Intravenous Fluid of Choice in Major Abdominal Surgery: A Systematic Review

Background Intravenous fluid therapy plays a role in maintaining the hemodynamic status for tissue perfusion and electrolyte hemostasis during surgery. Recent trials in critically ill patients reported serious side effects of some types of fluids. Since the most suitable type of fluid is debatable, a consensus in perioperative patients has not been reached. Method We performed a systematic review of randomized control trials (RCTs) that compared two or more types of fluids in major abdominal surgery. The outcomes were related to bleeding, hemodynamic status, length of hospital stay, and complications, such as kidney injury, electrolyte abnormality, major cardiac adverse event, nausea, vomiting, and mortality. A literature search was performed using Medline and EMBASE up to December 2019. The data were pooled to investigate the effect of fluid on macrocirculation and intravascular volume effect. Results Forty-three RCTs were included. Eighteen fluids were compared: nine were crystalloids and nine were colloids. The results were categorized into macrocirculation and intravascular volume effect, microcirculation, anti-inflammatory parameters, vascular permeability, renal function (colloids), renal function and electrolytes (crystalloids), coagulation and bleeding, return of bowel function, and postoperative nausea vomiting (PONV). We found that no specific type of fluid led to mortality and every type of colloid was equivalent in volume expansion and did not cause kidney injury. However, hydroxyethyl starch and dextran may lead to increased bleeding. Normal saline can cause kidney injury which can lead to renal replacement therapy, and dextrose fluid can decrease PONV. Conclusion In our opinion, it is safe to give a balanced crystalloid as the maintenance fluid and give a colloid, such as HES130/0.4, 4% gelatin, or human albumin, as a volume expander.


Introduction
Many factors affect the outcome of elective surgery. Beyond the nature of the primary disease and the surgical factors, intravenous fluid therapy and inotropic drugs play a role in maintaining the hemodynamic status for tissue perfusion and electrolyte hemostasis [1,2]. e first intravenous fluid was invented about 200 years ago and evolved progressively during world wars to replace blood plasma by adding a complex sugar, protein, and colloids. [3] While believing that 0.9% sodium chloride (NaCl) is physiologic [4] and synthetic colloids are more effective than crystalloids in restoring plasma volume [5], they are widely used for resuscitation and maintenance purposes.
During recent decades, previous knowledge has been questioned. First, it was discovered that endothelial glycocalyx is the key structure to regulate microvascular hemodynamics, not oncotic pressure. ese studies have led to a revised Starling principle and a new approach to vascular fluid dynamics [6]. Second, many large trials in critically ill patients and subsequent meta-analyses report potential clinical side effects of IV fluids, especially 0.9% NaCl which is associated with the development of metabolic acidosis that results in kidney injury and increases mortality rates [7,8]. Synthetic colloids were also reported to cause side effects in kidney function and hemostasis [9][10][11]. erefore, the use of all hydroxyethyl starches (HES) became restricted in critical illness, renal failure, or coagulopathy by the European Medicine Agency in 2013 and in sepsis patients by the Survival Sepsis Campaign 2012 [12]. Balanced crystalloids are currently the first choice of resuscitation in critically ill patients [2]. is knowledge has been applied to perioperative settings even though the results are inconsistent from the small number of studies with different physiological changes [1].
Our goal was to systematically review the latest evidence of perioperative intravenous fluid therapy in major abdominal surgery with a focus on the types of fluids. Volume, administration technique, and surgery beyond the abdominal field were not reviewed.

Methods
e Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [13] guideline was used to conduct this systematic review. e full search is included in Appendix S1. Search strategies were adapted for the other databases. e applied restrictions were randomized controlled trials (RCTs); English only; age more than 18 years; and human trial. e year of publication was not restricted.

Study Selection/Inclusion and Exclusion.
Two levels of screening were used independently by two reviewers (SN and OA). First, the titles and abstracts of the included studies were screened and then the full text was reviewed. e included studies followed these inclusion criteria: (1) the population of patients was more than 18 years old and had undergone elective major abdominal surgery which was defined as any operation with peritoneum cavity exposure with resection and/or anastomosis; (2) intervention using two or more types or doses of intravenous fluids; and (3) the reported outcomes related to bleeding, hemodynamic status, length of hospital stay, and complications such as kidney injury, electrolyte abnormality, major cardiac adverse event, nausea, vomiting, and mortality. e excluded articles were duplicate or retracted studies, organ donor or animal studies, case reports, and case series. Any difference of opinion was resolved by discussion.

Data Extraction and Quality Assessment.
Two authors (SN and OA) extracted data into a data sheet. e extracted data included type of surgery, number of patients, fluid regimen, and the primary and secondary outcomes of each paper. e quality of the studies was independently assessed with the Cochrane tool to assess the risk of bias for RCTs [14] in the following domains: randomization method; allocation concealment; blinding; data completeness; and publication bias. Any disagreement was resolved by discussion.

Data Analysis.
e studies that compared the microcirculation and intravascular volume effect between colloid and crystalloid were selected for analysis. e total intraoperative volume to achieve hemodynamic parameters was used to represent the effect of crystalloid and colloid on microcirculation. e standard mean difference (SMD) was used to demonstrate the effect size of the types of fluid.

Identification of Studies.
e initial search in Medline (PubMed) and EMBASE (Ovid) identified 1,412 articles of which 421 are duplications. A further 938 were excluded because they did not fulfill the selection criteria. Fifty-six articles were selected for full-text reading. irteen articles were then excluded for the reasons described in Figure 1.
ree additional RCT studies [15][16][17] were added after a cross-reference review.  Table 1, and the types of study fluids are shown in Table 2. Full data sheet is shown in Table S1.

Quality of the Included Studies.
e results of the quality assessments of all studies are shown in Figure 2. Ten studies were considered high risk for blinding of participants due to safety issues. Two studies had a high risk of detection bias due to the open-label study. Most of the trials followed patients for a short period; therefore, missing data or lost to follow-up rates were low. Table  3 shows the results of 16 trials [15-17, 21, 23, 24, 27, 30, 41, 47-51, 56, 57] that reported the volume effects of fluids. Lavu et al. [30] compared 3% NaCl to lactated Ringer's solution (LRS) in patients who underwent pancreaticoduodenectomy using the fluid restriction technique and found lower perioperative intake in the 3% NaCl group (278 vs. 315 mL/kg; p value � 0.017) to maintain hemodynamic status. Six studies compared HES 130/0.4 (Volulyte [15][16][17] and Voluven [23,27,50]) to crystalloids. All of them reported good volume expansion according to stable hemodynamic parameters and needed both lower amounts of intraoperative fluids and inotropes to maintain hemodynamic status. Yates et al. [16] and Zhang et al. [56] who used goal-directed fluid therapy also reported colloids at crystalloid ratios of 1.6 : 1 and 1.67 : 1 to maintain the same hemodynamics in their trials. Vogt et al. [51] reported 6% HES 200/0.5 was an economical alternative to 5% human albumin for resuscitation because they had the same volume expansion effects although a lower serum colloid osmotic pressure was reported in the HES group. Two studies [24,48] compared HES 130/0.4 to HES 200/0.5 and found no differences in the hemodynamic parameters, but HES 200/0.5 in one study [21] had a prolonged INR (1.25 ± 0.19 vs. 1.18 ± 0.09; p value<0.05). Ragaller et al. [41] reported HES 200 in 7.2% NaCl could restore the hemodynamics faster than HES 200 in 0.9% NaCl using pulmonary capillary wedge pressure guidance. Two studies [21,57] compared 4% gelatin to 4.5% NaCl in 7.6% HES 40. Deng el al. [21] favored hypertonic NaCl-HES due to a more stable systemic vascular resistance index but Zhu [57] reported no significant differences in the hemodynamics.

Anti-Inflammatory Parameters and Vascular
Permeability. Table 5 shows the results of six trials [16,18,33,34,46,47] that studied the effects of colloids on the inflammatory process. Rittoo et al. [46,47] and Mahmood et al. [33,34] compared the effects of HES 200/0.62 (Elohes) and HES 130/0.4 (Voluven) to 4% gelatin (Gelofusine) in four RCTs that were performed in patients who underwent open aortic aneurysm repair. Using CRP, IL-6, and the lung injury score as biomarkers of the inflammatory process and the microalbumin/creatinine (Cr) ratio to indicate glomerular microvascular permeability, they reported that Elohes could decrease the inflammatory process by decreasing the CRP level which led to decreased microalbumin and von Willebrand factor (vWF) levels. Two studies compared medium to low molecular weight HES (HES 130/0.4 (Volulyte) [16] and HES 70/0.5 (Hespander) [18]) to balanced crystalloids. ey found that both solutions did not significantly decrease the inflammation parameters and did not alter vascular permeability [18]. Table 6 shows the results of 10 RCTs that reported the effects of fluids on renal function. Five trials [15][16][17][18]27] compared colloids to crystalloids and five trials [20,26,33,38,47] compared HES to other colloids or human albumin. Ando et al. [18] compared low molecular weight HES (HES 70/0.5 or Hespander) to acetate Ringer's solution and found a significant difference in the glomerular filtration rate (GFR) and the urinary microalbumin/Cr ratio from intraoperative evaluations to discharge. Kancir et al. [27] reported no renal toxicity when HES 130/04 (Voluven) was compared to normal saline solution (NSS) using serum neutrophil gelatinase-associated lipocalin and Cr as the parameters.

Renal Function and Electrolyte Imbalance (Balanced Solutions vs. Saline Solution).
[32] compared chloride-depleted glucose solution 5%  Table 10 shows four studies that reported the effects of fluid on PONV [19,22,36,42]. Chaudary et al. [19] used preoperative intravenous volume loading by LRS and hetastarch. ey found that both fluids decreased the rate of PONV and vomiting at four hours after operation compared to the IV restricted group. Two studies [36,42] that compared LRS to 5% dextrose in laparoscopic cholecystectomy showed that 5% dextrose fluid decreased the rate of PONV by more than 50%. One study [22] showed that 6% HES 130/0.4 decreased the vomiting rate compared to LRS (11% vs. 3%; p value � 0.266) in gastrointestinal surgery.

Discussion
Nowadays, the type of fluid therapy in perioperative settings is still debatable concerning the risks and benefits. e data from small single-center studies are still inconsistent. is systematic review compares each type of fluid for perioperative fluid therapy in major abdominal surgery. We found large heterogeneous outcomes due to various types of fluids compared (both colloids and crystalloids), variations in the fluid therapy protocols, types of abdominal surgery, and different parameters in outcome measurement. We attempted to group them into topics of interest.
Restoring and maintaining tissue perfusion is the primary goal of fluid therapy. In the present review using the parameters of lower fluid intake and greater hemodynamic stability, the macrocirculation or volume expansion effect showed more positive results in the colloid group compared to the crystalloid group with SMD of −0.638 (95% CI −1.137 to −0.138, p � 0.012). A lower fluid balance can decrease the incidence of complications from volume overload such as ileus, pulmonary edema, and impaired wound healing. [58] Complications from higher colloid intake were demonstrated in the Senagore trial [49] which was the first study to demonstrate goal-directed therapy using colloids compared to crystalloids. It was reported that the hetastarch group had a significantly higher volume compared to the crystalloid results that resulted in a high frequency of total postoperative complications and longer length of stay. e authors could not identify the cause of this event. When each colloid was compared, there were no differences in hemodynamic outcomes. In our opinion, each colloid has its initial volume expansion, colloid oncotic pressure, and half-life [59]. Hypertonic saline [30] (also with HES in hypertonic saline [21,41]) demonstrates good volume expansion compared to an isotonic saline (and HES in 0.9% NaCl). Hypertonic saline draws water out of the intracellular compartment and into the intravascular space leading to restoration of the circulating volume with smaller volumes of fluid and reduced intracranial pressure in cases associated with traumatic brain injury [60]. However, a large trial in prehospital trauma patients demonstrated a nonsignificant higher mortality rate in the hypertonic saline group [61] which may also lead to coagulopathy, increased acidosis, hypothermia, kidney injury, and immunologic disorder [62]. Yates [16] and Zhang [24] studied the colloid to crystalloid ratios of 1 : 1.6 and 1 : 1.67, respectively, in perioperative settings. ese ratios were higher compared to sepsis settings (1:1-1:3) [9,10,63] where the previously accepted ratio was 1 : 3 [60].
is result can be explained by endothelial dysfunction and capillary leakage in the postoperative period and sepsis [64].
Since stability of the vital signs and a decrease in the lactate level reflect macrovascular status, but not microcirculation [65], acceptance of these parameters may not be enough [66]. For example, abnormal splanchnic microcirculation may present in hemorrhage, sepsis, laparoscopic procedures, and in aortic cross clamp in aortic repair. Gastric mucosal hypoperfusion increases the production of mucosal CO 2 (PgCO 2 ) and decreases gastric mucosal pH (GpHi) [67]. ese two parameters were used to demonstrate microcirculation in abdominal aortic aneurysm repair [34,35,46] during resuscitation with HES of different molecular weights, gelatin, and crystalloids. HES 130 and HES 200 were reported to have good properties to maintain microcirculation, especially HES 200. Two studies in laparoscopic colonic surgery attempted to compare gelatin to 4.5% NaCl in 7.6% HES 40. One study supported gelatin [57] while the other reported no difference [21]. e reason they did not use the same variables to report the results was because Deng [21] claimed that gastric pH is disturbed by carbon dioxide pneumoperitoneum. Most of the included trials supported using colloids because they were better for microcirculation. ese results were supported by Wu et al. [65] who compared NSS, 3% NaCl, 4% succinylated gelatin, and 6% HES 130/0.4 in the hemorrhagic shock rat model. is animal trial reported that all of these fluids stabilized the vital signs and renal blood flow, but only HES, gelatin, and 3% NaCl restored intestinal microcirculation that was demonstrated by laser speckle contrast imaging. Human albumin and dextran also reported effects in supporting microcirculation [66]. e release of inflammatory mediators during surgery, such as C-reactive protein and tumor necrosis factor, is one of the causes of impaired endothelial barrier function due to an increase of large pores in the endothelial lining and induced glycocalyx shedding [60] which results in capillary leakage and volume maldistribution [64]. In this review, we included the in vivo anti-inflammatory effects of colloids, mostly from abdominal aneurysm repair because this   show significant effects in decreasing inflammation. is type of surgery may not cause as much inflammation as aortic repair. Anti-inflammatory effects of HES that were demonstrated in animal ischemic-reperfusion model [68] found that HES inhibited firm adhesion and decreased surface expression of CD11b of leukocytes. Chen et al. [69] reported that HES 130/0.4 decreased the levels of reactive oxygen species and tumor necrosis factor, while gelatin and HES 200 did not have such effects.
Most of the studies in this review compared crystalloids to colloids, and most of the colloids were HES. We found that every colloid demonstrated abnormal clot firmness and platelet function, but none of them had an abnormal coagulogram. Abnormality in the TEG tended to increase in medium molecular weight HES compared to the lower molecular weight HES [31]. Only two trials [27,43] reported that HES 130/0.4 (Voluven) increased intraoperative hemorrhage compared to a crystalloid. However, both trials were in urological surgery which has a high chance of bleeding due to the raw surface. ese results were similar to the meta-analysis by Rasmussen et al. [70] which reported on human albumin and both high and medium molecular weight HES. Higher bleeding was found in the subgroup of noncardiac surgery using HES 130 but no significant decrease was found in the amount of bleeding compared to HES 200. After a multivariate analysis, two trials [44,45] reported that TEG MA is the only factor that could reflect the amount of intraoperative bleeding. e mechanism of impaired coagulation by colloids was reported by de Jonge and Levi [71] through dilutional effect, molecular weight dependent reduction of vWF (acquired von Willebrand disease), factor VIII, and clot firmness. Gelatin and albumin had the least effect on coagulation among the colloid solutions [60].
For a comparison of crystalloids in perioperative renal function, the information available was mainly from kidney  transplantation patients who have a very high risk for renal failure. Most studies compared a balanced crystalloid to NSS and reported similar results. NSS caused hyperchloremic metabolic acidosis and hyperkalemia in the intraoperative to postoperative periods. However, we did not find a significant difference in mortality rate, AKI, graft rejection, or kidney dysfunction. However, higher early postoperative renal replacement therapy (RRT) within 48 hours was needed to treat hyperkalemia in the Weinberg et al. trial [54]. A metaanalysis by Cochrane [72], which included 1,096 participants from 18 RCTs in major perioperative settings, also reported that increased serum creatinine, hyperkalemia, negative base excess, and low serum pH occurred in the postoperative period but most subsided within postoperative day 1. No significant incidence of long-term kidney dysfunction or mortality rate was reported. is was contrary to the results of the SALTED trial [73] (study in noncritical illness) and SMART trial [8] (study in critical illness). In these trials, resuscitation used NSS which significantly increased major adverse kidney event (compound outcome) within 30 days without a significant difference in mortality rates. A large volume of NSS was related to renal vasoconstriction [60]. All of the above information was compiled into a guideline that supports using balanced crystalloids for peri-interventional volume substitution [2]. However, there were some situations where NSS was indicated, such as the presence of cerebral edema and gastric outlet obstruction [60]. Following a report of osmotic nephrosis in kidney transplant recipients after administration of HES [74], renal function after the use of colloids became a concern. However, two studies found that HES administration had better tubular and glomerular function based on the RIFLE criteria and the level of serum Cr [33,47]. Also, another study found a lower incidence of AKI grade 1 compared to gelatin [20]. Other trials showed no significant difference in AKI using HES compared to albumin [26,38] or HES compared to crystalloids [18,27]. e ALBIOS trial [63] reported no difference in mortality rate or RRT when albumin was compared to colloids in sepsis patients. Many large multicenter trials reported a higher incidence of RRT [9][10][11] and mortality rate [10] in the HES groups compared to crystalloids in sepsis patients, but they had defects in methodology [75]. In 2013, the CRISTAL trial [76] compared crystalloids (isotonic or hypertonic saline and balanced solution) to colloids (gelatin, dextran, HES, and albumin) in patients with hypovolemic shock. ey reported a lower mortality rate at 30 days and lower need of vasopressor therapy in the colloid group. No differences were found in the incidence of RRT and AKI. Furthermore, the subgroups of each type of colloid still showed a lower mortality rate. A recent meta-analysis [77] which compared colloids to crystalloids reported a higher incidence of RRT and mortality rate in the pentastarch group. In a subgroup analysis of sepsis, colloids led to a higher incidence of RRT and mortality rate, but these outcomes were not significant in cardiac and general surgery. is might be explained by the mechanism of AKI in surgery where volume loss can be improved by adequate volume replacement. However, in septic AKI, microvascular dysfunction is the key mechanism [77]. Larger endothelial pores allow colloids to leak into the tissues leading to organ dysfunction, especially in the kidney [60]. Colloids with higher molecular weights, for example, pentastarch, are more harmful due to the long metabolism time.
In two trials, a solution of 5% dextrose fluid was compared to a nondextrose fluid to determine the incidence of PONV [36,42]. e results showed that the 5% dextrose fluid decreased the incidence of PONV. However, these two trials were performed in laparoscopic cholecystectomy and the IV fluid protocol required a postoperative loading of <1000 mL. A meta-analysis which focused on PONV using dextrose fluids also included uncomplicated surgeries (laparoscopic gynecological surgery and laparoscopic cholecystectomy). e results showed a decreased incidence of PONV and the need for antiemetics by a mechanism related to hyperglycemia [78]. Colloids can decrease PONV, vomiting, and the need for antiemetics compared to crystalloids [19,22] by increased mucosal perfusion [78].
Many factors can affect bowel function and the type of fluid is also one of them. In our review, colloids could enhance bowel function compared to crystalloids, but may not have clinical significance (7 [16], 9 [56], and 13 [22] hours). All of the indicated trials used a goal-directed protocol and found a significantly lower need for fluid in the colloid groups. Using more crystalloids to achieve the same clinical volume effect as colloids can be detrimental. Crystalloids have a propensity to filter across the capillary membrane. A greater expansion of extravascular volume leads to intestinal mucosal edema and delayed recovery in postoperative ileus [79]. However, colloids can generate oncotic pressure to maintain fluid in intravascular component [80].

Limitations.
e present systematic review has some limitations that should be considered when interpreting the results. First, we had too many primary outcomes which resulted in including various types of fluids, volume administration protocols, and types of surgical procedures, which may account for the high heterogeneity of our results. Second, the trials included in this systematic review were often small and single-center studies. ird, the volume of a given fluid that may affect the outcome was not included in our review. Fourth, only major abdominal surgery was our surgical type. erefore, the results may not apply to other types of surgery. Fifth, some types of fluid (dextran and gelatin) were restricted in Europe and America which resulted in low reliability of the data obtained. Sixth, most of the participants included were ASA class I-III. erefore, it may be incorrect to apply this information to an emergency condition or higher ASA class. Finally, there were some flaws in our search methods which caused some important trials to be missed. e strength of this review was we had many primary outcomes which resulted in including various types of fluids, volume administration protocols, and types of surgical procedures.

Conclusion
Perioperative fluid management depends on many factors such as patient status, type of operation, type of fluid, and administration technique. e colloids had an individual volume expansion effect, maintained microcirculation, and can be used interchangeably. Every colloid affected clot firmness and clot formation time, but only dextran significantly increased bleeding. NSS resulted in perioperative hyperchloremic metabolic acidosis and hyperkalemia which may lead to RRT compared to a balanced crystalloid. No specific type of fluid increased the mortality rate.

Data Availability
All data collected in this research are available for review.

Conflicts of Interest
e authors declare that there are no conflicts of interest regarding the publication of this paper.