A balanced view of balanced solutions

The present review of fluid therapy studies using balanced solutions versus isotonic saline fluids (both crystalloids and colloids) aims to address recent controversy in this topic. The change to the acid-base equilibrium based on fluid selection is described. Key terms such as dilutional-hyperchloraemic acidosis (correctly used instead of dilutional acidosis or hyperchloraemic metabolic acidosis to account for both the Henderson-Hasselbalch and Stewart equations), isotonic saline and balanced solutions are defined. The review concludes that dilutional-hyperchloraemic acidosis is a side effect, mainly observed after the administration of large volumes of isotonic saline as a crystalloid. Its effect is moderate and relatively transient, and is minimised by limiting crystalloid administration through the use of colloids (in any carrier). Convincing evidence for clinically relevant adverse effects of dilutional-hyperchloraemic acidosis on renal function, coagulation, blood loss, the need for transfusion, gastrointestinal function or mortality cannot be found. In view of the long-term use of isotonic saline either as a crystalloid or as a colloid carrier, the paucity of data documenting detrimental effects of dilutional-hyperchloraemic acidosis and the limited published information on the effects of balanced solutions on outcome, we cannot currently recommend changing fluid therapy to the use of a balanced colloid preparation.


Introduction
Normal saline solution has been used for over 50 years in a multitude of clinical situations as an intraoperative, resus citation and maintenance fl uid therapy. Neither normal nor physiological, however, saline solution is still a standard against which other solutions are measured.
Much attention has been given recently to so-called balanced solutions such as Ringer's lactate, and more recent derivatives. Colloids prepared in balanced electrolyte solutions have also been developed, alongside colloids in isotonic saline.
As one might expect, excessive use of saline has been observed to result in hyperchloraemic acidosis -which has been identifi ed as a potential side eff ect of salinebased solutions. Th ere is debate about the morbidity associated with this condition, although some consider the associated morbidity is probably low. It has been suggested that the use of balanced solutions may avoid this eff ect.
Th is acidosis eff ect was reviewed and highlighted in the British Consensus Guidelines on Intravenous Fluid Th erapy for Adult Surgical Patients [1]. Th ese guidelines clearly recommend the use of balanced crystalloids rather than saline -but they make no specifi c recommen dations regarding colloids, implying that they could be either standard or balanced. Th e publication of these guidelines has provoked strong reactions. In a British Medical Journal editorial, Liu and Finfer comment: ' Although administration of normal saline can cause hyperchloraemic acidosis, we do not know whether this is harmful to patients. Adopting this guideline is unlikely to harm patients, but may not have any tangible benefi t' [2].
Others have reviewed the physiological eff ects of acidosis. Handy and Soni noted that 'Th ere is little evidence that in 50 years of normal saline usage, there has been signifi cant morbidity from the use of this fl uid' [3]. Liu and Finfer continue: 'Th e danger in providing consensus guidelines endorsed by specialist societies is that clinicians may feel pressured to adopt interventions that may, in the longer term, be found to cost more and to do more harm than good. We agree with the recently expressed view that unless recommendations are based on high quality primary research, then perhaps guidelines should be avoided completely, and clinicians would be better off making clinical decisions on the basis of primary data' [4].
Given the obvious controversy that exists based on the interpretation of the available information, the entire topic should clearly be reviewed again. Accordingly, the present article reviews the available literature comparing balanced solutions with isotonic saline fl uids (both crystal loids and colloids) and investigates the scientifi c basis that should be taken into account in any future guidelines or recommendations.

The acid-base equilibrium: Henderson-Hasselbalch versus Stewart
It is vital to determine the mechanism for an acid-base disturbance in critically ill patients in order to administer appropriate treatment. Th e Henderson-Hasselbalch equation is still the standard method for interpreting acid-base equilibrium in clinical practice [5]: Th is equation describes how plasma CO 2 tension, plasma bicarbonate (HCO 3 -) concentration, the apparent dissociation constant for plasma carbonic acid (pK) and the solubility of CO 2 in plasma interact to determine plasma pH. Th e magnitude of the metabolic acidosis is generally quantifi ed by the base defi cit or base excess, which is defi ned as the amount of base (or acid) that must be added to a litre of blood to return the pH to 7.4 at a partial pressure of carbon dioxide (PCO 2 ) of 40 mmHg. Th e main conse quence of infusion of isotonic saline is a dilution of bicar bo nate. Th e dilution of albumin may also play a minor role. Accordingly, the observed disorder is reported as a dilutional acidosis, associating a base defi cit with a high chloride concentration.
A diff erent approach (the strong ion approach) to acidbase equilibrium was developed in 1983 by Stewart to account for fl uctuation of the variables that independently regulate plasma pH [6]. He proposed that plasma pH is aff ected by three independent factors: PCO 2 ; the strong ion diff erence (SID), which is the diff erence between the charge of plasma strong cations (sodium, potassium, magnesium and calcium) and strong anions (chloride, sulphate, lactate and others); and the sum of all anionic charges of weak plasma acids (A tot ), which is the total plasma concentration of nonvolatile buff ers (albumin, globulins, phosphate). More advanced explanations are available in a recent review by Yunos and colleagues [7]. Th e Stewart equation may be written in a similar form to the Henderson-Hasselbalch equation [8]: At the usual pH of plasma, part of the albumin complex carries a negative charge, which could therefore play a role in buff ering H + ions. Th e same applies to phosphate, although the concentration of phosphate in the plasma is too low to provide signifi cant buff ering. Accordingly, the Stewart approach emphasises the role of albumin, phosphate and other buff ers in acid-base equilibrium.
Th e Stewart approach can distinguish six primary acidbase disturbances instead of the four diff erentiated by the Henderson-Hasselbalch equation. Th is strong ion approach also provides a more comprehensive explanation of the role of chloride in acid-base equilibrium.
Th e SID of isotonic saline being 0, the infusion of large quantities will dilute the normal SID of plasma and decrease pH. Hyperchloraemic metabolic acidosis is there fore a decrease in SID associated with an increase in chloride. Th e Stewart equation also shows that the infusion of isotonic saline will also dilute albumin and decrease A tot , which tends to increase pH. Using the Stewart equation, a balanced solution with a physiological SID of 40 mEq/l would induce a metabolic alkalosis. Morgan and Venka tesh have calculated that a balanced solution should have a SID of 24 mEq/l in order to avoid this induction [9]. It should be noted that balanced solutions using organic anions (such as lactate, acetate, gluconate, pyruvate or malate) have an in vitro SID equal to 0, similar to isotonic saline. In vivo, the metabolism of these anions increases the SID and also decreases the osmolarity of the solution.
Th is equation, while comprehensive, is still complex for common use if used in its entirety, but a simplifi ed Stewart approach can be used to make a graphical interpretation of the acid-base equilibrium. Th is approach takes into account the eff ects of the most important substances aff ecting equilibrium: sodium, potassium, calcium and magnesium minus chloride and lactate. In this approach, the apparent SID is defi ned as follows (see Th e two acid-base equilibrium approaches are mathematically equivalent but are completely diff erent from a conceptual point of view. Both are subject to criticism. Th e Stewart approach has been criticised for incor porating bicarbonate as a dependent variable, the result of a calculation, while it is obvious that physiologically bicarbonate plays a central role and is regulated mainly by the kidneys. Conversely, the Henderson-Hasselbalch approach is centred on bicarbonate, which may refl ect the physiological reality better. In the dilution concept, metabolic acidosis following resuscitation with large volumes of isotonic saline is attributed to dilution of serum bicarbonate. Th e Stewart approach rejects this explanation, however, and off ers an alternative that is based on a decrease in SID. Th is mechanistic explanation is questioned by several authors for fundamental chemical reasons [10,11]. If correct, the Stewart approach is valid at the mathematical level but does not provide mechanistic insights. Th e quantifi cation and categorisation of acid-base disorders using the Stewart approach, however, may be helpful in clinical practice to understand some complex disorders. Th e intra-erythrocyte and interstitial space buff ers are not taken into account in either approach. Th ese buff ers play a major role in acid-base equilibrium and must be included, particularly in the case of isotonic saline administration [12] (Figure 2).
Th e most important consideration is the cause of the acidosis. Acidosis is often the consequence of a physiological disturbance or an iatrogenic event. Th e diffi culty lies in separating the eff ects of the pathophysiology driving the acidosis. For example, metabolic acidosis can be a sign of organ distress due to hypoperfusion or hypoxia (for example, shock, ketoacidosis or kidney disease) [3]. Th is will produce profound physiological eff ects that are all readily ascribed to the acidosis rather than to its cause. Correction of the pathology may correct the acidosis, but correction of the acidosis solely is unlikely to aff ect the pathology. Th erefore it is important to understand the mechanism causing the acidosis.

Defi nitions
In the present article, in an attempt to better describe disorders and solutions, we have used the following terms.

Dilutional-hyperchloraemic acidosis
Th e term dilutional-hyperchloraemic acidosis is used instead of dilutional acidosis or hyperchloraemic metabolic acidosis, in order to reconcile both theories (Henderson-Hasselbalch and Stewart). In reality, many articles on hyperchloraemic metabolic acidosis do not report SID changes and only mention base excess variations and chloride concentrations.

Isotonic saline
Isotonic saline describes the main property of 0.9% saline solution. Th e solution is neither normal, abnormal nor unbalanced. Sodium and chloride are partially active, the osmotic coeffi cient being 0.926. Th e actual osmolality of 0.9% saline is 287 mOsm/kg H 2 O, which is exactly the same as the plasma osmolality.

Balanced solutions
Used generally to describe diff erent solutions with diff erent electrolyte compositions close to plasma compo sition, balanced solutions are neither physiological nor plasmaadapted. Table 1 presents the electrolyte compo sition of commonly available crystalloids. Table 2 presents the electrolyte composition of commonly used colloids.

Quantitative eff ects of isotonic saline infusion on acid-base equilibrium
Th e eff ects of isotonic saline infusion are illustrated by Rehm and Finsterer in patients awaiting intra-abdominal surgery [13]. Patients received 40 ml/kg/hour of 0.9% isotonic saline, a total of 6 litres isotonic saline in 2 hours. Th e apparent SID decreased from 40 to 31 mEq/l, chloride signifi cantly increased from 105 to 115 mmol/l and a decrease in base excess of approximately 7 mmol/l was observed. Th ese data perfectly illustrate   [18] of the actual HCO 3 concentration (top curve), composed of the calculated HCO 3 values (fi lled triangles) from plasma dilution, plus the increments from the plasma proteins (Pr), the erythrocytes (E), and the interstitial fl uid (ISF) with corresponding buff ers. Adapted from Lang and Zander [12].
dilutional-hyperchloraemic acidosis following infusion of large volumes of isotonic saline in clinical practice. Before determining the clinical relevance of dilutional hyperchloraemic acidosis, it is important to quantify the respective contribution of crystalloids and colloids.
Several studies have reported the biological eff ects following infusion of crystalloids alone [14,15]. Boldt and colleagues provide an interesting illustration of the eff ects following infusion of very high doses of crystalloid (isotonic saline versus Ringer's lactate) [16]. In patients undergoing major abdominal surgery, they reported the intraoperative infusion of 8 litres of crystalloids, followed by a further 10 litres of postoperative infusion in 48 hours (Table 3), resulting in a total dose of 18 litres of either Ringer's lactate or isotonic saline. As shown in Table 3, these extreme doses of isotonic saline were associated with moderate and transient eff ects on acid-base equilibrium: a decrease in base excess of 5 mmol/l that lasted for 1 or 2 days.
A number of studies have also reported and compared the eff ects following the infusion of large volumes of colloids and crystalloids with isotonic saline or balanced solutions [17][18][19][20][21][22].
In patients undergoing abdominal surgery, Boldt and colleagues used colloid (HES 130/0.42) either in a balanced solution or in an isotonic saline solution. In this study, a total balanced fl uid therapy (colloid and crystalloid) was compared with a total isotonic saline-based strategy [18]. It is interesting to note that, despite the large volumes of fl uid used (>6 litres), the diff erence in chloride concentration was +8 mmol/l and the diff erence in base excess was -5 mmol/l between the groups (Table 4). Th ese changes were similar to or lower than those in other studies (Table 4).
O'Dell and colleagues established that there is an inverse linear relationship between chloride load and base excess [23]. According to this relationship, to decrease base excess by 10 mmol/l in a typical 70 kg   Overall, these studies suggest that when patients are treated with a combination of isotonic saline-based colloids and crystalloids, the eff ects on acid-base equilibrium are limited. Base and colleagues used a diff erent fl uid strategy in patients undergoing cardiac surgery. HES 130/0.4 was administered either in a balanced solution or a saline solution. Th e two groups also received the same balanced crystalloid, Ringer's lactate [17]. Th e chloride concentration at the end of surgery was 110 mmol/l in the group receiving HES in a balanced solution, compared with 112 mmol/l in the isotonic saline-based solution. Th e diff erence is statistically signifi cant but is not clinically relevant. Base excess decreased in both groups, but the maximum diff erence between the groups at any time point was around 2 mmol/l.
Th e respective role of crystalloids and colloids on acidbase equilibrium is perfectly illustrated by Boldt and colleagues in elderly patients undergoing abdominal surgery [24]. Th ree diff erent strategies were used: Ringer's lactate, isotonic saline, and HES 130/0.4 plus Ringer's lactate. Th e chloride and sodium loads and the eff ect on base excess are shown in Figure 3. Although the colloid used in this study was supplied in an isotonic saline carrier, overall the impact on base excess was similar to that of Ringer's lactate alone and remained within the normal range.
Overall, these studies suggest that large volumes of saline will increase the chloride concentration and reduce base excess in a dose-dependent manner, with the peak eff ect occurring a few hours post infusion. Th e eff ect is temporary, and levels generally return to normal within 1 or 2 days. When fl uid therapy is based on colloids in an isotonic saline carrier, together with a balanced crystalloid like Ringer's lactate, the eff ects on acid-base equilibrium appear limited. Owing to a lack of published clinical experience, it remains to be seen whether patients with pre-existing metabolic acidosis are more aff ected due to a reduced buff ering capacity. Transient isotonic saline-induced reduction of base excess should be considered when interpreting the acid-base status in unstable patients.

Is dilutional-hyperchloraemic acidosis clinically relevant?
While it is clear that dilutional-hyperchloraemic acidosis exists, it is important to examine whether it has any eff ect on organ function. Th e kidney, gastrointestinal tract and coagulation system have often been mentioned as possible targets.

Eff ects of dilutional-hyperchloraemic acidosis on renal function
Animal studies suggest that chloride may have eff ects on the kidney including renal vasoconstriction, an increase in renal vascular resistance, a decrease in glomerular fi ltration rate and a decrease in renin activity [25][26][27][28]. At normal and slightly high concentrations, however, the eff ects are small [29].
Diff erences in osmolarity between Ringer's lactate and isotonic saline have to be taken into account to understand the eff ects on renal function and urine output. Th e osmolarity of Ringer's lactate is 273 mOsm/l. In dilute physiological solutions, the values of osmolality and osmolarity are interchangeable. In vivo, however, the osmolality of Ringer's lactate is only 254 mOsm/kg. Th is discrepancy is due to incomplete ionisation of the solutes in Ringer's lactate. On the contrary, isotonic saline, which is completely ionised, has an osmolality similar to the calculated osmolarity of 308 mOsm/l. Compared with the osmolality of normal serum (285 to 295 mOsm/kg), therefore, Ringer's lactate is clearly hypotonic while 0.9% saline is isotonic. In a study with human volunteers, Williams and colleagues tested the hypothesis that infusion of large volumes of Ringer's lactate or isotonic saline may have diff erent eff ects on renal function and urine output [15].
Th ere was a signifi cant diff erence in mean time to urination, Ringer's lactate solution being associated with the shorter time to fi rst urine output. In fact, in the Ringer's lactate group a decrease in serum osmolality probably inhibited the release of antidiuretic hormone. Th e resulting diuresis of hypotonic urine causes the serum osmolality to return quickly to normal.
Th ese changes in osmolarity must be taken into account in the interpretation of clinical studies comparing Ringer's lactate with isotonic saline. In a similar study by Reid and colleagues, time to fi rst micturition was shorter in the Ringer's lactate group, and was associated with a decreased urine osmolarity [30]. Th is suggests that the free water clearance adjusted to changes in osmolality. In their study, the isotonic saline group retained a greater proportion of the sodium load than did the Ringer's lactate group, which may account for the diff erence in fl uid retention. Th ese results emphasise that diff erences in osmolality between balanced solutions and isotonic saline must be taken into account in the interpretation of renal function parameters such as time to micturition and urine output. O'Malley and colleagues compared Ringer's lactate with isotonic saline in patients undergoing renal transplantation. Th is study found that recipients undergoing kidney transplants had greater acidosis and higher potassium concentrations if they were given isotonic saline as opposed to Ringer's lactate [31]. Th ese eff ects are the consequence of acidosis mobilising potassium from the intracellular space in patients where renal function is unable to compensate for these changes. It is worth noting that there was no adverse eff ect of isotonic saline on renal function. Th ere is no evidence of this eff ect in other studies comparing isotonic saline with balanced salt crystalloids [31].
Boldt and colleagues published a series of articles in which a totally balanced strategy (balanced crystalloid and balanced colloid) was compared with a standard treatment (isotonic saline and colloid in isotonic saline carrier) ( Table 4). In one study, in patients undergoing major abdominal surgery there was no signifi cant diff erence in urine output and in serum creatinine on the fi rst postoperative day [18].
Another study in elderly patients undergoing cardiac surgery also reported no major impact on renal function [19]. For up to 60 days following surgery, there was no diff erence between the groups regarding plasma creatinine concentration. Levels of neutrophil gelatinase-associated lipocalin (NGAL) were also measured. Th ere was a small increase on the fi rst day after surgery in the isotonic saline-based group, but levels in both groups were nearnormal by the second day. Overall NGAL values were extremely low (around 20 ng/ml), signifi cantly below the threshold of 150 ng/ml that is considered an indicator of acute kidney injury.
Finally, a study investigating the eff ects of two colloid strategies in patients undergoing cardiopulmonary bypass was also reported by Boldt and colleagues [20]. Albumin in saline carrier was compared with an HESbased colloid in balanced solution. Th ere was no significant diff erence in serum creatinine following surgery; and although an increase in NGAL of 15 ng/ml was observed in the albumin group, values remained within the normal range.
It has been claimed that NGAL is an early biomarker of acute renal injury [32], but NGAL values can vary considerably even in the absence of adverse kidney eff ects. Using the same test as was used in the two previously mentioned studies, Wagener and colleagues reported rises of 165 to 1,490 ng/ml in cardiac surgery patients with and without acute kidney injury [33]. Th ese results suggest that values reported by Boldt and colleagues are very low and, although the type of solution signifi cantly infl uenced the NGAL values, there is no indication of signifi cant impairment in renal function.
In conclusion, no signifi cant diff erences in creatinine variations have been reported and only slight diff erences in NGAL, not clinically relevant, were observed. From these results one may conclude there is no convincing diff erence between isotonic saline strategies and balanced strategies in terms of renal function.

Eff ects of dilutional-hyperchloraemic acidosis on coagulation and bleeding
Data from in vitro studies suggest that balanced solutions may have fewer negative eff ects on coagulation parameters [34,35]. Th e authors acknowledge the inherent problems of in vitro studies, however, which include the eff ects of haemodilution, calcium dilution and the absence of physiological components such as the endothelium. Owing to these signifi cant limitations, no clinically relevant conclusions can be drawn from in vitro studies.
Clinical studies provide more relevant insights. Boldt and colleagues compared the eff ects of very high doses (around 18 litres in 48 hours) of Ringer's lactate and isotonic saline in patients undergoing abdominal surgery (Table 3) [16]. Th ere was no signifi cant diff erence in coagulation tests and in blood loss between the groups.
Waters and colleagues compared Ringer's lactate with isotonic saline in patients undergoing repair of abdominal and thoracoabdominal aortic aneurysm (Table 5) [36]. Th ere was a small but nonsignifi cant diff erence in blood loss in favour of the Ringer's lactate group (Table 5). Th ere was no signifi cant diff erence in the use of packed red blood cells or fresh-frozen plasma between the two groups. Th e only statistically signifi cant diff erence was a higher volume of platelet transfusion in the saline group. When all blood products were summed, the use of blood products was signifi cantly higher in the saline group. Both groups included patients with thoracoabdominal aneurysm, however, which may account for the high variability in blood loss and transfusion requirements. No signifi cant diff erence in morbidity or mortality was reported.
Studies investigating the use of colloids also found no diff erence in blood loss between colloids in balanced solutions and those in isotonic saline solutions. Kulla and colleagues did not observe diff erences in blood loss patients undergoing abdominal surgery, and all other coagulation parameters were not signifi cantly diff erent between the two groups [21]. A similar study by Boldt and colleagues also found no diff erence in blood loss between the two groups ( Table 5) [18].
Only one study reported diff erences between isotonic saline-based and balanced colloids. Comparing HES 130/0.42 in balanced solution with albumin in saline as a priming solution for cardiopulmonary bypass, Boldt and colleagues reported small but signifi cant diff erences in coagulation (Rotem, Pentapharm, Munich, Germany) in favour of the balanced HES. Th is observation was associated with signifi cantly lower blood loss [20]. Similarly, use of blood products throughout and after surgery was signifi cantly lower in the HES group (Table 5). Th e number of patients in each group was very small (n = 25), however, given that coagu lation and bleeding in cardiac surgery may be highly variable. A recent study performed by the same investi gators in the same setting (cardiac surgery), comparing a balanced HES with albumin, did not confi rm these results [22].
In conclusion, there is little evidence that large volumes of isotonic saline have a signifi cantly detrimental eff ect on coagulation, blood loss or transfusion.

Eff ects of dilutional-hyperchloraemic acidosis on gastrointestinal function
Several studies have investigated the eff ects of dilutionalhyperchloraemic acidosis on gastrointestinal function with controversial results.
Williams and colleagues reported that healthy volunteers receiving saline experienced more frequent abdominal discomfort than those receiving Ringer's lactate [15]. Wilkes and colleagues investigated the eff ects of 6% hetastarch in a balanced carrier plus Ringer's lactate versus hetastarch in saline plus isotonic saline in elderly surgical patients [37]. Th e only diff erence related to gastrointestinal function was a small diff erence in the gastric CO 2 gradient, which showed a larger increase in the saline group. Th e diff erence is small and probably not clinically relevant (0.3 ± 1.5 kPa in the Ringer's lactate group compared with 1.0 ± 0.7 kPa in the saline group), but may suggest a better gastric mucosal perfusion in the Ringer's lactate group. A nonsignifi cant trend towards more nausea and vomiting was observed in the saline group.
Moretti and colleagues reported diff erent results. Patients were randomised into three groups to compare the eff ects of hetastarch in isotonic saline, of hetastarch in balanced solution and of Ringer's lactate on postoperative outcomes [38]. While there was no signifi cant diff erence in the incidence of nausea and use of antiemetics between the hetastarch groups, both were signifi cantly lower than in the Ringer's lactate group (Table 6). Th e authors concluded that intraoperative fl uid resuscitation with colloids, compared with crystalloids, improved postoperative recovery with regards to postoperative nausea and vomiting. Th ese results suggest that fl uid volume may be more important than composition. Several other studies suggest that intraoperative crystalloid restriction may be associated with an improve ment in gastrointestinal function and a decrease in postoperative complications [39][40][41].
In conclusion, there is not suffi cient evidence from the available literature to suggest that dilutional-hyperchloraemic acidosis has a clinically relevant eff ect on gastrointestinal function. Some degree of intraoperative crystalloid restriction and colloid use may, however, be associated with an improvement in gastrointestinal function and outcome.

Eff ects of dilutional-hyperchloraemic acidosis on mortality
Metabolic acidosis is often associated with adverse outcomes; however, it is important to diff erentiate between the eff ects of acidosis itself and the conditions that cause it. In the clinical setting, metabolic acidosis arises from diff erent causes, within which hyperchloraemia may play a role. Following trauma, for example, major metabolic acidosis has been reported in relation to severe hypovolaemia, tissue hypoxia and shock. In this situation, it is very diffi cult to determine the specifi c role of isotonic saline administration and the potential impact of other mechanisms on outcome [42][43][44][45].
Experimental studies may therefore be useful to understand the impact of fl uid therapy on outcome. Short-term survival was measured in a model of experimental sepsis with rats resuscitated with a balanced hetastarch, Ringer's lactate or isotonic saline [46]. In terms of mortality, Ringer's lactate was no better than isotonic saline. Th e best survival was observed in the colloid group, suggesting that a colloid strategy may be favourable in sepsis.
Gunnerson and colleagues carried out an observational, retrospective review of hospital data of 9,799 critically ill patients admitted to the intensive care unit [47]. Th ey selected a cohort (n = 851) in which clinicians ordered a measurement of arterial lactate level; 584 patients (64%) had a metabolic acidosis, either related to lactate, a strong ion gap or hyperchloraemia. Mortality was highest in patients with lactate acidosis (56%). In patients with dilutional-hyperchloraemic acidosis, mortality was the same as in the control group without metabolic acidosis ( Figure 4). From this observational study, it may be concluded that patients with hyperchloraemic acidosis were not associated with an increased risk of mortality compared with critically ill patients without metabolic acidosis.

Study Group Blood loss (ml) P value between groups
Noritomi and colleagues performed an observational study in 60 patients with severe sepsis and septic shock [48]. In this group of patients, mortality was signifi cantly associated with an increased inorganic ion diff erence. Th e diff erence in plasma chloride concentrations between survivors and nonsurvivors was minimal (3 mEq/l). Of note in the Rivers study, a diff erence in base excess of 5 mEq/l after 6 hours of treatment was observed between optimised patients and controls, with a concomitant reduced mortality in the patients receiving the highest dose of colloids and crystalloids (6 litres versus 4.5 litres) [49]. In their study, however, several confounding variables might have infl uenced the acid-base status and the mortality is more related to the cause of acidosis rather than to transient dilutional-hyperchloraemic acidosis.
In a prospective observational study set in the paediatric intensive care unit following cardiac surgery, Hatherill and colleagues documented that dilutionalhyperchloraemic acidosis was associated with reduced requirement for adrenaline therapy [50]. It is suggested that dilutional-hyperchloraemic acidosis is a benign phenomenon that should not prompt escalation of haemodynamic support.
In another prospective observational trial, Brill and colleagues studied 75 consecutive surgical intensive care patients with base defi cits >2.0 mmol/l. Patients were divided into those with hyperchloraemic acidosis and those with acidosis from other causes. Th ere were no signifi cant diff erences in age, Acute Physiology and Chronic Health Evaluation II scores, or volumes of resuscitation between the hyperchloraemic group and the remaining patients. Th ere were four deaths (10.8%) in the hyperchloraemic group and 13 deaths (34.2%) in the remaining patients (P = 0.03). Th e authors concluded that hyperchloraemic acidosis is a common cause of base defi cit in the surgical intensive care unit, associated with lower mortality than base defi cit secondary to another cause [51]. Maciel and Park have reported similar results [52].

Conclusion
Th e current review has presented an extensive analysis of all available studies using balanced solutions. We conclude that dilutional-hyperchloraemic acidosis is a side eff ect, mainly observed after the administration of large volumes of isotonic saline as a crystalloid. In this particular setting, however, the eff ect remains moderate  . Hospital mortality associated with type of metabolic acidosis. Mortality associated with the major ion contributing to the metabolic acidosis. Hospital mortality associated with the various causes of metabolic acidosis (standard base excess (SBE) <-2). Mortality percentage is mortality within each subgroup, not a percentage of overall mortality. Lactate indicates that lactate contributes to at least 50% of the SBE; SIG, strong ion gap contributes to at least 50% of SBE (and not lactate); hyperchloraemic, absence of lactate or SIG acidosis and SBE <-2; none, no metabolic acidosis (SBE ≥-2 mEq/l). P <0.001 for the four-group comparison. Adapted from Gunnerson and colleagues [47].
and relatively transient (24 to 48 hours), and is minimised with the use of olloids, whatever the nature of the carrier. From the available literature, the evidence for adverse eff ects of dilutional-hyperchloraemic acidosis on organ function, morbidity or mortality remains of small importance. In addition, the use of colloids together with crystalloids allows a reduction of the total volume of fl uids used and considerably limits the chloride load.
In view of the substantial experimental and clinical information on the effi cacy and safety of various colloids, including third-generation HES (HES 130/0.4), and of the limited published information on the eff ects of balanced solutions on outcome, we cannot changing to a new generation of colloids until there is evidence suggesting genuine detriment from existing fl uids and clear evidence of benefi t with new solutions.

Key messages
• Th e term dilutional-hyperchloraemic accurately defi nes a decrease in base excess, or a decrease in SID, associated with hyperchloraemia and a normal anion gap. • Isotonic saline describes the main property of 0.9% saline solution. It is neither normal, abnormal nor unbalanced. Balanced solutions is a general term to describe diff erent solutions with diff erent electrolyte compositions. • Dilutional-hyperchloraemic acidosis is a moderate and relatively transient side eff ect, minimised or avoided by limiting crystalloid administration through the use of colloids in any carrier. • No convincing evidence for clinically relevant adverse eff ects of dilutional-hyperchloraemic acidosis on morbidity or mortality can be found. • Following extensive review, owing to the limited published information on the eff ects of balanced solutions on outcome, the change of practice from colloids in isotonic saline to balanced colloid use cannot be recom mended.