Choice of fluids in critically ill patients

The following section discusses the characteristics of most existing fluid solutions. The chemical composition of many of the solutions currently on the market is presented in Table 1.

Colloids contain macromolecules such as hydroxyethyl-starch (HES), gelatin, dextran, or albumin. In the past colloids were thought to be distributed primarily in the intravascular space and were therefore considered 3–4 times more effective than crystalloids for restoring intravascular volume. Clinical evidence supports the assumption of higher intravascular retention of colloids, albeit not to such extent. Administration of 1400–1800 ml of gelatin, albumin, and HES increases cardiac index by 25–44% in surgical patients while administration of the same amount of saline (1800 ml) does not affect cardiac index [16]. Clinical hemodynamic stabilization also seems to occur more rapidly and with smaller volumes of colloids compared to crystalloids [17]. Unfortunately, many studies yielding such evidence were not designed for this purpose, which limits the validity of their findings.

Today it is clear that the ratio of intravascular to administered volume of colloids is usually only 1:1.2 [4, 16‐19], far less than previously believed. Large multicentre, randomised trials have shown ratios < 1:2 [16‐19]. Furthermore, many trials noting decreased transfusion requirements with the use of colloids are being criticised for bias, as fluid therapy was often determined by the treating clinicians [17‐19].

The CMDh statement suggests that dextrans may be used as alternative fluid solutions in routine clinical practice [29]. This recommendation is somewhat questionable given the paucity of data regarding dextrans to-date.

Early trials studying EGDT either used no colloids at all [44] or were not explicit regarding the specific fluid solutions used [45‐47]. The 6S [19], VISEP [17], CHEST [4], and CRYSTMAS [16] studies included no fluid solutions that contain dextrans. In the CRISTAL trial only five of the 1414 patients receiving colloids were treated with solutions containing dextrans [48]. A search of PubMed using the keywords “dextran/TI AND volume/AB” (1998 to January 2018) yielded only 17 studies that describe the use of dextrans in humans and most of these were small studies focusing on dextran-based hyperoncotic therapy.

Summary statement:

Balanced crystalloids are generally the solutions of choice for intravascular fluid resuscitation of hypovolaemic patients. The evidence against isotonic saline remains inconclusive but the possible risks associated with its use are not balanced by any advantage in therapeutic efficacy or cost at this time. Colloids are probably more effective than crystalloids, albeit not as effective as theoretically expected. However, no additional benefit has been conclusively proven for colloids over crystalloids. Therefore, if a decision has been made to administer colloids, it should always follow crystalloid administration. The solutions to consider after failed treatment with crystalloids should be either albumin or MFG. Regardless of the choice of colloids to be used, colloid administration should be considered rescue therapy and remain limited to profound, acute hypovolaemia.

As noted above, there is accumulating evidence that specific disease states may require different fluid therapy. In the section below the data supporting this statement is presented for specific disease conditions often seen in the ICU (i.e. sepsis, major abdominal surgery, ARDS trauma and AKI). The guidelines published regarding fluid administration in these disease conditions are summarised in Table 2. Additional file 2 reports some of the most relevant articles retrieved by the systematic search.

Lactic acidosis is a major metabolic side effect of sepsis. As noted above, intravenous administration of 0.9% saline may cause iatrogenic hyperchloremic acidosis [12, 49]. Hyperchloremia has been associated with increase in mortality in both septic and non-septic patients [50]. However, most studies examining this issue were retrospective, which precludes derivation of a meaningful causative association between the two. Studies comparing solutions with high versus low-chloride concentrations have yielded conflicting results thus far. Reduced rates of mortality and AKI have been described with balanced solutions [12, 13, 15, 49, 51] therefore until more information from RCTs is available, balanced solutions remain preferred over 0.9% saline for the treatment of hemodynamically unstable septic patients.

ARDS was initially considered an inflammatory protein-rich pulmonary edema accompanied by leakage of protein-rich fluids into the interstitial space. The resultant increase in lung weight was thought to generate atelectasis with eventual impairment of lung mechanics and gas exchange [55, 56]. However, ARDS has both inflammatory edema and hydrostatic components [55, 57, 58]. Development of pulmonary hypertension may lead to an increase in hydrostatic pressure [55]. Activation of the renal aldosterone-angiotensin system during mechanical ventilation also generates high increased intrathoracic pressure which causes water and salt retention [59, 60]. Fluid loading may improve hemodynamics and oxygenation but it may also worsen lung aeration in patients with lung inflammation through several mechanisms [61]. Moreover, a positive fluid balance in patients with ARDS may increase mortality rate [62].

Data about the best type of fluid in patients with ARDS are scarce. A recent meta-analysis investigated the effect of colloids versus crystalloids in patients with ARDS. Three trials were included for a total of 206 patients. All the included studies compared albumin versus saline. The meta-analysis found improved oxygenation but no survival benefit in patients treated with albumin versus crystalloid [63]. However, the risk of bias of included trials ranged from unclear to high and the sample size was very low.

Summary statements:

Fluid administration is part of the perioperative routine in both elective and urgent major abdominal surgery but these two situations could not differ more. Elective major abdominal surgery is often accompanied by bowel preparation [70‐72], preoperative cardiac assessment when indicated and is performed on a patient that is hemodynamically stable and adequately hydrated. Conversely, patients undergoing urgent abdominal surgery often suffer severe intravascular fluid depletion due to both intestinal and extra-intestinal losses (e.g. vomiting, extra-vascular leakage), are often hemodynamically unstable, and have usually undergone little preoperative assessment.

Elective surgery - The sparse literature addressing perioperative fluid administration in patients undergoing major abdominal surgery refers to elective patients [73]. Although mechanical bowel preparation is no longer recommended [74], many patients still undergo drug-induced bowel preparation. Similarly the evidence-based recommendation to allow ingestion of clear fluids up to 2 h before surgery is often translated to fasting from midnight on the day before surgery [75]. Such practice may induce dehydration and electrolyte imbalance despite institution of corrective hydration.

In this clinical scenario, intraoperative hydration is generally titrated to cover the fluid deficit resulting from bowel preparation and fasting as well as routine fluid maintenance (2–3 ml/kg/h). With adequate preoperative preparation however, the fluid deficit in these patients rarely exceeds 2.5% of body weight. Yet, traditional rehydration during surgery has been shown to result in administration of 7 l of fluid on the day of surgery and a weight gain of 3–6 kg [76‐78]. Such practice has led to the current speculation regarding the impact of perioperative fluid administration (both volume and type) on patient physiology.

One ongoing treatment dilemma is whether adding vasopressor therapy to fluid administration is beneficial since such practice may decrease the amount of fluid administered. An early meta-analysis of intra-operative hemodynamic optimization achieved by combining fluids and vasopressors compared to fluids alone showed a decrease in both renal and gastrointestinal complications, but later multicentre trials have yielded mainly controversial results [79‐81]. Most of these studies follow patients either throughout admission or to 28 days after surgery. However, none present any data regarding post-operative fluid management, which may have determined the outcomes sought during this time frame.

Regarding the choice of fluids, most discussion still surrounds the issue of crystalloids versus colloids [82]. While newer data does not suffice as yet to support the use of colloids, neither does it suggest that risk is increased. Conversely, there is some evidence that gastro-intestinal outcomes may even be slightly better with colloids [83]. This finding is supported by animal studies suggesting that goal-directed colloid fluid therapy increases microcirculatory blood flow and tissue oxygen tension in healthy and injured peri-anastomotic colon compared to goal-directed or restricted crystalloid fluid therapy [84]. With regards to a direct comparison between balanced crystalloid solutions versus normal saline, even less literature exists. An RCT comparing these solutions in major abdominal surgery demonstrated that balanced solutions caused less electrolyte disturbances, acid-base disequilibrium and increases in NGAL levels and were associated with a stronger anti-inflammatory effect [85].

Urgent surgery - Patients undergoing urgent abdominal surgery often present with sepsis or septic shock. Therefore, the principles guiding fluid administration in sepsis should also guide perioperative fluid administration. An average patient with a hollow viscus perforation who presents to the department of emergency medicine is likely to receive at least 1–2 l of crystalloids before surgery and several litres more during induction of anaesthesia and throughout surgery. These should not be discounted when initiating fluid therapy in the ICU after surgery. The choice of fluids to be administered should be determined by timely information regarding acid-base and electrolyte balance with particular emphasis on avoidance of an unnecessary chloride load. In the setting of severe extravascular leakage, intravascular fluid repletion with crystalloids alone may decrease tissue capillary density, thereby worsening microcirculatory flow dynamics and oxygen delivery. An overload of crystalloid solution may decrease oncotic pressure and viscosity and exacerbate the inflammatory response [86]. Hence the importance of considering the type of fluid in further resuscitation.

Summary statements:

Recent years have seen some interesting changes in fluid management of trauma patients. Although severe bleeding is the lead cause of death in trauma patients [87], the European guidelines for management of major bleeding and coagulopathy following trauma strongly recommend restricting volume replacement during initial trauma resuscitation [88]. This recommendation is based on data showing not only the feasibility of this approach but also its advantages in term of both process (e.g. hospital length of stay) and outcomes (e.g. survival) [89, 90].

For many years treatment with colloids was considered particularly efficacious in trauma patients. This concept was based on the assumption that the vascular endothelium remains intact after trauma (contrary to septic shock) [91]. Early experimental data supported this assumption, showing that resuscitation with HES 130/0.4 was superior to lactated Ringer [92]. In humans, an exploratory study of patients monitored with a pulmonary artery catheter showed similar hemodynamic outcomes with a lower volume of colloids than crystalloids [93]. However, subgroup analyses of trauma patients included in the RCTs comparing colloids and crystalloids have since failed to confirm this assumption with regards to wither mortality [48] or transfusion requirements [94]. In patients with TBI, mortality was actually higher with albumin than with saline, probably due to the greater increase in intracranial pressure observed during administration of albumin [95]. The European guidelines for management of major bleeding and coagulopathy following trauma therefore recommend isotonic crystalloids rather than colloids for initial resuscitation of hypotensive bleeding trauma patients [88].

Among crystalloid solutions, the respective roles of balanced solutions and saline remain controversial. Unsurprisingly, administration of lactated Ringer solution increases plasma lactate concentrations, whereas normal saline increases the base deficit [96]. In patients with severe TBI, hypotonic solutions (including lactated Ringer) should be avoided as they exacerbate cerebral edema. Conversely, balanced solutions cause less hyperchloremic acidosis than saline in these patients [97]. A RCT of adult trauma patients requiring blood transfusion, intubation, or operation within 60 min of arrival showed that pre-hospital resuscitation with Plasma-Lyte A yielded better acid-base status and less hyperchloremia 24-h after injury compared with saline [98]. To summarise - the use of balanced solutions seems promising for trauma resuscitation but currently remains under investigation [99].

There is ongoing debate regarding intravascular volume expansion with hypertonic saline in trauma patients [100, 101]. Han et al. randomized 294 patients with hypovolemic shock after trauma to receive 3% hypertonic saline (n = 82), 7.5% hypertonic saline (n = 80), or lactated Ringer (n = 84) [102]. Although baseline population characteristics were similar in the three groups, patients receiving hypertonic solutions (3% or 7.5%) were given about half the amount of fluids than those given lactated Ringer within the first hour, a difference which disappeared within 24 h. Some side effects (e.g. arrhythmia, hypernatremia) were more commonly observed in patients receiving 7.5% hypertonic saline, whereas others (e.g. renal failure, coagulopathy, pulmonary edema) were more prevalent among patients receiving lactated Ringer. The authors concluded that among the solutions examined 3% hypertonic saline has the best safety and efficacy profile [102]. With regards to colloids - the relative contribution of micro-circulatory abnormalities, endothelial dysfunction, local and systemic inflammatory processes and oxidative stress differs between hemorrhagic and septic shock. Decreased tissue perfusion is a major component of haemorrhagic shock whereas inflammatory processes are likely more predominant in septic shock. Hence the effects of HES may also differ. Evidence supporting the presence of a difference includes three meta-analysis showing that the use of HES was not associated with renal effects or clinically significant coagulopathy in the OR [27, 28, 103]. Similarly, no study found deleterious effects of HES in early resuscitation of trauma patients [104]. The European Medicine Agency decided that HES can still be used in surgical patients, and for management of hemorrhagic shock following an initial fluid challenge with crystalloids that has failed. However the clinician should be aware that colloids have not been associated with an improvement in survival in patients with trauma, burns or following surgery [31].

Summary statements:

Fluid administration is one of the cornerstones of prevention of AKI. As with any other body organ, the goal of fluid therapy in this clinical scenario is restoration of intravascular volume with secondary improvement in kidney perfusion pressures and a resultant improvement in local tissue oxygenation. However, the precise relationship between hypo/hypervolemia and AKI remains unclear. Studies differ substantially in case mix, fluid volumes and types and the timing of fluid administration. Regardless of the cause and/or mechanism of AKI, macro-circulation alterations (i.e. changes in renal blood flow) are associated with micro-circulation abnormalities (tissue perfusion), endothelial dysfunction, local and systemic inflammatory processes and oxidative stress [105]. The relative contribution of each of these to the development of AKI differs dependent on the cause of renal injury [105]. Whereas decreased tissue perfusion is a major component of haemorrhagic shock, inflammatory processes may be more predominant in AKI caused by septic shock [105]. Patients with sepsis seem particularly susceptible to the deleterious effects of hypervolemia on kidney function [106]. The importance of microcirculatory changes in this clinical scenario makes the choice of fluids all the more crucial.

In many patients stabilization of the systemic hemodynamic condition is not immediately accompanied by improvement in microcirculatory parameters. This situation may persist for hours or days, indicating long-lasting tissue ischemia [113]. Ongoing microcirculatory derangement is associated with increased morbidity and mortality, even when global hemodynamics are compensated [114]. Studies incorporating data on the effect of various fluids on the microcirculation are needed [115]. Dark-field microscopy, a new technique for measuring microcirculation, may offer important information regarding the microcirculatory changes occurring during administration of various fluids in specific disease conditions [116].