"ALBUMIN INDUCED HYPERCOAGULABILITY DOES NOT REDUCE BLOOD LOSS"
Scandinavian Journal of Surgery 94: 227–232, 2005 ALBUMIN INDUCED HYPERCOAGULABILITY DOES NOT REDUCE BLOOD LOSS IN PATIENTS UNDERGOING TOTAL HIP ARTHROPLASTY T. T. Niemi1, M. Silvanto2, P. H. Rosenberg1 1 Department of Anaesthesiology and Intensive Care Medicine, Helsinki University Central Hospital, Helsinki, Finland 2 Department of Anaesthesia, Orton Orthopaedic Hospital, Invalid Foundation, Helsinki, Finland ABSTRACT Background and Aims: Albumin may enhance and hydroxyethyl starch (HES) may im- pair haemostasis. While the effects are also dependent on haemodilution we minimized it by early structured transfusion therapy, and compared albumin and HES regarding blood loss and coagulation parameters in hip arthroplasty patients. Material and Methods: 101 patients undergoing primary hip arthroplasty received in random order 4 % albumin (n = 48) or HES (average Mw 120 kDa/molar substitution ra- tio 0.7, n = 53). The administration of colloid, red blood cell (RBC), fresh frozen plasma and platetet concentrates begun after a 6–8 %, 12–16 %, 60 % and 100 % blood loss of the patient’s calculated blood volume respectively. Explanatory risk factors for blood loss were modelled by regression analysis. Results and Conclusions: Administration of albumin or HES 1200 ml (500–2000 and 500–1800) [median (range) respectively] did not affect blood loss. The vWF antigen was higher in the albumin group (p = 0.04) postoperatively. Haematocrit value, platelet count, bleeding time, prothrombin time value, activated thromboplastin time, FV activity and fibrinogen concentration were comparable between the groups. Long operation time was associated with great blood loss (p < 0.001). In hip arthroplasty patients with near nor- mal levels of haematocrit albumin enhanced coagulation without altering blood loss. Key words: Albumin; HES; coagulation; blood loss INTRODUCTION these are mainly the results of haemodilution per se or of an effect by the type of the infused solution used The infusion of a crystalloid solution (1) or albumin for the compensation of the surgical blood loss, e.g., (2, 3) may enhance coagulation while hydroxyethyl the characteristics of the HES molecule (4), remains starch (HES) solutions impair haemostasis both by open. dilution and a deterioration of coagulation factor In our previous in vitro studies dilution of human function in plasma of surgical patients (4). Whether blood by 20 % with 4 % albumin clearly increased and 6 % low-molecular weight HES decreased the coagulation efficacy (5). To what extent these find- Correspondence: Tomi Niemi, M.D. ings can be translated into clinical practice, e.g., Department of Anaesthesiology and Intensive whether perioperative haemorrhage and coagulation Care Medicine are influenced by the solutions to a clinically signifi- Helsinki University Central Hospital cant extent, was investigated in patients undergoing PO Box 340, FIN - 00029 HUS, Finland hip arthroplasty. In order to minimize a dilutional Email: email@example.com influence on coagulation function we applied an ac- 228 T. T. Niemi, M. Silvanto, P. H. Rosenberg tive transfusion strategy including early transfusion Blood Transfusion Service, Helsinki, Finland) or 6 % low compartment therapy followed by administration of molecular weight HES (HES group n = 53) (average Mw 120 whole blood. kDa/molar substitution ratio 0.7, Plasmafusin®, Medipolar, Oulu, Finland) for intravascular fluid replacement. The schematic presentation of transfusion management is shown in Table 1. The blood volume was estimated as MATERIAL AND METHODS 72 ml/kg for female and 77 ml/kg for male patients. Intraoperative blood loss was measured from the con- After approval of the local Ethics committee and written tent of suction bottles and the increase in weight of surgi- informed consent a total of 101 patients undergoing uni- cal swabs. In the recovery room and during the three post- lateral primary noncemented total hip replacement (THR) operative days blood loss estimation was based on the vo- were enrolled in the study. Patients having haemoglobin lume in the vacuum drainage bags. In addition, blood loss concentration below 120 g/l, coagulation disorder or tak- was evaluated as a percent value of the patient’s calculat- ing warfarin were excluded. Acetylsalicylic acid and dipy- ed blood volume. The units of transfused red blood cell ridamol were not allowed within 7 days before surgery, concentrate (RBC, SAGM, Finnish Red Cross Blood Trans- and nonsteroidal anti-inflammatory analgesic drugs fusion Service, Helsinki, Finland) and whole blood (WB, (NSAIDs) within 2 days preoperatively. Low molecular CPD whole blood, Finnish Red Cross Blood Transfusion weight heparin, enoxaparin 40 mg once a day, beginning Service, Helsinki, Finland) and their weight were also re- 11–16 h before anaesthesia, was given for thromboproph- gistered. ylaxis. COLLOIDS ANAESTHESIA, SURGERY AND POSTOPERATIVE ANALGESIA The colloid administration was begun after a 6–8 % blood loss of the patient’s calculated blood volume. Colloid (al- The patients were premedicated with oral diazepam 10– bumin or HES) was given in a volume of 250 ml with one 20 mg 45–60 min before anesthesia. An epidural catheter RBC unit after every 450-ml blood loss. was placed at the L1–L4 level. Thereafter spinal anaesthe- sia was induced at the L2–4 level with plain bupivacaine 5 mg/ml 15–20 mg (Marcain spinal, Astra, Södertälje, Swe- RED BLOOD CELL CONCENTRATES AND WHOLE BLOOD den). During surgery additional doses of bupivacaine 5 mg/ml were given epidurally as needed. The upper body RBC concentrates were given after 12–16 % blood loss. One of the patients was covered by warm blankets to prevent RBC unit was given for a blood loss of 450 ml. After trans- heat loss. fusion of 4–6 RBC units (according to the weight 4 RBC The great majority of the operations (n = 86) were done units for 45–60 kg, 5 RBC units for 61–75 kg, 6 RBC units by senior orthopaedic surgeons with all the patients in the for 76–90 kg) the blood loss was replaced by giving one lateral decubitus position. A noncemented surgical tech- WB unit for a blood loss of 450 ml. nique was used. The patients were given diclofenac 25 mg or 50 mg oral- RINGER’S AND 5 % GLUCOSE SOLUTION ly three times a day beginning on the first postoperative day. Additional postoperative pain management included After every 450-ml blood loss 250 ml of Ringer’s acetated intramuscular oxycodone and epidural analgesia with solution (Ringersteril®, Medipolar, Oulu, Finland) was bupivacaine 5 mg/l, as needed. given with one RBC unit. Additionally, the patients were given 1000 ml of Ringer’s acetated solution and 2000 ml of TRANSFUSION AND FLUID THERAPY DURING SURGERY balanced 5 % glucose solution during the day of surgery. AND IN THE RECOVERY ROOM The patients weighing over 60 kg were given additional 250–500 ml of Ringer’s solution for every 10 kg increase of The patients were randomized to receive either 4 % albu- weight. If there were signs of hypovolemia, 500–1000 ml min (ALB group, n = 48) (Albumin 4 %®, Finnish Red Cross of Ringer’s solution was given. TABLE 1 Schematic presentation of transfusion management. The arrows show the approximate start point of the administration of colloids (albumin or HES), red blood cell concentrates (RBC), whole blood (WB), fresh frozen plasma (FFP) and platelet concentrates (PC) according to the estimated blood loss of the patient’s calculated blood volume. Ringer’s solution and colloids were given 250 ml with each RBC unit. In addition the patients were given 1000 ml of Ringer’s solution and 2000 ml of 5 % glucose solution during the day of surgery. 6–8 % 12–16 % 12–16 % 60 % blood loss 100 % blood loss blood loss blood loss blood loss Colloid →→→→ →→→→ →→→→→→→→ →→→→→→→→ →→→→→→→→ RBC one unit for a blood loss of 450 ml ➡➡➡➡ WB one unit for a blood loss of 450 ml ➔➔➔➔➔➔➔ ➔➔➔➔➔➔➔ ➔➔➔➔➔➔➔ FFP ➡➡➡➡➡➡➡➡ ➡➡➡➡➡➡➡➡ PC ➜➜➜➜➜➜➜ Albumin enhances coagulation 229 FRESH FROZEN PLASMA AND PLATELETS terassay precision was 6.1 %. A frozen plasma pool of more than 50 normal donors was used as a standard in FV and The indication for fresh frozen plasma (FFP) transfusion vWF antigen assays. APTT was measured using standard was a blood loss over 60 %. If the platelet count was be- techniques and PT was measured by the P+P principle us- low the level of 100 × 109/l or the blood loss was estimat- ing commercial reagents. Anti-FXa-activity was deter- ed to be over 100 % of the calculated blood volume, eight mined by a commercial method (Kabi Coatest, Heparin). units of platelets (Platelet concetrates, Finnish Red Cross Bleeding time was measured on the volar surface of the Blood Transfusion Service, Helsinki, Finland) average vol- upper part ot the forearm with a transverse incision using ume 85 ml/unit) were given. Transfusions and fluid ther- the Simplate II R device (BioMeriéux, North Carolina, apy on the 1st–3rd postoperative day were given using the USA). same principles. STATISTICS HAEMATOLOGICAL AND COAGULATION STUDIES The patient number needeed was estimated to show a 25 % Haemoglobin concentration (Hb, g/l), haematocrit value difference in intraoperative blood loss. The number of (Hct, %) and platelet count (PC, × 109/l) in whole blood coagulation measurements was based on the data of vWF were determined in all the study patients preoperatively, antigen concentrations (30 % difference) shown in the pre- in the recovery room, on the first, second and third post- vious study (6). With a power of 80 % (α-error 5 %) the operative day with a Coulter Counter T880 blood cell coun- sample size was 40 for blood loss and 16 for coagulation ter (Coulter Electronics Inc., Hialeah, Florida, USA). measurements in both groups. The results were analyzed Coagulation determinations were done in random order by the two-way analysis of variance (ANOVA) for repeat- in 19 albumin treated and 21 HES treated patients. The de- ed measurements. The Tukey multiple comparison test was terminations included fibrinogen concentration (g/l), co- used a posteriori to identify the differences between the agulation factor V activity (FV, IU/ml), plasma von Wille- groups. Patient characteristics, blood loss and transfusion brand factor antigen (vWF, IU/ml), activated partial data were compared using the student t-test, the Mann- thromboplastin time (APTT, s), prothrombin time value Whitney Rank Sum test, the χ2 -analysis or the Fisher Exact (PT, %) and IVY bleeding time (BT, s). The venous blood test as appropriate. The values are presented as means (SD) samples were taken the day before surgery (fibrinogen, FV, or medians (range). The possible variables affecting blood vWF, APTT, PT, BT), before anaesthesia (fibrinogen, FV, loss was modelled with multiple linear regression. The α vWF), postoperatively in the recovery room (fibrinogen, level was 0.05 for all statistical tests. FV, vWF, APTT, PT, BT) and on the first postoperative day (fibrinogen, FV, APTT, PT, BT). In addition, enoxaparin concentrations were determined as anti-FXa-activity/ml before anaesthesia. RESULTS For the coagulation measurements, 9 volumes of blood was mixed with 1 volume of 0.129-M trisodium citrate. Af- The patient groups had comparable demographic ter centrifugation the plasma was aliquoted, snap-frozen data (Table 2). The duration of surgery was longer and stored at –70 ˚C until assayed. Levels of fibrinogen in the ALB group (p = 0.004). The frequency of the were measured using with ACL method (IL TestTM PT- preoperative use of NSAIDs did not differ between Fibrinogen, Instrumentation Laboratory, Italy). A lot of IL the groups (p = 0.065). Test Calibration plasma was used as a standard. The inte- rassay precision was 1.9 %. FV activity was measured us- ing the one-stage method with rabbit brain thromboplas- BLOOD LOSS AND TRANSFUSIONS tin (Thromboplastin-IS, Baxter-dade, USA) with ACL co- agulometer. The interassay precision was 3.5 %. Plasma The perioperative blood loss, the amount of trans- vWF antigen was assayed with an ELISA method. The in- fused RBC, WB and intravascular fluids were com- TABLE 2 Patient characteristics. Values are mean (SD) unless otherwise indicated. ALB n = 48 HES n = 53 Age, years 53 (10) 56 (8) Sex F/M (number) 19/29 19/34 Weight, kg 73 (13) 75 (12) Height, cm 167 (10) 167 (9) ASA physical status [(median (range)] I (I–II) I (I–III) Calculated blood volume, ml 5404 (1022) 5561 (999) Preoperative use of NSAID# (number) 22 35 Arterial hypertension (number) 6 13 Ischaemic heart disease (number) 1 2 Diabetes (number) 3 1 Intraoperative bupivacaine dose, mg [median(range)] 20 (16–83) 20 (17–145) Duration of surgery min [median (range)] 119 (60–225)* 104 (52–281)* * P = 0.004 between the groups (Mann-Whitney Rank Sum test). # stopped two days before surgery, at the latest. ASA American Society of Anesthesiologists physical status NSAID Nonsteroidal anti-inflammatory drug 230 T. T. Niemi, M. Silvanto, P. H. Rosenberg TABLE 3 Blood loss. The values presented are milliliters as medians (range). There are no significant differences between the groups. Group Intraoperative blood Blood loss in the Additional blood loss Additional blood loss loss, ml recovery room, ml until first postoperative until first postoperative morning, ml evening, ml ALB n = 48 1400 (400–3500) 1000 (200–2600) 300 (0–1100) 100 (0–570) HES n = 53 1200 (400–8300) 1150 (300–3400) 300 (0–1000) 150 (0–600) TABLE 4 The amounts of transfused red blood cell concentrate (RBC), whole blood (WB) and colloid in the albumin group (ALB) and hydroxyethyl starch group (HES). The values presented are medians (range). There are no significant differences between the groups. Group During surgery In the recovery room RBC, g WB, g Colloid, ml RBC, g WB, g Colloid, ml ALB n = 48 1158 (0–1935) 0 (0–1800) 725 (400–1600) 400 (0–1905) 595 (0–4095) 475 (0–1100) HES n = 53 1135 (0–2270) 0 (0–4775) 700 (450–1500) 595 (0–1840) 595 (0–4715) 450 (0–1100) Group Intraoperatively and during the first postoperative day RBC, g WB, g Colloid, ml ALB n = 48 1848 (1420–2245) 1148 (0–5385) 1200 (500–2000) HES n = 53 1810 (735–2955) 920 (0–5385) 1200 (500–1800) parable between the groups (Table 3 and 4). The postoperatively in albumin treated patients, indicat- average intraoperative administration of albumin 725 ing an enhancement of coagulation, is in accordance ml (400–1600) [median (range)] or HES 700 ml (450– with our in vitro results (5) and a recent in vivo pre- 1500) did not influence intraoperative blood loss and liminary study (3). Other coagulation parameters the amount of transfused RBC and WB units. Long changed to a similar extent as in the HES group sug- operation time, but not the patient preoperative char- gesting a dilutional effect. acteristics, was associated with high blood loss as The type of replacement fluid used and the degree shown by multiple linear regression analysis (p < of haemodilution seem to determine the efficacy of 0.001). All the patients received RBC transfusion by coagulation during haemodilution (7). Low haema- the end of the recovery room period. tocrit may compromise blood coagulation (8), and al- logeneic blood transfusions have been recommend- LABORATORY FINDINGS ed for raising the haematocrit at least to 30 % just to optimize blood coagulation. Furthermore, during The measured clotting variables decreased to a simi- massive transfusion correction of haematocrit is es- lar extent in both groups as compared with preoper- sential for sustaining adequate hemostasis (9), and ative values (Table 5). The vWF antigen concentra- patients with low haematocrit levels may bleed more tion was higher in the ALB group postoperatively in than patients having haematocrit level above 33 % the recovery room (p = 0.04, Tukey test) (Table 5). after cardiac surgery (10). Before anaesthesia the anti-FXa-activity was 0.2 (0.06) In the current study the median intraoperative U/ml [(mean, (SD)] in both groups (p = 0.88). The blood loss was relatively high, 1400 ml (ALB group) Hb concentration (Fig. 1), Hct value and PC (data not and 1200 ml (HES group), and totally 2860 ml and shown) decreased (p < 0.001) similarily in the study 2950 ml, respectively, until the end of the first post- groups. operative day. There were however no clinical rele- vant abnormalities in the measured coagulation pa- rameters which could aggravate haemorrhage. Usu- DISCUSSION ally in orthopaedic surgery RBC transfusions with high blood loss are most common in patients with This study demonstrates that the administration of reduced preoperative red blood cell reserve, female 4 % albumin and 6 % HES (Mw 120 kDa/ molar sub- gender, advancing age, small body size and frequent stitution ratio 0.70) have comparable effects on blood comorbid diseases (11). However, we could not iden- loss intra- and postoperatively in primary total hip tify those factors as risk for blood loss in the present replacement patients with modest haemodilution. study which is in accordance with a recent study in The finding of a higher vWF antigen concentration patients undergoing total hip arthroplasty (12). Albumin enhances coagulation 231 TABLE 5 Mean (SD) clotting variables. The p-values indicate the significance level as compared to the preoperative value within both groups. The concentra- tion of vWF antigen was higher postoperatively in the ALB group (*P = 0.04 between the groups, Tukey test). Otherwise there are no significant differences between the groups. Variable ALB group P-value HES group P-value Fibrinogen, g/l n = 19 n = 21 Day before surgery 3.6 (1.1) 3.4 (0.6) Before anaesthesia 3.3 (1.1) 0.15 3.2 (0.5) 0.56 Recovery room 2.2 (0.5) < 0.001 2.3 (0.3) < 0.001 First postoperative morning 4.0 (0.7) 0.003 3.9 (0.5) < 0.001 Factor V, IU/ml n = 19 n=2 Day before surgery 114 (14) 1115 (18) Before anaesthesia 102 (16) < 0.001 0107 (18) < 0.001 Recovery room 070 (16) < 0.001 0067 (11) < 0.001 First postoperative morning 083 (15) < 0.001 0078 (11) < 0.001 PT, % n = 48 n = 48 Day before surgery 99 (25) 102 (20) Recovery room 65 (12) < 0.001 060 (11) < 0.001 First postoperative morning 65 (14) < 0.001 059 (10) < 0.001 APTT, s Day before surgery 28 (9) n = 48 260(2) n = 48 Recovery room 27 (3) n = 48 0.41 280(3) n = 53 0.09 First postoperative morning 28 (2) n = 48 1.0 290(5) n = 53 0.004 vWF antigen, IU/ml n = 19 n = 21 Day before surgery 112 (33) 124 (35) Before anaesthesia 107 (30) 0.85 123 (35) 1.0 Recovery room 134 (53)* 0.04 109 (31)* 0.20 Bleeding time, min Day before surgery 6.8 (1.9) n = 48 6.8 (1.9) n = 48 Recovery room 8.4 (2.8) n = 37 < 0.001 8.6 (2.9) n = 41 < 0.001 First postoperative morning 7.4 (2.3) n = 48 0.26 7.9 (2.3) n = 53 0.005 The findings concerning HES in the present study Hb (g/l) are in accordance with a previous study in volunteers 180 where a small dose of a low molecular weight HES (125 kDa) with high substitution ratio (0.7) caused no, or only minimal changes in blood coagulation 160 * * * * (13). However, had a larger dose of HES had been n=48 given, a decrease in coagulation factor levels might 140 n=48 n=48 n=48 n=48 have been seen (14). The small postoperative de- n=48 creases in fibrinogen concentration, FV activity and 120 n=53 PT value as well as the lenghtening of bleeding time n=53 n=53 were similar in both groups and probably the result n=53 of the modest dilution by the intravenous fluids. 100 The transfusion algorithm designed particularly for the present study is certainly too liberal, and ex- 80 pensive and we do not recommend it for routine cli- 0 1 2 3 4 5 nical practice. Guidelines for blood component thera- Sample number py suggest to accept lower haemoglobin concentra- ALB tions, coagulation factor levels and platelet counts HES than traditionally have been considered safe (15). Fig. 1. Haemoglobin concentration (Hb) (mean, SD) in whole Whole blood should be used only under special cir- blood. *P < 0.001 in comparison with preoperative value within cumstances such as in pediatric cardiac or craniofa- both groups (Tukey test). The sample numbers are 1 = preopera- cial surgery and possibly in war situations (16, 17). tively, 2 = in the recovery room, 3 = first postoperative morning, 4 = second postoperative morning, 5 = third postoperative morn- This is also partly supported by our findings in the ing. current study where no clinical significant enhance- ment in coagulation or reduced blood loss could be demonstrated although relatively high red blood cell mass was maintained. The findings in this study do fusion is indicated when patients with normal initial not challenge the current recommendation for FFP level of coagulation factors have been transfused administration either, which suggests that FFP trans- with more than one blood volume (15). 232 T. T. Niemi, M. Silvanto, P. H. Rosenberg In summary, the administration of albumin to pa- 08. Turitto VT, Weiss HJ: Red blood cells: their dual role in throm- tients undergoing surgery offers no advantage in re- bus formation. Science 1980;207 (4430):541–543 09. Hardy JF, De Moerloose P, Samama M; Groupe d’interet en spect to blood loss in comparison with low molecu- Hemostase Perioperatoire: Massive transfusion and coagulop- lar weight HES. The disturbances in coagulation athy: pathophysiology and implications for clinical manage- were minimal due to the early and aggressive re- ment. Can J Anaesth 2004;51:293–310 placement of blood loss by red blood cell concen- 10. Klass O, Mehlhorn U, Zilkens K, Kroner A, Hekmat K, Geissler HJ, de Vivie ER: Impact of hematocrit value after coronary trates, modified whole blood, fresh frozen plasma artery surgery on perioperative myocardial infarction rate. and platelet concentrates. In spite of that, the amount Thorac Cardiovasc Surg 2002;50:259–265 of intra- and postoperative haemorrhage was consid- 11. Khanna MP, Hebert PC, Fergusson DA: Review of the clinical erable which raises the challenge of the need for an practice literature on patient characteristics associated with perioperative allogeneic red blood cell transfusion. Transfus improvement also in intraoperative surgical haemos- Med Rew 2003;17:110–119 tasis. 12. Pola E, Papaleo P, Santoliquido A, Gasparini G, Aulisa L, De Santis E: Clinical factors associated with an increased risk of perioperative blood transfusion in nonanemic patients under- REFERENCES going total hip arthroplasty. J Bone Joint Surg Am 2004;86-A: 57–61 13. Korttila K, Gröhn P, Gordin A, Sundberg S, Salo H, Nissinen 01. Ng K, Lo J: The development of hypercoagulability state, as E, Mattila MA: Effect of hydroxyethyl starch and dextran on measured by thrombelastography, associated with intraoper- plasma volume and blood hemostasis and coagulation. J Clin ative surgical blood loss. Anaesth Intens Care 1996;24:20–25 Pharmacol 1984;24:273–282 02. McCammon AT, Wright JP, Figueroa M, Nielsen VG: Hemodi- 14. Vogt N, Bothner U, Brinkmann A, de Petriconi R, Georgieff lution with albumin, but not Hextend, results in hypercoagu- M: Peri-operative tolerance to large-dose 6 % HES 200/0.5 in lability as assessed by thrombelastography in rabbits: role of major urological procedures compared with 5 % human albu- heparin-dependent serpins and factor VIII complex. Anesth min. Anaesthesia 1999;54:121–127 Analg 2002;95:844–850 15. Stehling LC, Doherty DC, Faust RF, Greenburg AG, Harrison 03. Jones SB, Whitten CW, Despotis GJ, Monk TG: The influence CR, Landers DF: Practice guidelines for blood component ther- of crystalloid and colloid replacement solutions in acute nor- apy. A report by the American Society of Anesthesiologists movolemic hemodilution: a preliminary survey of hemostatic task force on blood component therapy. Anesthesiology 1996; markers. Anesth Analg 2003;96:363–368 84:732–747 04. de Jonge E, Levi M: Effects of different plasma substitutes on 16. Grosso SM, Keenan JO: Whole blood transfusion for exsan- blood coagulation: a comparative review. Crit Care Med 2001; guinating coagulopathy in a U.S. field surgical hospital in 29:1261–1267 postwar Kosovo. J Trauma 2000;49:145–148 05. Niemi TT, Kuitunen AH: Hydroxyethyl starch impairs in vit- 17. Maclennan S, Murphy MF: Survey of the use of whole blood ro coagulation. Acta Anaesthesiol Scand 1998;42:1104–1109 in current blood transfusion practice. Clin Lab Haematol 2001; 06. Kuitunen A, Hynynen M, Salmenperä M, Heinonen J, Vahtera 23:391–396 E, Verkkala K, Myllyla G: Hydroxyethyl starch as a prime for cardiopulmonary bypass: effects of two different solutions on haemostasis. Acta Anaesthesiol Scand 1993;37:652–658 07. Jamnicki M, Kocian R, van der Linden P, Zaugg M, Spahn DR: Acute normovolemic hemodilution: physiology, limitations, Received: October 8, 2004 and clinical use. J Cardiothorac Vasc Anesth 2003;17:747–754 Accepted: March 8, 2005