Elderly patients have a diminished ability to respond to the hemodynamic and metabolic demands of anesthesia and surgery (1). Inappropriate IV fluid management seems to be one of the greatest problems associated with poor outcome in elderly surgical patients (2). Perioperative intravascular volume optimization, guided by gastric tonometry or transesophageal Doppler ultrasonography, may improve outcome after major surgery (3,4). There is, however, little consensus as to the preferred choice of IV fluid. Crystalloid solutions satisfy basic fluid requirements and compensate for insensible losses during open surgical procedures. Colloid solutions maintain the circulating blood volume when used as plasma volume expanders. Sodium chloride 0.9% solution is often administered because it is isotonic with plasma and is initially distributed in the extracellular compartment. Its nonphysiologic levels of chloride, however, have been linked to the development of a metabolic acidosis and the possible impairment of splanchnic perfusion, as judged by reduced urine flow and abdominal discomfort in healthy volunteers (5). The relevance of this type of metabolic acidosis to patients is unclear (6,7). Our trial was designed to compare balanced crystalloid and colloid solutions with sodium chloride-based solutions in relation to postoperative biochemical status, acid-base balance, and clinically available indices of splanchnic and peripheral perfusion in elderly surgical patients.
This prospective, randomized, double-blinded and controlled clinical trial was conducted at two teaching hospitals (University College London Hospitals and Royal Free Hospital, London). Ethics committee approval was granted. Patients >60 yr old, ASA status I–III, who were scheduled for elective, open surgical procedures with an anticipated blood loss of >500 mL were screened for eligibility. Exclusion criteria were hypersensitivity to hydroxyethyl starch, neurosurgery, cardiac surgery, clinically significant renal dysfunction, congestive heart failure, preexisting bleeding diathesis, repeat surgery, the use of intraoperative cell salvage or intentional hemodilution, and the absence of written, informed consent.
After recruitment, patients were randomly allocated to the Balanced Fluid group or the Saline group. Randomization and stratification were instituted by using permuted blocks with a size of 4. All clinicians involved in the care of the patients were blinded to patients’ group allocation. Patients in the Balanced Fluid group received Hartmann’s solution and 6% hetastarch in balanced electrolyte and glucose injection (Hextend®; BioTime, Berkeley, CA), and patients in the Saline group were given 0.9% sodium chloride solution and 6% hetastarch in 0.9% sodium chloride solution (Hespan®; Abbott Laboratories, North Chicago, IL). Table 1 illustrates the composition of these fluids. Both study groups received a 500-mL bolus of colloid at the induction of general anesthesia. Initial fluid loading was followed by a crystalloid infusion of 7 mL · kg−1 · h−1 during surgery. Further IV fluids were given according to a clinical algorithm (Fig. 1).
Blood samples were obtained from an indwelling arterial cannula for biochemical and hematologic analysis and for determination of acid-base balance throughout the surgical procedure at 30-min intervals. Physiologic variables for hemodynamic assessment (heart rate, arterial blood pressure, central venous pressure) and indices of organ perfusion (gastric tonometry, urine output, core peripheral temperature gradient) were recorded every 30 min. For gastric tonometry we used an automated air method (Tonocap®; Datex-Ohmeda, Helsinki, Finland) (8). In brief, this technology uses a modified nasogastric tube with a gastric balloon that is automatically filled with room air at 10-min intervals. The air is withdrawn, and the partial pressure of carbon dioxide (the PgCO2) is measured with an infrared sensor that is often used for end-tidal CO2 measurement. The Pg-aCO2, or CO2 gap, can then be calculated after determination of the Paco2 from arterial blood gas analysis. The Pg-aCO2 is a measure of the relationship between gastric mucosal blood flow, metabolism, and alveolar ventilation. Automated air tonometry is the only validated and clinically available method of indirectly assessing gastric perfusion, with Pg-aCO2 values >1 kPa considered abnormal (8).
Sedative premedication was administered as required. General anesthesia was induced with an IV anesthetic and maintained with inhaled anesthetics. Intraoperative monitoring included electrocardiography, pulse oximetry, invasive blood pressure measurement, and central venous pressure measurement, which constituted part of the clinical algorithm for study drug infusion (Fig. 1).
The objective of this trial was to investigate whether the composition of intraoperatively administered fluids had an effect on postoperative chloride levels and other plasma electrolytes, glucose, acid-base balance, and measures of organ perfusion. A chloride level >110 mmol/L plus a base deficit of <−2.0 mmol/L was defined a priori as hyperchloremic metabolic acidosis on the basis of the reference ranges of the respective analyses.
Adverse events (AEs) and clinical interventions were recorded for 5 days after the operation by applying a standard AE reporting system often used in drug evaluation studies. After 28 days, patients were contacted by telephone to identify any further AEs. All observers and clinicians were blinded to the patients’ group allocation for the duration of the study period until after all case report forms were closed. The AE reporting was rigorous and externally audited, but it was primarily intended for safety screening purposes rather than outcome analysis, because Hextend is a new colloid solution not yet approved for clinical use in the United Kingdom. Special emphasis was placed on postoperative nausea and vomiting as a clinical manifestation of poor gastric perfusion, but it was anticipated that the power of our study would not permit the meaningful statistical evaluation of such events.
Clinical data management was performed independently by Clinical Data Care in Lund, AB, Sweden. The data management plan and all outcome variables were agreed with the statistical team a priori. No post hoc analyses are reported here. All data were entered twice (double-data entry) by using Microsoft Access 97 (Microsoft, Redmond, WA). Data were then transferred to SAS® version 6.12 by using SAS® ODBC (SAS, Cary, NC) under Windows 95.
Descriptive statistics, such as the number of observations, mean, sd, and 95% confidence intervals, were calculated for quantitative variables, whereas qualitative variables were described with frequency tables. All statistical tests were two-sided and performed at the 5% significance level. Postoperative chloride levels were analyzed with an analysis of covariance technique. For comparison of means or means of differences of secondary variables, Student’s t-test was applied. For secondary variables showing numbers of patients, Fisher’s exact test was used. It was calculated that 62 patients were needed to detect a 5% difference in the primary outcome variable (postoperative chloride levels) with a level of significance of 5% and a power of at least 99%.
After 47 patients were enrolled in the trial, physicians involved in the care of a study patient expressed concern that this patient might have experienced AEs as a result of intraoperative metabolic derangement. The study was halted, and the safety committee reviewed all patient records and, as a result, requested a blinded interim analysis of the primary outcome variables. This revealed a highly statistically significant difference between the study groups in plasma chloride and acid-base levels at the end of surgery. Because the power of the study was so great and it was considered possible that hyperchloremic metabolic acidosis might be associated with an increased incidence of patient AEs, the safety committee recommended that the study be stopped.
Twenty-three patients had been randomly allocated to the Balanced Fluid group and 24 to the Saline group. The groups were similar with regard to demographic patient characteristics and type and duration of surgical procedures. There were no significant differences in volumes of study fluids administered and postoperative hemodynamic indices (Table 2).
Two patients withdrew their consent before participation in the trial, one patient’s operation was canceled, one patient suffered a cardiac arrest before surgery, and one patient’s operation was rescheduled, leaving 42 patients to be analyzed (21 in the Balanced Fluid group and 21 in the Saline group).
The mean increase in chloride levels from baseline to the postoperative sample was significantly larger in the Saline group than the Balanced Fluid group (Table 3). The ranges for postoperative chloride levels were 106 to 125 mmol/L in the Saline group and 99 to 114 mmol/L in the Balanced Fluid group. Acid-base analysis demonstrated a significantly larger decline in all indices, i.e., standard base excess, plasma bicarbonate, and plasma pH, in the Saline group compared with the Balanced Fluid group (Table 4). In the Saline group, six patients developed marked acidosis with a standard base excess of <−4.0 mmol/L (range, −9.7 to 1.3 mmol/L) at the end of surgery, compared with none in the Balanced Fluid group (range, −3.5 to 3.1 mmol/L). In the Saline group, eight patients had a pH <7.32 at the end of surgery (range, 7.24 to 7.48), compared with none in the Balanced Fluid group (range, 7.34 to 7.49).
Fourteen patients in the Saline group (67%), but none (0%) in the Balanced Fluid group, developed hyperchloremic metabolic acidosis (P = 0.0001). These 14 patients were of similar age (71.3 ± 6.6 yr vs 72.4 ± 6.8 yr, P = 0.536) and had a similar body weight (67.3 ± 17.3 kg vs 70.3 ± 15.3 kg, P = 0.731), compared with all other patients. Duration of surgery in this group did not differ significantly from other patients (193 ± 70 min vs 184 ± 94 min, P = 0.942).
Gastric tonometry demonstrated a significantly larger increase in Pg-aCO2 from start to end of surgery in the Saline group (1.7 ± 0.5 kPa, end of surgery; range, 0 to 6.3 kPa) compared with the Balanced Fluid group (0.9 ± 1.1 kPa, end of surgery; range, −0.6 to 2.2 kPa;P = 0.0394), suggesting better gastric perfusion in the Balanced Fluid group. Differences in mean intraoperative urine output between the Balanced Fluid group (1.68 ± 1.3 mL · kg−1 · h−1) and the Saline group (0.96 ± 0.7 mL · kg−1 · h−1) did not reach statistical significance (P = 0.0787). Similarly, the perioperative change of core to peripheral temperature gradient at the end of surgery, although smaller in the Balanced Fluid group (−0.3°C ± 1.8°C) than the Saline group (−1.2°C ± 1.5°C), was not statistically different (P = 0.0859).
Descriptive analysis of the AEs, and in particular AEs related to the study drug, revealed a frequent incidence in patients in the Saline group (overall, 379 AEs in the Saline group versus 272 in the Balanced Fluid group; possibly or probably related AEs, 198 in the Saline group versus 116 in the Balanced Fluid group). The combination of nausea and vomiting was recorded 23 times in the Saline group compared with 12 times in the Balanced Fluid group. Eight patients (38%) in the Saline group experienced postoperative vomiting compared with three patients (14%) in the Balanced Fluid group. Seven patients (33%) in the Saline group were given rescue antiemetics compared with four patients (19%) in the Balanced Fluid group.
IV solutions with a balanced electrolyte formulation match the biochemical composition of human plasma more closely than sodium chloride-based fluids. Lactated Ringer’s solution was first constituted more than 100 years ago. Alexis Hartmann later added lactate to produce lactated Ringer’s solution (Hartmann’s solution) in the 1930s (9). Hextend is a new plasma volume expander containing 6% hetastarch suspended in a solution of balanced electrolytes, lactate, and glucose (10,11).
Although the acidifying potential of large volumes of sodium chloride solutions on plasma has been recognized in two randomized controlled trials (6,7) and a series of case reports and letters (12–18), controversy has surrounded both the etiology and the clinical relevance of this phenomenon. The term “dilutional acidosis” implied that plasma expansion and consecutive dilutional reduction of plasma bicarbonate were the underlying mechanisms (13–18). In contrast, the Stewart model emphasizes the importance of hyperchloremia resulting in a reduction of the strong ion difference (SID) (19). The SID is calculated as [Na+] + [K+] − [Cl−] − [lactate−]. This approach considers Paco2 and SID as independent variables regulating acid-base balance, whereas hydrogen ion and bicarbonate concentrations are dependent variables.
In our study, two-thirds of patients that were given Hespan and sodium chloride 0.9% developed a hyperchloremic metabolic acidosis at the end of surgery. This phenomenon was absent in patients who were given Hextend and Hartmann’s solution. When McFarlane and Lee (6) administered sodium chloride 0.9% or Plasmalyte 148 (Baxter Healthcare, Deerfield, IL), 15 mL · kg−1 · h−1, to 30 patients undergoing major hepatobiliary or pancreatic surgery, they demonstrated a tendency to metabolic acidosis in their Saline group. Scheingraber et al. (7) gave 24 patients undergoing gynecologic surgery either sodium chloride 0.9% or lactated Ringer’s solution, 30 mL · kg−1 · h−1. Their results confirmed that large volumes of sodium chloride solutions led to a metabolic acidosis that was associated with a decrease in the SID. Consistent with previous studies of lactated IV solutions, the increase in serum lactate seen in our study in the Balanced Fluid group reflected the lactate content of Hextend and Hartmann’s solution (6). The increase in glucose, although statistically significant, was thought not to be clinically relevant in this group of patients.
One of the risks of hyperchloremic metabolic acidosis may lie in its inappropriate clinical management. An intraoperative metabolic acidosis that persists despite repeated fluid challenges can be misinterpreted as undertreated hypovolemia and can lead to further IV infusions. It is not always recognized that hyperchloremia is the cause of the metabolic acidosis and that the administration of sodium chloride infusion may indeed exacerbate rather than ameliorate the abnormality. In an editorial, Prough and Bidani (20) emphasized the need to differentiate hyperchloremic metabolic acidosis from lactic acidosis and a primary lactatemia.
Hyperchloremic metabolic acidosis has often been considered to be benign and self-limiting (6,7,13). The evidence is now increasing that it may impair end organ perfusion and interfere with cellular exchange mechanisms: Williams et al. (5) found a significantly increased time to first urination and a greater frequency of abdominal discomfort in 18 healthy volunteers who developed metabolic acidosis after the administration of 50 mL/kg sodium chloride 0.9%, compared with lactated Ringer’s solution. Hyperchloremia alone has a negative effect on renal blood flow and glomerular filtration rate (21). Acidosis in anesthetized pigs has been associated with impaired gastropyloric motility (22). The authors speculated that this mechanism may be implicated in perioperative complications such as gastroparesis and vomiting. Another animal model demonstrated better survival after massive hemorrhage when lactated Ringer’s solution instead of sodium chloride solution was used for resuscitation (23).
We found better indices of gastric mucosal perfusion in the Balanced Fluid group than the Saline group. Gastric tonometry is an important predictor of outcome after major surgery (3). Poor gastric perfusion may also be a risk factor for postoperative nausea and vomiting (24). Fewer events of postoperative nausea and vomiting and a less frequent use of rescue antiemetics were reported in patients in the Balanced Fluid group in our study. The results for urine output and core peripheral temperature difference showed trends in favor of the Balanced Fluid group, but the findings were not statistically significant.
In conclusion, our study has shown that, in elderly surgical patients, the use of balanced IV infusions can avoid the development of hyperchloremic metabolic acidosis and is associated with better indices of gastric mucosal perfusion than saline-based fluids. There is increasing evidence, both in human and animal experiments, to suggest that the use of saline-based fluids may be associated with clinically relevant sequelae. These findings warrant further investigation in appropriately designed prospective, randomized, controlled trials.
The authors thank the manufacturing pharmacies at University College London Hospitals and the Royal Free Hospital for the provision of study fluids and the randomization of patients. The authors gratefully acknowledge the help with planning and statistical analysis provided by Clinical Data Care in Lund, AB, Sweden.
1. Traynor C. Anaesthesia in the elderly. In: Morgan M, Hall GM, eds. Short practice of anaesthesia. London, 1998.
2. Report of the National Confidential Enquiry into Perioperative Deaths (NCEPOD) 1997/1998. London: Chapman & Hall, 1998.
3. Mythen MG, Webb AR. Perioperative plasma volume expansion reduces the incidence of gut mucosal hypoperfusion during cardiac surgery. Arch Surg 1995; 130: 423–9.
4. Sinclair S, James S, Singer M. Intraoperative intravascular volume optimisation and length of hospital stay after repair of proximal femoral fracture: randomised controlled trial. BMJ 1997; 315: 909–12.
5. Williams EL, Hildebrand KL, McCormick SA, Bedel MJ. The effect of intravenous lactated Ringer’s solution versus 0.9% sodium chloride solution on serum osmolality in human volunteers. Anesth Analg 1999; 88: 999–1003.
6. McFarlane C, Lee A. A comparison of Plasmalyte 148 and 0.9% saline for intra-operative fluid replacement. Anaesthesia 1994; 49: 779–81.
7. Scheingraber S, Rehm M, Sehmisch C, Finsterer U. Rapid saline infusion produces hyperchloremic acidosis in patients undergoing gynecologic surgery. Anesthesiology 1999; 90: 1265–70.
8. Chapman MV, Webb AR, Vincent JL, Mythen MG. Gastrointestinal tonometry: state of the art. Intensive Care Med 2000; 26: 613–22.
9. Lee JA. Sydney Ringer (1834–1910) and Alexis Hartmann (1898–1964). Anaesthesia 1981; 36: 1115–21.
(6% hetastarch in lactated electrolyte injection) [package insert]. North Chicago, IL: Abbott Laboratories, 1999.<
11. Woolf RL, Chapman MV, Mythen MG. Early clinical experience with a newly formulated hydroxyethyl starch: Hextend [letter]. Br J Anaesth 1999; 82: 299–300.
12. Miller LR, Waters JH, Provost C. Mechanism of hyperchloremic metabolic acidosis. Anesthesiology 1996; 84: 482–3.
13. Mathes DD, Morell RC, Rohr MS. Dilutional acidosis: is it a real clinical entity? Anesthesiology 1997; 86: 501–3.
14. Russo MA. Dilutional acidosis: a nonentity? Anesthesiology 1997; 87: 1010–1.
15. Dorje P, Adhikary G, McLaren ID, Bogush S. Dilutional acidosis or altered strong ion difference. Anesthesiology 1997; 87: 1011–2.
16. Storey DA, Bellomo R. Is Hartmann’s the solution? Anaesthesia 1997; 52: 1022–3.
17. Waters JH, Miller LR, Clack S, Kim JV. Cause of metabolic acidosis in prolonged surgery. Crit Care Med 1999; 27: 2298–9.
18. Dorje P, Adhikary G, Tempe DK. Avoiding iatrogenic hyperchloremic acidosis: call for a new crystalloid fluid. Anesthesiology 2000; 92: 625–6.
19. Stewart PA. Modern quantitative acid-base chemistry. Can J Physiol Pharmacol 1983; 61: 1444–61.
20. Prough DS, Bidani A. Hyperchloremic metabolic acidosis is a predictable consequence of intraoperative infusion of 0.9% saline [editorial]. Anesthesiology 1999; 90: 1247–9.
21. Wilcox CS. Regulation of renal blood flow by plasma chloride. J Clin Invest 1983; 71: 726–35.
22. Tournadre JP, Allaouchiche B, Malbert CH, Chassard D. Metabolic acidosis and respiratory acidosis impair gastro-pyloric motility in anesthetized pigs. Anesth Analg 2000; 90: 74–9.
23. Healey MA, Davis RE, Liu FC, et al. Lactated Ringer’s is superior to normal saline in a model of massive hemorrhage and resuscitation. J Trauma 1998; 45: 894–9.
24. Gan TJ, Mythen MG, Glass PSA. Intraoperative gut hypoperfusion may be a risk factor for postoperative nausea and vomiting. Br J Anaesth 1997; 78: 476.