Albumin was initially compared to saline in ICU patients who required fluid resuscitation (a comparison of albumin and saline for fluid resuscitation in the ICU [SAFE study]) [11]. The decision to initiate fluid resuscitation was made by the treating physicians. However, patients should have at least one of the following clinical objective signs of intravascular volume depletion: tachycardia >90/min, low systolic blood pressure or mean arterial pressure, low central venous pressure or pulmonary capillary wedge pressure, delayed capillary refill time, or decreased urine output. The SAFE study enrolled 6,997 patients, and the patients were randomly allocated to the albumin (4% albumin) or saline group (normal saline [NS]). Baseline characteristics were well balanced between the two groups except for a higher central venous pressure in the albumin group than in the NS group (9.0±4.7 mm Hg vs. 8.6±4.6 mm Hg, P=0.03). ICU admission diagnoses (trauma, sepsis, and acute respiratory distress syndrome) did not differ between the two groups. However, 28-day mortality was not significantly different (albumin group, 20.9% vs. NS group, 21.1%). Interestingly, higher mortality was observed in the trauma subgroup of the albumin group than that of the NS group (relative risk [RR], 1.36). This was mainly due to the considerable death among patients with traumatic brain injury (TBI) in the albumin group. This result raised concerns about the use of albumin in patients with TBI, and triggered a post-hoc follow-up study (SAFE-TBI) study [12]. The primary purpose of SAFE-TBI study was to validate whether albumin was unsafe for patients with TBI, even when controlling for important baseline confounders that the SAFE study did not consider, and to compare long term functional outcomes at 24 months because the follow-up period of 28 days in the SAFE study was too short to compare clinically meaningful differences in patients with TBI. The investigators identified 515 patients (492 from the SAFE study and 23 from the SAFE screening database) and excluded 55 patients. As a result, 460 patients with TBI were allocated to two groups (albumin group, n=231 vs. NS group, n=229), and they were followed for up to 24 months. As expected, mortality at 24 months was higher in the albumin group (33.2% vs. 20.4%, P= 0.003). Interestingly, most deaths occurred within 28 days (85.9% in albumin group and 85.7% in NS group), and the difference mainly originated in patients with severe TBI (Glasgow Coma Scale [GCS] score, 3–8) (RR in albumin group, 1.88). In patients with mild to moderate TBI (GCS score, 9–12), the mortality rates were similar between the albumin group (16.0%) and NS group (21.6%) (P=0.5) [12]. To investigate the underlying mechanism of albumin-related excessive mortality in patients with severe TBI, another sub-study from SAFE-TBI study was performed [13]. From the SAFE-TBI database (n=460), patients who underwent intracranial pressure (ICP) monitoring (n=321) were screened, and hourly ICP values were compared between the two groups. In general, the ICP did not differ significantly between the albumin and NS groups over 14 days. However, the use of albumin treatment was associated with an increase in ICP and with additional interventions to treat ICP crisis, particularly sedatives, analgesics, and vasopressors during the first week after injury. The primary mechanism of ICP aggravation in the albumin group was disruption of the blood–brain barrier. In patients with severe TBI and contusions, infused albumin was extravasated, leading to aggravated osmotic pressure in the brain parenchyme [14]. It is not clear whether the albumin itself or tonicity was responsible for this. In the SAFE-TBI study, the concentration of albumin was 278 mOsm/kg (Albumex) in 4% solution, and was hypotonic. Another meta-analysis on the use of hypertonic albumin (20%–25%) in TBI patients showed that albumin reduced mortality in four small controlled clinical trials [15]. The use of albumin in patients with TBI and the association of its tonicity with mortality is still under debate and requires further large-scale randomized controlled trials.

HES is a hydrolyzed derivative from natural starch, and diluted in saline or balanced solutions (Voluven or Volulyte) [16]. Many different HES products are available, varying in mean molecular weight, molar substitution (MS), and C2/C6 ratio. In general, small molecules below the renal clearance threshold (50–60 kDa) are rapidly removed [17]. Therefore, HES with a larger molecular weight tends to stay longer in the vascular spaces and play a more potent role as a volume expander. The MS ratio is also important. The MS ratio is calculated as the average number of hydroxyethyl residues per glucose subunit. For example, an MS of 0.6 indicates that there are six hydroxyethyl residues exist on average per 10 glucose subunits; hence, it is called a hexastarch [17]. Substituted glucose units are resistant to enzymatic breakdown by amylase. Therefore, a higher MS ratio indicates a higher resistance to degradation and greater potency as a volume expander. In addition, hydroxyethylation predominantly occurs at the C2 and C6 positions, and hydroxyethylation at the C2 position is more resistant to degradation. Therefore, a higher C2/C6 ratio suggests delayed degradation and clearance [17]. However, resistance to degradation does not always translate to improved efficacy and safety. Prolonged retention of HES in the vasculature may trigger coagulopathy or renal injuries [18,19].

Patients with sepsis or septic shock are more likely to require a large volume of fluid resuscitation to maintain their blood pressure and mitigate organ failure. Therefore, mortality and adverse renal events are typically compared to evaluate the efficacy and safety of HES in these patients. A German study (Efficacy of volume substitution and insulin therapy in severe sepsis; n=537) tested the effect of intensive insulin therapy compared with conventional insulin therapy and 10% pentastarch vs. Ringer’s lactate (RL) for fluid resuscitation using 2×2 factorial design. Regardless of the effect of the insulin on sepsis, HES use was associated with a higher rate of acute renal injury (34.9% vs. 22.8%, P=0.002) and renal replacement therapy (RRT) than RL [20]. A post-hoc analysis showed a direct correlation between the cumulative dose of HES and both the need for RRT and the death rate at 90 days. In another randomized trial (n=804), HES and Ringer’s acetate were compared in patients with severe sepsis requiring fluid resuscitation in the ICU [21]. HES increased mortality or dependence on dialysis at 90 days compared with RL (51% vs. 43%, P=0.03). However, only one patient in each group was dependent on dialysis at 90 days. Therefore, the difference in primary outcome was due to an increase in mortality in the HES group. An additional clinical trial tested the efficacy and safety of HES in a large number of patients (n=7,000) who required fluid resuscitation in the ICU [22]. Although patients with TBI were included, those with documented intracranial hemorrhage on computed tomography were excluded. The mortality rate at 90 days did not differ between the HES and NS groups (18.0% vs. 17.0%, P=0.26). However, HES treatment significantly increased the risk of adverse renal outcomes and use of RRT (7.0% vs. 5.8%, P=0.04).

In line with a series of large clinical trials that did not show any mortality benefit of HES compared to crystalloids, but did increase the risk of renal injury or RRT, the Food and Drug Administration changed the labeling on mortality, renal injury, and bleeding risk with the use of HES on July 7, 2021 [23]. In addition, the European Medical Agency also withdrew marketing authorizations in European countries in June 2022, and HES is no longer available in those countries [24].

Many neurointensivists believe that “normal physiologic saline” is a misnomer. The origin of this name can be traced back to the Dutch chemist Hartog Jacob Hamburger, who tested the optimal concentration of saline to replace human serum, limiting hemolysis. He found that the freezing point of human serum was –0.52 °C, which was identical to that of 0.9% saline; therefore, it was termed as “normal and physiological.” He did not publish a paper at that time, but the result was published in 1896 by W.S. Lazarus-Barlow, who explained that Dr. Hamburger suggested 0.92% saline as being “normal” for mammalian blood [25]. Later in 1921, Hamburger reiterated the significance of 0.9% saline as normal and physiological saline [26]. How 0.9% saline became the main crystalloid in modern medicine is unclear; however, 0.9% saline has much higher concentration of sodium and chloride than plasma. Therefore, infusion of 0.9% saline inadvertently leads to hyperchloremia and hypernatremia. Unwanted hyperchloremia has two detrimental effects including deterioration in renal function and metabolic acidosis in addition to the effect of acidity due to 0.9% saline (pH 5.6). As Peter Stewart highlighted, three independent factors play important roles in the acid-base physiology: PaCO₂, total weak acid concentration, and strong ion difference (SID). SID represents the difference between strong cations (Na⁺, K⁺, Ca⁺, and Mg⁺) and strong anions (Cl^-, lactate, and uric acid). To maintain electrical neutrality, an increase in chloride concentration leads to a decrease in SID, resulting in a compensatory reduction in negatively charged bicarbonate, provided that the PaCO₂ and total weak acid concentrations remain constant. Therefore, infusing a large volume of 0.9% saline results in hyperchloremic metabolic acidosis.

Balanced crystalloids or buffered crystalloids, are solutions that have sodium, potassium, and chloride content similar to that of extracellular fluid and help replace a part of chloride anions with alternative anions, such as lactate, acetate, and gluconate [27]. Some examples of balanced crystalloids are RL, Hartmann’s solution, and Plasma-Lyte.

In 1882, Dr. Sydney Ringer published an article on frog heart contraction [28]. He had been studying blood substitutes that could sustain heart function in frogs, with a focus on sodium and potassium. He found that by infusing 0.75% saline solution, he could keep a frog’s heart beating [29]. One day, his technician, Mr. Fielder, was on leave, and Dr. Ringer had to prepare the experiment on his own, but he failed to obtain the same results. When his technician came back to work, Dr. Ringer found that Mr. Fielder used new river tap water instead of distilled water for reconstituting 0.75% saline; thus, he speculated that inorganic solutes in the new river tap water played a significant role in keeping the frog’s heart beating. The following year, he published an additional paper showing that a mixture of 0.75% sodium bicarbonate, 0.1% calcium chloride, and 1% potassium chloride was the optimal solution to keep the heart beating [30].

Alexis Frank Hartmann was a pediatrician who treated pediatric patients with diabetes and acidosis. He was attempting to develop an isotonic alkalizing solution and found that 1/6 molar sodium lactate was isotonic and easily sterilizable by heat [31]. Furthermore, infused lactate would be converted to bicarbonate in the liver via Cori’s cycle for approximately 2 hours. Because Hartmann’s solution is mainly based on Ringer’s solution with 1/6 molar lactate, it has lower chloride and additional calcium content than 0.9% saline. Plasma-Lyte is one of the balanced crystalloids, developed to closely represent actual plasma and thus a “physiologic” or “balanced” solution [32]. It does not contain calcium, but contains magnesium to correct hypomagnesemia, which is common in patients with severe illness. As a bicarbonate precursor, Plasma-Lyte has acetate and gluconate instead of lactate, as in Hartmann’s solution. Acetate is metabolized by most tissues in the body, but lactate requires an intact liver or kidney for bicarbonate production while consuming adenosine triphosphate [32]. Therefore, Hartmann’s solution does not correct metabolic acidosis as expected when patients are too sick with liver dysfunction.

With the increasing use of balanced crystalloids, many randomized clinical trials have compared the effects of balanced crystalloids with 0.9% saline. Most trials have focused on patients with sepsis because they are usually targeted for rigorous IV fluid therapy. In the 0.9% saline vs. Plasma-Lyte 148 (PL-148) for ICU fluid therapy (SPLIT) trial, 2,278 patients who required fluid therapy when clinically indicated were randomized to 0.9% saline and Plasma-Lyte [33]. The primary outcome was acute kidney injury (AKI) at 90 days, which did not differ between the Plasma-Lyte and 0.9% saline groups (9.6% vs. 9.2%, P=0.77). Additionally, in-hospital mortality and RRT use did not differ between the two groups. The SPLIT trial’s patient cohort had relatively mild disease severity compared with those in other clinical trials, as it included patients who had undergone elective surgery. Only 4% of the patients had sepsis, with an average mortality of approximately 7%. The average Acute Physiology and Chronic Health Evaluation (APACHE) II score in the SPLIT trial was 14, which was lower than that in the CHEST trial (17). Therefore, the beneficial effects of balanced crystalloids may not have been manifest in patients with mild disease severity [33].

The Isotonic Solutions and Major Adverse Renal Events Trial (SMART) enrolled 15,802 patients in five ICUs at Vanderbilt University who required IV fluid therapy [34]. Approximately 15% of the patients had sepsis, and 50% of the patients were admitted from the operating room. The primary outcome of major adverse kidney events (composite of death, new receipt of RRT, or persistent renal dysfunction) after 30 days was lower in the balanced crystalloid group than in the NS group (14.3% vs. 15.4%, P=0.04). In prespecified subgroup analysis, the difference in the rate of the primary outcome between the balanced crystalloid and NS groups was more significant in patients who received larger volumes of crystalloids and in those with sepsis [34]. In the subgroup of patients with sepsis, in-hospital mortality was 25.2% in the balanced crystalloid group and 29.4% in NS group (P=0.02). Six months later, a similar clinical trial was conducted to compare the effect of balanced crystalloids with NS in non-critically ill patients who visited the ER and were treated with IV fluids at the same institutions included in the SMART trial (Saline against Lactated Ringer’s or Plasma-Lyte in the Emergency Department [SALT-ED]) [35]. Among the 13,347 patients, the primary outcome (median hospital-free days to day 28) did not differ between the balanced crystalloid and NS group (24 days vs. 24 days, P=0.41). However, the rate of major adverse kidney events (composite death, new RRT, or persistent renal dysfunction) at 30 days was lower in the balanced crystalloid group (4.7% vs. 5.6%, P=0.01).

In the Balanced Solutions in Intensive Care Study (BaSICS) trial conducted in Brazil, 10,520 patients admitted to the ICUs with a risk factor for AKI were enrolled [36]. Patients in the balanced crystalloid and NS groups were administered an average of 1.5 liters of IV fluids during the study periods. The primary outcome of mortality at 90 days did not differ between balanced crystalloid and NS groups (26.4% vs. 27.2%, P=0.47). Prespecified subgroup analysis for primary outcomes did not show any major interactions, except for the presence of TBI and type of fluid. For patients with TBI, balanced crystalloid led to a higher mortality rate at 90 days than NS (31.3% vs. 21.1%). The exact mechanism is not clearly understood; however, the use of relatively hypotonic balanced crystalloids might be associated with an increase in mortality, which requires further studies with a large number of patients.

In the Plasma-Lyte 148 versus Saline (PLUS) study, 5,037 critically ill patients in a multicenter ICU requiring fluid resuscitation were enrolled [37]. Patients with TBI or those considered at risk of developing cerebral edema were excluded. The admission diagnosis were mainly sepsis (43%) and trauma (8%). The primary outcome (mortality at 90 days) did not differ between the Plasma-Lyte group and NS group (21.8% vs. 22%, P=0.90). There was no interaction between the pre-specified subgroups and treatment fluid. NS infusion was associated with low pH and high chloride levels, but had no adverse effect on renal outcomes.

Given that multiple clinical trials showed heterogenous results regarding mortality and major adverse renal events, a meta-analysis of six studies with a low risk of bias, involving 34,450 patients, was performed [38]. The RR of all-cause of mortality in the Plasma-Lyte group at 90 days was 0.96 (95% CI, 0.91–1.01), which was not statistically significant. Although the 95% CI crossed 1.0, the authors concluded that the estimated effect of using balanced crystalloids in a heterogeneous population of critically ill adults ranged from a 9% relative reduction to a 1% relative increase in death at 90 days. In addition, it is highly likely that the average treatment effect of using balanced crystalloids reduced the mortality rate [38]. Another meta-analysis using the data of 34,685 patients (mean age, 58.8; men, 57.9%) showed that compared with the use of saline, the use of balanced crystalloids was associated with a reduced in-hospital mortality rate (17.3% vs. 16.8%; odds ratio [OR], 0.962; 95% CI, 0.909–1.019; absolute difference, –0.4 percentage points [–1.5 to 0.2]) [39]. The need for new RRT was lower in the balanced crystalloids group (5.6%) than in the NS group (5.9%) (OR, 0.931; 95% CI, 0.849–1.020; absolute risk difference, –0.4 percentage points [–1.3 to 0.1].