An appreciation of the significance of oliguria, and its early recognition is essential if steps are to be taken to preserve renal function in the critically ill patient. The minimum effective urine volume is determined by the obligate solute load which has to be excreted each day. In healthy individuals, about 800 mmol can be eliminated in as little as 650 ml of urine. Even in those taking a high carbohydrate, low sodium, low nitrogen diet, any degree of renal impairment greatly increases the minimum obligate urine volume. As a rough guide, a urine requirement of 1 ml/kg.h provides adequate solute clearance. At 0.5 ml/kg.h (840 ml for a 70-kg patient) azotaemia will only be avoided if renal function is normal. Urinary tract catheterization is often necessary, but is invasive. Having accepted the potential complications associated with urinary catheters it is all the more important to monitor urine volumes constantly and to take appropriate action if necessary. Table 1 56 outlines a simple approach to the evaluation and treatment of oliguria. Complete anuria should always, of course, prompt consideration of a blocked urinary catheter.

Relying on only a limited number of the criteria listed in Table 1 56 may lead to an erroneous conclusion regarding the volume status of the patient. Several other pitfalls may mislead the unwary. In the patient with sepsis, oedema (extravascular fluid) may coexist with marked intravascular hypovolaemia, indicated by a low central venous pressure and positive response to a fluid challenge. Conversely, the central venous pressure and pulmonary artery wedge pressure can easily be overestimated in patients on positive pressure ventilation if measurements are not made at end expiration.

There can be no substitute for personally measuring the venous pressure and, in doing so, confirming the calibration and zeroing of the equipment. Over-reliance upon fluid balance records, even when accurately maintained, can mislead since the compliance of the major capacitance vessels can change.

Several urinary measurements may help to distinguish potentially reversible prerenal azotaemia from established intrinsic renal failure: these include sodium concentration (U&subN;&suba;), osmolality (U&subo;&subs;&subm;), urine to plasma urea nitrogen ratio, urine to plasma creatinine ratio (U&subc;&subr;/ P&subc;&subr;, fractional excretion of sodium (FE&subN;&suba;), and renal failure index (RFI). Table 2 57 shows the differences between these indices for prerenal and renal azotaemia.

Filtered and excreted sodium can be calculated as follows: Equation 15

where GFR is the glomerular filtration rate (ml), P&subN;&suba; is the plasma sodium concentration (mmol/1), U&subN;&suba; is the urine sodium concentration (mmol/1), P&subc;&subr; is the plasma creatinine concentration (mmol/1), U&subc;&subr; is the urine creatinine concentration (mmol/1), and V is the urine flow rate (ml/min).

Since creatinine is essentially not reabsorbed or secreted, the urinary concentration of creatinine is a function of water reabsorption. Sodium, however, is actively reabsorbed by the tubules, such that the final U&subN;&suba; depends upon sodium reabsorption as well as the amount of water reabsorbed by the tubules. The renal failure index (RFI) offers no advantages over the FE&subN;&suba; but may be calculated thus: Equation 16

In the oliguric patient, a FE&subN;&suba; below 1 per cent is consistent with prerenal azotaemia. A FE&subN;&suba; of less than 0.4 per cent is even more specific. Conversely, an FE&subN;&suba; of greater than 1 per cent may be consistent with renal damage, although a FE&subN;&suba; greater than 3 per cent is much more indicative of intrinsic renal disease. Values between 1 per cent and 3 per cent are therefore less conclusive. A low FE&subN;&suba; may, under certain circumstances, be an unreliable guide since patients with intrinsic renal failure occasionally have a FE&subN;&suba; of 1 per cent. Conversely, there are hypovolaemic conditions in which the FE&subN;&suba; can be paradoxically high (Table 3) 58.

The increase in the urine sodium produced by diuretics persists much longer following ingestion of thiazides than after frusemide, ethacrynic acid, or the osmotic diuretics. By 24 h after the last dose of diuretic, FE&subN;&suba; can again be adopted as an indicator of prerenal azotaemia.

There are rare situations where the sodium cation is an unreliable indicator of the nature of renal dysfunction. This is the case when there is sodium wasting such as occurs in the patient with a metabolic alkalosis due to the loss of upper gastrointestinal tract secretions. In such a situation, a urinary chloride level of less than 20 mmol/1 (or FE&subC;&subl; less than 1 per cent) is indicative of volume depletion.

Once the clinician is reasonably confident that the observed oliguria is due to volume depletion the obvious response is to administer fluid, the nature of which is dictated by the degree of urgency, electrolyte status and the haemoglobin concentration of the patient. Colloidal solutions expand the intravascular space rapidly, while electrolyte solutions are distributed across intra- and extracellular compartments. Blood is preferred when the patient is anaemic. If fluid is infused rapidly (500 ml in 30 min) the response to volume expansion is similar regardless of the type of fluid. The haemodynamic and renal response to a fluid bolus will often resolve any doubts as to the intravascular volume status of the patient. Fluid resuscitation should be performed quickly and under the personal supervision of the clinician. It is insufficient simply to increase the infusion rates of existing fluids and re-evaluate the patient 6 to 12 h later.

The administration of dopamine in so called ‘renovascular’ doses (about 2.5 &mgr;g/kg.min) may reverse the renal vasoconstriction that is a feature of the volume depleted state. Provided that sufficient fluid resuscitation has been undertaken, renovascular dopamine may promote a diuresis. Whether this approach can prevent the onset of vasomotor nephropathy remains to be conclusively proven.

Vasoconstrictor agents have often been considered likely to contribute to renal vasoconstriction and have therefore been contraindicated in prerenal azotaemic states. Contrary to this opinion, recent experience in sepsis has suggested that pressor agents such as noradrenaline, by increasing systemic blood pressure, may encourage a diuresis more effectively than inotropes such as dobutamine that have peripheral vasodilator properties.

In the patient with sepsis syndrome or primary cardiac failure the insertion of a pulmonary artery catheter to optimize preload, combined with the use of inotropes or pressors, may be necessary to ensure the preservation of renal function.

Overall mortality from acute renal failure remains in excess of 50 per cent, although there has been an indication of improved survival in recent years. Less than 10 per cent of patients with uncomplicated acute renal failure die, but in those with multiple organ system failure the mortality rate from acute renal failure rises to above 70 per cent, and may approach 100 per cent when four organ systems fail. Renal failure therefore represents a serious complication in the critically ill patient.

Once renal failure is established, additional renal insult must be avoided and consideration paid to the reduction in fluid and drug elimination. Acute renal failure may be grouped into four broad categories: ischaemic injury, nephrotoxicity, glomerular disorders, and vascular disorders (Table 4) 59.

Although more than 80 per cent of renal failure in the critical care unit patient will be due to vasomotor nephropathy (acute tubular necrosis) the clinician must remain alert to the possibility of other aetiologies. It may be necessary to rule out an obstructive uropathy or an active glomerulitis. Although intravenous urography with nephrotomograms will define renal size and detect caliceal and ureteric dilation of obstruction, the renal ultrasonogram visualizes the kidney without the risk of exacerbating renal failure, particularly in the diabetic patient. Retrograde urography may still be necessary to identify accurately the level of obstruction and facilitate the placement of stents. Occasionally, renal biopsy may be indicated when the aetiology is obscure and the urine sediment is consistent with a glomerulopathy.

The most immediate clinical problem for the patient who has recently developed renal failure is the maintenance of an appropriate intravascular volume in the face of a large obligate fluid input due to drug administration and total parenteral nutrition. By concentrating the nutrients, reducing drug diluent volumes, and using syringe pumps instead of volumetric pumps, fluid overload van be minimized. Hyperkalaemia can be temporarily controlled by the use of glucose and insulin infusion (1 unit soluble insulin to each 2 g of glucose) or ion exchange resins. Although high levels of urea and related metabolites are well tolerated in the ambulant or stable patient with renal failure, an early and aggressive approach to renal support with haemofiltration or haemodialysis is preferable in the patient with multiple organ system failure. Thus, the threshold for embarking upon haemofiltration or haemodialysis should be much lower in the critically ill than in the uncomplicated renal failure patient. Instead of waiting for volume overload, bleeding, uraemic pericarditis, hyperkalaemia, or acidosis, renal replacement therapy should be started early to maintain metabolic state as near normal as possible. Such use of renal replacement therapy is associated with an improved outcome, and the overall approach to patient management is greatly simplified once haemofiltration is established. Fluid balance is readily achieved and, provided adequate caloric intake is available, nitrogen intake does not need to be restricted. Full profile amino acid solutions are probably as effective as the special solutions of essential and branch chain amino acids. Energy requirements should be met using a balance of carbohydrate and lipid to deliver 45 to 50 kcal/kg.day. Where renal function is impaired for longer than 2 to 3 weeks some consideration has to be given to supplementation with vitamins and other essential elements such as zinc.

Recovery from a reversible renal pathology such as vasomotor nephritis is typically heralded by a polyuric phase followed by a plateau in serum creatinine lasting a few days, before creatinine levels consistently fall.

Renal replacement therapy can take the form of peritoneal dialysis, intermittent haemodialysis, continuous ultrafiltration, continuous haemofiltration, or continuous haemodiafiltration. Continuous arteriovenous or venovenous haemofiltration provide relatively simple and effective renal replacement therapy which is particularly well suited to the critical care patient. Although haemodialysis removes certain solutes, such as potassium, more efficiently, haemofiltration causes less haemodynamic disturbance and facilitates the regulation of fluid balance.

Peritoneal dialysis is suitable for patients with acute renal failure who have not had abdominal surgery and are not excessively catabolic. However, peritoneal dialysis has been largely superseded by haemofiltration, although it may still be preferred in patients suspected of carrying hepatitis B virus or HIV, or in very young children.

The need for renal replacement therapy must be evaluated in each individual patient. Although there are threshold values for hyperkalaemia (>6.0 mmol/l), blood urea (>40 mmol/l), or severe acidosis (pH < 7.2) which would justify renal replacement therapy, the overall clinical status, including the need to make room for drugs and nutrition, all need to be considered. The aim of renal replacement therapy is to remove uraemic toxins and maintain electrolyte, acid/base and fluid balance.

In the patient with uncomplicated renal failure, maintaining the plasma urea below 30 mmol/l would generally be considered acceptable. In the critically ill patient lower target values approaching 20 mmol/l are preferred.

Continuous haemofiltration is achieved by passing heparinized blood at flow rates of between 100 and 200 ml/min through a highly permeable haemofilter. The modern high-flux, biocompatible haemofilters, made from polyamide, polyacrylonitrile, or polysulphone, have negligible effects on platelets, neutrophils, or complement.

Plasma water passes through the haemofilter membrane and is drained into a collecting system (Fig. 1) 58. Blood cells and proteins are not filtered through the membrane and are returned to the patient. The haemofiltrate contains all of the water soluble components of plasma and thus enough urea and creatinine are removed to control the patient's biochemistry. At haemofilter blood flows of 200 ml/min, about 1 litre of filtrate will be produced each hour. Since the clearance of creatinine and urea by the membrane is close to 100 per cent, a filtration rate of 1 l/h (24 l/day) results in a clearance of about 17 ml/min, which is adequate for all but the most catabolic patient. Since the plasma concentration of potassium is very low, only about 100 mmol can be removed each day. Haemofiltration is therefore not the most efficient treatment for hyperkalaemia although any correction of metabolic acidosis lessens the risks associated with an elevated serum potassium.

Several techniques are available to the clinician depending upon the availability of local expertise and the needs of the patient. The simplest mode is spontaneous continuous ultrafiltration which requires arterial and venous access such as that provided by a Scribner shunt (Fig. 1(a)) 58. Filtered volume is not replaced and relief from fluid overload can be achieved quickly.

The volume of filtrate removed by the haemofilter can be replaced by a suitable haemofiltration replacement fluid which contains a lactate buffer. The replacement fluid may be infused through the return (venous) blood line or proximal to the haemofilter (predilution). The latter technique improves blood flow through the filter and increases haemofiltration efficiency. The reinfusion of a suitable haemofiltration fluid converts spontaneous continuous ultrafiltration to continuous arteriovenous haemofiltration (Fig. 1(b)) 58. The filtration rate is regulated by a gate clamp on the filtrate outflow tubing. Tightening of this clamp reduces the filtration rate as does raising the level of collection bag.

A feature of these two techniques is that as the blood pressure falls in response to fluid withdrawal, the rate of blood flow and therefore filtration is reduced. The absence of a blood pump further adds to the safety and simplicity of this technique.

An alternative to continuous arteriovenous haemofiltration is continuous venovenous haemofiltration (Fig. 1(c)) 58, which requires the insertion, by the Seldinger technique, of a double lumen venous catheter in the jugular, subclavian, or femoral vein. Infection of these catheters is a potential hazard and they should not be used for any other purpose. All access sites, whether arteriovenous or venovenous, should always be kept uncovered and in direct view of the attending nurse so that disconnections are immediately apparent. Pumped continuous venovenous haemofiltration can maintain clearances approaching 17 ml/min but requires close monitoring and safety alarms. The relative efficiency of this technique and ease of vascular access make it suitable for routine intensive care renal replacement.

Continuous arteriovenous or venovenous haemodiafiltration (Fig. 1(d)) 58 offers yet another alternative. Haemofiltration replacement fluid is pumped by a volumetric infusion pump in a countercurrent direction, to the filtrate side of the haemofilter membrane. Dialysate fluid flow rates of only 1 to 2 l/h are required to obtain clearance rates that are over twice that of haemofiltration alone.

The entire extracorporeal circuit must be thoroughly flushed through with 1 to 2 l of heparinized saline (5 units of heparin/ml) to exclude all air from the system. This is a simple procedure for continuous arteriovenous haemofiltration but rather more involved with continuous venovenous haemofiltration or haemodiafiltration, which require blood pumps. Suitably trained nurses or specialists in intensive care can undertake all aspects of haemofiltration if specialized renal unit personnel are unavailable. Some recommend administration of 5000 units of heparin to the patient, but this is generally not necessary. Maintenance of fluid balance is achieved by either adding replacement fluid intermittently each hour or by using a mechanical or microprocessor controlled balance which will ensure a preset fluid balance (such as zero, − 500 ml, − 1000 ml). With the mechanical balance system, fluid balance goals can be changed at the end of each 4-l cycle if necessary. Microprocessor controlled balances allows continuous adjustment of fluid balance. Gradual slowing of filtration rate, as evidence by a lengthening of the cycle time usually indicates impending failure of the filter due to blood clots.

Continuous infusion of 500 to 1000 units of heparin per hour into the haemofilter input blood line will generally prevent clotting within the extracorporeal circuit. Since there is significant filtration of the heparin, this produces very little systemic anticoagulation: the dose of heparin represents a compromise between the need to prevent the haemofilter clotting and the avoidance of systemic effects. In patients at great risk from bleeding the use of low molecular weight heparin or prostacyclin should be considered. In the uraemic patient, the extracorporeal circuit can often be maintained without any anticoagulant for some days. The viability of haemofilter circuits is difficult to predict even with heparin. Some remain functioning for 5 or 6 days while other repeatedly fail after 24 h.

Most haemofiltration replacement fluids contain lactate as a buffer. Conversion to bicarbonate is impaired in patients with renal and hepatic disease and correction of the metabolic acidosis may be therefore be slower than expected. If necessary, non-buffered replacement fluid can be used and bicarbonate infused separately to correct the acidosis. In contrast, a marked metabolic alkalosis may be precipitated by lactate buffer, particularly if the patient is small and the filtration rates high.

Hypothermia (temperature <35°C) can develop with high replacement fluid flow rates unless the fluid is warmed with a blood warmer. Even with warmed fluid, the fever associated with sepsis may be abated, suggesting that circulating cytokines are actively removed by the filtration process.

There are few data regarding the clearance of commonly used non-protein bound drugs, including antibiotics. Efficient venovenous haemofiltration provides the equivalent of a creatinine clearance of almost 20 ml, a value that is often quoted in drug package inserts as a guide for drug dosage. Wherever possible drug levels should be monitored. Once haemofiltration is established dietary nitrogen restriction is no longer necessary and standard parenteral and enteral nutrition preparations can be used.

The anion gap can be a useful guide in resolving biochemical disturbances in the critically ill patient. The anion gap is calculated by sum of the concentrations of the principal anions, chloride and bicarbonate, subtracted from the sodium concentration i.e. [Na⫀] − ([Cl&supminus;] + [HCO&sub3;&supminus;). The normal value is in the range 7 to 18 mmol/l. Low values are seen in a variety of conditions, including hypoalbuminaemia, bromide or iodide toxicity, and myeloma. However, the changes are sufficiently small as to render the recognition of low anion gaps to be of little clinical use. In contrast, a raised anion gap, due to excess of anions other than chloride and bicarbonate, may be of clinical value. These anions include lactate, citrate, and acetate, present usually as the sodium salts, but may also be inorganic or organic acids. When this is the case the increased anion gap is associated with a metabolic acidosis (Table 5) 60. Not all metabolic acidosis is associated with a raised anion gap, however. When acidosis is caused by the primary loss of bicarbonate or the gain of hydrochloric acid, there are no foreign anions present and the sodium ions are balanced by appropriate though abnormal amounts of chloride or bicarbonate (Table 6) 61.

In high anion gap acidosis, the fall in bicarbonate approximates the rise in anion gap. In mixed acid–base disorders the change in anion gap may be more or less than that expected from the change in bicarbonate: an oversimplistic approach to the application of anion gap theory may therefore mislead the clinician. Any consensus of opinion reached by the use of anion gap analysis must pass the test of being consistent with the overall clinical picture.

The factor of 2.0 applied to the sodium concentration (in mmol/l) allows for the osmolal contribution of the anions that balance sodium cations. Estimates of osmolal gap produce values of 0 &plusmin; 6 mosmol/l. The clinical utility of the osmolal gap rests upon identifying patients with toxins, drugs or certain plasma constituents present in excessive amounts in their serum. Ethanol, methanol, and ethylene glycol are common toxins associated with a raised osmolal gap. The osmolal gap will be raised in patients receiving a mannitol for the treatment of raised intracranial pressure and the size of the gap parallels the plasma mannitol concentration.

As with the anion gap, the osmolal gap may help elucidate the underlying cause of certain electrolyte disturbances. However, neither are a substitute for a careful and systematic clinical evaluation of the patient.

Cameron JS. Acute renal failure in the intensive care unit today. Intensive Care Med, 1986; 12: 64–70.

Emmett M, Narins RG. Clinical use of the anion gap. Medicine, 1977; 56: 38–54.

Gennari FJ. Serum osmolality uses and limitations. N. Engl J Med, 1984; 310: 102–5.

Golper TA, Pulliman J, Bennett WM. Removal of therapeutic drugs by continuous arteriovenous haemofiltration. Arch Intern Med, 1985; 145: 1651–2.

Henderson LW, Colton CK, Ford CA. Kinetics of haemodiafiltration: II. Clinical characterisation of a new blood cleansing modality. J Lab Clin Med, 1975; 85: 372–91.

Kramer P, Wigger W, Rieger J, Mathaei D, Scheler F. Arteriovenous haemofiltration: a new and simple method for treatment of overhydrated patients resistant to diuretics. Klin Wochenschr, 1977; 55: 1121–2.

Rainford D, Sweny P, eds. Acute renal failure. London: Farrand Press, 1990.

Smithline N, Gardner KD Jnr. Gaps anionic and osmolal. JAMA, 1976; 236: 1594.

Steiner RW. Interpreting the fractional excretion of sodium. Am J Med, 1984; 77: 699–702.

Stevens PE, Riley B, Davies SP, Gower PE, Brown EA, Kox W. Continuous arteriovenous haemodialysis in critically ill patients. Lancet 1988; ii: 150–2.

Zarich S, Fang LST, Diamond JR: Fractional excretion of sodium: exceptions to its diagnostic value. Arch Intern Med, 1985; 145: 108–12.