metabolic effects of plasma expanders

Metabolic effects of plasma expanderstatm_1137 10..21

LAURENT MULLER, MD, MSC & JEAN-YVES LEFRANT, MD

Department of Surgical Intensive Care,Nîmes University Hospital, Nîmes,France

Correspondence to:Dr L. Muller, Service de RéanimationChirurgicale, Pôle AnesthésieRéanimation Douleur Urgences,Hôpital Universitaire Carémeau, Placedu Professeur Robert Debré, 30029Nîmes Cedex 9, FranceE-mail: [email protected]

Publication dataSubmitted: 9 April 2010Resubmitted: 8 June 2010Accepted: 9 June 2010

Keywords• Colloids• Crystalloids• Plasma expansion• Safety• Volume replacement

SUMMARY

All plasma expanders exert metabolic effects – either favorable effectssuch as correction of hypovolemia-induced lactic acidosis or renal failure,or unwanted adverse effects such as hypotonic solution-inducedhyponatremia, acid-base disorders such as hyperchloremic acidosis andeffects related to buffers associated with plasma expanders (lactate,acetate). The use of crystalloids alone is associated with a risk of inter-stitial fluid overload responsible for organ dysfunction in anesthesia andintensive care. The exclusive and prolonged use of high doses of colloidsis associated with a risk of severe renal failure. Crystalloids appear to besufficient to correct tissue hypoperfusion induced by moderate hypov-olemia. Alternating prescription of isotonic saline and balanced plasmaexpanders should be able to avoid the metabolic complications of thesetwo types of crystalloids. Administration of colloids is safe when therecommended maximum doses are observed. When administration of acolloid is indicated, a latest generation hydroxyethyl starch solutionappears to present the best benefit/risk ratio.

INTRODUCT ION

All plasma expanders exert metabolic effects – eitherfavorable effects such as correction of hypovolemia-induced lactic acidosis or renal failure, or unwantedadverse effects such as hypotonic solution-inducedhyponatremia, acid-base disorders such as hyper-chloremic acidosis and effects related to buffersassociated with plasma expanders (lactate, acetate).This review will focus on tissue perfusion disordersinduced by excess crystalloids and on renal toxicity ofcolloids.

AVAILABLE PLASMA EXPANDERS ,THEORET ICAL PLASMA-EXPANDINGPROPERT IES AND EXPECTED BENEF ITS :A COMPLEX ISSUE

The plasma expanders most widely used are isotoniccrystalloids and synthetic colloids.1–3 Albumin, due to itscost and the traceability requirements inherent to allblood-derived products, is rarely prescribed as first-linetreatment.3,4 The synthetic colloids most widely used arehydroxyethyl starch (HES) solutions and gelatins.2–4 Inmany countries, dextrans are no longer marketed

Transfusion Alternatives in Transfusion Medicine TATM

© 2010 The AuthorsTransfusion Alternatives in Transfusion Medicine © 2010 Medical Education Global Solutions

doi: 10.1111/j.1778-428X.2010.01137.x

10

because of their adverse effects, especially anaphylacticreactions.3,4

It should be noted that nonionic 5% or 10% dextrosesolution is not a plasma expander, as the volume remain-ing in the vascular compartment after intravenous infu-sion is insignificant because of rapid diffusion to allcompartments of the body (Table 1).5 These solutionsbehave like free water and infusion of large volumesinduces a risk of hyponatremia and water intoxicationwith potentially serious cerebral consequences.

The low cost and few apparent adverse effects ofcrystalloids justify their widespread use despite a plasmaexpansion property of the order of 20% (Table 2). Syn-thetic colloids provide a plasma expansion property ofclose to 100% (80%–120% depending on the product;Table 2) but with a risk of anaphylaxis, renal failure andclotting disorders.

The plasma expansion property of a solution theoreti-cally has direct metabolic effects. A product with a highexpansion property corrects blood volume more effec-tively, limiting the risks of tissue hypoperfusion respon-sible for lactic acidosis. The expected benefit of a colloidshould therefore be logically greater than that of a crys-talloid, but this superiority has never been demonstrated.

This absence of clear-cut superiority of a particularplasma expander can be explained by the followingelements:

• Despite the apparent safety of crystalloids, largevolumes must be infused when these compounds areused alone, with a risk of interstitial fluid overloadthat can be harmful during the perioperativeperiod.6–8

• Furthermore, the plasma expansion property of acrystalloid can be 20% higher in certain conditions,9

especially in the case of moderate hypovolemia.• Finally, despite the toxic effects related to massive

use of colloids, compliance with the maximumrecommended dose and the development of lowmolar substitution HES limit the risks of adverseeffects.

These points will be discussed below.

I SOTONIC CRYSTALLOIDS : ADVANTAGES ANDDISADVANTAGES

The available isotonic crystalloids are isotonic salinesolution (ISS) and Ringer’s lactate (RL). Although thesetwo solutions have comparable plasma expansion prop-erties, their chemical composition is very different(Table 3), accounting for marked clinical differences.

Advantages

The low plasma volume expansion property of crystal-loids contrasts with their widespread use in routineclinical practice.3 This paradox can probably be partlyexplained by the results of studies conducted on akinetic model of vascular elimination of RL in healthyvolunteers.9 When RL was administered to consciousnormovolemic subjects, the elimination rate constant(i.e. the rate of diffusion to the interstitial compartment)was 133 mL/min. When subjects were made hypov-olemic after withdrawing 450 mL and then 900 mL ofblood, this constant decreased to 100 mL/min and then

Table 1. Comparative volumes of the various solutions necessary to obtain plasma volume expansion of 1000 mL inhealthy subjects

Plasma volumeexpansion (mL)

Volumeinfused

Modification ofinterstitial volume (mL)

Expansion of theintracellular compartment

5% albumin 1000 1,00025% albumin 1000 250 -7505% dextrose 1000 14,000 +3700 9300Ringer’s lactate/isotonic saline 1000 4,700 +3700HES 130/0.4 1000 1,000

Adapted from Zornow and Prough.5

MULLER & LEFRANT | METABOLIC EFFECTS OF PLASMA EXPANDERS 11

© 2010 The AuthorsTransfusion Alternatives in Transfusion Medicine © 2010 Medical Education Global Solutions • 11 (Suppl. 3), 10–21

34 mL/min, respectively, i.e. a fourfold decrease of theelimination constant. The plasma-expanding efficacy ofa crystalloid is therefore greater than the classical valueof 20% during the acute phase of hypovolemia and canexplain the high level of confidence of practitioners inthese plasma expanders during situations of moderatehypovolemia. This is no longer true in patients withcapillary leak syndrome in the context of a systemicinflammatory response regardless of the origin, as oldstudies10 suggest that the efficacy of crystalloidsdecreases with the volume infused during hemorrhagicshock, probably as a result of increased capillary leak.

Disadvantages

The tissue edema (particularly gastrointestinal) inherentto administration of large volumes of crystalloids couldalter tissue oxygenation and participate in the develop-ment of multiple organ dysfunction syndrome (MODS).In a rat model of hemorrhage, Moon et al.11 showed thatplasma expansion by crystalloids induced severe edemaof cardiac tissue and the gastrointestinal tract, the latterbeing the principal factor incriminated clinically in thedevelopment of MODS. In a similar model, Wang et al.12

showed that plasma expansion with LR of four times the

Table 2. General characteristics of the main plasma expanders

Type of solutionOsmolarity(mmol/L) Composition of solution

Plasma expansion property(1 = 100% of volume infusedremains in the intravascularcompartment)

Duration ofaction (hour)

Isotonic crystalloidsRL 277 NaCl 0.6% NaCl + lactate 0.19 –0.9% NaCl 308 0.9% NaCl 0.22 –Isofundine® 304 0.8% NaCl + acetate 0.2

Hypertonic crystalloidsHyperhes® 2464 7.2% NaCl + HES 2–3 –RescueFlow® 2400 7.5% NaCl + dextran 70 2–3 –

Natural colloid: albumin4% 250–350 0.9% NaCl 0.7 1–320% 300 0.9% NaCl 3.5 1–3

Synthetic colloidsGelatins

Modified fluid gelatins3%: Geloplasma® 320 RL 0.8–1 33%: Plasmagel® 350 0.9% NaCl 0.8–1 32.5%: Plasmagel® sodium-free 320 5% D 0.8–1 34%: Gelofusine® 308 0.9% NaCl 0.8–1 3

Urea-linked gelatin3.5%: Haemaccel® 300 0.9% NaCl 0.8–1 3

HESsHigh molecular weight (� 200 kD)

6%: Elohes® 304 0.9% NaCl 1–1.4 12–246%: HAES-steril® 308 0.9% NaCl 1–1.4 3–66%: Hemohes® 310 0.9% NaCl 1–1.4 3–610%: Hemohes® 310 0.9% NaCl 1.2–1.5 3–6

Low molecular weight (130 kD)6%: Voluven® (maize starch) 308 0.9% NaCl 1 66%: Venofundin® (potato starch) 309 0.9% NaCl 1 6

HES, hydroxyethyl starch; kD, kiloDalton; RL, Ringer’s lactate.

12 MULLER & LEFRANT | METABOLIC EFFECTS OF PLASMA EXPANDERS


volume lost did not correct the hepatic, renal, splenic,muscle and intestinal microcirculation evaluated bylaser Doppler, despite correction of filling pressures(doubling of central venous pressure). In a rat model ofhead injury,13 plasma expansion by crystalloids (saline)worsened cerebral edema and induced muscle andmesenteric edema. This effect was less marked when amacromolecule or whole blood was used.

A study performed in healthy subjects showed thatinfusion of 40 mL/kg of RL induced significant limita-tion of peak expiratory flow rate and persistent weightgain at the 24th hour,14 suggesting tissue retention ofcrystalloids that a healthy body is unable to completelyeliminate in 24 hours. In pathological situations, a posi-tive fluid balance is associated with excess mortality inpatients with acute respiratory distress syndrome oracute renal failure.15,16 In traumatology, supranormalresuscitation with RL in patients with multiple injuries isresponsible for mesenteric edema with abdominal com-partment syndrome and excess mortality because ofMODS.17 Similarly, the volume of crystalloids adminis-tered in the pre-hospital setting and in the emergencydepartment to patients admitted for complicated frac-tures of the lower limbs is an independent risk factor fordevelopment of abdominal compartment syndrome,even in the absence of associated abdominal trauma.18

In elective abdominal surgery, restriction of the volumesof crystalloids during the perioperative period improveshealing, reduces the incidence of nausea and vomiting,accelerates return of bowel function and shortens thehospital stay.7,8,19–21

According to Stewart’s model, high doses (> 30 mL/kg)of ISS induce hyperchloremic acidosis because of thesupraphysiological chloride concentration (154 mmol/L)(strong anions).22 Although the clinical importance ofhyperchloremic acidosis has not been formally demon-strated, it could be a risk factor for hemodynamic dete-rioration in patients with shock.23 Hyperchloremicacidosis is not observed with the same doses of RL asa result of a chloride concentration close to that ofplasma24 and the presence of lactate ions. RL is conse-quently considered to be a balanced plasma expander.As the lactate ion is a base, the use of large volumes ofRL carries a risk of metabolic alkalosis. As RL containspotassium, this solution is classically contraindicated inpatients with hyperkalemia. A new balanced isotoniccrystalloid (Ringerfundin®, B. Braun, Melsungen,Germany) has recently been released onto the market,using a buffer composed of acetate and malate, precur-sors of bicarbonate. The superiority of these new plasmaexpanders compared with RL or ISS has not beendemonstrated. Nevertheless, hyperchloremic acidosisinduced by ISS can also be responsible for hyperkale-mia. In a randomized study comparing plasma expan-sion with RL versus ISS in 51 patients undergoing renaltransplantation, the incidence of serious hyperkalemiawas significantly higher in the ISS group because of thehigh rate of acidosis.25 Finally, RL is associated with arisk of hyponatremia (sodium concentration: 6 g/L) andhypoosmolarity (osmolarity: 277 mmol/L) that could bepotentially harmful in acute cerebral diseases.26

HYPERTONIC CRYSTALLOIDS

Hypertonic crystalloids are essentially represented by7.5% hypertonic saline solution (HSS). This solution ismarketed in combination with a macromolecule theo-retically designed to increase its duration of action. Theassociated macromolecule can be an HES (HyperHAES®,Fresenius Kabi, Bad Homburg, Germany) or a 70 kDdextran (RescueFlow®, BioPhausia, Knivsta, Sweden). Inview of the anaphylactic risks related to dextrans, thechoice of HSS with HES appears to be more logical.Because of the hypertonicity of these solutions, infused

Table 3. Respective compositions of plasma, isotonicsaline solution (ISS) and Ringer’s lactate (RL)

Concentrations (mmol/L) Plasma ISS RL

Na+ 142 154 130K+ 5 0 4Cl- 103 154 108Ca++ 2.5 0 0.91Mg++ 1 0 0HCO3

- 27 0 0Lactate 5 0 27.6Osmolarity 295 308 277*

ISS is slightly hypertonic inducing a risk of hyperchloremicacidosis because of its supraphysiological concentration ofstrong chloride anions. RL does not induce acidosis in view ofits physiological chloride concentration (balanced crystalloid)but is associated with a risk of metabolic alkalosis at highdoses. RL is a hypotonic solution, which must be used cau-tiously in patients with neurological lesions. *Osmolarity canvary according to the manufacturer.



volumes greater than 250 mL can induce serum sodiumconcentrations higher than 160 mmol/L, responsible forserious complications.27 Experimentally, HSS exertsmany favorable effects on the microcirculation andimmune functions.28,29 On the macrocirculation, HSSinduces plasma expansion two to three times the infusedvolume with an increase of filling pressures and cardiacoutput during hemorrhagic shock30 and septic shock.31–33

HSS also exerts direct positive inotropic effects.33 Exces-sively rapid administration is associated with hypoten-sion and arrhythmias. The hemodynamic effects of HSSare only transient, as the duration of action of HSS is2–3 hours during septic shock, even when a macromol-ecule is associated, suggesting a limited effect of addi-tion of a macromolecule.31,32 A cohort study published in1997 suggested that hypotensive head injury patientshad a twofold higher survival when they received acombination of HSS-dextran, but the power of thisstudy was limited34 and these results were not con-firmed.35 With a low level of proof, the most severelyinjured patients, particularly patients with penetratinginjuries, therefore, appear to benefit from a combinationof HSS-macromolecule, particularly in the pre-hospitalsetting.36,37

COLLOIDS : ADVANTAGES ANDDISADVANTAGES OF THE VARIOUSPLASMA EXPANDERS

Many practice studies have shown that despite a heatedcontroversy, colloids are very frequently used.1–4 Themost widely prescribed colloids are HES and gelatins.3,4

Albumin is rarely used as first-line plasma expander anddextrans have been abandoned in many countriesbecause of their adverse effects.3,4

Albumin

Albumin is a small endogenous protein. The total quan-tity of albumin in the body is 4–5 g/kg, one-third ofwhich is located in the vascular compartment and 2/3 inthe interstitial compartment. Despite its small size,albumin does not cross the vessel wall in healthy sub-jects. This fragile equilibrium appears to be easilybroken in disease states, as albumin leakage is fre-quently observed during hypovolemic shock or majorsurgery.38 These findings also apply to sepsis, in whichadministration of radioisotope-labeled albumin in septic

patients shows major albumin leakage into the intersti-tial compartment with the formation of edema.39

Infusion of 4% albumin is responsible for little or noelevation of the oncotic pressure in intensive carepatients.40 Administration of albumin in intensive carepatients therefore appears to be logical, as patients withhypoalbuminemia have a poor prognosis.41,42 A meta-analysis43 of 30 randomized studies (1419 patients) con-cluded that 6% albumin was associated with excessmortality: one attributable death per 17 patients, butthese conclusions were refuted by three recent studies.The randomized, controlled Saline versus Albumin FluidEvaluation (SAFE)44 study based on 6997 intensive carepatients compared plasma expansion with isotonicsaline versus 4% albumin. No mortality difference wasobserved between the two groups. This study showedthat albumin does not induce any specific mortality witha comparable plasma expansion efficacy to that of ISS.Another meta-analysis published in 2004 concludedthat albumin is the colloid that induces the fewestadverse effects.45 Finally, a recent review of the litera-ture, based on 71 randomized, controlled trials (3782patients) comparing albumin with other plasma expand-ers, suggested that administration of albumin decreasesthe morbidity of patients admitted to the intensive careunit.41 This effect was more marked in patient groupswith the lowest serum albumin (burns, cirrhosis, etc.)than in surgery or traumatology patients. The authorsemphasized ‘the obvious protective effect’ of maintain-ing normal oncotic pressure by administration ofalbumin in terms of fluid balance.

The effects of albumin correspond to both plasmaexpansion and correction of hypoalbuminemia (‘albuminmedicinal product’). In a randomized study published in2006, correction of hypoalbuminemia in intensive carepatients decreased organ dysfunction.46 However, con-centrated albumin must be used cautiously. A recentcohort study suggested that administration of hyper-oncotic albumin in patients in shock could induce renalfailure and may be associated with excess mortality.47

In summary, isotonic albumin presents an excellentsafety, but its plasma expansion property decreasesrapidly in patients with capillary leak syndrome. Nomortality has been attributed to the use of this molecule.However, because of the high cost of this product, therisk of transmission of prion diseases and its moderateexpansion property, there are now very few indicationsfor the use of albumin for emergency plasma expansion.



The use of albumin to correct hypoalbuminemia inintensive care patients is probably more legitimate thanfor plasma expansion. Controlled studies are currentlyunderway to clarify this issue.

Gelatins

Gelatins are derived from breakdown of bovine bonecollagen. These products have a short duration of action(2–3 hours) and are essentially eliminated by glomerularfiltration. Their plasma expansion property is limited(80%), as about 20% of the quantity administeredrapidly enters the interstitial compartment. The effectson coagulation are more limited than those of dextrans,but in vitro aggregability tests are modified by gelatins.Gelatins can even be considered to be procoagulant.48 InFrance, gelatins are responsible for 95% of anaphylacticreactions involving a colloid.49

HESs (Table 4)

HESs are natural polysaccharides structurally close toglycogen, composed of long chains of branched glucosepolymers (amylopectin) substituted by a hydroxylradical either at the C2 position or the C6 position. TheC2 substitution rate is higher because C6 (C2 > C6) is abranch site of the molecule. Linear branching occurs atC4. The unbranched molecule is called amylose and thebranched molecule is called amylopectin. In vivo,plasma amylase degrades the initial molecule, with an invitro molecular weight (MW) ranging between 100 and1000 kiloDaltons (kD), into lower MW molecules thatremain oncotically active. The in vivo MW results fromprogressive metabolism of the initial molecule and con-stitutes the main determinant of the plasma expansionproperty and safety.

The number of substituted hydroxyl radicals, whichprevent breakdown by amylase, is expressed by molarsubstitution rate, with values ranging from 0 to 1. Theinitial industrial strategy was therefore to createintensely substituted high MW HES in order to ensure amaximum and prolonged expansion property. This pos-tulate was subsequently rejected because it was respon-sible for numerous and severe adverse effects such asrenal failure and acquired von Willebrand coagulopa-thies. It has now been clearly demonstrated that thedeterminant of clinical HES-induced coagulopathy is ahigh molar substitution rather than a high MW.50

Inversely, a high MW largely determines the nephro-toxic effects of HES. The manufacturers’ strategy wastherefore to find a way of synthesizing low MW HES(130 kD) while preserving their plasma expansion prop-erty and a duration of action lasting several hours. Asamylase acts at position C1, any substitution close tothis site (especially C2) will limit breakdown of themolecule, thereby preserving its plasma expansionproperty and its half-life.

The C2/C6 ratio, representing the number of carbonssubstituted at C2, determines the in vivo stability of theproduct. For older generation HES, the C2/C6 ratio wasof the order of 5/1 to 5/3. Voluven® (Fresenius Kabi, BadHomburg, Germany) has a C2/C6 ratio of 9/1, explainingits plasma expansion property and duration of actioncomparable to those of earlier products despite a lowerMW (130 kD versus 200–450 previously) and a lowermolar substitution (0.4 versus 0.5).

The renal safety provided by a low MW remains to bedemonstrated on great series, although several recentstudies argue in favor of the excellent renal safety of thelatest generation HES.51–53 The available HESs wereusually obtained from corn, as this substrate has a highcontent of branched molecules such as amylopectin,

Table 4. Simplified physicochemical classification of HES solutions

Origin MW (kD) Concentration (%) Molar substitution (OH/glucose) C2/C6 ratio (OH C2/OH C6)

Voluven® Corn 130 6 0.4 9/1Venofundin® Potato 130 6 0.42 6/1Hemohes® Corn 200 6 or 10 0.5HAES-steril® Corn 200 6 0.5 5/1Elohes® Corn 200 6 0.6 9/1

C2, carbon 2; C6, carbon 6; HES, hydroxyethyl starch; kD, kiloDaltons; MW, molecular weight; OH, number of hydroxyl radicals.



analogous to glycogen. A new HES obtained frompotato starch (Venofundin®, B. Braun, Melsungen,Germany) has been very recently marketed by B.Braunwith the same apparent characteristics (6% HES130:0.42 6/1) as its direct competitor, Voluven (6% HES,130:0.4 9/1), marketed for several years by FreseniusKabi. The main difference between the two productsconcerns the C2/C6 ratio, which is 6/1 for Venofundin(versus 9/1 for Voluven), which would logically confer ashorter life span to this new HES. Venofundin alsocomprises fewer branched chains (less amylopectin,more amylose), which could confer allergenic propertiesand interferences with certain substances, but this pointneeds to be clarified. Finally, an experimental studysuggested more marked alteration of coagulation withpotato HES (200:0.5) compared with corn HES(200:0.5),54 but the results of this already old study havenot been confirmed. The characteristics of the variousHESs are summarized in Table 4.

HESs are the least allergenic of all colloids. In a 1-yearFrench study conducted in 2000 on 518 anaphylacticreactions, 21 cases were attributed to colloids with onlyone case related to HES.49

The main controversy raised by the use of HES con-cerns their renal safety. The renal toxicity of HES isrelated to the hyperoncotic nature of the solutionsresponsible for decreased glomerular filtration47 andtubular deposits of HES, demonstrated by the presenceof vacuoles in proximal tubules.55 An increase of urinaryviscosity at the tubular level could constitute a thirdmechanism of toxicity, although its role has been lessclearly demonstrated. All studies performed over the last15 years show that renal complications of HES arerelated to prolonged administration of high doses ofHES at high concentrations (> 6%) and/or with a highMW (200 kD) and high molar substitution (� 0.5). Thispoint is clearly illustrated by the results of the veryrecent Efficacy of Volume Substitution and InsulinTherapy in Severe Sepsis56 study, which showed thatprolonged use (21 days) of a highly concentrated HES(10%) with a high MW (200 kD) and high molar substi-tution (0.5) in patients with septic shock induced anincreased rate of extrarenal clearance, which is directlycorrelated with the cumulative dose. However, a retro-spective study on more than 3000 intensive carepatients failed to demonstrate any harmful effect of HESon renal function.4 A very recent study in the context ofrenal transplantation showed that the use of 6% HES

130:0.4 9/1 (Voluven) in donors appeared to limit theharmful effects on recipient renal function57 initiallyobserved with high MW, strongly substituted HES (6%HES 200:0.6).55 Similarly, four very recent studies in theperioperative period of major cardiovascular surgerydemonstrated the absence of any harmful effects of HES130:0.4 compared with other natural or synthetic mac-romolecules.51,52,58,59 Overall, in the perioperative setting,administration of latest generation HES (HES 130:0.49/1) must be considered to be safe when the recom-mended doses (50 mL/kg) are observed. In patients withseptic shock, when it is decided to administer a colloid,6% HES 130:0.4 presents the best benefit–risk balancewhen the maximum dose is observed.60

Like all colloids, HESs interfere with coagulation, to agreater extent than albumin and gelatins and to a lesserextent than dextrans.61 The effects on coagulation aredirectly proportional to MW and molar substitution.62

This hypocoagulability is related to a disturbance ofplatelet aggregability because of interference of HESwith factor VIII:C.63–65 In acute settings, HESs do notsignificantly affect coagulation, even at doses as high as38 mL/kg for HES 200:0.566 and 50 mL/kg for HES130:0.4.67,68 A recent review of the literature based onseven trials and 449 patients showed that, when thesemaximum doses are observed, this coagulopathy has noclinical consequences and is biologically reduced by theuse of latest generation HES (6% 130:0.4).65

Finally, HES could exert favorable effects on themicrocirculation. Two very recent experimental studiessuggested that targeted intraoperative prescription ofHES 130:0.4 could have favorable microcirculatoryeffects on the splanchnic circulation and could preservethe vitality of gastrointestinal anastomoses.69,70

In summary, HESs are the least allergenic colloids.Their adverse effects on renal function and coagulationhave no clinical consequences during the perioperativeperiod when the recommended maximum doses areobserved. The validated clinical experience of HES 130/0.4/9-1 based on numerous publications offers a supple-mentary safety margin justifying an increase of therecommended maximum dose from 33 to 50 mL/kg/day.

CHOICE BETWEEN CRYSTALLOIDSAND COLLOIDS

The essential advantage of a colloid is that its expansionproperty is always higher71,72 and more rapid73 than that



of a crystalloid, which justifies the prescription of col-loids in situations of severe hypovolemia. Even if, inpractice, the efficacy ratio between the two types ofplasma enhancers is not 1/4 or 1/5, as predicted byTable 1, a colloid will always be at least twice as effec-tive in terms of plasma expansion compared with acrystalloid.71 The concept of severity of hypovolemiawas adopted as a criterion of choice of a colloid by the1997 guidelines. In these guidelines, colloids were indi-cated as first-line treatment when systolic blood pres-sure was less than 80 mmHg and/or for blood lossestimated to be more than 20% of the blood mass.74

However, the concept of hemodynamic severity israrely reported in studies comparing crystalloids andcolloids.

More than a hundred studies designed to answer thisquestion have been conducted over the last 30 years,without providing a clear answer.75 A first meta-analysis, published in 1989 by Velanovich76 on septicand trauma patients, demonstrated a overall mortalitydifference of 5.7% in favor of crystalloids. In thetrauma patient subgroup, a mortality difference of12.3% was observed in favor of crystalloids, whichappears to corroborate the previous idea that the effi-cacy of crystalloids in hypovolemic patients is greaterthan that observed in healthy subjects. In the septicpatient group, a mortality difference of 7.8% wasobserved in favor of colloids, which is only logical inview of the capillary permeability disorders character-istic of sepsis. In 1998, Schierhout and Roberts analyzed26 studies including 1622 patients receiving crystal-loids, 1422 receiving colloids and 38 receiving hyper-tonic solutions in the settings of trauma, burns, sepsisor major surgery.77 The main result was an excess mor-tality of 4% in patients receiving colloids. In 1999, Choiet al. reviewed 17 studies comprising a total of 814patients75 and did not observe any difference for theincidence of pulmonary edema or the length of hospitalstay between the colloid group and the crystalloidgroup. In the trauma patient subgroup, a mortality dif-ference was observed in favor of crystalloids(OR = 0.39; CI = 0.17–0.89). One of the major limita-tions of these two meta-analyses is that they includedold studies (before 1980) corresponding to products andtreatment practices that are very different from currentstandard practice. A systematic review of the literaturepublished in 2004 concluded that there are no strongarguments to recommend either type of plasma

expander in sepsis.78 The Cochrane group published twometa-analyses in 2004 and 2007 comprising almost8000 trauma, surgery and burns patients.79,80 The con-clusions of this study were the absence of proof of thesuperiority of colloids in this type of patient. Theauthors of this study added the deliberately provocativesentence to their conclusion: ‘As colloids are not asso-ciated with an improvement in survival, and as they aremore expensive than crystalloids, it is hard to see howtheir continued use in these patients can be justifiedoutside the context of randomized controlled trials.’ TheSAFE study, published in 2004,44 which constitutes thelargest patient series (more than 7000 patients) includedin a randomized controlled study designed to comparetwo plasma expanders did not show any differencebetween a crystalloid (ISS) and a colloid (albumin).

The failure of published studies to clearly resolve thisissue may be due to three reasons. First, both the SAFE44

study and the meta-analysis cited above76–80 studiedheterogeneous patient groups. Second, the main end-point of these studies was 28-day mortality, but, withinthis timeframe, the effect of a plasma expander inter-feres with many other factors influencing survival ofpatients with acute circulatory insufficiency. Third, thelevel of severity was not taken into account. No studyhas been designed to determine whether administrationof a colloid during severe hypovolemia (systolic bloodpressure < 80 mmHg for example) corrects blood pres-sure more rapidly and more effectively than a crystal-loid. The results could be more informative if severitywere taken into account. To illustrate this point of view,patients included in the SAFE study had a low level ofseverity (more than 50% did not present any organdysfunction) and few frank criteria of hypovolemia werepresent on inclusion (more than 50% of patients had amean blood pressure > 75 mmHg, a pulse < 100 bpm, anhourly urine output > 90 mL/hour and central venouspressure > 8 mmHg). This moderate severity couldexplain the absence of observed difference between thetwo products. In a study based on a very homogeneousgroup of patients (children with severe sepsis secondaryto dengue fever) in which the primary endpoint was theefficacy of plasma expansion during the first days oftreatment, the authors showed that use of a crystalloid(RL) in moderately severe patients (seen rapidly) was aseffective as use of a colloid, although hemodynamicparameters were corrected more rapidly with a colloid.73

Two types of colloids were used in this study: dextran 70



and HES 200/0.5. The comparable efficacy of the twocolloids but the higher incidence of allergic reactionswith dextran led to the conclusion that HESs are themacromolecules of choice in the most seriously illpatients.

All of the disappointing results reported above con-trast with those of studies conducted during the peri-operative period, in which administration of colloidsappears to be associated with a reduction of postop-erative complications. By definition, these studiesincluded less severe patients, presenting almost pureintraoperative hypovolemia. These patient groups weretherefore more homogeneous than intensive carepatients and presented fewer confounding factors. Theendpoint was not late mortality but the postoperativecomplication rate and length of hospital stay. Use ofHES during caesarean sections significantly decreasedthe incidence and severity of hypotension, limitedreflex tachycardia and decreased adrenaline consump-tion by one-third.81 The study by Venn et al.,82 con-ducted in patients operated for hip fracture, showedthat intraoperative monitoring of blood volume limitedpostoperative complications and shortened the lengthof hospital stay. This more favorable prognosis wasaccompanied by greater plasma expansion, including asignificantly higher proportion of colloids. Similarresults were reported after heart surgery.83 In a studyconducted in major non-cardiac surgery, administra-tion of HES (balanced or not) compared with RLallowed a reduction of total volumes administered, asignificant reduction of nausea and vomiting, a reduc-tion of edema and blurred vision and a markedimprovement of pain scores.21

CONCLUS ION

The use of crystalloids alone is associated with a risk ofinterstitial fluid overload responsible for organ dysfunc-tion, in anesthesia21 and intensive care.18 The exclusiveand prolonged use of high doses of colloids is associatedwith a risk of severe renal failure.56 Crystalloids appearto be sufficient to correct tissue hypoperfusion inducedby moderate hypovolemia. Alternating prescription ofISS and balanced plasma expanders should be able toavoid the metabolic complications of these two types ofcrystalloids. Administration of colloids is safe when therecommended maximum doses are observed. Targetedprescription of colloids in situations of severe hypo-volemia is justified because of their rapid action andhigh and prolonged plasma expansion properties. Whenadministration of a colloid is indicated, a latest genera-tion HES (130/0.4) appears to present the best benefit/risk balance. The quest for the ideal plasma expander isprobably vain. As for other therapeutic groups, such asantibiotics or vasoactive agents, the availability of alarge range of plasma expanders should allow prescrip-tion adapted to various clinical settings. The best meta-bolic yield (relationship between plasma expansionbenefits and adverse effects) of a plasma expander canbe obtained only by weighing up, for each prescription,the type of clinical situation (simple dehydration ormassive blood loss), the level of severity (assessed bystrict monitoring) and the volumes already infused.

CONFL ICT OF INTEREST

The authors declare no conflict of interest.

REFERENCES

1 Kastrup M, Markewitz A, Spies C, et al.Current practice of hemodynamic moni-toring and vasopressor and inotropictherapy in postoperative cardiac surgerypatients in Germany: results from apostal survey. Acta Anaesthesiol Scand2007; 51: 347–58.

2 Fluid Study Investigators for the Scandi-navian Critical Care Trials Group. Prefer-ences for colloid use in Scandinavianintensive care units. Acta AnaesthesiolScand 2008; 52: 750–8.

3 Schortgen F, Deye N, Brochard L.Preferred plasma volume expanders forcritically ill patients: results of an inter-national survey. Intensive Care Med2004; 30: 2222–9.

4 Sakr Y, Payen D, Reinhart K, et al. Effectsof hydroxyethyl starch administration onrenal function in critically ill patients. BrJ Anaesth 2007; 98: 216–24.

5 Zornow MH, Prough DS. Fluid manage-ment in patients with traumatic braininjury. New Horiz 1995; 3: 488–98.

6 Gan TJ, Soppitt A, Maroof M, et al. Goal-directed intraoperative fluid administra-tion reduces length of hospital stay aftermajor surgery. Anesthesiology 2002; 97:820–6.

7 Lobo DN, Bostock KA, Neal KR, et al. Ef-fect of salt and water balance on recoveryof gastrointestinal function after electivecolonic resection: a randomised con-trolled trial. Lancet 2002; 359: 1812–18.

8 Brandstrup B, Tonnesen H, Beier-Holgersen R, et al. Effects of intravenous



fluid restriction on postoperative compli-cations: comparison of two perioperativefluid regimens: a randomized assessor-blinded multicenter trial. Ann Surg 2003;238: 641–8.

9 Drobin D, Hahn RG. Volume kinetics ofRinger’s solution in hypovolemic volun-teers. Anesthesiology 1999; 90: 81–91.

10 Cervera ALMG. Crystalloid distributionfollowing hemorrhage and hemodilution:mathematical model and prediction of op-timum volumes for equilibration at nor-movolemia. J Trauma 1974; 14: 506–20.

11 Moon PF, Hollyfield-Gilbert MA, MyersTL, Kramer GC. Effects of isotonic crys-talloid resuscitation on fluid compart-ments in hemorrhaged rats. Shock 1994;2: 355–61.

12 Wang P, Hauptman JG, Chaudry IH.Hemorrhage produces depression inmicrovascular blood flow which persistsdespite fluid resuscitation. Circ Shock1990; 32: 307–18.

13 Drummond JC, Patel PM, Cole DJ, KellyPJ. The effect of the reduction of colloidoncotic pressure, with and withoutreduction of osmolality, on post-traumatic cerebral edema. Anesthesiology1998; 88: 993–1002.

14 Holte K, Jensen P, Kehlet H. Physiologiceffects of intravenous fluid administra-tion in healthy volunteers. Anesth Analg2003; 96: 1504–9; table of contents.

15 Sakr Y, Vincent JL, Reinhart K, et al.High tidal volume and positive fluidbalance are associated with worseoutcome in acute lung injury. Chest2005; 128: 3098–108.

16 Payen D, de Pont AC, Sakr Y, et al. Apositive fluid balance is associated with aworse outcome in patients with acuterenal failure. Crit Care 2008; 12: R74.

17 Balogh Z, McKinley BA, Cocanour CS,et al. Supranormal trauma resuscitationcauses more cases of abdominal com-partment syndrome. Arch Surg 2003;138: 637–42; discussion 642–3.

18 Madigan MC, Kemp CD, Johnson JC,Cotton BA. Secondary abdominal com-partment syndrome after severe extrem-ity injury: are early, aggressive fluidresuscitation strategies to blame? JTrauma 2008; 64: 280–5.

19 Holte K, Klarskov B, Christensen DS,et al. Liberal versus restrictive fluidadministration to improve recovery afterlaparoscopic cholecystectomy: a ran-domized, double-blind study. Ann Surg2004; 240: 892–9.

20 Nisanevich V, Felsenstein I, Almogy G,et al. Effect of intraoperative fluid man-agement on outcome after intraabdomi-nal surgery. Anesthesiology 2005; 103:25–32.

21 Moretti EW, Robertson KM, El-MoalemH, Gan TJ. Intraoperative colloid admin-istration reduces postoperative nauseaand vomiting and improves postopera-tive outcomes compared with crystalloidadministration. Anesth Analg 2003; 96:611–17; table of contents.

22 Scheingraber Rehm MSC, Finsterer U.Rapid saline infusion produces hyper-chloremic acidosis in patients undergo-ing gynecologic surgery. Anesthesiology1999; 90: 1265–70.

23 Kellum JA, Song M, Venkataraman R.Effects of hyperchloremic acidosis onarterial pressure and circulating inflam-matory molecules in experimental sepsis.Chest 2004; 125: 243–8.

24 Scheingraber S, Rehm M, Sehmisch C,Finsterer U. Rapid saline infusion pro-duces hyperchloremic acidosis in patientsundergoing gynecologic surgery. Anes-thesiology 1999; 90: 1265–70.

25 O’Malley CM, Frumento RJ, Hardy MA,et al. A randomized, double-blind com-parison of lactated Ringer’s solution and0.9% NaCl during renal transplantation.Anesth Analg 2005; 100: 1518–24; tableof contents.

26 Réanimation. R-SFdA: remplissage vas-culaire au cours des hypovolémiesabsolues ou relatives. Agence nationalepour le développement de l’évaluationmédicale. Texte des recommandations.Réan Urg 1997; 6: 335–41.

27 Dominguez TE, Priestley MA, Huh JW.Caution should be exercised whenmaintaining a serum sodium level> 160 meq/L. Crit Care Med 2004; 32:1438–9; author reply 1439–40.

28 de Carvalho H, Matos JA, Bouskela E,Svensjo E. Vascular permeability increaseand plasma volume loss induced byendotoxin was attenuated by hypertonicsaline with or without dextran. Shock1999; 12: 75–80.

29 Pascual JL, Ferri LE, Seely AJ, et al.Hypertonic saline resuscitation of hem-orrhagic shock diminishes neutrophilrolling and adherence to endotheliumand reduces in vivo vascular leakage.Ann Surg 2002; 236: 634–42.

30 Chiara O, Pelosi P, Brazzi L, et al. Resus-citation from hemorrhagic shock: experi-mental model comparing normal saline,

dextran, and hypertonic saline solutions.Crit Care Med 2003; 31: 1915–22.

31 Muller L, Lefrant JY, Jaber S, et al. [Shortterm effects of hypertonic saline duringsevere sepsis and septic shock]. Ann FrAnesth Reanim 2004; 23: 575–80.

32 Oliveira RP, Weingartner R, Ribas EO,Moraes RS, Friedman G. Acute haemody-namic effects of a hypertonic saline/dextran solution in stable patients withsevere sepsis. Intensive Care Med 2002;28: 1574–81.

33 Goertz AW, Mehl T, Lindner KH, et al.Effect of 7.2% hypertonic saline/6%hetastarch on left ventricular contractil-ity in anesthetized humans. Anesthesiol-ogy 1995; 82: 1389–95.

34 Wade CE, Grady JJ, Kramer GC, et al.Individual patient cohort analysis of theefficacy of hypertonic saline/dextran inpatients with traumatic brain injuryand hypotension. J Trauma 1997; 42:S61–5.

35 Cooper DJ, Myles PS, McDermott FT,et al. Prehospital hypertonic saline resus-citation of patients with hypotension andsevere traumatic brain injury: a random-ized controlled trial. JAMA 2004; 291:1350–7.

36 Wade CE, Grady JJ, Kramer GC. Efficacyof hypertonic saline dextran fluid resus-citation for patients with hypotensionfrom penetrating trauma. J Trauma 2003;54: S144–8.

37 Wade CE, Kramer GC, Grady JJ, FabianTC, Younes RN. Efficacy of hypertonic7.5% saline and 6% dextran-70 in treat-ing trauma: a meta-analysis of controlledclinical studies. Surgery 1997; 122: 609–16.

38 Hoye RC, Bennett SH, Geelhoed GW,Gorschboth C. Fluid volume and albuminkinetics occurring with major surgery.JAMA 1972; 222: 1255–61.

39 Ernest D, Belzberg AS, Dodek PM. Distri-bution of normal saline and 5% albumininfusions in septic patients. Crit CareMed 1999; 27: 46–50.

40 Mbaba Mena J, De Backer D, Vincent JL.Effects of a hydroxyethylstarch solutionon plasma colloid osmotic pressure inacutely ill patients. Acta AnaesthesiolBelg 2000; 51: 39–42.

41 Vincent JL, Dubois MJ, Navickis RJ,Wilkes MM. Hypoalbuminemia in acuteillness: is there a rationale for interven-tion? A meta-analysis of cohort studiesand controlled trials. Ann Surg 2003;237: 319–34.



42 Vincent JL, Sakr Y, Reinhart K, et al. Isalbumin administration in the acutely illassociated with increased mortality?Results of the SOAP study. Crit Care2005; 9: R745–54.

43 Cochrane Injuries Group AlbuminReviewers. Human albumin administra-tion in critically ill patients: systematicreview of randomised controlled trials.BMJ 1998; 317: 235–40.

44 Finfer S, Bellomo R, Boyce N, et al. Acomparison of albumin and saline forfluid resuscitation in the intensivecare unit. N Engl J Med 2004; 350:2247–56.

45 Barron ME, Wilkes MM, Navickis RJ. Asystematic review of the comparativesafety of colloids. Arch Surg 2004; 139:552–63.

46 Dubois MJ, Orellana-Jimenez C, Melot C,et al. Albumin administration improvesorgan function in critically ill hypo-albuminemic patients: a prospective,randomized, controlled, pilot study. CritCare Med 2006; 34: 2536–40.

47 Schortgen F, Girou E, Deye N, BrochardL. The risk associated with hyperoncoticcolloids in patients with shock. IntensiveCare Med 2008; 34: 2157–68.

48 Karoutsos S, Nathan N, Lahrimi A, et al.Thrombelastogram reveals hypercoagu-lability after administration of gelatinsolution. Br J Anaesth 1999; 82: 175–7.

49 Mertes PM, Laxenaire MC, Alla F. Ana-phylactic and anaphylactoid reactionsoccurring during anesthesia in France in1999–2000. Anesthesiology 2003; 99:536–45.

50 Kozek-Langenecker S, Scharbert G. Com-paring the effects of colloids on clot for-mation. Anaesthesia 2008; 63: 673–4;author reply 674.

51 Boldt J, Brosch C, Rohm K, Papsdorf M,Mengistu A. Comparison of the effects ofgelatin and a modern hydroxyethylstarch solution on renal function andinflammatory response in elderly cardiacsurgery patients. Br J Anaesth 2008; 100:457–64.

52 Mahmood A, Gosling P, Vohra RK. Ran-domized clinical trial comparing theeffects on renal function of hydroxyethylstarch or gelatine during aortic aneurysmsurgery. Br J Surg 2007; 94: 427–33.

53 Jungheinrich C, Scharpf R, Wargenau M,Bepperling F, Baron JF. The pharmacoki-netics and tolerability of an intravenousinfusion of the new hydroxyethyl starch130/0.4 (6%, 500 mL) in mild-to-severe

renal impairment. Anesth Analg 2002;95: 544–51; table of contents.

54 Jamnicki M, Zollinger A, Seifert B, et al.The effect of potato starch derived andcorn starch derived hydroxyethyl starchon in vitro blood coagulation. Anaesthe-sia 1998; 53: 638–44.

55 Cittanova ML, Leblanc I, Legendre C,et al. Effect of hydroxyethylstarch inbrain-dead kidney donors on renal func-tion in kidney-transplant recipients.Lancet 1996; 348: 1620–2.

56 Brunkhorst FM, Engel C, Bloos F, et al.Intensive insulin therapy and pentastarchresuscitation in severe sepsis. N Engl JMed 2008; 358: 125–39.

57 Blasco V, Leone M, Antonini F, et al.Comparison of the novel hydroxyethyl-starch 130/0.4 and hydroxyethylstarch200/0.6 in brain-dead donor resuscita-tion on renal function after transplanta-tion. Br J Anaesth 2008; 100: 504–8.

58 Hanart C, Khalife M, De Ville A, et al.Perioperative volume replacement inchildren undergoing cardiac surgery:albumin versus hydroxyethyl starch 130/0.4. Crit Care Med 2009; 37: 696–701.

59 Godet G, Lehot JJ, Janvier G, et al. Safetyof HES 130/0.4 (Voluven(R)) in patientswith preoperative renal dysfunctionundergoing abdominal aortic surgery: aprospective, randomized, controlled,parallel-group multicentre trial. Eur JAnaesthesiol 2008; 25: 986–94.

60 SFAR/SRLF: Conférence de consensus surla prise en charge hémodynamique dusepsis grave. 2005.

61 Egli GA, Zollinger A, Seifert B, et al.Effect of progressive haemodilution withhydroxyethyl starch, gelatin and albuminon blood coagulation. Br J Anaesth1997; 78: 684–9.

62 Franz A, Braunlich P, Gamsjager T, et al.The effects of hydroxyethyl starches ofvarying molecular weights on plateletfunction. Anesth Analg 2001; 92:1402–7.

63 Turkan H, Ural AU, Beyan C, Yalcin A.Effects of hydroxyethyl starch on bloodcoagulation profile. Eur J Anaesthesiol1999; 16: 156–9.

64 Omar MN, Shouk TA, Khaleq MA. Activ-ity of blood coagulation and fibrinolysisduring and after hydroxyethyl starch(HES) colloidal volume replacement. ClinBiochem 1999; 32: 269–74.

65 Kozek-Langenecker SA, Jungheinrich C,Sauermann W, Van der Linden P. Theeffects of hydroxyethyl starch 130/0.4

(6%) on blood loss and use of bloodproducts in major surgery: a pooledanalysis of randomized clinical trials.Anesth Analg 2008; 107: 382–90.

66 Vogt N, Bothner U, Brinkmann A, dePetriconi R, Georgieff M. Peri-operativetolerance to large-dose 6% HES 200/0.5in major urological procedures comparedwith 5% human albumin. Anaesthesia1999; 54: 121–7.

67 Kasper SM, Meinert P, Kampe S, et al.Large-dose hydroxyethyl starch 130/0.4does not increase blood loss and transfu-sion requirements in coronary arterybypass surgery compared with hydroxy-ethyl starch 200/0.5 at recommendeddoses. Anesthesiology 2003; 99: 42–7.

68 Neff TA, Doelberg M, Jungheinrich C,et al. Repetitive large-dose infusion ofthe novel hydroxyethyl starch 130/0.4 inpatients with severe head injury. AnesthAnalg 2003; 96: 1453–9; table ofcontents.

69 Hiltebrand LB, Kimberger O, ArnbergerM, et al. Crystalloids versus colloids forgoal-directed fluid therapy in majorsurgery. Crit Care 2009; 13: R40.

70 Kimberger O, Arnberger M, Brandt S,et al. Goal-directed colloid administra-tion improves the microcirculation ofhealthy and perianastomotic colon.Anesthesiology 2009; 110: 496–504.

71 McIlroy DR, Kharasch ED. Acute intra-vascular volume expansion with rapidlyadministered crystalloid or colloid in thesetting of moderate hypovolemia. AnesthAnalg 2003; 96: 1572–7; table ofcontents.

72 Verheij J, van Lingen A, Beishuizen A,et al. Cardiac response is greater forcolloid than saline fluid loading aftercardiac or vascular surgery. IntensiveCare Med 2006; 32: 1030–8.

73 Wills BA, Nguyen MD, Ha TL, et al. Com-parison of three fluid solutions for resus-citation in dengue shock syndrome. NEngl J Med 2005; 353: 877–89.

74 Réanimation. AnplddlémTdrSdrdlFSFdA:Remplissage vasculaire au cours deshypovolémies absolues ou relatives.Reanimation et médecine d’urgence1997.

75 Choi PT, Yip G, Quinonez LG, Cook DJ.Crystalloids vs. colloids in fluid resusci-tation: a systematic review. Crit CareMed 1999; 27: 200–10.

76 Velanovich V. Crystalloid versus colloidfluid resuscitation: a meta-analysis ofmortality. Surgery 1989; 105: 65–71.



77 Schierhout G, Roberts I. Fluid resuscita-tion with colloid or crystalloid solutionsin critically ill patients: a systematicreview of randomised trials. BMJ 1998;316: 961–4.

78 Vincent JL, Gerlach H. Fluid resuscitationin severe sepsis and septic shock: anevidence-based review. Crit Care Med2004; 32: S451–4.

79 Roberts I, Alderson P, Bunn F, et al.Colloids versus crystalloids for fluidresuscitation in critically ill patients.

Cochrane Database Syst Rev 2004;CD000567.

80 Perel P, Roberts I. Colloids versus crys-talloids for fluid resuscitation in criticallyill patients. Cochrane Database Syst Rev2007; CD000567.

81 Siddik SM, Aouad MT, Kai GE, Sfeir MM,Baraka AS. Hydroxyethylstarch 10% issuperior to Ringer’s solution for preload-ing before spinal anesthesia for Cesareansection. Can J Anaesth 2000; 47: 616–21.

82 Venn R, Steele A, Richardson P, et al.Randomized controlled trial to investi-gate influence of the fluid challenge onduration of hospital stay and periopera-tive morbidity in patients with hip frac-tures. Br J Anaesth 2002; 88: 65–71.

83 McKendry M, McGloin H, Saberi D, et al.Randomised controlled trial assessing theimpact of a nurse delivered, flow moni-tored protocol for optimisation of circu-latory status after cardiac surgery. BMJ2004; 329: 258.



metabolic effects of plasma expanders

Documents