Isotonic crystalloid solutions: a structured review of the literature

Transcription

1 British Journal of Anaesthesia 112 (6): (2014) Advance Access publication 15 April doi: /bja/aeu047 REVIEW ARTICLES Isotonic crystalloid solutions: a structured review of the literature D. Orbegozo Cortés, A. Rayo Bonor and J. L. Vincent* Department of Intensive Care, Erasme Hospital, Université Libre de Bruxelles, Route de Lennik 808, 1070 Brussels, Belgium * Corresponding author. jlvincen@ulb.ac.be Editor s key points There is renewed interest in perioperative fluid therapy. The authors have systematically reviewed literature related to crystalloid solutions. Despite similar haemodynamic profiles, different crystalloid solutions have varying effects on electrolyte balance and end-organ function. Background. Several different crystalloid solutions are available for i.v. fluid administration but there is little information about their specific advantages and disadvantages. Methods. We performed a systematic search of MEDLINE, EMBASE, and CENTRAL up until May 17, 2012, selecting all prospective human studies that directly compared any near-isotonic crystalloids and reported any outcome. Results. From the 5060 articles retrieved in the search, only 28 met the selection criteria. There was considerable heterogeneity among the studies. Several articles reported an increased incidence of hyperchloraemic acidosis with the use of normal saline, and others an increase in blood lactate levels when large amounts of Ringer s lactate solutions were infused. From the limited data available, normal saline administration appears to be associated with increased blood loss and greater red blood cell transfusion volumes in high-risk populations compared to Ringer s lactate. Possible effects of the different solutions on renal function, inflammatory response, temperature, hepatic function, glucose metabolism, and splanchnic perfusion are also reported. The haemodynamic profiles of all the solutions were similar. Conclusions. Different solutions have different effects on acid base status, electrolyte levels, coagulation, renal, and hepatic function. Whether these differences have clinical consequences remains unclear. Keywords: crystalloid solutions; infusions, intravenous; isotonic solutions Crystalloids are the most widely used i.v. solutions and are administered at some time to almost all acutely ill patients. A primary goal of i.v. fluid therapy is to restore and maintain sufficient blood volume to guarantee adequate transport of oxygen and nutrients to the tissues. Each type of i.v. fluid has certain advantages compared with other fluids but also some shortcomings. Recent studies have focused on the safety of various colloid solutions, 1 4 but less attention has been paid to the effects of different isotonic crystalloid solutions. The most widely used i.v. crystalloid solutions differ considerably from human plasma in composition, tonicity, or both. 5 The most frequently prescribed crystalloid solutions are normal saline and Ringer s lactate solutions. Possible negative effects of these solutions on acid base status and plasma tonicity prompted the development of so-called balanced solutions, the most widely used being PlasmaLyte (Baxter, Deerfield, IL, USA) in which some of the negative ions (chloride and lactate) are replaced with acetate and gluconate. 5 Nevertheless, none of these solutions is completely physiological and all may have unwanted effects. 6 Because of the potential clinical differences between solutions and the lack of definitive data, we conducted a systematic review of all prospective, randomized, controlled trials that have compared different crystalloid solutions in order to assess the available evidence. Methods We conducted a search of the literature using MEDLINE, EMBASE, and CENTRAL databases, on May 17, 2012, looking for all articles that had prospectively compared different nearisotonic crystalloid solutions. The details of the search strategy are presented in the Supplementary material. The inclusion criteria were: (i) human studies in the adult population; (ii) prospective randomized clinical trials (RCT) of two or more pure cohorts of near-isotonic i.v. solutions; (iii) results reported for anyoutcome parameters; and (iv) at least 20 subjects included. In studies with additional cohorts comparing colloids, hyperoncotic, or hypertonic solutions, we attempted to extract the data for the pure near-isotonic crystalloid cohorts. We excluded from our analysis studies of hypertonic solutions or hyperosmolar solutions, studies in children, studies using topical orother routes of application, studies comparing specific ingredients, such as magnesium or bicarbonate, and studies & The Author [2014]. Published by Oxford University Press on behalf of the British Journal of Anaesthesia. All rights reserved. For Permissions, please journals.permissions@oup.com

2 Crystalloid solutions: a review of the literature BJA of parenteral nutrition. Study selection was performed without limits of date of publication or language. If the article wasnot in English, it was translated by a native speaking colleague. We completed the search by reviewing the bibliographies of every selected article to look for possible additional articles that had not been retrieved by the search. Two authors (D.O.C. and A.R.B.) independently selected the articles based on the criteria above; any discrepancy was resolved by consensus. The reference data, populations, and outcomes were then extracted from the articles into prespecified tables. The quality of all studies was evaluated using the Jadad scale. 7 There was considerable heterogeneity in the design of the studies in terms of the type and amount of fluid given, study population, fluid indication (maintenance and resuscitation), duration of fluid administration, and outcome studied, making it impossible to conduct a formal meta-analysis of all the outcomes. However, for some outcomes (ph and lactate levels, red blood cell transfusion, blood loss, creatinine levels, and urine output) obtained from studies comparing normal saline with Ringer s lactate, a meta-analysis was performed because they were reported systematically in these populations. The estimated mean effect of each study of these outcomes was calculated with the respective 95% confidence intervals and the global effect was then assessed using a Der- Simonian Laird random effects model. 8 Results are presented as Forest plots with estimates of heterogeneity. All analyses were performed using MetaEasy (Statanalysis, UK). Results We identified atotal of 5060 articles but only 28 studies met the selection criteria and were retained for the descriptive analysis (Fig. 1). Seven of the articles 9 15 were not in English and were translated before extracting the data. The included studies are given in Table 1. Types of fluids compared Some studies had threetreatment cohorts so they were included in more than one group. The differences between normal saline and Ringer s lactate were explored in 21 studies, between PlasmaLyteandanother near-isotonic solution in 7 studies, and between Ringer s acetate and another near-isotonic solution in 3 studies Acomplete description of the study comparisons is given in Table 2. The administered doses and duration of exposure were extremely variable (Table 1): doses ranged from to ml and the duration between 20 min 17 and 48 h. 29 Studied populations Only six studies were performed in non-surgical populations with cohorts of between and patients. The other 22 studies were performed in diverse surgical patients undergoing scheduled procedures with cohorts of between and patients. The most commonly studied population was renal-transplant patients who were included in five studies. Total Definitive 28 Fig 1 Study selection tree. Effects on outcome measures Duplicated 828 Rejection by title 3944 Other topics or solutions 252 Retracted Haemodynamics All studies showed similar haemodynamic effects with each of the crystalloid solutions studied. Ten studies reported no significant differences between crystalloids in amount of volume required, the need for vasoactive drugs, or the time to achieve resuscitation. In a study by Kashimoto and colleagues 19 comparing Ringer s lactate with Ringer s acetate, there was a trend towards a higher fingertip blood flow with Ringer s acetate, which may reflect the vasodilating properties of acetate, but interpretation is limited by the small population studied. Acid base state and electrolyte concentration Twenty-two of the 23 articles that explored the effects on acid base status showed that the administration of normal saline was associated with a decrease in blood ph, an increase in chloride levels, and a decrease in bicarbonate or strong ion difference (SID); the study by Deng and colleagues 12 reported a similar trend, but without statistical significance. Most of the fluids were administered over a short period and these changes resolved in the subsequent hours. The largest difference in ph levels was reported in the study by Scheingraber and colleagues, 20 who noted a ph of 7.28 with normal saline compared with 7.40 with Ringer s lactate after infusing 60 ml kg 21 of fluid over 2 h. Takil and colleagues 22 randomized 30 patients undergoing elective spinal surgical procedures to preoperative Ringer s lactate or normal saline. These authors reported two main findings: transient intraoperative hyperchloraemic acidosis in 2 Less than 20 subjects 6 969

3 970 Table 1 The structured review of results from the selected articles. NS, normal saline; RL, Ringer s lactate; PL, PlasmaLyte; RA, Ringer s acetate; BE, Base excess; HCO 2 3, bicarbonate; CRP, C-reactive protein; AST, aspartate transaminase; ALT, alanine transaminase; ALP, alkaline phosphatase; BA, before anaesthetic induction; HR, heart rate; MAP, mean arterial pressure; CVP, central venous pressure; CK, creatine kinase; AG, anion gap; SID, strong ion difference; ATOT, total weak non-volatile acids; Na, sodium; K, potassium; Cl 2, chloride; INR, International Normalized Ratio; LOS, length of stay; SIRS, systemic inflammatory response syndrome; AT III, antithrombin III; TBSA, total body surface area Study Population Groups Crystalloid administration protocol Principal outcome Secondary outcomes Jadad scale Ramanathan and Elective Caesarean NS (15) colleagues 16 RL (15) PL (15) Veroli and Benhamou 17 McFarlane and Lee 18 Butscher and colleagues 9 Kashimoto and colleagues 19 Scheingraber and colleagues 20 Deng 10 and colleagues 21 Takil and colleagues 22 Minor orthopaedic surgery with lumbar epidural anaesthesia Scheduled elective major hepatobiliary or pancreatic surgery Scheduled lumbar intervertebral disc surgery General surgery lasting more than 2 h using isoflurane Lower gynaecological system procedures Hysterectomy and myomectomy Aortic reconstructive surgery Elective major spine surgery NS (10) RL (10) NS (15) PL (15) NS (10) RL (10) RA (15) RL (15) NS (12) RL (12) NS (12) RA (12) NS (33) RL (33) NS (15) RL (15) O Malley and Renal transplantation NS (26) colleagues 23 RL (25) BA: 1200 ml in 1 h NS decrease in ph No differences in glucose, pyruvate, arterial pressure, or ephedrine needs. Greater increase in lactate levels in RL and PL groups BA: 20 min bolus with same amount of sodium (15 ml kg 21 with RL and 13 ml kg 21 with NS) Intraoperative: 15 ml kg 21 h 21 Intraoperative: 2000 ml in 2 h Intraoperative: 8 ml kg 21 in the first hour and 5 ml kg 21 in the second hour Intraoperative: 30 ml kg 21 h 21 No differences in required volumes Greater increase in postoperative chloride using NS Greater decrease in osmolarity after RL Higher central temperature with RA Lower ph at second hour with NS Same doses of ephedrine, similar HR and MAP in both groups Greater decrease in BE and HCO 2 3 with NS concordant with hyperchloraemic acidosis. No differences in chloride levels, Na or K levels, Pa CO2, lactate concentrations, dose of crystalloids or bleeding after surgery No differences in Na, glucose or protein concentrations after crystalloid infusion No differences in digital and forearm temperature, or fingertip blood flow Lower values of BE, lactate, and calculated SIDat 2 h with NS. Lower Na and Cl 2 at 2 h with RL. No differences in P CO2, blood loss or diuresis at 2 h Intraoperative: 20 ml kg 21 h 21 Lower ph at 120 min with NS Lower BE and higher Na and Cl 2 levels with NS at 120 min. Higher lactate levels with RA at 120 min. Lower Pa CO2 at 30 min in the NS group Intraoperative: titrated during surgery. The mean volume of 6871 ml with RL and 7000 ml with NS Intraoperative and 12 h postoperative: 20 ml kg 21 h 21 intraoperative and 2.5 ml kg 21 h 21 thereafter Intraoperative: titrated during surgery. Mean volume of 5600 ml with RL and 6100 ml with NS Greater decrease in intraoperative ph with NS Transient respiratory acidosis in first postoperative hours with RL No differences in creatinine at postoperative third day, 1 week or 6 months Greater decrease in intraoperative HCO 2 3 and BE, higher levels of Cl 2, greater need for blood product transfusions and greater urine output with NS. No significant changes in lactate levels, ventilation hours or LOS Transient hyperchloraemic acidosis (ph, HCO 2 3, and BE) intraoperative and until 12 h postoperative with NS. Mild hyponatraemia in first postoperative hours with RL. No significant differences in crystalloid volume, urinary output, and blood losses More cases of hyperkalaemia, use of intraoperative HCO 3 and higher Cl 2 levels at end of surgery with NS. No significant differences in blood loss or patients transfused BJA Orbegozo Cortés et al.

4 971 Chin and colleagues 24 Karaca and colleagues 11 Elective surgery in non-diabetic patients Transuretheral prostatic procedures under spinal anaesthesia NS (16) RL (16) NS (20) RL (20) Chanimov and Elective Caesarean NS (20) colleagues 25 RL (20) Cho and colleagues 26 Deng 12 Tellan and colleagues 13 Rhabdomyolysis induced by doxylamine intoxication Orthotopic liver transplantation Sigmoid diverticulosis undergoing left hemicolectomy NS (13) RL (15) NS (30) RL (30) NS (16) RL (16) NS+RL (15) Khajavi and Renal transplantation NS (26) colleagues 27 RL (26) Hadimioglu and colleagues 28 Liu Xiaoyan 14 Related kidney transplantation Scheduled gastrointestinal surgery NS (30) RL (30) PL (30) RL (20) PL (20) Aoki and Burns.30% of TBSA RL (10) colleagues 29 RA (10) Nuraei and Renal transplantation NS (54) colleagues 15 RL (54) Shin and colleagues 30 Mahler and colleagues 31 Right hepatectomy for liver transplantation Moderate or severe diabetic ketoacidosis RL (52) PL (52) NS (23) PL (22) Wu and Acute pancreatitis NS (21) colleagues 32 RL (19) Heidari and Lower abdominal surgery NS (30) colleagues 33 RL (30) Intraoperative: 500 ml in min Glucose similar in both groups No differences in urea, Na, K, and osmolarity 3 Intraoperative: bolus of 20 ml kg 21 and then 2 ml kg 21 h 21 Intraoperative: bolus of 2000 ml and 2 ml kg 21 h 21 thereafter Postoperative hearing loss similar in both groups Similar crystalloid volume administered No differences in number of transoperative hypotension episodes, postoperative dizziness, vomiting, or nausea No differences in mg of ephedrine used or surgical time Rehydration protocol: initiate with Urine ph at 12 h lower with NS More hyperchloraemic acidosis and longer time to 120 ml h 21. Max 400 ml h 21 CK,200 with NS. No differences in Na or K levels Intraoperative: titrated during surgery. Mean volume of 2263 ml with RL and 2570 ml with NS Intraoperative: titrated during surgery. Mean volume of 3710 ml with RL, 3643 ml with NS and 3894 ml with NS+RL Intraoperative: protocol to achieve 60 ml kg 21. Intraoperative: protocol to maintain CVP infusing ml kg 21 h 21 BA: 8 ml kg 21 in 30 min At least 48 h: protocol to maintain MAP and diuresis Intraoperative: protocol to maintain CVP. Mean volume of 3125 ml with RL and 3005 ml with NS Intraoperative: titrated during surgery. Mean volume of 3407 ml with RL and 3302 ml with PL No differences in lactate levels intraoperative or at the end of surgery Lower ph in the NS group More acidosis at end of surgery with NS Lower ph at end of surgery with NS Higher lactate levels with RL at the end of surgery Lower gastric P CO2 gap with RA No differences in ph at the end of surgery Lower values of bilirubin at first day with PL No differences in the given amount of fluids, blood loss, ph or BE intraoperative, or at the end of surgery Higher Cl 2 values with NS. Lower P CO2,HCO 2 3, BE, AG, and SID with NS. No differences in ATOT levels, blood loss, or patients transfused Greater increase in K at end of surgery with NS. No differences in third day creatinine, transfused blood, blood loss, or vascular thrombosis Lower BE and higher chloride levels with NS. Higher lactate levels with RL and diuresis with NS. Similar K levels and renal function tests in all groups No differences in amount of fluids, glucose, ph, P CO2,HCO 2 3, BE at the end of surgery Lower gastric ph gap with RA. No differences in gastric P CO2, gastric ph, arterial lactate, arterial BE, AT III, orarterial ketones in the first 72 h No significant changes at the end of surgery in creatinine, P CO2, Na, K, or HCO 3. No differences in LOS, creatinine, or BUN at third and seventh postoperative days No differences in AST, ALT, or ALP levels. Higher albumin and lower INR with PL in first days that disappeared at fifth day. Higher lactate onlyat first postoperative hour with RL Rehydration protocol (max 24 h). Initial Hyperchloraemia with NS Decrease in HCO 2 3 values with NS 5 bolus of 20 ml kg 21 Rehydration protocol (24 h). Boluses of Higher SIRS reduction at 24 h 20 ml kg 21 and infusion of 3 ml kg 21 h 21 with RL Lower HCO 3 2 decrease and lower CRP levels at 24 h with RL. No differences in new onset respiratory distress, hypotension, need for renal replacement therapies, requirement of ICU admission, necrosis, or infection BA: 15 ml kg 21 h 21 bolus Less vomiting at 6 h with RL No differences in vomiting after 6 h, nausea and vomiting scores, or Na levels Continued Crystalloid solutions: a review of the literature BJA

5 BJA Orbegozo Cortés et al. Table 1 Continued Study Population Groups Crystalloid administration protocol Principal outcome Secondary outcomes Jadad scale 5 Lower HCO3 2 levels with NS. No differences in Na, K, or Cl 2 levels Venous peripheral ph at 120 min lower with NS and higher with PL Rehydration protocol. 20 ml kg 21 h 21 for 2 h NS (32) RL (31) PL (33) Hasman and colleagues 34 Moderately and severely dehydrated patients 4 Higher K and Cl 2 levels at fourth intraoperative hour with NS. No differences in intraoperative infused volume or creatinine at first postoperative day Lower HCO3 at fourth intraoperative hour with NS Intraoperative: protocol to maintain CVP. Mean volume of 5250 ml with RL and 5100 ml with NS NS (37) RL (37) Modi and colleagues 35 Live related renal transplantation 5 Lower insulin units used in 6 first hours with NS. No differences in time to reach a ph of 7.32, time to reach a HCO3 2 of 18, Na or Cl 2 levels after ketoacidosis correction Rehydration protocol until glucose Shorter time to achieve a,14 mmol litre 21 glucose,14 mmol litre 21 with NS NS (27) RL (27) Van Zyl and colleagues 36 Diabetic ketoacidosis Table 2 Solutions compared in the clinical trials. NS, normal saline; RL, Ringer s lactate; RA, Ringer s acetate; PL, PlasmaLyte Comparison NS vs RL 17 NS vs RL vs PL 3 PL vs NS 2 PL vs RL 2 RA vs RL 2 RA vs NS 1 NS vs RL vs mixed NS-RL 1 Number of studies the normal saline group and an early postoperative increase in Pa CO2 and bicarbonate levels in the Ringer s lactate group; the latter finding was interpreted to be the result of lactate metabolism to CO 2. Cho and colleagues 26 showed that the hyperchloraemic acidosis was accompanied by a concomitant decrease in urinary ph in patients treated with normal saline for rhabdomyolysis. These changes can be explained by the need for the kidney to eliminate the chloride load. In a meta-analysis of the studies comparing normal saline with Ringer s lactate that reported this outcome before and after fluid administration, use of normal saline was correlated with the development of hyperchloraemic acidosis (Fig. 2) and this effect was similar when considering only studies of surgical populations or patients undergoing kidney transplantation (Supplementary material, Figs S1 and S2). Blood lactate concentrations Five studies showed an increase in lactate levels with Ringer s lactate compared with other solutions. Two studies showed a trend in the same direction but without statistical significance and one study 12 showed no change in lactate levels when comparing Ringer s lactate with normal saline in a population of hepatic transplant patients. McFarlane and Lee 18 compared the lactate-free solutions, PlasmaLyte and normal saline, in patients undergoing scheduled major hepatobiliary or pancreatic surgery and reported no significant differences in blood lactate levels. Deng and colleagues 10 reported higher lactate levels comparing Ringer s acetate with normal saline when used in women undergoing scheduled hysterectomy. The largest difference in lactate levels was reported in a study by Scheingraber and colleagues 20 who noted, after infusing 60 ml kg 21 of fluid over 2 h, a serum lactate of 0.4 mmol litre 21 for patients given normal saline compared with 2 mmol litre 21 for Ringer s lactate. In a meta-analysis of data from the studies comparing normal saline with Ringer s lactate that reported this outcome, the same association (i.e. greater lactate levels with Ringer s lactate solutions) was found (Fig. 3). Coagulation and bleeding Ten studies investigated the effects on bleeding, need for transfusion of blood products, or other aspects of coagulation. In patients undergoing major 972

6 Crystalloid solutions: a review of the literature BJA A Scheingraber Takil O Malley Deng Tellan Khajavi Hadimioglu Nuraei Hasman TOTAL [95% CI] 0.11 [ 0.05, 0.28] Q = 7.22, df = 9 (P = 0.61); I 2 = 0% 1.5 Normal saline (274 patients) Ringer s lactate (272 patients) B Ramanathan Scheingraber Takil O Malley Deng Tellan Hadimioglu Nuraei Hasman TOTAL [95% CI] 1.35 [ 1.89, 0.82] Q = 84.1, df = 9 (P < 0.01); I 2 = 89% Normal saline (263 patients) Ringer's lactate (261 patients) Fig 2 Forest plot of blood ph before (A) and after (B) fluid administration in studies comparing normal saline with Ringer s lactate. scheduled hepatobiliary or pancreatic surgery, McFarlane and Lee 18 reported an identical bleeding risk with PlasmaLyte and normal saline. Seven studies comparing normal saline with Ringer s lactate found no significant differences between the fluids in blood loss or need for transfusion; and colleagues 21 reported, in a high-risk population of patients undergoing aortic reconstructive procedures, an increased need for blood product transfusions in those who received normal saline. In a meta-analysis of the data from the studies comparing normal saline with Ringer s lactate that reported this outcome in millilitres, there were no differences regarding the volume of blood lost. We also performed a subgroup analysis of these studies, dividing them according to the volumes of blood loss reported in each study into those that included high-risk and those with low-risk patients, using a cut-off point of 800 ml, which is roughly 15% of the average blood volume (the limit of a Grade II hemorrhage). In this subgroup analysis, there was increased blood loss with 973

7 BJA Orbegozo Cortés et al. A Ramanathan Scheingraber Deng Hadimioglu TOTAL [95% CI] 0.28 [ 0.19, 0.74] Q = 12.77, df = 4 (P = 0.01); I 2 = 69% Normal saline (120 patients) Ringer s lactate (120 patients) B Ramanathan Scheingraber Deng Hadimioglu TOTAL [95% CI] 1.99 [0.40, 3.58] Q = , df = 4 (P < 0.01); I 2 = 97% Normal saline (120 patients) Ringer s lactate (120 patients) Fig 3 Forest plot of lactate levels before (A) and after (B) fluid administration in studies comparing normal saline with Ringer s lactate. normal saline (Fig. 4) in the high-risk patients. No differences were found when grouping studies according to renal transplant or non-renal transplant populations (Supplementary material, Figs S3 and S4). When considering data regarding the volume of transfused red blood cells reported in millilitres, greater transfusion volumes were reported in patients receiving normal saline (Fig. 5). Just two studies reported effects on measures of coagulation: Shin and colleagues 30 noted an increase in the International Normalized Ratio (INR) after infusion of Ringer s lactate compared with PlasmaLyte, and Aoki and colleagues 29 found no differences in antithrombin (AT) III levels after infusion of Ringer s acetate or Ringer s lactate. Renal function Eleven studies explored renal function after exposure to different crystalloids. Five studies in high-risk post-renal-transplant patients reported no significant differences between normal saline, Ringer s lactate, and Plasma- Lyte When evaluating only comparisons of normal saline and Ringer s lactate, seven studies reported data on diuresis and two showed significant opposite differences in urine output. For meta-analysis purposes, we took the first urine output measure expressed in millilitres or the first creatinine level reported after the first 24 h of fluid administration. There were no differences between solutions for any outcome (Fig. 6). Results of meta-analyses of the data showed no differences between solutions in renaltransplant or non-renal transplant patients (Supplementary material, Figs S5 S7). Meta-analysis of studies reporting creatinine levels 3 days after renal transplantation showed no differences between the two solutions (Supplementary material, Fig. S8). Inflammation Wu and colleagues 32 reported a more rapid decrease in the prevalence of the systemic inflammatory response syndrome after the first 24 h and a greater decrease in blood C-reactive protein (CRP) levels with Ringer s lactate than with normal saline in patients with acute pancreatitis. No other studies were found on this topic. 974

8 Crystalloid solutions: a review of the literature BJA A Scheingraber Takil O Malley Tellan Khajavi TOTAL [95% Cl] 0.09 [ 0.24, 0.43] Q = 8.84, df = 5 (P = 0.12); I 2 = 43% Ringer s lactate (127 patients) Normal saline (128 patients) B O Malley Tellan Khajavi TOTAL [95% CI] 0.21 [ 0.55, 0.12] Q = 1.58, df = 2 (P = 0.45); I 2 = 0% Ringer s lactate (67 patients) Normal saline (68 patients) C Scheingraber Takil TOTAL [95% Cl] 0.42 [0.06, 0.78] Q = 0.88, df = 2 (P = 0.65); I 2 = 0% Ringer s lactate (60 patients) 0.5 Normal saline (60 patients) Fig 4 Forest plot of blood loss volumes in studies comparing normal saline with Ringer s lactate when considering all patients (A), patients with a low risk of bleeding (B), and patients with a high risk of bleeding (C). 975

9 BJA Orbegozo Cortés et al. Takil Deng TOTAL [95% CI] 0.42 [0.11, 0.73] Q = 0.12, df = 2 (P = 0.94); I 2 = 0% Ringer's lactate (78 patients) Normal saline (78 patients) Fig 5 Forest plot of red blood cell transfusion volumes in studies comparing normal saline with Ringer s lactate. Glucose concentration Four studies found no difference in blood sugar levels after infusion of anystudysolution in non-diabetic populations. Two studies compared different crystalloids in patients with ketoacidosis: Mahler and colleagues 31 reported a greater degree of hyperchloraemic acidosis with normal saline than with PlasmaLyte; and Van Zyl and colleagues 36 observed that the time to reach normalization of glycaemia was shorter with normal saline than with Ringer s lactate. Liver metabolism In donors undergoing right hepatectomy, Shin and colleagues 30 reported that, compared with PlasmaLyte, Ringer s lactate may have been associated with hepatic dysfunction, as measured by a transient increase in bilirubin and mild prolongation of prothrombin time during the first 2 days and a decrease in albumin levels during the first 4 days. These transient changes were, however, not clinically relevant; there were no differences in liver enzymes, and no increased need for transfusion. Splanchnic dysoxia Aoki and colleagues 29 used gastric tonometry to evaluate the degree of splanchnic dysoxia during the resuscitation of adult burns patients. The gastric P CO2 gap wasreduced in the first 72 h in patients resuscitated with Ringer s acetate compared with those resuscitated with Ringer s lactate. There were no differences in the amounts of fluid administered, urinary output, or mean arterial pressure. The study was limited by the small sample size (20 patients) and a randomization process by alternate allocation. Temperature Kashimoto and colleagues 19 reported a smaller decrease in central body temperature with Ringer s acetate than with Ringer s lactate during isoflurane anaesthesia. The physiological explanation for this result is unclear, but could be related to acetate-induced vasodilatation or peripheral metabolism and heat production under acetate perfusion. Nausea and vomiting Two studies explored the development of nausea and vomiting after anaesthesia. Heidari and colleagues 33 reported less vomiting in the first 6 postoperative hours with Ringer s lactate compared with normal saline, but Karaca and colleagues 11 reported no differences. Discussion I.V. fluid administration is one of the most widely used interventions in hospitalized patients and yet our literature review underlines the paucity of good trials on this important topic. The studies were small and generally of low quality (15 of the 28 studies had a Jadad score of,3). The studies were also very heterogeneous in terms of patient populations, amounts and duration of fluid given, timing of fluid in relation to surgery when performed, the type of crystalloid used, and the outcome reported. It was, therefore, not possible to make any global statement related to effects on mortality or morbidity, including length of stay, need for organ support, haemostasis, or gastrointestinal complications. There is a clear association between normal saline infusion and the development of hyperchloraemic acidosis. Studies in healthy volunteers have shown the same association. This metabolic effect is expected, because the administration 976

10 Crystalloid solutions: a review of the literature BJA A Scheingraber Takil O Malley Khajavi Hadimioglu Nuraei TOTAL [95% CI] 0.21 [ 0.17, 0.59] Q = 20.86, df = 6 (P < 0.01); I 2 = 71% Ringer s lactate (195 patients) Normal saline (196 patients) B O Malley Khajavi Hadimioglu Nuraei Modi TOTAL [95% CI] 0.02 [ 0.28, 0.33] Q = 12.18, df = 5 (P < 0.03); I 2 = 59% Ringer s lactate (205 patients) Normal saline (206 patients) Fig 6 Forest plot of first reported urinary output volume (A) or creatinine level (B) after the first 24 h of fluid administration in studies comparing normal saline with Ringer s lactate. of chloride ions diminishes the anion gap and bicarbonate is used as a buffer to compensate; 18 alternatively, the increase in chloride levels decreases the SID, 13 which affects the dissociation of water into hydrogen ions Several studies defined hyperchloraemic acidosis as a principal outcome measure, but none tried to correlate its appearance with other clinically important outcomes such as length of stay, renal function, inflammation, or mortality. In a retrospective study, Gunnerson and colleagues 42 tried to quantify the impact of different types of acidosis on the outcomes of intensive care unit (ICU) patients. These authors collected data from a period of 18 months and a total of 851 patients were analysed. There was a clear association of lactic acidosis but not of hyperchloraemic acidosis with mortality. 42 There is some suggestion that hyperchloraemic acidosis may impact on inflammation and coagulation. Animal data have linked Ringer s lactate to increased neutrophil activation, production of reactive oxygen species (ROS), overexpression of leucocyte inflammatory genes, and increased cytokine levels In vitro dilutional studies using different thromboelastography (TEG) techniques have shown that the use of normal saline may be associated with a hypocoagulable state and Ringer s lactate with a hypercoagulable state, as expressed by the clotting time, the amplitude of coagulation, or thrombin generation In an uncontrolled haemorrhagic pig model, Kiraly and colleagues 48 reported an increase in blood losses, prolongation of coagulation times and more TEG alterations with the use of normal saline compared with Ringer s lactate. Whether these in vitro and animal studies can be translated into clinical practice is unclear, but our data do tend to support a greater risk of bleeding with normal saline, perhaps particularly in populations at high risk of bleeding; nevertheless, our results must be interpreted with caution because of the methodological limitations already discussed. More prospective studies are needed on this topic. Hyperchloraemic acidosis may also result in renal vasoconstriction. In a denervated and autotransplanted kidney model 977

11 BJA Orbegozo Cortés et al. in dogs, Wilcox 49 showed transient renal vasoconstriction related to the chloride content of different hypertonic solutions. The administration of chloride results in free water retention. Interestingly, a delayed initiation of diuresis was observed in studies that compared normal saline with Ringer s lactate in healthy volunteers However, in the studies included in our review, there were no major differences in renal function between solutions in different populations, even after renal transplantation Nevertheless, whether renal function may deteriorate secondary to hyperchloraemic acidosis induced by normal saline administration deserves further study. Romanelli and colleagues 50 injected Ringer s lactate or normal saline for 5 s into the coronary arteries of patients undergoing coronary angiograms and showed more electrocardiogram alterations in the normal saline than in the Ringer s lactate group. Unal and colleagues 51 used the same solutions topically after nasal septal surgery and demonstrated that normal saline impaired mucociliary function more than Ringer s lactate. Clearly, more research is needed to define the effects of hyperchloraemic acidosis induced by normal saline on metabolism, inflammation, renal function, coagulation, and even mortality. The replacement of some of the chloride by lactate in Ringer s lactate solution was proposed by Dr Hartmann, a paediatrician, so that it is not surprising that the solution is hypotonic. Blood lactate levels increase with administration of large amounts of Ringer s lactate, and Jackson and colleagues 52 noted the possible spurious hyperlactataemia when blood is drawn from the line where the solution is infused. However, Didwania and colleagues 53 reported no differences in the measured levels of lactate in healthy adult volunteers receiving normal Ringer s lactate infusion rates. There is a clear association of lactate levels with mortality, but no evidence that lactate induces damage. In contrast, lactate may be used as a fuel for the myocardium, the neurones, red blood cells, and other cells. Normal lactate in the body appears as the L-isomer and humans are probably unable to metabolize the D-isomer easily. In the studies in our review, the exact composition of the Ringer s lactate used was not defined although many commercially available solutions use a racemic mixture of both isomers. Some studies have compared the two isomers in infused solutions and reported potential alterations in the lung and inflammatory system with the D-isomer, but these findings are not conclusive. We know relatively little about the effects of the acetate and gluconate that replace some of the chloride content in some fluids. Ringer s acetate solutions may create metabolic alterations. Inoue and colleagues 59 explored acetate metabolism using a liver rat model and showed that Ringer s acetate was associated with pyrophosphate accumulation inside the mitochondria, a phenomenon that could potentially lead to mitochondrial dysfunction. In the past, Ringer s acetate was used routinely in haemodialysis units, but these limitations and studies suggesting a vasodilatory and cardio-depressant effect of acetate have restricted its use. The haemodynamic effects of PlasmaLyte have been less well studied than those of the other crystalloid solutions, and further studies are needed to evaluate the possible deleterious effect of the acetate when PlasmaLyte is used in large amounts or for prolonged periods of time. PlasmaLyte also includes gluconate, which is probably the reason for false-positive galactomannan tests, a biomarker used for early diagnosis of invasive pulmonary aspergillosis, in patients who have received Plasma- Lyte. 65 This effect may not be detrimental, but we actually know very little about the metabolism and safety of gluconate. Some early studies explored the oral administration of gluconic acid and showed that its metabolites were excreted in the urine and associated with urinary acidification in humans Gluconate appears spontaneously in nature and seems to be very well tolerated by living organisms. 68 Bacteria have a specialized metabolic cycle known as the Entner Doudoroff pathway, which enables them to use gluconate as a source of fuel, 69 but although it is a molecule similar to glucose, there is no clear evidence that human beings can incorporate it as a fuel source. If we cannot introduce it into the glycolysis pathway, then hepatic metabolism or urinary excretion is expected. However, these aspects need to be clarified. Although PlasmaLyte may offer some advantages over Ringer s lactate or normal saline in terms of acid base status, there are concerns about its safety profile. Some data from the industry have suggested different energetic costs with the different solutions, as expected by their different metabolism, but other studies suggested no differences, such as that by Zadak and colleagues, 70 who directly measured oxygen consumption (VO 2 ) by indirect calorimetry in healthy subjects and found no significant differences between Plasma- Lyte and Ringer s acetate. More relevant are data from a pig model of severe haemorrhagic shock showing that PlasmaLyte was associated with a higher fatality rate compared with normal saline or Ringer s lactate. 71 These investigators suggested that the acetate or magnesium contents of PlasmaLyte were a possible explanation for these findings. Our review was limited to prospective RCTs. However, it is worth mentioning two recent non-randomized studies. Using a propensity analysis, Shaw and colleagues 72 suggested, in a retrospective evaluation of patients undergoing open major abdominal surgery, that patients who received normal saline had increased development of infections and renal failure requiring dialysis, increased blood transfusion requirements, increased electrolyte disturbances, and increased acidosis treatment compared with those who received PlasmaLyte. Yunos and colleagues 73 compared two periods, one in which chloride-liberal solutions were given and the other in which chloride-restrictive solutions were used, and reported that the chloride-restrictive period was associated with a decrease in the development of acute renal failure. If we assume that the differences reside only in the administered crystalloid, a good explanation for these differences could be related to the development of hyperchloraemic acidosis. The main limitation of these nonrandomized studies lies in the fact that we cannot exclude the possibility that those who received PlasmaLyte were less ill, with a lower risk of complications. This extensive review included all publications without language restrictions, making it the most complete on this topic. 978

12 Crystalloid solutions: a review of the literature BJA However, we still know very little about the effects of the crystalloids that we infuse into so many patients. Do we, therefore, need a large RCT comparing the effects of crystalloids on morbidity and mortality? We are not convinced, for two reasons: first this would require a large number of patients to detect any potential difference; secondly and more importantly, crystalloids may not be one size fits all fluids and should be considered more as drugs that need to be individually prescribed based on the patient s inflammatory, acid base, renal, and coagulation profiles. There is currently no completely physiological solution without side-effects that we can administer safely and indiscriminately to all patients. It may be more appropriate to study specific questions, such as the effects on renal function or haemostasis, in well-defined groups of patients. Conclusions Each near-isotonic crystalloid solution has a different profile in terms of impact on acid base status, electrolyte levels, coagulation, inflammation, renal, and hepatic function. Whether these differences lead to deleterious effects or affect prognosis remains unresolved. The choice of the best near-isotonic solution foreach patient resides in a complete understanding of the expected response to each solution and the patient s risk factors. Supplementary material Supplementary material is available at British Journal of Anaesthesia online. Authors contributions D.O.C.: study design, literature search and synthesis, and first draft of manuscript. A.R.B.: literature search and synthesis, and first draft of manuscript. J.L.V.: study design and critical review of manuscript. Declaration of interest None declared. Funding This work was supported by institutional funds only. References 1 Finfer S, Bellomo R, Boyce N, French J, Myburgh J, Norton R. A comparison of albumin and saline for fluid resuscitation in the intensive care unit. N Engl J Med 2004; 350: Hartog C, Reinhart K. CONTRA: hydroxyethyl starch solutions are unsafe in critically ill patients. Intensive Care Med 2009; 35: Perner A, Haase N, Guttormsen AB, et al. Hydroxyethyl starch 130/ 0.42 versus Ringer s acetate in severe sepsis. N Engl J Med 2012; 367: Zarychanski R, Turgeon AF, Fergusson DA, et al. Renal outcomes and mortality following hydroxyethyl starch resuscitation of critically ill patients: systematic review and meta-analysis of randomized trials. Open Med 2009; 3: E Guidet B, Soni N, Della RG, et al. A balanced view of balanced solutions. Crit Care 2010; 14: Kaplan LJ, Kellum JA. Fluids, ph, ions and electrolytes. Curr Opin Crit Care 2010; 16: Jadad AR, Moore RA, Carroll D, et al. Assessing the quality of reports of randomized clinical trials: is blinding necessary? Control Clin Trials 1996; 17: DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986; 7: Butscher K, Jamali S, Talib R, Ravussin P, Poggi-Bach J, Ecoffey C. Effects of different loading solutions on plasma osmolality. Ann Fr Anesth Reanim 1996; 15: Deng LH. Effects of intraoperative fast infusion of three different kinds of crystalloid fluids on acid base balance in patients with hysteromyoma. J Guangdong Med Coll 2001; 19: Karaca B, Solak M, Yildiz TS, Ozkarakas H, Toker K. Effects of various loading solutions on postspinal hearing loss. J Turk Anesthesiol Reanim Soc 2006; 34: Deng LQ. Influence of transfusion of lactated Ringer s solution on arterial blood lactate concentration and acid base balance during orthotopic liver transplantation. Chin J Mod Nurs 2008; 14: Tellan G, Antonucci A, Marandola M, et al. Postoperative metabolic acidosis: use of three different fluid therapy models. Chir Ital 2008; 60: Liu Xiaoyan SL. The influence of PlasmaLyte A on glycometabolism in type 2 diabetic patients during operation. J Med Res 2009; 38: Nuraei N, Khajenouri R, Soleimani M, Dabbagh A. The effects of intraoperative normal saline versus lactated ringer solution on clinical outcomes and laboratory findings in renal transplant patients. Tehran Univ Med J 2010; 68: Ramanathan S, Masih AK, Ashok U, Arismendy J, Turndorf H. Concentrations of lactate and pyruvate in maternal and neonatal blood with different intravenous fluids used for prehydration before epidural anesthesia. Anesth Analg 1984; 63: Veroli P, Benhamou D. Comparison of hypertonic saline (5%), isotonic saline and Ringer s lactate solutions for fluid preloading before lumbar extradural anaesthesia. Br J Anaesth 1992; 69: McFarlane C, Lee A. A comparison of PlasmaLyte 148 and 0.9% saline for intra-operative fluid replacement. Anaesthesia 1994; 49: Kashimoto S, Narumi Y, Matsukawa T, Oguchi T, Kumazawa T. Comparative effects of Ringer s acetate and lactate solutions on intraoperative central and peripheral temperatures. J Clin Anesth 1998; 10: Scheingraber S, Rehm M, Sehmisch C, Finsterer U. Rapid saline infusion produces hyperchloremic acidosis in patients undergoing gynecologic surgery. Anesthesiology 1999; 90: JH, Gottlieb A, Schoenwald P, Popovich MJ, Sprung J, Nelson DR. Normal saline versus lactated Ringer s solution for intraoperative fluid management in patients undergoing abdominal aortic aneurysm repair: an outcome study. Anesth Analg 2001; 93: Takil A, Eti Z, Irmak P, Yilmaz GF. Early postoperative respiratory acidosis after large intravascular volume infusion of lactated ringer s solution during major spine surgery. Anesth Analg 2002; 95: O Malley CM, Frumento RJ, Hardy MA, et al. A randomized, doubleblind comparison of lactated Ringer s solution and 0.9% NaCl during renal transplantation. Anesth Analg 2005; 100:

13 BJA Orbegozo Cortés et al. 24 Chin KJ, Macachor J, Ong KC, Ong BC. A comparison of 5% dextrose in 0.9% normal saline versus non-dextrose-containing crystalloids as the initial intravenous replacement fluid in elective surgery. Anaesth Intensive Care 2006; 34: Chanimov M, Gershfeld S, Cohen ML, Sherman D, Bahar M. Fluid preload before spinal anaesthesia in Caesarean section: the effect on neonatal acid-base status. Eur J Anaesthesiol 2006; 23: Cho YS, Lim H, Kim SH. Comparison of lactated Ringer s solution and 0.9% saline in the treatment of rhabdomyolysis induced by doxylamine intoxication. Emerg Med J 2007; 24: Khajavi MR, Etezadi F, Moharari RS, et al. Effects of normal saline vs. lactated ringer s during renal transplantation. Ren Fail 2008; 30: Hadimioglu N, Saadawy I, Saglam T, Ertug Z, Dinckan A. The effect of different crystalloid solutions on acid-base balance and early kidney function after kidney transplantation. Anesth Analg 2008; 107: Aoki K, Yoshino A, Yoh K, SekineK, YamazakiM, Aikawa N. A comparison of Ringer s lactate and acetate solutions and resuscitative effects on splanchnic dysoxia in patients with extensive burns. Burns 2010; 36: Shin WJ, Kim YK, Bang JY, Cho SK, Han SM, Hwang GS. Lactate and liver function tests after living donor right hepatectomy: a comparison of solutions with and without lactate. Acta Anaesthesiol Scand 2011; 55: Mahler SA, Conrad SA, Wang H, Arnold TC. Resuscitation with balanced electrolyte solution prevents hyperchloremic metabolic acidosis in patients with diabetic ketoacidosis. Am J Emerg Med 2011; 29: Wu BU, Hwang JQ, Gardner TH, et al. Lactated Ringer s solution reduces systemic inflammation compared with saline in patients with acute pancreatitis. Clin Gastroenterol Hepatol 2011; 9: Heidari SM, Saryazdi H, Shafa A, Arefpour R. Comparison of the effect of preoperative administration of Ringer s solution, normal saline and hypertonic saline 5% on postoperative nausea and vomiting: a randomized, double blinded clinical study. Pak J Med Sci 2011; 27: Hasman H, Cinar O, Uzun A, Cevik E, Jay L, Comert B. A randomized clinical trial comparing the effect of rapidly infused crystalloids on acid-base status in dehydrated patients in the emergency department. Int J Med Sci 2012; 9: Modi MP, Vora KS, Parikh GP, Shah VR. A comparative studyof impact of infusion of Ringer s lactate solution versus normal saline on acid-base balance and serum electrolytes during live related renal transplantation. Saudi J Kidney Dis Transpl 2012; 23: Van Zyl DG, Rheeder P, Delport E. Fluid management in diabetic-acidosis Ringer s lactate versus normal saline: a randomized controlled trial. QJM 2012; 105: 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: Reid F, Lobo DN, Williams RN, Rowlands BJ, Allison SP. (Ab)normal saline and physiological Hartmann s solution: a randomized double-blind crossover study. Clin Sci (Lond) 2003; 104: Kamp-Jensen M, Olesen KL, Bach V, Schutten HJ, Engquist A. Changes in serum electrolyte and atrial natriuretic peptide concentrations, acid-base and haemodynamic status after rapid infusion of isotonic saline and Ringer lactate solution in healthy volunteers. Br J Anaesth 1990; 64: Kellum JA. Clinical review: reunification of acid-base physiology. Crit Care 2005; 9: Kellum JA. Determinants of plasma acid-base balance. Crit Care Clin 2005; 21: Gunnerson KJ, Saul M, He S, Kellum JA. Lactate versus non-lactate metabolic acidosis: a retrospective outcome evaluation of critically ill patients. Crit Care 2006; 10: R22 43 Rhee P, Burris D, Kaufmann C, et al. Lactated Ringer s solution resuscitation causes neutrophil activation after hemorrhagic shock. J Trauma 1998; 44: Koustova E, Stanton K, Gushchin V, Alam HB, Stegalkina S, Rhee PM. Effects of lactated Ringer s solutions on human leukocytes. J Trauma 2002; 52: Kellum JA, Song M, Almasri E. Hyperchloremic acidosis increases circulating inflammatory molecules in experimental sepsis. Chest 2006; 130: Roche AM, James MF, Bennett-Guerrero E, Mythen MG. A head-to-head comparison of the in vitro coagulation effects of saline-based and balanced electrolyte crystalloid and colloid intravenous fluids. Anesth Analg 2006; 102: Brummel-Ziedins K, Whelihan MF, Ziedins EG, Mann KG. The resuscitative fluid you choose may potentiate bleeding. J Trauma 2006; 61: Kiraly LN, Differding JA, Enomoto TM, et al. Resuscitation with normal saline (NS) vs. lactated ringers (LR) modulates hypercoagulability and leads to increased blood loss in an uncontrolled hemorrhagic shock swine model. J Trauma 2006; 61: Wilcox CS. Regulation of renal blood flow by plasma chloride. J Clin Invest 1983; 71: Romanelli MF, Meissner MD, Fromm BS, Spears JR. Prominent ECG repolarization changes associated with intracoronary infusion of normal saline: comparisons with alternate coronary catheter flush solutions. Catheter Cardiovasc Interv 1999; 48: Unal M, Gorur K, Ozcan C. Ringer-lactate solution versus isotonic saline solution on mucociliary function after nasal septal surgery. J Laryngol Otol 2001; 115: Jackson EV Jr, Wiese J, Sigal B, et al. Effects of crystalloid solutions on circulating lactate concentrations: Part 1. Implications for the proper handling of blood specimens obtained from critically ill patients. Crit Care Med 1997; 25: Didwania A, Miller J, Kassel D, Jackson EV Jr, Chernow B. Effect of intravenous lactated Ringer s solution infusion on the circulating lactate concentration: Part 3. Results of a prospective, randomized, double-blind, placebo-controlled trial. Crit Care Med 1997; 25: Hamada T, Yamamoto M, Nakamaru K, Iwaki K, Ito Y, Koizumi T. The pharmacokinetics of D-lactate, L-lactate and acetate in humans. Masui 1997; 46: Murakami M, Nomiyama S, Tanaka R. The effects of L- and DL-lactated Ringer s solution on lactate and glucose metabolism and endocrine response during upper abdominal surgery. Masui 1988; 37: Kono K, Inoue K, Tsuda S, Ishida H. Intraoperative use of L-lactated Ringer s solution. Masui 1984; 33: Etoh H, Yano H, Ohsumi A, Sakai T, Fujiwara K, Takaori M. Study on metabolic effects of lactated Ringer s solution-l-lactate vs DL-lactate sodium. Masui 1984; 33: Kitamura S, Ikegaki J, Hoshino Y, et al. The effects of L- and D.L-lactated Ringer s solutions on body fluids and lactate metabolism during surgery. Masui 1984; 33: Inoue T, Yamada T, Furuya E, Tagawa K. Ca2+-induced accumulation of pyrophosphate in mitochondria during acetate metabolism. Biochem J 1989; 262: