Complications of dialysis in the ICU

Though this chapter focuses primarily on the complications of continuous renal replacement therapy, one must acknowledge that there are other forms of dialysis, and one might occasionallyu encounter patients undergoing such therapy in the ICU; and therefore an ICU specialist should probably have some idea of their complications. The CRRT-centric focus is owed to the fact that until Question 15 from the second paper of 2018, this had never been asked about before. However, it was felt that one should know all about RRT complications for the purposes of viva practice. And in general, if you are planning not to offer somebody a lifesaving therapy, it is important to appreciate why it is being withheld.

"List the complications associated with renal replacement therapy", demands Question 15,  and "consider in your answer continuous renal replacement therapy, peritoneal dialysis and chronic intermittent dialysis."  Excellent chapters in Critical Care Nephrology (2009, 2nd ed) exist, but are perhaps too long for the time-poor exam candidate. Some effort has been spent to summarise these into the tables presented below. Other excellent free-to-read articles are available on the intrawebs, containing vast lists of complications. Additionally, for the sanity-conserving purpose of rendering this list more easy to remember, Benjamin Gladwin has submitted a mnemonic for complications of haemodialysis: I BATHE HAM

  • Inflammatory response
  • Blood loss
  • Access complications
  • Thrombocytopenia
  • Hypoxia
  • Electrolyte disturbance
  • Hypothermia
  • Anaemic (Haemolytic) complications.
  • Malnutrition due to adsorption of useful molecules
Complications of Renal Replacement Therapy
Domain Details
Generic complications for all types of RRT

All RRT requires access of some sort.
Be it fistula or vas cath, there are risks:

  • Bleeding
  • Vessel damage
  • Bloodstream or localised infection
  • Air embolis

Hypoxia: Activation of complement and the inflammatory mechanisms leads to an increase in the activity of nitric oxide synthase, which counteracts the normal mechanisms of hypoxic pulmonary vasoconstriction. Increased shunt develops; therefore hypoxia ensues.

Hypocapnea: The dialyser membrane is no obstacle for the highly water-soluble CO2; some CO2 will diffuse through the membrane and into the dialysate.


Haemodynamic instability: with all modalities of RRT, one can expect some fluid and electrolyte shifts, and therefore some effect on the haemodynamics of the patient. Of these effects, some patients will be quite tolerant, even in the face of wildly erratic dialysis prescriptions. In other circumstances, even the most careful CRRT regimen will produce severe cardiovascular bewilderment. In short, though this is mostly seen with IHD, haemodynamic instability can be regarded as a generic feature of RRT.

Hypothermia: Because a large volume of blood (roughly 5-10% of the blood volume) spends every minute outside the body, it is exposed to the ambient temperature, which in the ICU is typically rather chilly. The returning blood is usually cool. The patient may become hypothermic as a result. This phenomenon may obscure the presence of a fever, or it may result in a clinically significant drop in the core body temperature.


Dialysis disequilibrium syndrome: This is the movement of small solutes so rapid and in such massive volume, that the concentration of chronically accumulated uraemic wastes in the brain becomes substantially greater than the extracellular fluid. The resulting osmotic movement of water into the brain can give rise to cerebral oedema, which manifests at first as confusion, progressing into seizures and unconsciousness. It is usually only seen with careless IHD, but one needs to acknowledge that an insane CRRT prescription can also produce this complication. According to some cruel dog experiments by Arieff et al (1973), one would need to generate a brain-serum urea gradient of around 45 mmol/L in order to create cerebral oedema. The investigators ligated the animals' ureters to make them uraemic. When the urea concentration increased to over 70 in both brain and blood, they dialysed some of the dogs rapidly, dropping their blood urea to around 25 mmol/L (with the brain urea dropping to ~ 40 mmol/L). The dogs were then killed. Their brain water was found to be significantly increased (by 15%), which was apparent grossly as brain swelling and a raised ICP (31 cm H2O), with "visible bulging of the brain through a trephine opening in the skull". That sort of gradient is not going to be produced in somebody with a serum urea under 40, and normal-dose CRRT is in any case unlikely to drop your urea so quickly.


Electrolyte disturbance, which can be described as "hypo-everything-aemia". Unintelligently prescribed dialysis can lead to electrolyte depletion. If you have prescribed a dialysate or replacement fluid which is completely free of potassium, you should not be surprised that the patient becomes dramatically hypokalemic.

Hyper-electrolytaemia is also a possibility, in the event that you have removed too much fluid and haemoconcentrated the patient (or, alternatively, if you have prescribed some sort of unusual dialysate with an excess of electrolytes in it).


Delayed renal recovery is possible.Renal recovery may be delayed by the very use of dialysis, or it may never occur at all. This may be counterproductive if you suspect the patient will not be offered long-term dialysis. The following mechanisms have been implicated as causes of this "dialysis-induced dialysis dependence":

  • Haemodynamic instability
  • Haemofilter membrane-induced complement and cytokine activation, with subsequent "cytotoxic" tubular injury (analogous to septic nephropathy).
  • Trophic hormone depletion (missing paracrine triggers for nephron regeneration)

Malnutrition due to dialytic nutrient loss: The bloodstream is a necessary destination for all the absorbed nutrients, as well as for TPN. Dialysis removes many of the useful nutrient molecules. Specific easily cleared nutrients are amino acids (all highly water-soluble small molecules) and water-soluble vitamins. Depending on one's ultrafiltration volume, the total amino acid loss may be around 10-20g/day. If on TPN, up to 10% of infused protein content may end up in the effluent bags.


Haemolytic complications: All RRT filters tend to eat red cells. This is a complication of forcing blood to rub against a cheesegrater-like porous membrane.

Blood loss due to circuit loss: If a filter clots, the whole thing is discarded, together with whatever blood is in the circuit. This could be a little or a lot, depending on the filter and circuit. Usually, the amount of blood lost is no greater than 200-300ml, equivalent to a drop of 10g/L of haemoglobin.


Inflammatory response: The dialyser membrane is a proinflammatory surface. Modern membranes are a massive improvement, but some inflammatory reaction (particularly complement activation) is to be expected. Additionally, one's bloodstream becomes showered with the shredded remains of red blood cells, which exerts its own proinflammatory effect.

Specific to IHD

Fistula complications  are in some ways unlike complications from a vas cath, and include the following list of problems:

  • Infection of the fistula
  • Stenosis of the fistula
  • High output cardiac failure
  • Low diastolic pressure
  • Air emboli
  • Steal phenomena

Haemodynamic instability: Consider that with an IHD or SLEDD session, one is removing 2-4 litres of fluid from the patient over 3-8 hours, whereas with CRRT one is removing the same amount over the course of 24 hours. Naturally, patients who are preload-sensitive will not enjoy such rapid fluid movements.

Haemodynamic instability may also occur in situations where the patient is dependent on high levels of vasopressor/inotrope support. In such circumstances, one can assume that the infused catecholamines are being cleared by the circuit.

Renal Amyloidosis develops in chronic haemodialysis patients because of the formation  of small extracellular tissue deposits, made up of low-molecular-weight subunits of a variety of proteins. 
Specific to CRRT
Electrolyte Hypocalcemia with citrate toxicity is a particularly interesting cause of metabolic acidosis (and then alkalosis) which is associated with CRRT 

Complications related to anticoagulation: Unlike IHD circuits, CRRT circuits run over long periods, and are disadvantaged by a rather sluggish blood flow. Consequently, they must be anticoagulated. This is usually achieved by using heparin; although other forms of anticoagulation are available, they are not in routine use. Anyway; circuit anticoagulation usually results in at least some degree of systemic anticoagulation, which in turn results in bleeding complications.

There is a risk of HITTS each time heparin is used. Complications related to citrate anticoagulation are even more interesting, and are discussed in greater detail elsewhere.

Specific to PD


Complications associated with establishing access: bowel perforation

Complications related to routine regular access: peritonitis

Complications related to chronic PD: 

Respiratory` Pleural effusions and ascites:  these impact on respiratory function much like a gravid uterus might, i.e. by displacing the diaphragm and bases of lungs cephalad the extra water ends up decreasing FRC and increasing the work of breathing.
Gastrointestinal Ileus: chronic PD gives rise to adhesions and peritoneal thickening; bowel function suffers mechanically.


Chapter (pp. 540) 48   Renal  replacement  therapy, also by Rinaldo  Bellomo

Bellomo, R., and C. Ronco. "Renal replacement therapy in the intensive care unit." Intensive Care Med (1999) 25: 781±789

Zepeda-Orozco, Diana, and Raymond Quigley. "Dialysis disequilibrium syndrome." Pediatric nephrology 27.12 (2012): 2205-2211.

Arieff, Allen I., et al. "Brain water and electrolyte metabolism in uremia: effects of slow and rapid hemodialysis." Kidney international 4.3 (1973): 177-187.

Bender, Filitsa H. "Successful treatment of severe hyponatremia in a patient with renal failure using continuous venovenous hemodialysis." American journal of kidney diseases 32.5 (1998): 829-831.

Wendland, Erik M., and Andre A. Kaplan. "A Proposed Approach to the Dialysis Prescription in Severely Hyponatremic Patients with End‐Stage Renal Disease." Seminars in dialysis. Vol. 25. No. 1. Oxford, UK: Blackwell Publishing Ltd, 2012.