increasing. According to the UK Renal Registry, in the UK, at the end of 2009, there were 49, 080 patients receiving renal replacement therapy (RRT) – 48% of these were renal transplant patients, and 44% were haemodialysis (HD) patients (23% Hospital HD, 20% Satellite HD, 1% Home HD), and 8% on peritoneal dialysis (Steenkamp, 2010).

The median age of prevalent patients was 57.7 years (HD 65.9 years, PD 61.2 years and transplant 50.8 years).

Key point: although transplantation is preferred, haemodialysis will continue to be the commonest form of non-transplant renal replacement therapy. Although it is a successful life-saving and life-sustaining therapy, the technique only partially replaces one aspect of renal function, ie water and solute excretion; and provides approximately 5% GFR. Consequently life expectancy is still significantly reduced.

Key point: mortality rates are high, but prolonged survival on dialysis is possible.



Principles and Prescription

In essence, dialysis involves the movement of solutes and water across a semi-permeable membrane. There are two principle forces involved in the process – diffusion and convection.


This is the random movement of particles driven by their thermal kinetic energy. This movement tends to disperse solute moleciles from regions of higher concentration to regions of lower concentration. The kinetic energy is temperature. The instantanous velocity is therefore proportional to temperature and inversely proportional to the square root of the molecule's mass.

The instantaneous velocity of smaller solute molecules is very high (several hundred km/hour), but random changes of direction reduces overall progress after travelling a short distance. The time taken for the transfer of solute mass by diffusion is proportional to distance squared. For this reason, diffusion is only an effective transport mechanism over very small distances. In haemodialysis, diffusion occurs accross the dialyser mebrane, which must be very thin. Solute must be carried to the membrane surface by blood flow, and from the other side of the membrane by dialysate flow.

The rate at which solute mass passes across a dialysis membrane by diffusion (assuming very high blood and dialsysate flow rates) is proportional to:

  • Absolute Temperature
  • Concentration difference between blood and dialysis fluid
  • Total surface area of the pores accessible to the molecule

And inversely proportional to:

  • Membrane thickness squared
  • Radius of the molecule
  • Viscosity of the blood water

Lager solutes such as albumin cannot access the pores and do not pass through. Some membranes, being negatively charged, repel anions such as phosphate, limiting clearance. Solute molecules may interact with water, increasing their effective radius. Phosphate diffusion is reduced for this reason as well.

In a modern dialyser, almost all of the urea diffuses from the blood water as it flows through the dialyser. Therefore the urea concentration at the blood outlet is very low, typically less than 10% for its level at the inlet. Unlike other solutes, urea is also cleared from the erythrocytes, due to the presence of specific urea channels in the cell membrane.

Toxins other than urea are cleared to a lesser extent, as solute within the erythrocyte water is inacessible to the dialysis process. Larger toxins such as β2-microglobulin diffuse more slowly and may be partly excluded from the pores, due to their greater radius. The concentration of β2-microglobulin in the plasma water at the dialyser outlet is typically around 50% of that at the inlet.

Diffusion also transfers solute in the opposite direction across the dialyser mebrane from dialysis fluid to blood. Renal replacement therapy includes the infusion of bicarbonate, by this route. The composition of  the dialysis fluid must be controlled to ensure the appropriate amount of solute transfer.


This refers to the movement of solutes, carried by fluid flow. It is the mechanism by which solute is carried around the body by the circulation. In the dialysis system, solute is carried to the dialyser membrane by convection due to the extracorporeal blood flow. The extracorporeeal blood flow and the dialysis fluid flow has the greatest influence on clearance by dialysis.

Unlike diffusion, convection is effective over long distances and is independent of temperature, viscosity and molecular radius. Convection is the main mechanism for solute transport in the human body, including the glomerulus of the kidney.

The clearance of solute molecules can be enhanced by convection across the dialyser membrane. This is achieved by creating fluid flow through the pores from blood to dialysis fluid in the process of ultrafiltration. Convection across the membrane is particularly effective at enhancing the clearance of larger solutes, which diffuse slowly.

Convection will remove all solute molecules at the same rate regardless of their radius as long as they are small enough to access the pores. The ability of solute to access the pores is quantified as the sieving coefficient or the ratio of concentration in the ultrafiltrate to the plasma water. A sieving coefficient of 1 means that the pores are fully accessed and the solute passes through the membrane unimpeded.

Clearance by convection across the membrane is limited by the presence of blood components such as cells and plasma proteins which cannot access the pores. Ultrafiltration concentrates these components, increasing viscosity. In practice, this limits the ultrafiltration rate to 20-30% of plasma flow. Higher ultrafiltration rates are possible by diluting the blood before it enters the dialyser (pre-dilution). However this reduces efficiency by reducing solute concentration in the dialyser and cancels out the effect of the increased ultrafiltration rate.

Convection is not used on its own in intermittent dialysis due to the limited clearance that it can achieve. However, convection is routinely used to augment the clearance of larger toxins in haemodiafiltration (HDF), dialysis with combined ultrafiltration of 20-30% of blood flow.

Haemodiafiltration has been shown to achieve a lower mortality rate, compared to dialysis.

Haemodialysis Machine

The two principle components of a haemodialysis machine system are the dialyser (artificial kidney) and the extra-corporeal system. The 'dialyser’ is a series of semi-permeable membranes, arranged to form paths for blood to pass next to dialysis fluid on opposite sides of the membrane, flowing in opposite directions.

The 'extracorporeal system’ refers to blood being drawn from a needle (‘A’-needle) by a pump, passing through the dialyser and returning to the patient through another needle (‘V’-needle).

The system has a series of fail-safe mechanisms to prevent a number of possible complications. There is an arterial pressure monitor to protect the fistula from excess negative pressure. There is a bubble trap to prevent air embolus and a venous pressure monitor to detect and prevent blood loss.

The flow rates of blood and dialysate fluid, plus composition of dialysate and length of dialysis are individualised for each patient. The dialysis fluid is made up of specific (individualised) concentrations of electrolytes with treated water. Meticulous preparation of water for dialysis is essential as contamination of water with microorganisms and chemicals is dangerous.

Types of Haemodialysis

Modern dialysis now uses high-flux haemodialysis – the basic principles are the same as those used in conventional HD. But (as the name suggests) the membranes are highly permeable; and therefore able to combine diffusion and convective clearance to give balanced clearance of small and middle molecules.

Haemofiltration – compared with conventional HD, this technique relies entirely on convection. The membranes used are highly permeable (high flux), allowing large volume ultrafiltration. In contrast to conventional HD, middle molecule clearance is excellent, but small molecule clearance is poor; and it is for this reason that the technique is not appropriate for long-term treatment for ESRD. It is used very effectively in the acute care setting for patients with acute kidney injury (AKI).

Haemodiafiltration – this as a combination of all of the above. It is essentially haemofiltration in addition to the technique of high-flux HD. The extra convective element of haemofiltration improves middle molecule and small solute clearance and is a viable alternative for long term therapy in ESRD.

Prescription of Haemodialysis

Urea clearance rate (K)
This depends on four main prescription variables:

  • Duration and Frequency: these are the most important variables to achieve adequacy. The time likely to be required needs to be explained to patients before they choose haemodialysis. Increasing dialysis time at a later date is hard
  • Blood flow rate (Qb): 200-500 ml/min
  • Dialyser surface area
  • Dialysate flow rate (Qd): 500-800ml/min possible (NB the difference this makes is not quantified and likely to be small, see below)

There has been an increasing trend to prescribe higher blood flow rates (>400 ml/min) and use dialysers with a higher surface area (more porous) to provide higher efficient dialysis. There is not good controlled evidence to support this change.

The interdialytic fluid (weight) gain is removed under volumetric control by the HD machine which adjusts the transmembrane pressure to achieve the prescribed UF rate.

Factors Affecting Urea Clearance

Blood and dialysate flow rate
The most important rate-limiting factor for urea clearance is blood flow rate, which is largely determined by the vascular access. Increasing the dialyser flow rate has relatively little effect on clearance; eg an increase in dialysate from 500 to 800 ml/min (60%) only produces a 5-10% increase in urea clearance.

Duration of dialysis
This has some effect on urea clearance. But >75% of urea clearance occurs within the first two hours, as shown by this graph of KT/V (equivalent to urea clearance) and time on dialysis:

This observation may explain the main conclusion of National Co-operative Dialysis Study (NCDS) (Gotch, 1985). In this study patients were randomised to receive short (2.5-3.5h) or long (4.5-5h) dialysis times, and two levels of time averaged urea concentrations. Longer dialysis gave a better but statistically insignificant outcome. More recently, the HEMO Study (Eknoyan, 2002), was carried out. This is the only large RCT that has studied the effect of dialysis dose on outcome. In the study, a larger dialysis dose did not affect survival.

Frequency of dialysis
This graph also suggests that the frequency of weekly solute removal may be more important than duration of dialysis. For this reasons, the FHN (Frequent Haemodialysis Network) Trial (2010) was carried out. Unfortunately, although some end-points were better in the group that received HD 6x per week compared to the thrice-weekly group, there was no effect on all-cause mortality. Unlike the HEMO Study, it was not sufficiently powered to demonstrate such an effect.

A thrice-weekly schedule evolved from a belief that patients would need adequate 'recovery time' between dialysis sessions, to tolerate the procedure longterm.

Membrane permeability
Membrane permeability affects urea clearance. But Locatelli (2009), in the Membrane Permeability Outcome (MPO) Study, was unable to detect a significant survival benefit with either high-flux or low-flux membranes in the population overall. But the use of high-flux membranes conferred a significant survival benefit among patients with a low serum albumin or diabetes.

Urea Clearance by Different Methods

For very crude comparison of small molecule clearance by continuous versus intermittent treatments, the following figures are provided. HD figures are for urea clearance, ignoring UF

Modality  Urea Clearance
Normal GFR 150 L/day
Daily intermittent HF 15-25 L/day
Continuous HF at 1L/hr 24 L/day
Continuous HF at 2L/hr 48 L/day
Daily HD (4hr), QB = 200 ml/min 46 L/day

When Prescribed vs Delivered Dialysis Differ

If the prescribed vs delivered dialysis is very different, consider: access recirculation, interrupted or shortened dialysis, slowed pump speeds, clotting dialysers, delayed re-equilibration of urea (eg in shock or cardiac failure), and errors in assumptions about V, which will often tend to reduce actual dialysis dose.









Vascular Access

Key point: maintenance of good vascular access is essential for effective management of HD patients

There are three main types of medium-longterm vascular access:

  • Primary arteriovenous (AV) fistula (long-term and most effective)
  • Central venous catheter. There are two types: temporary (short-term); or dual lumen cuffed (tunnelled) catheters (medium-term)
  • Polytetrafluoroethylene (PTFE) grafts (long-term) – detail is not included here, but the workup is similar to AV fistulae. The major advantage is that they can be used sooner, but long term results are not as good as for fistulae

Arteriovenous Fistula

A fistula is created by connecting an artery and a vein, creating a surgical shunt from the artery to the vein. They are most commonly created at the wrist and the antecubital fossa. The increase in pressure on the venous side causes thickening of the vein wall of the fistula, and enlargement of the lumen – this is called arterialisation.

The surgical procedure anastomoses the radial, brachial, or femoral artery to an adjacent vein in an end-of-the-vein to the side-of-the-artery fashion. When the adjacent vein is not suitable for access creation, a piece of prosthetic graft is used. For patients who have poor veins, an autogenous saphenous vein graft is also an option.

AV fistulae are the preferred method for vascular access in HD patients because of better survival (60-90% functioning at 3 years) and a lower risk of infection, compared to tunnelled lines and arteriovenous PTFE grafts (AVGs). A good fistula can also tolerate higher blood flow rates, giving efficient dialysis. Unfortunately not all patients have suitable veins, and therefore require one of the other forms of vascular access.

When and where to create a fistula

Key point: a newly created fistula may take 6 to 8 weeks to mature and be useable, so in patients with CKD, the fistula should be created early, when GFR is between 25 and 30 mL/min (ie in the early stage of CKD4)

Preparations should be made for renal replacement therapy (RRT) when the patient is in early CKD stage 4 (GFR 25-30 mL/min), with patient education about all forms of RRT, including haemodialysis. In patients who are in CKD Stage 4, it should be made clear to patients (and clinicians) that forearm veins should be avoided for IV cannulation and venepuncture.

If the patient, guided by the nephrologist and multi-disciplinary team, chooses haemodialysis, vein mapping (using Doppler ultrasound or occasionally a venogram) should be carried out, and a fistula created. As fistulae (if they work) take 6-8 weeks to mature, the operation should occur at least 6 months before initiation of HD. This time gap allows for the 10-25% failure rate of the procedure.

In the UK, the procedure is usually done by a surgeon (although a few physicians do them) and is often a day case procedure, which can be done under local or general anaesthesia. The procedure takes between 1 and 2 hours.

Pre-op assessment

Patients need pre-op assessment for fitness for surgery as for any other surgical procedure, and it is important to find out whether the patient has any history of:

  • Multiple previous accesses in an extremity planned as an access site
  • Previous subclavian catheter placement in venous drainage of planned access
  • Previous arm, chest or neck trauma or surgery
  • Current or previous transvenous pacemaker in venous drainage of planned access

Additionally on examination, it is important to check for:

  • Oedema
  • Collateral veins
  • Any differential extremity size

Dual Lumen HD Catheters

The primary disadvantages of central vein catheters are: a relatively narrow calibre that does not allow for blood flow high enough to achieve optimal clearance; and a high risk of catheter site infection and thrombosis. There are 2 types:

  • Temporary lines, which are dual lumen, non-cuffed and non-tunnelled - these are inserted as a bridge to more permanent access, or where long term dialysis is not anticipated. If vascular access is likely to be required for more than three weeks, it it preferable to use:
  • ‘Permanent’ lines, which are dual lumen, cuffed and tunnelled - the cuff of the catheter is inflated in the subcutaneous tissues and (as a consequence of tissue in-growth) secures the catheter in place. Compared to temporary lines, the have higher blood flow rates and a reduced incidence of infection

HD catheters are placed in either the internal jugular vein, or the femoral vein. They are NOT placed in the subclavian vein as there is a high incidence of venous stenoses at this site.

The 2 main complications of HD catheters are:

  • Infection - all lines are at risk from this, and it is a serious problem, carrying significant morbidity and mortality. The exit site, the tunnelled tract or the blood stream can become infected and common organisms are Staphylococcus aureus and Staphylococcus epidermidis. Femoral lines may also lead to gram negative bacterial infections
  • Thrombosis - this may lead to complete occlusion, or partial occlusion and poor flows that do not permit adequate dialysis (catheter dysfunction - defined as extracorporeal blood flow of less than 300ml/min)

Temporary Line Infection

If this is suspected:

  • Take blood cultures from a peripheral vein
  • Ensure there are no other sources of infection (history, examination, FBC, U+E, CRP and CXR)
  • If the patient has signs and symptoms of severe sepsis, then the line should be removed immediately, antibiotics should be given and the tip should be sent for culture
  • Antibiotics should be given empirically according to local guidelines after cultures have been sent
  • If sepsis resolves, wait as long as possible before placing a fresh line
  • Exclude endocarditis and osteomyelitis/discitis in patients with persistent fevers, raised CRP, or clinical suspicion

Permanent Line Infection

The issues here are more complicated than in temporary lines. In the first instance, it is important to treat the same way as in temporary lines (look for other sources, blood cultures etc), but also take blood cultures through the catheter. If no other sources of infection can be found, then treat with IV antibiotics, once cultures have been sent. Exit site infections should be treated with topical antibiotics.

The important decision is when to take the line out, which depends on various factors: how important the line is; how many previous lines they have had; how difficult further access will be; when do they need to dialyse; and most importantly, how ill is the patient at present. Consultants or experienced registrars should make these decisions and each case will be different and judged on the clinical scenario, but basic principles are:

  • If the patient is systemically well, give IV antibiotics, watch closely over the next day or so to see if they get better. If they do, the patient should receive 1-2 weeks of antibiotics. If the patient doesn’t get better, or deteriorates, see below
  • If the patient is unwell, give IV antibiotics and leave the line in and keep under close review. If the patient doesn’t get better in next 12 hours, remove the line and send the tip off. Give a protracted course of antibiotics regardless of whether the line is removed or not

If the patient is very unwell, remove the line ASAP.

HD Catheter Thrombosis

This is the commonest cause of catheter dysfunction, but the incidence is reduced by filling each lumen ('locking the line') with heparin at the end of each dialysis (as per local protocol).

In some cases of permanent catheter thrombosis, it may be appropriate to administer intraluminal urokinase - although this is not done in all units, or in every case of catheter thrombosis. Thrombolytic agents should only be administered by experienced clinicians.

Taking tunnelled lines out is tricky – don’t attempt this unless you have been taught how.

Adequacy of Dialysis


A concept of dialysis adequacy is essential for successful treatment by haemodialysis. Adequacy refers to the 'dose' of dialysis appropriate to the needs of the patient. Too little dialysis and the patient will fail to become rehabilitated or even survive. Too much or too rapid dialysis can be fatal in some acute situations. This is similar to the concept of dosing by a drug.

Adequacy considers 'dose', or amount of dialysis delivered to the patient, and 'requirement', which depends on the condition of the patient. The dose can be controlled and measured exactly, as it is delivered by a machine, whose function is engineered and obeys physical laws. On the other hand, the patient’s requirement for dialysis depends on the individual patient’s condition and physiology. While these physiological processes also behave according to physical laws, there are many processes in play, which are variable between patients and are often incompletely understood.

Key Point: Dialysis adequacy relates to both the dialysis dose and the clinical requirement of the patient, to maintain health and quality of life

A composite quantification of adequacy, combining dose and requirement, can be obtained by history, examination and measuring variables in the patient.

Key Point: Since dialysis performs multiple functions of the failed kidneys, multiple parameters must be measured. An adequate URR alone does not mean adequate dialysis

These parameters include the serum concentrations of a range of uraemic toxins and electrolytes, an assessment of fluid balance, BP, serum bicarbonate and haemoglobin. If dialysis is performed intermittently, some understanding of the dynamic, un-physiological changes during dialysis sessions is required; to avoid harmful disequilibrium during dialysis or intolerable solute concentrations, or fluid content at the beginning or end of the sessions.

How much dialysis is ‘adequate’?

An adequate dialysis could be considered as one which controls the accumulation of uraemic toxins and the balance of acid/base, fluid and electrolytes to a level which is physiologically tolerable in the long term and which can be delivered without intolerable symptoms.

So for example, for a patient with GFR>10 ml/minute/1.73m2, no dialysis at all would be ‘adequate’ as the kidneys already achieve optimal control, compared to the best dialysis. For anuric patients treated by 3 sessions of carefully supervised haemodialysis, 3hrs, 10 minutes each per week, with a urea clearance 240 ml/min/1.73m2, combined with dietary and pharmaceutical intervention is ‘adequate’.

These adequacy levels have been proven by RCT, as higher doses have not resulted in measurably improved outcome. For patients treated other than by 3 sessions and 9.5 hours per week, and those with GFR in the range 1-6, an optimal dose has not been established by RCT, but could be predicted by analysis of existing data and an understanding of the physical process involved. It is generally considered that patients would gain benefit from a higher dose of dialysis than the established 'adequate' levels, especially if achieved through longer treatment times, and more porous dialysis membranes.

Can a patient have 'too much' dialysis?

Haemodialysis can clear solute and fluid from the patient at a rate which is many times higher than normal kidney function. It has the potential to cause dehydration, solute depletion and disequilibrium which could be harmful or fatal. In a typical intermittent haemoduialysis schedule, the 9.5-12 hours per week of treatment has the overall effect on solute clearance equivalent to less than 10% of normal continuous renal function.

This is normally insufficient to cause clinically important solute depletion in well-nourished patients, except, possibly, of water-soluble vitamins. However, in patients who are malnourished or treated by prolonged dialysis sessions (e.g. nocturnal dialysis), depletion of potassium or phosphate is possible. In nocturnal dialysis, phosphate and additional potassium is added to the dialysis fluid to limit clearance of these solutes.

When a patient presents with severe uraemia or hypernatraemia, standard dialysis causes a rapid fall in plasma osmolality and the disequilibrium syndrome, which can be fatal. For this reason, dialysis should be provided with low blood flow and short duration in this situation. Where serum osmolaity is high due to hypernatraemia, dialysis should be avoided if possible until sodium has been normalised by other means.

Where dialysis is provided continuously in critically ill patients, high clearance rates have the potential to cause multiple adverse effects. These include disequilibrium due to over rapid correction of electrolyte abnormalities, especially sodium, alkalosis and solute depletion, especially of magnesium, phosphate and potassium.

Dialysis time

During dialysis, all of the solute and fluid which has accumulated in the patient since the previous dialysis must be removed. Therefore, the absolute difference between the concentrations of solute at the beginning and the end of dialysis depends on their rate of accumulation and the time since the previous dialysis - independent of the duration of dialysis. With shorter dialysis sessions, the rate at which this reduction is achieved must be greater than with longer sessions. This change of concentration and fluid content results in a disequilibrium between fluid and solute; this tends to be retained in body compartments peripheral to the main circulation. The rate of change of fluid content, concentrations, and resulting disequilibrum is inversely proportional to dialysis time.
There is RCT evidence that increasing dialysis time above 3hrs,10 minutes, and/or increasing urea clearance above 240 ml/min/1.73m2, does not improve outcome. Separate RCTs have failed to demonstrate improved outcome with overnight or more frequent dialysis (though there are benefits in surrogate outcomes such as blood pressure and phosphate control, which could benefit some patients). A European observational study suggested that longer dialysis time up to 4 hours per treatment was associated with improved survival.

Despite the lack of RCT evidence in favour of longer dialysis, there are significant theoretical benefits. The mass of toxin removed by dialysis (especially of larger toxins) is increased as the length of the dialysis session increases. Longer dialysis results in lower ultrafiltration rates, which may be better tolerated and makes it easier to achieve target weight. Finally, with very long treatments (e.g. overnight) or increased frequency, the length of the interval between sessions is reduced, reducing the accumulation of toxins between sessions.

On the other hand, increasing dialysis time or frequency increases cost and can reduce quality of life. The majority of patients prefer shorter treatments, even at the expense of symptoms and theoretically reduced health in the long term. Prescribing haemodlalysis needs to incorporate the patient's view.

Clearance of higher molecular weight toxin

There are hundreds of different substances which are known to accumulate in renal failure and are associated with uraemic toxicity. Almost all of these toxins have a higher molecular weight than urea, and are cleared less well by haemodialysis. The use of high flux dialysis membranes improves the clearance of high-molecular weight toxins. The addition of convection to dialysis in haemodiafiltration (HDF) further increases the clearance of high molecular weight toxins. Recently, dialysers with improved design and larger, more porous membranes have become available which clear larger toxins at a rate intermediate between high-flux haemodialysis and HDF.

Recent RCTs has demonstrated improved survival with high-flux dialysis compared to low flux and with HDF using high convection volumes over 40litres per session, compared to high-flux dialysis.

Salt and water

Anuric patients lack any effective mechanism for removing excess salt and water from the body. A typical dialysis patient will maintain a constant serum sodium concentration by drinking 1 litre of water for every 8.5g of dietary salt. The excess salt is removed during dialysis without significant changes in serum sodium concentration by ultrafiltration of the excess fluid. Where a dialysis patient restricts fluid intake without restricting salt, they may develop significant hypernatraemia. It is important to emphasise this to the patient.

During dialysis the serum sodium concentration falls rapidly, causing osmotic disequilibrium. Failure to remove the excess fluid results in chronic fluid overload, hypertension, left ventricular hypertrophy, peripheral and pulmonary oedema. These reduce quality of life and contribute to the high mortality due to cardiovascular causes, experienced by dialysis patients.

Excessive or too rapid removal of fluid can cause hypotension and increased blood viscosity. This reduces tissue perfusion, potentially causing infarction or myocardial stunning.

Adequate dialysis would avoid harmful fluid overload between dialysis sessions without causing significant changes in serum sodium or tissue perfusion.

Diet and fluid intake

The accumulation of toxic solutes can be influenced by dietary or pharmaceutical intervention. However significant dietary restrictions may compromise the patient’s nutrition. Typically, phosphate and intake is restricted by the use of oral phosphate binders. Sodium and potassium intake are limited by dietary restriction (to some extent) in most patients.

Fluid intake is not normally restricted independently of a sodium restriction as this could cause hypernatraemia at the start of the dialysis sessions, increasing disequilibrium. Fluid intake is usually driven by thirst in response to rising plasma osmolality. By restricting sodium, both thirst and fluid intake will be reduced.

The urea kinetic model (UKM)

UKM is a tool for predicting the clearance of urea by any dialysis prescription, and for quantifying both the dose and the requirement for dialysis. UKM uses pharmacokinetic principles to calculate clearance and generation rates of urea from the changing (i.e. kinetic) urea concentrations in the blood.

UKM considers urea to be generated continuously (G) into the body water volume (V), which varies due to ultrafiltration and fluid accumulation between dialysis sessions. Urea is continuously cleared by renal function (Kr) and intermittently by dialysis (Kd) during the dialysis session length (t). Using UKM, the concentration of urea at any time can be calculated, from known values of V, G, Kr, Kd and t. Alternatively G/V and Kt/V (where K = Kr+Kd) can be calculated from blood samples (after allowing for the effects of Kr in the interval between dialysis session, G during dialysis and variations in V).

  • G/V = rate of rise in urea concentrations between dialysis sessions
  • Kt/V = ln(pre/post), where pre and post are the urea concentrations before and after the dialysis sessions

G/V is a body size normalised urea generation rate, proportional to the normalized protein catabolic rate (nPCR), an important measure of dietary intake. A well nourished patient has nPCR>1 g/Kg Ideal weight/day. In an anuric patient this would be equivalent to arise in serum urea concentration of 10 mmol/l/day or 28 mg/dl/day.

Kt/V is a body size-normalised clearance expressed per session. An adequate dialysis for an anuric patient would have Kt/V at least 1.3 per session, three times per week. This would be achieved by a reduction of urea concentration by 67%, for a 4 hour treatment.

Kd can also be predicted from the dialyser mass transfer area coefficient, blood and dialysis fluid flow rates. V can be measured using bioimpedance. Kr is measured using urine collections. This independent calculation of Kt/V allows quality control and trouble-shooting of the dialysis process.

Urea Reduction Ratio (URR)

The URR can be calculated easily from pre- and post-dialysis blood urea concentrations as URR = (pre-post)/pre. URR approximates to the mass of urea removed by a dialysis session as a proportion of the total mass of urea in the patient. Where dialysis has been provided in a standard way, to anuric patients using low-flux dialysis, three times weekly and fixed dialysis time, URR has been shown to relate to outcome. A URR of >67% is considered adequate for an anuric patient treated by 3x weekly dialysis sessions of 3 hours 10 minutes. This is equivalent to Kt/V of 1.3. In this ‘standard’ dialysis, URR reflects extracorporeal blood flow rate and compliance with the prescribed dialysis time.

There is no evidence that URR reflects outcome when prospectively varied or when dialysis is provided according to contemporary standards (e.g. high-flux dialysis). In longer dialysis, or where there is significant renal function, URR is meaningless. Patients treated by continuous dialysis or those with adequate renal function to avoid dialysis altogether have URR of zero despite good outcome. Short, rapid treatments would achieve very high URR, but have been shown to have poor outcome.


Kt/V is the ratio of the volume of blood cleared by dialysis (Kt) as a proportion of the total body water (V). This is calculated or predicted from URR using a urea kinetic model. Unlike URR, Kt/V can take ultrafiltration, urea generation during dialysis and post-dialysis rebound into account. Like URR, Kt/V has been shown to relate to outcome in standard dialysis sessions in the past. However, Kt/V is strongly dependent on dialysis duration and blood flow rate. But Kt/V has never been shown to relate to outcome, independent of dialysis time. Prospective changes in Kt/V have not been shown to affect outcome. Kt/V does not take account of the number of sessions of dialysis per week and cannot realistically account for residual renal function.

With Kt/V, clearance is expressed as a proportion of body water volume. In contrast, the generation rate of uraemic toxins is proportional to surface area. To maintain toxin levels a tolerable level, clearance needs to be in proportion to surface area. GFR is normally proportional to body surface area. For this reason, clearance in dialysis should be prescribed in proportion to surface area, not V. There is RCT evidence that smaller patients require higher Kt/V (due to their higher BSA/V ratio).

While Kt/V is very poor or useless predictor clinical outcome, it is very useful in the quality assurance of the dialysis process: to prove that the prescribed dialysis has been delivered. A urea kinetic model can predict Kt/V accurately form the dialysis prescription. The difference between prescribed and actual Kt/V has been shown to predict outcome, regardless of the prescribed Kt/V.

A simplified type of Kt/V (ignoring any post-dialysis rebound) can now be measured without blood samples and at minimal cost by the dialysis machine using online conductivity clearance monitoring (OCM) and bioimpedance. With OCM, the machine uses built-in dialysate conductivity sensors to calculate the rate of ion transfer across the dialysis membrane, in response to controlled changes in dialysate solute concentration.

This information can be used to calculate conductivity clearance, closely related to urea clearance (K). The dialysis machine integrates repeated measurements of K over time (t) to calculate Kt. An additional input of body water volume (V), calculated from occasional measurements using bioimpedance, is required to calculate Kt/V.

Why urea?

Urea is used to track and quantify the dialysis process, because it is relatively stable in plasma, is present in relatively high concentration, is easy to measure and diffuses easily through red cell membranes. This last property is unique to urea. All other toxins diffuse more slowly across cell membranes, so the concentration in plasma during dialysis does not reflect the concentration in blood or other extracellular water. The concentrations of toxins other than urea cannot easily be assessed from samples taken at the end of dialysis for this reason.

Urea is a very atypical uraemic toxin. Cell membranes have specific urea channels which allow urea to diffuse freely throughout the body water. Within the dialyser, urea is cleared equally from red blood cell water as from the plasma. In contrast, all other uraemic solutes are cleared only from plasma in the dialyser and intracellular solute may not be cleared to any significant extent.

Accounting for the removal of toxins other than urea

Fortunately, it is not necessary to measure and control the removal of each toxin by dialysis. This is because it is possible to predict the removal of each toxin by dialysis, through understanding of its physical properties and the conditions of the dialysis process. The predictions are calibrated by the urea kinetic model. By using a high flux membrane and adding convection (e.g. HDF), the clearance of all uraemic toxins becomes closer to that of urea.

The concentration of toxins depend on the generation rate in the patient as much as clearance by kidneys and/or dialysis. If there was a fixed relationship between the generation rates of all toxins, then dialysis adequacy could be assured by controlling the plasma level of one key toxin, e.g. urea. Unfortunately, this is not the case. The generation rate of urea is uniquely proportional to protein catabolic rate, in turn dependent on the dietary protein intake; and the mismatch between the types of dietary amino acids and those required for protein synthesis. The effect of variations in generation rate is compensated for by quantifying dose as a function of urea clearance (i.e. as Kt/V or URR).

The generation of certain key toxins into plasma can be independently variable and may result in high levels in plasma, despite otherwise adequate dialysis. These include sodium, water, potassium, parathyroid hormone, phosphate and hydrogen ions. Therefore these must be measured regularly and controlled independently if necessary by dietary or pharmacological treatment.

The dialysis prescription

The prescription consists of dialyser type, dialysis time, frequency, composition of dialysis fluid and flow rate of blood, dialysis fluid and any ultrafiltration. Target maximum values for body weight and ultrafiltration rate are specified, depending on assessments of fluid overload and comorbidity. Likewise minimum values for Kt/V are specified, depending on Kr and body weight (smaller patients need higher Kt/V).

How to provide an adequate haemodialysis in CKD

Prescribe an adequate dialysis

  1. Minimum dialysis time: e.g. 12 hours per week minus 1.5 hours for each ml/min/1.73m2 of residual renal function (GFR).
  2. Minimum number of sessions per week e.g. 3 sessions per week minus one session for each 2.7 ml/minute of GFR.
  3. Minimum target blood flow rate e.g. 200 ml/min per m2 patient surface area.
  4. Minimum dialysate flow rate e.g. 1.8 x blood flow rate.
  5. Minimum high-flux dialyser membrane surface area e.g. equal to patient surface area.
  6. Online haemodiafiltration (HDF; if available) effective ultrafiltration rate 0.2 times blood flow rate. Online HDF adds minimal cost to the treatment if already included in the dialysis machine. There is RCT evidence that HDF is associated with improved outcome if high volumes are used.
  7. If HDF is not available, use the largest, most porous high-flux dialyser available. These dialysers are no more expensive to manufacture and clearance of larger molecular weight toxins is increased.
  8. Calculate the target blood volume processed for each treatment as target blood flow rate times minimum treatment time.
  9. Calculate target minimal prescribed Kt/V per treatment, using UKM. Since dialysis is prescribed in proportion to surface area, Kt/V will be relatively higher in smaller patients (due to a higher BSA/V ration)
  10. Estimate target weight, ideally using bioimpedance.
  11. Estimate maximum tolerable target intradialytic weight gain: e.g. 2.5% of body weight.
  12. Estimate maximum tolerable ultrafiltration rate: e.g. 0.625% of body weight.
  13. Set targets for bicarbonate, potassium, phosphate.

Quality control, each treatment

  1. Record actual time of effective dialysis (displayed on the machine at the end of dialysis). Compare with prescribed time. investigate any shortfall.
  2. Record actual blood volume processed by the dialysis machine (displayed on the machine at the end of treatment). Compare with the target minimum blood volume processed. Investigate any shortfall.
  3. If online clearance available, compare Kt/V from OCM with the target minimum per-treatment Kt/V. Investigate any shortfall.
  4. Record post-dialysis weight. Compare with the prescribed target weight. Investigate any difference
  5. Brief clinical assessment of the patient, including symptoms, shortness of breath, ankle swelling. Nurses should ask about these symptoms on each dialysis session

Quality control, monthly

  1. Fluid assessment, ideally using bioimpedance, and adjust target weight if necessary.
  2. If average pre-dialysis blood pressure is > target, consider increasing antihypertensive medications, reducing target weight, increasing weekly dialysis time or restricting sodium intake. Avoid vasodilating antihypertensives (as they reduce tolerance to ultrafiltration). If there is significant renal function, consider treatment with diuretic, avoid excessive dehydration (as dehydration can reduce renal function and renal function improves outcome).
  3. If average intradialytic weight gain > target, consider increasing dialysis frequency or restricting dietary sodium intake.
  4. If average ultrafiltration rate > target , consider restricting dietary sodium intake or increasing weekly dialysis time.
  5. If dialysis prescription includes GFR, repeat urine collections and adjust prescription if indicated.
  6. Calculate delivered Kt/V from pre- and post-dialysis urea. Calculate theoretical delivered Kt/V from HD session records. Compare each with prescribed minimum Kt/V and investigate any differences. Where delivered Kt/V is < prescribed and target treated blood volumes have been delivered, access recirculation should be suspected.
  7. If serum phosphate > target, consider increasing weekly dialysis time, adding or increasing dose of phosphate binders.
  8. If serum bicarbonate < target, consider increasing dialysis time or increasing dialysate bicarbonate concentration.

Key Point: URR of > 65% is generally accepted as a biochemically adequate dialysis session in an anuric patient, treated with around 4 hours of dialysis

How URR is Measured

The ‘pre’ sample is taken immediately after cannulation of the fistula with a dry needle and before the dialysis starts. For central venous cannulae, a sample can be drawn from the line after the heparin lock has been removed. Take care not to contaminate the samples with saline.

The post sample is taken at the end of dialysis after reducing blood flow to 50 ml/minute for 15 seconds. This is required to allow any recirculated blood to clear the fistula.

Some dialysis units stop the dialysate flow for 5 minutes before sampling. The URR calculated by this method will be significantly lower than when using the stop-flow method. This is because there is a significant post-dialysis urea rebound as urea re-equilibriates between body compartments. The URR calculated using the stop-dialysate flow is also more variable as the rate of rebound is maximal at minutes post dialysis and, therefore influenced by errors in the timing of the sample.

Problems with URR and KT/V

Small molecule clearance is not the only or even always the most important factor in determining dialysis adequacy:

  • Fluid balance is important for mortality and is not measured by KT/V or URR.
  • URR and Kt/V acount only for urea clearance. Urea has low toxicity and its behaviour is not representative of other, more toxic uraemic substances.
  • URR and Kt/V are functions of V (volume of distribution of urea). V is low in wasted, malnourished patients, resulting in high KT/V and URR values, even if clearance is low. Dose should be scaled according to surface area, not V.
  • URR and Kt/V have only been shown to be related with survival in historic, observational studies using relatively low dose.
  • URR does not take urea generation, or ultrafiltration into account (Owen, 1993; Held, 1996)
  • URR and Kt/V do not take frequency of dialysis and any residual renal function into account.

Renal Association Guidelines

The RA has published adequacy standards for HD patients. It is recommended that patients should receive HD at least three times per week, and that they should consistently have a URR of > 65%, or an equilibrated Kt/V of > 1.2. It also recommends that each session of dialysis should be at least 4 hours, and patients should not receive less than this without careful consideration. ‘Adequate HD’ is therefore defined as:

  • Minimum dialysis dose (URR > 65% or eKt/V > 1.2)
  • AND minimum treatment time per session of 240 minutes

They also suggest that dialysis frequency should be increased in patients with:

  • Refractory fluid overload
  • Uncontrolled hypertension
  • Hyperphosphataemia
  • Malnutrition
  • Cardiovascular disease

These guidelines offer a good starting point for managing dialysis, but meticulous attention to detail in individual patients is the only way to ensure they receive the treatment they deserve.

Dry Weight

One of the key functions of HD is to remove fluid that has accumulated between sessions. The concept of ‘dry weight’ refers to the weight at which the patient is oedema free and below which hypotension would occur on further fluid removal – the patient is euvolaemic.

Dry weight changes over time, falling (for example) when patients become ill, but increasing when they recover. It requires regular review. Dry weight should, ideally, also be assessed objectively (e.g. by bioimpedance).


It is best to think about these in terms of complications that are technique-related and arise a. during dialysis, or b. in the longterm. As part of the latter, there are also non-technique related complications, ie ones that affect patients on peritoneal dialysis or a failing transplant.

Complications Arising During Dialysis

The three commonest acute problems are hypotension, haematoma ('blown fistula') and fever:

Hypotension (sometimes called 'going flat')

The subject has been reviewed by Palmer in 2008. Symptomatic dialysis-related hypotension is common. It can also shorten treatment times, thus reducing the delivered dose. Although usually due to excessive ultrafiltration, consider other causes:

  • Cardiac disease (arrhythmia, MI, tamponade etc)
  • Autonomic neuropathy
  • Vasodilator drugs (especcially calcium channel blockers, alpha blockers)

Patients may complain of light-headedness, dizziness, syncope, nausea/vomiting or even cramps. It usually results from intravascular volume depletion from excessive ultrafiltration, and is common in patients with cardiovascular problems, or those on (aggressive) anti-hypertensive therapies. Be wary of sticking too rigidly to a previous dry weight estimation, as if the patient has put on lean body mass, then dialysing them down to this weight will mean intravascular volume depletion – dry weight needs frequent clinical assessment.

To manage this problem, start by reviewing dry weight. Consider longer, slower dialysis (unpopular). Consider serial ultrafiltration followed by isovolaemic dialysis (lengthens dialysis again; can be used for a time to get nearer to dry weight).  Review haemoglobin (effect of anaemia possibly via cardiac oxygenation). Consider providing oxygen during dialysis.

There are also tricks that can be done with the dialysate: (1) cooling (causes increased peripheral resistance); (2) sodium profiling, or ramping, in which the dialysate sodium is altered during dialysis. A higher dialysate Na reduces hypotension (probably by maintaining ECF osmolality) - but reduces Na removal. Start high, lower later helps. Ultrafiltration rate can also be profiled on some machines.

Other management includes advice on avoiding eating and drinking before/during dialysis (reduces peripheral resistance by causing splanchnic vasodilation). Omit hypotensive agents on the morning (or evening) before dialysis. Consider oral midodrine, an a1 adrenergic agonist (currently available in the UK on a named-patient basis). Consider haemofiltration or haemodiafiltration (different membranes, but in the case of haemofiltration, also usually a slower treatment - and possibly with more cooling of blood).

Haematoma (or 'Blown fistula')

Excessive haematoma around the AVF is a relatively frequent complication, which occurs most commonly in older patients. If the access has been assessed as mature for venupuncture, poor cannulation skills are often the cause.


Consider bactaeremia, water-borne pyrogens, or overheated dialysate. Pyrexial reactions used to occur more commonly when water used for dialysis was not purified to the same extent. Intradialytic pyrexial reactions are now more likely to be due to infection from vascular access (tunnelled lines etc) than from contamination of the dialysis fluid.

Air Embolus

A dialysis machine includes at least two independent sensor and control mechanisms to prevent air in the extracorporeal circuit entering the patient. Leaks or disconnections are more likely to cause blood loss, rather than air embolism.

Air can enter via a leak at the arterial access or blood lines upstream of the blood pump, where pressure may be below atmospheric. This air is turned to foam as it passes through blood pump and dialyser and should be easily visible in the venous bubble trap. In the unlikely event of failure of the protection systems, the foam could cause low output cardiac failure as it fills the heart.

A more likely cause of air embolism is via an incorrecly occluded or connected central line, independently of dialysis.

If suspected, the blood pump should be stopped, the venous line clamped, and the patient placed in the recumbent position, with the head tilted down. 

Dialyser Reaction

Chemical sterilisation of dialysers and tubing with ethylene oxide has been associated with anaphylactoid reactions in the past. This no longer occurs in Europe as ethyline oxide is no longer used to sterilise dialysers.

Anaphylactoid reactions mediated by bradykinin have been reported in patients on ACEi using AN69 dialysers in the past (Verresen, 1994). Modern versions of this membrane absorb bradykinin, so this problem also no longer occurs.


This is uncommon, but should be considered if the patient develops backache, chest pain or shortness of breath during dialysis. With modern highly porous dialysers, the free hamoglobin may pass through the membrane and trigger the machine's blood leak alarm. The free haemoglobin may be visible as a red colour in any separated plasma in the venous bubble trap or blood samples.

Haemolysis may be casued by chemical conatmination of the dialysis fluid (e.g. failure of the water treatment system or an incorrect or contaminated dialysis fluid concentrate), overheating of the dialysis fluid, or kinking of the blood lines (especially in the segment between blood pump and dialyser) where the pressure is not monitored.


Consider excessive urea clearance (dialysis disequilibrium syndrome, especially on first dialysis), air embolus or severe hypertension. If a first fit, the patient should be investigated in the normal way (admission, CT head etc).

Dialysis Disequilibrium Syndrome

This is rare but important. It occurs in severely uraemic patients who are dialysed too aggressively at the initiation of dialysis. When the initial dialysis is too fast, it causes a dramatic reduction in serum osmolality, which in turn causes a paradoxical acidosis of the cerebrospinal fluid and cerebral oedema. Clinically the patient may complain of headache and become confused, with restlessness and tremors. Occasionally they may have a seizure or even coma. This can happen after dialysis has finished.

Key point: the syndrome can be avoided by increasing the dialysis dose gradually in patients starting haemodialysis, eg 2h, 3h, 4h for the first three treatments


This may occur from AVF, AVG or in the GI tract. May be due excessive heparin given during HD, or other anticoagulant problems. An acute bleed should be treated as an emergency. For patients with an increased risk of bleeding, anticoagulation should be avoided or kept to a minimum by using a high blood flow rate and regular flushing of the extracorporeal circuit with saline every 15-30 min. Alternatively heparin can be replaced with prostacyclin or regional citrate anticoagulation.

Circuit Clotting

Can occur due to insufficient anticoagulation of the dialysis circuit – although this is seen less commonly now.

Muscle Cramps

Muscle cramps are very common during dialysis and can be of sufficient severity that they result in termination of the procedure. Their cause is unclear but the majority occur towards the end of the procedure after a significant volume of fluid has been removed. Their aetiology is postulated to involve volume depletion and tissue hypoxia. They are associated with large requirements for fluid removal. Acute management often involves the administration of hypertonic fluid, most commonly 50% dextrose (50mls), in order to raise plasma osmolality.

Other treatments include: Quinine sulphate 2-300mg before dialysis or at bedtime can be tried; this is unproven. Oral agents such as clonazepam, vitamin E, carnitine, or anti-convulsants are sometimes used as prophylaxis but their benefits are even less certain.

Most agree that it is important to limit of inter-dialytic weight gains, ensuring that post-dialysis dry weight is correct. Use of an appropriate dialysate sodium is also useful. Remember that a higher dialysate sodium will reduce intra-dialytic symptoms at the expense of thirst and weight gains; the converse holds true for a lower dialysate sodium. Sodium profiling may be of benefit.

Patients who experience cramps at night may benefit from muscle-stretching for a minute or two. Heat and massage for the camping muscle can help.

Cardiac Arrhythmias

Can occur due to the flux of electrolytes that occur during dialysis (especially serum potassium changes). They occur more regularly than you might think, especially in those with underlying cardiac dysfunction (particularly left ventricular hypertrophy and coronary artery disease).

Dialysis-Related Critical Incidents

The most serious acute events include air embolus, line disconnection leading to haemorrhage, acute haemolysis or toxicity related to line kinking or dialysis contamination, and acute allergic reactions to dialysers or sterilants (eg the 'first-use' syndrome attributed to antibody formation to ethylene oxide). If any such crisis occurs and the explanation is not entirely clear, in addition to all the necessary supportive measures:

  • Stop dialysis
  • Take samples from venous and arterial lines - look for alterations in haematology and biochemistry
  • Disconnect the patient. Record their weight and routine observations
  • Keep a sample of dialysate
  • Keep the used dialyser
  • Take the machine out of use. Inform the dialysis technicians that it was in use when an incident involving a patient occurred (eg by a prominent notice on the machine), so that an investigation can be made and evidence preserved
  • Record all the details, including the precise circumstances (patient's position, first symptoms, full history)
  • Fill in the appropriate incident report form

Longterm Complications

The commonest longterm complications are AVF thrombosis, and infection; and thrombosis is the most common cause of vascular access loss. Thrombosis and infection occur more frequently in arteriovenous grafts (AVG) and dialysis catheters than in arteriovenous fistulae (AVF).


This is the commonest cause of fistula loss. 80-90% of thromboses are caused by venous stenosis, but hypotension, prolonged compression and intravascular volume depletion can contribute.

Both AVFs and AVGs are vulnerable to thrombosis. The Dialysis Outcomes and Practice Patterns Study (DOPPS) reports that AVG are 3.8 times more likely to require thrombectomy and 3.0 times more likely to require access intervention than AVF (Young, 2002). AVF thrombosis rates remain in the range of 0.2 to 0.8 per patient year and AVG thrombosis rates are typically in the range of 0.6 to 1.2 per patient year (Sands, 2009).

United States Renal Data System (USRDS) data confirm that AVF have the lowest complication rates of any available vascular access (0.64 procedures per patient year versus 1.61 for AVG (Sands, 2009). Once a primary fistula is established, thrombosis is the leading cause of failure in approximately 40% of cases (Albers, 1994)

Key point: if you think a fistula may have (acutely) thrombosed, rapid intervention may save the fistula. But if it is left, it will be lost. This is a renal sub-emergency

Key point: patients should be told that if they stop feeling the buzz in their fistula, they should must seek medical attention immediately – day or night


Infectious complications of vascular access are a major source of morbidity and mortality among HD patients. Previous studies have reported infection as a common cause of death; accounting for 9.5 to 36% of deaths in HD patients (Dhingra, 2001). Vascular access infections (most commonly found in patients using dialysis catheters) are reported to be the source in up to 48 to 73% of all bacteraemias in HD patients (Nassar, 2001).

The risk of bacteraemia with tunnelled dialysis catheters averages 2.3 per 1000 catheter days. This translates into an approximate 20 to 25% bacteraemia risk over the average duration of use (Saad, 1999). This compares to a risk of infection of 10% in AVGs, and 2-5% in AVFs.

Arm and Hand Oedema

The ‘fistula arm’ is commonly 2-3 cm larger in diameter to the non-fistula arm. Any larger increases in size could suggest venous hypertension caused by venous outlet stenosis. Rapid/painful increases in size should prompt urgent investigation and senior review to rule out thrombosis (see above).

Ischaemic 'Steal Syndrome'

This occurs secondary to a HD arteriovenous access occurs in approximately 5 to 10% of cases. The pathophysiological basis of this condition is a marked decrease or reversal of flow in the arterial segment distal to the AVF or AVG, induced by the low resistance of the fistula outflow (Schanzer, 2004).

Because of the AV shunt created by the anastamosis, the distal part of the limb can suffer from inadequate blood supply. Small amounts of steal are normal and tolerated well. So, mild cases can be observed closely, as most of them will reverse in a few weeks, or be tolerated by the patient.

But if the limb is cold, weak, numb or painful .. take it seriously. The patient should be reviewed by a consultant or experienced registrar ASAP; as early recognition and reversal should preserve limb function. Otherwise, severe ischaemic complications including ischaemic neuropathy and ischaemic gangrene can occur - with the potential need for amputation.

Several surgical and endovascular treatments have been used, including: access ligation, banding, elongation, distal arterial ligation, and distal revascularization-interval ligation. The best reported results, with maintenance of access function and reversal of symptoms, have been obtained with the distal revascularization-interval ligation (DRIL) (Schanzer, 2004) and the endoluminal-assisted revision (MILLER) procedures (Goel, 2006).


These are almost always occurs on the venous side, and occurs due to intimal fibromuscular hyperplasia in the first 2-3cm of the venous anastamosis.

Aneursyms and Pseudoaneurysms

These may be false/true aneurysms and usually require no action. But if they become very large and the overlying skin becomes thin/tense, the segment may need excising.They result from improper needle site rotation or as a complication of more proximal stenosis.

Key point: an aneursimal AVF can rupture, leading to profuse bleeding, which requires emergency surgical intervention

Appropriate selection of dialysis staff for access cannulation together with cannulation training and education for staff members and patients may reduce the risk of this complication. In addition, a visibly tortuous access shape may be a major cosmetic concern for some patients .

SVC Obstruction

Venous hypertension occurs in approximately 3% of fistulas and grafts and is usually related to central vein stenosis (CVS), usually the SVC. Percutaneous transluminal angioplasty of a CVS, supplemented by stent placement as needed, is effective and considered the primary treatment for such lesions due to the lack of viable and safe surgical options (Levitt, 2006).

High-Output Heart Failure

This occurs from fistula placement occurs if fistula flow exceeds 20% of cardiac output. This complication is rare (less than 1% of patients). It may require closure of the AVF.

Prolonged Access Bleeding

This should not be overlooked, and should raise suspicion of high intra-access pressure, outflow stenosis or local inflammation. Prolonged bleeding may also be caused by excessive heparinisation of the blood circuit, access laceration during previous cannulation or skin atrophy. Clinical examination of the site should be performed and venous pressure measurements should be made.

Other (Non-Technique) Chronic Complications

Cardiovascular disease
CV disease is a common in ESRD, especially those on HD. It is caused by many inter-related factors, including hypertension and left ventricular hypertrophy, plus accelerated atheroma, vascular calcification and anaemia.

Dialysis related amyloid
This is a long-term complication of HD. β2-microglobulin is poorly cleared by conventional HD with low flux membranes, and is contained in amyloid fibrils. β2-microglobulin is a large molecular weight molecule (MW 11,600) which is released into the circulation in normal cell turnover but is not excreted in renal failure or removed by cellulose membranes. The most common clinical presentations include:

  • Carpal tunnel syndrome
  • Joint pains (especially hands, arms and shoulders) after >10y of HD
  • Tenosynovitis of tendons in hands
  • Pathological fractures due to amyloid bone cysts
  • Destructive spondyloarthropathy

Haemodiafiltration and high flux HD improve the clearance of β2-microglobulin and delays onset of amyoid. Transplantation may improve symptoms of dialysis-amyloid.

This is now rare. But in the past, incomplete removal of aluminium from dialysate water, prescription of aluminium antacids, contributed to the problem.

CKD-MBD and Anaemia
These are discussed in those sections of the website.




Top Tip: Maintenance of good vascular access is essential

  1. Currently, over 1 million people in the world receive haemodialysis
  2. Although transplantation is preferred, haemodialysis will continue to be the commonest form of non-transplant renal replacement therapy
  3. Mortality rates are high, but indefinite survival is possible; the commonest causes of mortality are CV disease, and infection
  4. Maintenance of good vascular access is essential for effective management of HD patients
  5. A newly created fistula takes 6-8 weeks to mature and be useable. So, in patients with CKD, the fistula should be created early, when GFR is between 25 and 30 mL/min (ie early CKD4)
  6. A URR of > 65% is generally accepted as a biochemically adequate dialysis session in an anuric patient, treated with around 4 hours of dialysis
  7. But dialysis adequacy relates to both the dialysis dose and the clinical requirement of the patient, to maintain health and quality of life
  8. Since dialysis performs multiple functions of the failed kidneys, multiple parameters must be measured. An adequate URR alone does not mean adequate dialysis
  9. Dialysis disequlibrium syndrome can be avoided by increasing the dialysis dose gradually in patients starting haemodialysis, eg 2h, 3h, 4h for the first three treatments
  10. If you think a fistula may have (acutely) thrombosed, rapid intervention may save the fistula. This is a renal sub-emergency
  11. Patients should be told that if they stop feeling the buzz in their fistula, they should must seek medical attention immediately – day or night
  12. An aneursimal AVF can rupture, leading to profuse bleeding, which requires emergency surgical intervention



Abel J, Roundtree L, Turner B. On the removal of diffusible substances from the circulating blood of living animals by dialysis. J Pharmacol Exp Ther 1914; 5: 275-316

Albers FJ. Causes of hemodialysis access failure. Adv Ren Replace Ther 1994; 1: 107-118

Alwall N. On the artificial kidney I: Apparatus for dialysis of blood in vivo. Acta Med Scand 1947; 128: 317-325

Alwall N, Bergsten B, Gedda P, Norvitt L and Steins AM. On the artificial Kidney IV. The technique in animal experiments. Acta Med Scand 1949; 132: 392-411

Alwall N, Norvitt L, Steins AM. On the artificial kidney. VII: Clinical experiences of dialytic treatment of uremia. Acta Med Scand 1949; 132: 587

Alwall N. Therapeutic and diagnostic problems in severe renal failure. Stockholm: Scandanavian University Books, 1963, 2, 11

Arieff A. Dialysis disequilibrium syndrome: current concepts on pathogenesis and prevention. Kidney Int 1994; 45: 629-635

Benedum J. Pioneer of dialysis, George Haas (1886-1971). Med Hist 1979; 14: 196-217

Bigelow G. Collodion membranes. J Am Chem Soc 1907; 29: 1576-1589

Blagg CR. Belding Hibbard Scribner—Better Known as Scrib. CJASN 2010; 5 (12): 2146-2149
A description of the life and achievements of Scribner, who with Wayne Quinton, made mainetance haemodialysis possible with the AV shunt

Brescia M, Cimino JE, Appel K, Hurwich BJ. Chronic hemodialysis using venipuncture and a surgically created arteriovenous fistula. N Engl J Med 1966; 275: 1089–1092

Bywaters EGL, Joekes AM. The Artificial Kidney: Its Clinical Application in the Treatment of Traumatic Anuria. Proc R Soc Med. 1948; 41(7): 420–426

Cameron JS. Practical haemodialysis began with cellophane and heparin: the crucial role of William Thalhimer (1884–1961). Nephrol Dial Transplant 2000; 15(7): 1086-1091

Crowther SM, L A Reynolds LA, Tansey EM. History of Dialysis in the UK (c1950–1980), 2009. Witness Seminar, on 26 February 2008, held at the Wellcome Trust Centre for the History of Medicine at UCL, London
This is a fascinating transcript of the seminar, well worth reading

Dhingra RK, Young EW, Hulbert-Shearon TE, Leavey S, Port FK. Type of vascular access and mortality in US hemodialysis patients. Kidney Int 2001; 60: 1443-1451

Eggerth AH. The preparation and standardization of collodion membranes. J Bio Chem 1921; 48: 203-221

Eknoyan G et al. The Hemodialysis Study Group: Effect of Dialysis Dose and Membrane Flux in Maintenance Hemodialysis.  N Engl J Med 2002; 347: 2010–2019

Ferry J. Ultrafilter membranes and ultrafiltration. Chem Rev 18:3, 1936

Fick A. Ueber Diffusion. Annalen der Physik 1855; 170(1): 59-86

The FHN Trial Group. In-Center Hemodialysis Six Times per Week versus Three Times per Week. N Engl J Med 2010; 363: 2287-2300

Fogazzi GB. May 1954: the first ever symposium on the artificial kidney. Nephrol Dial Transplant 2003; 18(9): 1726-1730

Goel N, Miller GA, Jotwani MC, Licht J, Schur Iand Arnold WP. Minimally invasive limited ligation endoluminal-assisted revision (MILLER) for treatment of dialysis access-associated steal syndrome. Kidney Int 2006; 70: 765–770

Gotch FA, Sargent JA. A mechanistic analysis of the National Cooperative Dialysis Study (NCDS). Kidney Int 1985; 28(3): 526-34

Graham T. The Bakerian lecture: Osmotic force. Philos Trans R Soc Lond 1854; 144: 117-128

Graham T. Liquid diffusion applied to analysis. Philos Trans R Soc Lond 1861; 151: 183

Haas G. Versuche der Blutauswaschung am Lebenden mit Hilfe der Dialyse. Klin Wochenschrift 1925; 4: 13

Haycraft JB. On the action of a secretion obtained from the medicinal leech on the coagulation of the blood. Proceedings of the Royal Society of London 1884; 36, 478-487.

Held PJ, Port FK, Wolfe RA, Stannard DC, Carroll CE, Daugirdas JT, Bloembergen WE, Greer JW, Hakim RM. The dose of hemodialysis and patient mortality. Kidney Int 1996; 50: 550–556

Hess J, McGuigan W. The condition of the sugar in the blood. Pharmacology 1914; 6: 45-55

Kiil F. Development of a parallel flow artificial kidney in plastics. Acta Chir Scand Suppl 1960; 253: 140-142

Jackson RC, Bunker NV, Elder WJ, Joekes AM. Treatment of renal failure in a unit with facilities for haemodialysis (twin-coil artificial kidney). Results with 80 patients. Br Med J 1960; 2:1909-16

Kolff WJ, Berk HTJ. De kunstmatige nier: een dialysator met groot oppervlak. Ned Tijdschr Geneeskd 1943; 87: 1684

Kolff WJ, Berk HTJ. The artificial kidney: A dialyzer with a great area. Acta Med Scand 1944; 117: 121-134

Kolff WJ, Watschinger B. Further development of the coil kidney. J Lab Clin Med 1956; 47: 969-977

Kolff WJ. Academy of Achievement Interview (1991)
A very interesting interview

Levit RD, Cohen RM, Kwak A, Shlansky-Goldberg RD, Clark TWI, Patel AA, Stavropoulos SW, Mondschein JI, Solomon JA, Tuite CM, Trerotola SO. Asymptomatic central venous stenosis in hemodialysis patients. Radiology 2006; 238: 1051-1056

Locatelli F et al. Effect of Membrane Permeability on Survival of Hemodialysis Patients. JASN 2009; 20(3): 645-654

Mannucci PM. Historical Review. Venous thrombosis and anticoagulant therapy. The first case of venous thrombosis. British Journal of Haematology 2001; 114, 258-270
A good historical review of thrombosis and anticoagulant therapy, discussing role of Haycraft (Hirudin) and Mclean (Heparin)

McLean J. The thromboplastic action of cephalin. American Journal of Physiology 1916; 41: 250-257

McKellar S. Gordon Murray and the artificial kidney in Canada. Nephrol Dial Transplant 1999; 14: 2766-2770

Nassar GM and Ayus JC. Infectious complications of the hemodialysis access. Kidney Int 2001; 60: 1-13
This a good review article regarding infection and HD

Necheles H. Ueber dialysieren des strömendes Blutes am Lebenden. Klin Wochenschr 1923; 2: 1257

Owen WF, Lew NL, Liu Y, Lowrie EG, Lazarus JM. The Urea Reduction Ratio and Serum Albumin Concentration as Predictors of Mortality in Patients Undergoing Hemodialysis. N Engl J Med 1993; 329: 1001–1006

Palmer BF, Henrich WL. Recent Advances in the Prevention and Management of Intradialytic Hypotension. JASN 2008; 19(1): 8-11 

Palmer RS, Rutherford PS. Kidney substitutes on uraemia; the use of Kolff’s dialyser in two cases. CMAJ 1949; 60: 261-266

Parsons FM, McCracken BH. The artificial kidney. Br J Urol 1957; 29: 424-33

Quinton W, Dillard D, Scribner BH. Cannulation of blood vessels for prolonged hemodialysis. Trans ASAIO 1960; 6: 104-107

Richardson BW. Practical studies in animal dialysis. Asclepiad 1889, 6: 331-332

Saad TF. Bacteremia associated with tunneled, cuffed hemodialysis catheters. Am J Kidney Dis 1999; 34: 1114-1118

Sands JJ. Vascular Access: The Past, Present and Future. Blood Purif 2009; 27(1): 22-7 This is a good review article on vascular access

Schanzer H and Eisenberg D. Management of steal syndrome resulting from dialysis access. Sem in Vascular Surgery 2004; 17: 45-49

Skeggs, LT. Persistence … and Prayer: From the Artificial Kidney to the AutoAnalyzer. Clinical Chemistry 2000; 46(9): 1425-1436
A very interesting account of his life. Leonard T Skeggs, with Jack Leonards, devised the Skeggs-Leonard dialyser; and later Skeggs devised the first auto-analyser

Stanley Shaldon S. Personal history of vascular access. 3rd Congress of Nephrology in Internet (CIN). 2003
Interesting paper that emphasises the role of Nils Alwall, as pioneer of the AV shunt in 1949

Steenkamp R et al. Renal Registry 2010. Chapter 2. UK RRT Prevalence in 2009: national and centre-specific analyses

Teschan, PE et al. Posttraumatic renal insufficiency in military casualties. I. Clinical characteristics. Am J Med 1955; 18: 172-86
Note: the history of renal care in the US military has been described by Macken DL and Knepshield JH

Thalheimer W. Experimental exchange transfusion for reducing azotemia. Use of the artificial kidney for this purpose. Proc Soc Exp Biol Med 1937; 37: 641-643

Thalhimer W, Solandt DY, Best CH. Experimental exchange transfusion using purified heparin. Lancet 1938; 2: 554–557

Verresen L et al. Bradykinin is a mediator of anaphylactoid reactions during hemodialysis with AN69 membranes. Kidney Int 1994; 45: 1497-1503

Young EW, Dykstra DM, Goodkin DA, Mapes DL, Wolfe RA, Held PJ. Hemodialysis vascular access preferences and outcomes in the Dialysis Outcomes and Practice Patterns Study (DOPPS). Kidney Int 2002; 61: 2266-2271


Australasia/CARI CARI Dialysis Guidelines: Dialysis Adequacy. Oct 2005

Canada/CSN CSN Hemodialysis Adequacy in Adults (Chapter 1) 2006

Europe/EBPG EBPG European Best Practice Guidelines for Haemodialysis (Part 1). July 2002

UK Renal Association Clinical Practice Guidelines (Fifth Edition). Dec 2009. Mactier, Hoenich N, Breen C

US/KDOQI KDOQI Clinical Practice Guidelines and Clinical Practice Recommendations. 2006 Updates


Pictures of early dialysis equipment in the 'Home dialysis central website: museum'