During the last two decades, the use of red cell concentrates (RCCs) for the replacement of surgical blood loss has practically eliminated whole blood transfusions. This change in transfusion practice has also led to a reevaluation of the recommended substitution schemes created during the era when whole blood was still used extensively [1-4]. Until recently, only a few studies have concentrated on the change of hemostatic factors when packed red blood cells are used to replace major surgical blood loss [5,6]. The purpose of this study was to investigate the change of the hemostatic factors when RCCs were used with colloids to replace major surgical blood loss. The aim was also to define the threshold blood loss for the use of fresh frozen plasma (FFP) and platelet concentrates (PCs).
The study was approved by our institutional ethics committee. Informed consent was obtained from 60 ASA physical status I or II patients scheduled for major urologic or abdominal surgery anticipated to cause blood loss exceeding 20% of the calculated blood volume (CBV). If the blood loss was less than 20% of CBV at the end of the recovery room phase, the patient was disqualified. Patients undergoing liver resection, or with abnormal hemostatic status, or who were presumed to have platelet dysfunction due to a drug effect, were also excluded.
All patients had general anesthesia. Before induction, a radial artery was cannulated to allow invasive blood pressure measurement and to secure easy access to blood samples. Anesthesia was induced with thiopental and fentanyl intravenously and maintained with enflurane (vaporizer set to 0.3%-1.0%) in nitrous oxide/oxygen (70%/30%) supplemented with fentanyl. Vecuronium was used for muscle relaxation guided by train-of-four responses. Standard perioperative monitoring including arterial and central venous pressures was used.
The blood volume of each patient was calculated according to a formula based on height, weight, and sex . If the initial hematocrit was 40% or more, the first 20% blood loss was replaced with colloid plasma substitute only, and thereafter with equal volumes of buffy-coat-free RCC and colloid. If the initial hematocrit was less than 40%, the red cell replacement combined with colloid was begun after only a 10% blood loss. The hematocrit of the buffy-coat-free RCC was 60%, containing approximately 100 mL of salineadenine-glucose-mannitol medium and approximately 20 mL of residual plasma.
The colloid plasma substitutes used in the study were 5% albumin, 3% dextran 70, and 4% and 6% hydroxyethyl starch (HES 120/0.7) solutions. If the maximum dose of the synthetic preparation was reached, 4% albumin solution was given thereafter. There were 15 patients in each colloid group, as a study investigating the clinical effect of hypooncotic and conventional plasma substitutes was performed simultaneously. The perioperative volume effects of the studied colloids proved to be quite similar.
Blood loss during surgery was determined by the weight change of the surgical sponges and by the volume change in the suction reservoirs. The replacements were given without delay to avoid major fluctuation of the blood volume. The weight change of the infusion units was measured repeatedly in order to have the replacement volume of plasma substitute and RCC match the blood loss as closely as possible.
Postoperatively, in the recovery room, the accumulation of blood in the drains was measured and replaced. At this stage, equal volumes of colloid and RCC were also used to increase blood volume without verified losses, if indicated by hemodynamic variables. For ethical reasons, once started, red cell units were infused to completion.
When the blood loss exceeded the CBV, or if any sign of abnormal bleeding was observed by the surgical team, 2-3 U of FFP were given. Abnormal hemostasis was defined as persistent oozing of wound edges or surgical field unaffected by conventional procedures. Platelet substitution was given if the platelet count was less than 50 times 103/mm3, and if hemostasis was still considered inadequate after FFP substitution.
For coagulation factor assay, blood was collected into 0.129 M trisodium citrate (one part anticoagulant, nine parts blood). For the measurement of platelet and hemoglobin concentrations, blood was drawn into a vacuum tube anticoagulated with EDTA (Venoject Registered Trademark; Terumo, Leuven, Belgium). Blood samples for platelet count, fibrinogen concentration, and prothrombin and coagulation factors V and VII activities were drawn before induction, prior to transfusion of FFP, and at the end of the recovery room period. If the patient received FFP, the corresponding blood loss was registered and the patient was subsequently excluded from further analyses pertaining to coagulation factors. The same exclusion criteria were applied in the case of platelet replacement.
After the blood samples for the coagulation factors were drawn, they were immediately submerged in an ice bath, centrifuged within 30-60 min, and deep frozen to -70 degrees C for later analysis at the Finnish Red Cross Blood Transfusion Service. The fibrinogen concentration was determined with the ACL 300R coagulometer (Instrumentation Laboratory, Milan, Italy). The method is based on the measurement of light scattered from a reaction mixture during the prothrombin test. The activity of prothrombin and factors V and VII was determined by a one-stage test based on prothrombin time.
Analysis of variance was used to detect differences in the hemostatic factors between the four colloid groups in the course of the study. As there were no significant differences, the data of the four groups were pooled. The relative values of the hemostatic variables were calculated and a logarithmic transformation was applied. Regression equations were calculated for each hemostatic factor and for the proportional blood loss. Analysis of variance was used to test the significance of the regressions.
A method described by Zar  was applied to calculate 95% confidence limits (confidence bands) for all points on the regression lines. For a given blood loss, the 95% confidence limits of the corresponding concentration or activity were calculated. The calculations were repeated at an interval of 10%. The resulting points formed the limits for the confidence bands. The critical level of the hemostatic factors and the corresponding blood loss with 95% confidence intervals (95% Cl) was determined by a method of inverse prediction .
The changes of platelet and fibrinogen concentrations in the course of blood loss are shown in Figure 1. The decline of fibrinogen closely resembles the mathematical washout model. The simple linear regression calculated for the logarithms of the relative values of platelet and fibrinogen is shown in Figure 2. The coefficient of determination (r2) for the regression of fibrinogen was 0.90 and 0.60 for the regression of platelets. The respective coefficients for prothrombin, factor V, and factor VII were 0.80, 0.63, and 0.82, respectively. All the regressions were highly significant. The regression coefficient (slope) was -0.30 for platelets, -0.39 for fibrinogen, -0.35 for prothrombin, -0.26 for factor V, and -0.29 for factor VII.
Deficiency of fibrinogen appeared first when the replacement was performed according to the study protocol. The critical fibrinogen concentration of 1.0 g/L was reached when the blood loss was 1.42 (95% Cl 1.17-1.69) of the CBV. Blood loss in excess of two CBVs caused the deficiency of prothrombin, factor V, platelets, and factor VII, in this order Table 1.
Ten patients were treated with FFP. Blood loss was 118% +/- 17% of the CBV at the time of FFP supplementation. The platelet count was 118 +/- 51 times 10 (3/mm)3 and the fibrinogen concentration was 1.2 +/- 0.4 g/L prior to FFP infusion. The activities of prothrombin and factors V and VII were 34% +/- 5%, 43% +/- 18%, and 35% +/- 10%, respectively.
Only one patient received eight units of PCs to treat persistent postoperative bleeding from abdominal drains, although no other sign suggesting bleeding diathesis was observed. Prior to platelet infusion, the measured blood loss was 183% of the CBV and the platelet count was 45 times 103/mm3, and increased to 179 times 103/mm3 after the transfusion. Subsequently, this patient returned to surgery and a surgical hemorrhage was verified and treated without complications. Two additional patients were also returned to surgery, one for a postoperative surgical bleeding and the other for an acute compartment syndrome of the calves. The hemostatic data of these three patients were included in the statistical analysis until violations of the study protocol occurred.
The rapid decay of fibrinogen concentration was the most significant result of this study. Approximately 90% of the variation could be explained by blood loss, as proved by the coefficient of determination. According to the literature, the minimal concentration of fibrinogen for adequate hemostasis is between 0.5 and 1.0 g/L [5,6,9]. However, it is impossible to determine the definite critical level of fibrinogen because the clinical circumstances vary, and because coinciding deficits of other hemostatic factors may also modify the results. In this study the critical level was prudently set at 1.0 g/L, since the mean initial fibrinogen concentration was 3.7 g/L, a value found at the upper end of the normal range of 2.0-4.0 g/L. More than one third of the patients were operated on for ulcerative colitis, an inflammatory bowel disease capable of inducing the production of acute phase proteins. This could explain the higher initial fibrinogen concentration found in these patients, which encouraged the use of 1.0 g/L rather than 0.5 g/L as the critical concentration in the calculations based on relative values. According to the inverse prediction, the critical concentration of 1.0 g/L was reached when blood loss exceeded 1.42 (95% Cl 1.17-1.69) of CBV Table 1, Figure 2.
A blood loss matching twice the CBV would lead close to the critical level of prothrombin, and a further loss of one third of the CBV would sufficiently compromise factor VII activity Table 1. Even though more than 80% of the decay of prothrombin and factor VII was related to blood loss, the decline was less steep than the decline of fibrinogen. This can be explained by the different roles of these components in the coagulation process. Prothrombin and factor VII are proenzymes rapidly converted to their active forms as part of the amplification system. Fibrinogen is the raw material for fibrin strands, the end-product of the coagulation cascade, ultimately forming the clot that mechanically impedes blood loss. The consumption of these hemostatic factors may be different, depending on the nature and the extent of the surgical insult causing blood loss.
The labile factor V and the platelet count were the most unpredictable hemostatic factors in this study. The great variance found in factor V activities probably was caused by the different capacity of individual endogenous sources. In some investigations this variation is further increased by the varying plasma content of transfused blood products . Although the RCC used in this study was virtually free of plasma, more than two blood volumes had to be lost before the hemostatic activity of factor V was compromised.
A decision to set the critical concentration of platelets at 50 times 103/mm3 was based on recent literature [2,3,5,6,10,11]. It has also been recommended that the platelet concentration of surgical patients should be more than 100 times 103/mm3[4,12]. Figure 1 shows that several of the patients had platelet counts less than 100 times 103/mm3 during the study. However, only one of the patients received PCs, which proved to be ineffective, since the cause of the bleeding was surgical. It is, therefore, believed that the critical level of platelets is closer to 50 times 103/mm3 than to 100 times 103/mm3, when elective, noncardiac surgeries are considered. The loss of platelets in this study was in accordance with the results of clinical investigations where packed red blood cells were used as the sole or principle source of red cells [5,6]. The level of 50 times 103/mm3 was reached late in the course of blood loss, but to a great extent the individual ability to recruit platelets was unpredictable.
The replacement therapy in this study was strictly based on measured losses in order to make it resemble the washout model as closely as possible, and to allow a better estimation of the dilutional effect. A special effort was made to measure and replace losses without delay to minimize major fluctuation of the blood volume, which was considered the major source of error when results were processed and interpreted according to the assumptions of the model. The behavior of the fibrinogen concentration proved that the model could be tested in clinical circumstances with acceptable accuracy Figure 1. It would be reasonable to expect that the model used in this prospective study showed the relationship between individual hemostatic factors and blood loss more precisely than models based on transfused blood units.
The different composition and combination of blood components used in the clinical studies concerning hemostatic changes produce varying results [9,13-15]. The patients in this study received buffy-coat-free RCC preserved in 100 mL of saline-adenine-glucosemannitol medium and containing approximately 20 mL of residual plasma per unit . Consequently, the exogenous source of hemostatic factors was minimal. Most clinical investigations of this kind are also performed on trauma or emergency patients, and many of these studies are based on retrospective data. For these reasons, our results can be compared with only a few clinical studies.
The prospective study by Murray et al.  also showed the decrease in fibrinogen when RCC and lactated Ringer's solution were used to replace blood loss caused by major elective surgeries. Four of 12 of their patients developed a bleeding diathesis and two of them responded to platelet transfusion. The two other patients required, in addition to several units of PC, 2-6 U of FFP to reach a satisfactory hemostasis. Although the hemostasis was still judged to be inadequate and the platelet counts were more than 50 times 103/mm3, the highest fibrinogen concentrations were 0.42 and 0.60 g/L. An increase to 0.88 and to 1.15 g/L, respectively, stopped the bleeding.
Murray et al.  recently published another study investigating dilutional coagulopathy encountered when surgical blood loss was replaced with RCC and lactated Ringer's solution. Of the 32 patients studied, 17 developed clinical coagulopathy when blood loss was 1.14 +/- 0.28 of the CBV, and 14 of these patients were treated successfully with FFP. The activated partial thromboplastin time of the coagulopathy group was significantly prolonged when compared to the group with normal hemostasis. The fibrinogen concentration was 1.28 +/- 0.47 g/L and the platelet count was 178 +/- 62 times 103/mm3 at the time of coagulopathy. These figures regarding fibrinogen and platelets should be compatible with normal hemostasis, as they were in the previous study of Murray et al.  for the eight patients with comparable blood loss and with a mean fibrinogen concentration of 1.43 g/L and a platelet count of 170 times 103/mm3. Which coagulation factor deficit needed correction in this particular patient population of posterior spinal stabilization compared to the patients of the previous study who had general surgery? The explanation to these conflicting results is speculative, although the special patient population and subjective assessment of hemostasis may have had an impact on the results .
Ten patients in our present study received FFP during the perioperative period. Two patients had postoperative surgical bleeding and were returned to surgery. The other patients were given FFP to treat abnormal hemostasis. FFP was transfused at the same blood loss level as Murray et al.  did in their recent study. Also the concentrations of platelets and fibrinogen were quite similar. However, the decision-making process triggering FFP transfusion may have been biased in our present study. The opinion of the surgeon was decisive when the nature of bleeding was determined. The briskly bleeding veins of the pelvis were involved in more than two thirds of the operations, and in many cases it was difficult to verify the actual nature of bleeding. It is therefore believed that some patients were treated with FFP while the true signs of pathologic bleeding were still missing.
Platelet concentrates have been advocated widely as the first choice to treat coagulopathy encountered in massive transfusion [1-5,9-12]. Evidently, this practice is appropriate providing the replacement components contain sufficient amounts of plasma to maintain the coagulation factors above adequate hemostatic levels. The PCs also contain a significant amount of plasma, which distorts the evaluation of the hemostatic effect in some cases. By transfusing enough platelets, both plasma- and platelet-derived coagulopathies are treated. The results of this study indicate the priority of FFP supplementation when plasmapoor RCC and colloid plasma substitutes are used to replace major blood loss caused by elective surgery. Leslie and Toy  came to the same conclusion in their retrospective study of patients resuscitated mainly with RCC. They suggested that coagulation factor replacement is needed in patients who receive 12 U or more of RCC and platelet replacement becomes necessary after 20 U.
We conclude that, in elective surgeries, the deficiency of fibrinogen develops earlier than any other hemostatic derangement when major blood loss is replaced with plasma-poor RCC and colloid plasma substitutes. We believe that this finding necessitates the use of FFP as the first choice to treat coagulopathy manifesting under these clinical circumstances, provided that the platelet count is more than the acceptable level and that the function of platelets is normal.
We are grateful to Professor Eero Ikkala for sound advice in the final preparation of this manuscript. We also thank Eija Kalso, MD, and Lola Lucke, RN, for their assistance in correcting the manuscript.
1. NIH Consensus Conference. Fresh frozen plasma. Indications and risks. JAMA 1985;253:551-3.
2. NIH Consensus Conference. Platelet transfusion therapy. JAMA 1987;257:1777-80.
3. Gravlee PG. Optimal use of blood components. Int Anesthesiol Clin 1990;28:216-22.
4. Crosby ET. Perioperative haemotherapy. I. Indications for blood component therapy. Can J Anaesth 1992;39:695-707.
5. Murray DJ, Olson J, Strauss R, Tinker JH. Coagulation changes during packed red cell replacement of major blood loss. Anesthesiology 1988;69:839-45.
6. Leslie SD, Toy PTCY. Laboratory hemostatic abnormalities in massively transfused patients given red blood cells and crystalloid. Am J Clin Pathol 1991;96:770-3.
7. Nadler SB, Hidalgo JU, Bloch T. Prediction of blood volume in normal human adults. Surgery 1962;51:224-32.
8. Zar JH. Biostatistical analysis. 2nd ed. Englewood Cliffs, NJ: Prentice-Hall, 1984.
9. Counts RB, Haisch C, Simon TL, et al. Hemostasis in massively transfused trauma patients. Ann Surg 1979;190:91-9.
10. Phillips TF, Soulier G, Wilson RF. Outcome of massive transfusion exceeding two blood volumes in trauma and emergency surgery. J Trauma 1987;27:903-10.
11. Fakhry SM, Sheldon GF. Massive transfusion in the surgical patient. In: Jefferies LC, Brecher ME, eds. Massive transfusion. Bethesda, MD: American Association of Blood Banks, 1994:17-42.
12. Miller RD. Transfusion therapy. In: Miller RD. Anesthesia. New York: Churchill Livingstone, 1990:1467-99.
13. Ciavarella D, Reed RL, Counts RB, et al. Clotting factors levels and the risk of diffuse microvascular bleeding in the massively transfused patient. Br J Haematol 1987;67:365-8.
14. Miller RD, Robbins TO, Tong MJ, Barton SL. Coagulation defects associated with massive blood transfusions. Ann Surg 1971;174:794-801.
15. Simmons RL, Collins JA, Heisterkamp CA, et al. Coagulation disorders in combat casualties. Ann Surg 1969;169:455-82.
16. Hogman CF, Akerblom O, Hedlund K, et al. Red blood cell suspensions in SAGM medium. Vox Sang 1983;45:217-23.
17. Murray DJ, Pernell BJ, Weinstein SL, Olson JD. Packed red blood cells in acute blood loss: dilutional coagulopathy as a cause of surgical bleeding. Anesth Analg 1995;80:336-42.
© 1995 International Anesthesia Research Society
18. Miller RD. Coagulation and packed red blood cell transfusions. Anesth Analg 1995;80:215-6.