MORE than 22 million blood components are transfused each year in the United States. [1]Many of these transfusions are administered to surgical and obstetric patients. The transfusion of red blood cells (RBCs), platelets, fresh-frozen plasma (FFP), and cryoprecipitate has the potential of improving clinical outcomes in perioperative and peripartum settings. These benefits include improved tissue oxygenation and decreased bleeding. However, transfusions are not without risks or costs. The transmission of infectious diseases (e.g., hepatitis, human immunodeficiency virus (HIV) infection), hemolytic and nonhemolytic transfusion reactions, immunosuppression, alloimmunization, and other complications are potential sequelae of blood component therapy.
A number of groups have issued clinical practice guidelines for blood component therapy in an effort to improve transfusion practices, minimize the incidence of adverse transfusion reactions, and decrease costs. In the 1980s, the National Institutes of Health convened consensus conferences and published recommendations for RBC transfusion, platelet therapy, and the administration of FFP. [2–4]In 1984, the American College of Obstetricians and Gynecologists (ACOG) issued recommendations on blood component therapy. [5]In 1990, the Transfusion Practices Committee of the American Association of Blood Banks issued guidelines for transfusion of patients undergoing coronary artery bypass surgery. [6]In 1992, the American College of Physicians (ACP) issued recommendations for RBC transfusion. [7]In 1994, the College of American Pathologists (CAP) published a practice parameter for FFP, cryoprecipitate, and platelet transfusion. [8]Guidelines for blood utilization review were published in the same year by the American Association of Blood Banks. [9]Although these documents include sections on the use of blood components in the surgical setting, no group has issued current and comprehensive recommendations on perioperative and peripartum blood component therapy.
In 1994, the American Society of Anesthesiologists convened the Task Force on Blood Component Therapy to develop evidence-based guidelines on the proper indications for perioperative and peripartum administration of RBCs, platelets, FFP, and cryoprecipitate. The task force included nine anesthesiologists (in private and university-based practice); one physician representative from the American College of Surgeons, College of American Pathologists, and ACOG; and a methodologist.
Before the guidelines were developed, the task force reviewed published evidence regarding the clinical effectiveness of perioperative and peripartum blood component therapy. A total of 1,417 articles were retrieved in a computerized and manual literature search conducted in mid-1994. The computerized search sought all English-language literature published in any country on the use of RBCs, platelets, FFP, or cryoprecipitate in the perioperative or peripartum setting. A total of 160 articles were considered relevant. Published evidence was considered relevant if it addressed the perioperative or peripartum use of the above blood components and measured effectiveness in terms of clinical outcomes. The strength of evidence was classified by study design category, using the scale in Table 1. Further details about the literature review methodology are available on request. Input from practicing anesthesiologists was obtained at an open forum held in October 1994. The document was sent to members of the American Society of Anesthesiologists House of Delegates, Board of Directors, and Component Society Presidents for review. The overall guideline development process is reviewed elsewhere. [10,11].
This article summarizes the results of the literature review and the recommendations of the task force. Recommendations apply to typical surgical and obstetric patients. Infants, children, and special clinical settings (e.g., liver transplantation, sickle cell anemia) are beyond the scope of the report. Thus, the task force has not considered the transfusion of neonatal, infant, or pediatric patients or the recommendations on this topic that have been published by other groups. [12,13]The recommended indications are based on scientific evidence and expert opinion regarding the effectiveness of the intervention. Effectiveness was judged by considering the potential clinical benefits, adverse effects, and costs of blood component therapy.
Adverse Effects
Transfusion Reactions
Nonhemolytic transfusion reactions, often manifested in awake patients by fever, chills, or urticaria, are the most common adverse effects of RBC transfusion, but these signs may not be detectable during anesthesia. Nonhemolytic transfusion reactions occur in approximately 1–5% of all transfusions. [4]Hemolytic reactions due to administration of incompatible blood can be life-threatening. The estimated risk of ABO-incompatible transfusion is 1:33,000 RBC transfusions. [14,15]As with nonhemolytic reactions, general anesthesia may mask the symptoms of hemolytic reactions, and many of the signs (hypotension, tachycardia, hemoglobinuria and microvascular bleeding) may be attributed erroneously to other causes. The probability of a fatal hemolytic transfusion reaction is uncertain, with estimates ranging from 1:500,000 to 1:800,000. [15]Between 1976 and 1985, the U.S. Food and Drug Administration was notified of 131 fatal ABO-incompatible transfusions. [16].
Infectious Diseases
The incidence of post-transfusion hepatitis, more than 90% of which is due to hepatitis C virus, has decreased since the introduction of testing for the virus in 1990. [17]The reported incidence of hepatitis C virus seroconversion is 0.03% per unit transfused. [18]However, the actual incidence is believed to be lower because of improved testing introduced in 1992. [19]The risk of transmission of hepatitis B is estimated to be 1:200,000 units. [20].
The risk of exposure to HIV through blood transfusion is uncertain. Although a range of incidence rates has been reported, [21–23]recent estimates suggest that the mean infectious window period (the period between viral infection and its detection by tests for the presence of antibodies) is approximately 22 days [24]and that the current risk of HIV infection in the United States is 1:450,000–1:660,000 per transfused unit of blood. [25]Implementation of donor screening tests for HIV-1 antigen is expected to prevent up to 25% of the window period cases, or five to ten cases per year. [25]Higher rates may occur in areas with increased HIV prevalence. [26].
Perhaps the most common viral agent transmitted by blood transfusion is cytomegalovirus. Most infections are subclinical, although immunocompromised patients may develop severe morbidity. Parasitic and bacterial agents can be transmitted by blood components, but the incidence of clinically significant disease in the United States is low, possibly 1:1,000,000 units of blood. [27]Twenty-six deaths due to bacterial contamination of blood components were reported to the Food and Drug Administration between 1976 and 1985. [16].
Immunosuppression and Blood Transfusion
Some studies suggest that patients with colorectal, breast, prostate, and certain other cancers may experience earlier recurrence and lower survival rates if they receive allogeneic (homologous) blood transfusions in the perioperative period, [28,29]but other data challenge the association. [30]Higher rates of postoperative infections have been reported in patients who received perioperative allogeneic transfusions than in those who were not transfused or who received only autologous blood. [31–34]Other studies, however, have not confirmed this relationship. [35,36].
Costs
Although the costs of blood component therapy are substantial, exact figures are lacking. A study at one medical center estimated that the base cost for providing one unit of allogeneic RBCs was $114 and that the direct and indirect services involved in transfusing the unit increased the cost to $151. [37]A study at 18 hospitals estimated that average hospital costs were $155 per transfused unit of whole blood or RBCs, or $397 per patient for all blood components. [38,39]With an estimated 12 million units transfused each year in the United States, this amounts to an annual cost of at least 2 billion dollars, with some estimating costs as high as 5–7 billion dollars for all transfused components. [40]Such estimates are of limited validity, because they do not include the cost of all blood components, administrative expenses, and indirect transfusion costs, [40]and they do not reflect the economic benefits of transfusion therapy. Nonetheless, the resource implications of blood component therapy must be recognized, and the role of improved transfusion practices in reducing costs must be considered. Almost 25% of the costs of RBC transfusions may be attributable to inappropriate transfusions. [39]These costs can be reduced through the adoption of more appropriate transfusion practices. For example, one large teaching hospital was able to reduce transfusion costs by $1.6 million over 3 yr by adopting new transfusion guidelines. [41].
Red blood cells
Approximately 12,000,000 units of RBCs are transfused each year in the United States. [1]Although most of these transfusions are performed for appropriate reasons, studies have documented substantial rates of unnecessary transfusions. Inappropriate rates of 18–57% have been reported. [42–50]The proper objective of RBC transfusion is the improvement of inadequate oxygen delivery. The scientific argument for perioperative RBC transfusion therefore rests on two principal assumptions:(1) surgical patients experience adverse outcomes as a result of diminished oxygen-carrying capacity, and (2) RBC transfusions, by enhancing oxygen-carrying capacity, can prevent these adverse outcomes. Evidence to support these assumptions is reviewed below.
Adverse Effects of Diminished Oxygen-carrying Capacity
Diminished oxygenation due to inadequate oxygen-carrying capacity can have serious clinical implications, primarily because of ischemic effects on the myocardium and brain. Oxygen delivery (DO2) is defined as the product of cardiac output (Qt) and arterial oxygen content (Ca sub O2) The latter is a function of hemoglobin saturation (SaOsub 2), hemoglobin concentration (Hb), and the amount of oxygen physically dissolved in arterial blood:Equation 1. Although an increase in cardiac output is the primary compensation for reduced oxygen-carrying capacity, changes in the microcirculation can affect oxygen transport at the tissue level. For example, during periods of blood loss, the autonomic nervous system can restrict blood flow and oxygen delivery to skin, muscle, and the abdominal viscera to preserve oxygen delivery to the central nervous system and heart.
The effects of anemia must be separated from those of hypovolemia, although both can interfere with oxygen transport. The clinical manifestations of hypovolemia are well known, and a classification based on blood loss has been established by the American College of Surgeons. [51]A loss of up to 15% of total blood volume (class I hemorrhage) usually has little hemodynamic effect other than vasoconstriction and mild tachycardia. A loss of 15–30% of blood volume (class II hemorrhage) produces tachycardia and decreased pulse pressure; unanesthetized patients may also exhibit anxiety or restlessness. A loss of 30–40%(class III hemorrhage) produces increasing signs of hypovolemia, including marked tachycardia, tachypnea, and systolic hypotension; unanesthetized patients demonstrate altered mental status. Experience has shown that, in young healthy patients, losses of up to 30–40% of blood volume usually can be treated adequately with crystalloid therapy. Loss of more than 40% of total blood volume (class IV hemorrhage) is life-threatening and accompanied by marked tachycardia and hypotension, very narrow pulse pressure, and low urine output; mental status is markedly depressed.
The lower limit of human tolerance to acute normovolemic anemia has not been established. It is believed that oxygen delivery is adequate in most individuals at hemoglobin concentrations as low as 7 g/dL. [4]In healthy, normovolemic individuals, tissue oxygenation is maintained and anemia tolerated at hematocrit values as low as 18–25%. [52,53]The heart does not begin producing lactic acid until a hematocrit of 15–20% is reached. [54,55]Myocardial lactate flux does not appear to be affected at hemoglobin concentrations as low as 6 g/dL. [56]Heart failure usually does not occur until the hematocrit reaches 10%. [57,58].
Chronic anemia is better tolerated than acute anemia. Oxygen delivery is facilitated through increases in 2,3-diphosphoglycerate levels in RBCs. In patients with chronic anemia, cardiac output usually does not change until the hemoglobin concentration falls below 7 g/dL. Significant symptoms are unusual unless the RBC mass is decreased by approximately 50%. [58]Chronic anemia has special implications for pregnant women. Obstetric patients usually tolerate chronic anemia without significant adverse maternal or fetal effects. A review of 17 studies of obstetric patients revealed no effect of hemoglobin concentration on the incidence of stillbirth or intrauterine growth retardation, [59]whereas another found increased complications of pregnancy associated with both low (< 10.4 g/dL) and high (> 13.2 g/dL) hemoglobin concentrations. [60]Studies of acute isovolemic anemia in animals suggest that fetal oxygen extraction is maintained until the maternal hematocrit is reduced by more than 50%. [61]In a study of pregnant sheep with chronic anemia (hematocrit less than 14% for 6 days), decreased oxygen delivery to the placenta did not reduce fetal oxygen consumption. [62].
In acute anemia, reductions in arterial oxygen content usually are well tolerated because of compensatory increases in cardiac output. This compensatory mechanism may be affected by several factors, however, such as left ventricular dysfunction and vasoactive pharmacologic agents (e.g., beta-adrenergic or calcium channel blockade), necessitating a higher hemoglobin concentration for adequate oxygen delivery. Human tolerance of acute anemia is further affected by certain pharmacologic agents, such as anesthetics, hypnotics, and neuromuscular blocking drugs, and by intraoperative conditions (e.g., hypothermia). Anesthetics have important cardiovascular and endocrine actions that influence oxygen transport and consumption and the physiologic response to acute anemia. Most anesthetics cause myocardial depression and decrease arterial blood pressure, cardiac output, stroke volume, peripheral vascular resistance, total-body oxygen consumption, and cerebral and myocardial oxygen demands. [63]The magnitude of these effects varies among anesthetics and as a function of anesthetic depth. In addition, anesthetics differ in their effects on hepatic blood flow, and thus they may differ in how they influence the development of systemic lactic acidosis and base-deficit in patients with anemia or impaired oxygen transport.
The physiologic limit of oxygen transport is not known in either awake humans or those under general anesthesia. Case reports suggest that humans may tolerate lower hemoglobin concentration and oxygen transport during anesthesia than when awake. [64]This may be due to an anesthetic- and neuromuscular blockade-induced reduction of oxygen consumption. However, there are no controlled prospective studies addressing this important issue. The impact of regional anesthesia on oxygen transport is also unclear.
These physiologic principles have been reinforced by clinical studies demonstrating inconsistent associations between anemia and adverse perioperative or peripartum outcomes. Case series reports of Jehovah's Witnesses indicate that some patients tolerate very low hemoglobin concentrations (less than 6–8 g/dL) in the perioperative period without an increase in mortality. [65–67]A review of 16 series published between 1983 and 1990, involving 1,404 operations on Jehovah's Witnesses, found that lack of blood was implicated as the primary cause of death in only 8 (0.6%) patients and as a contributor to death in an additional 12 (0.9%) patients. [68]Another review of 61 reports of 4,722 Jehovah's Witnesses identified 23 deaths due to anemia, all but 2 of which occurred at hemoglobin concentrations less than 5 g/dL. [64]A statistical analysis of one series of Jehovah's Witnesses found that hemoglobin alone was not a statistically significant predictor of outcome unless it was less than 3 g/dL. [69]Both hemoglobin and intraoperative blood loss must be taken into consideration. [65]This body of literature represents self-selected reports, however, in which clinicians are more likely to report survivors than nonsurvivors. The information provided in most reports is insufficient to allow for independent conclusions regarding the degree to which profound anemia contributed to morbidity or mortality.
Decisions regarding perioperative transfusion are often difficult, necessitating clinical judgment. There is little scientific support for relying on a specific hemoglobin or hematocrit value as a "transfusion trigger," such as the outdated "10/30 rule" that transfusion is necessary in patients with a hemoglobin concentration less than 10 g/dL or an hematocrit less than 30%. [70]Estimates of blood volume are also unreliable, because of inaccuracies of intraoperative blood loss measurement, intercompartmental fluid shifts during surgery, and the dilutional effects of crystalloid therapy. Although often useful, intraoperative hemoglobin determinations can be misleading. Alterations of intravascular volume due to the concomitant administration of colloids and crystalloids can produce artificially lowered or elevated hemoglobin concentrations.
Intraoperative estimates of blood volume are indirect, being inferred from pressure measurements obtained at various locations (arterial, central venous, or pulmonary capillary wedge pressures). Whole-body oxygen consumption, oxygen extraction ratio, and oxygen delivery have been used to estimate the need for RBC transfusion. [71–73]These measurements require invasive monitoring (e.g., arterial, pulmonary artery), have not been independently verified, and are global (not organ-specific) measures of oxygen utilization. In the clinical setting, it is not possible to directly measure the adequacy of oxygen transport to specific organs or to regions within these organs.
The perioperative decision-making process regarding transfusion is complicated by knowledge that myocardial ischemia is often silent and most frequent in the postoperative period, when monitoring is less intense. [74,75]A patient's oxygen transport needs can increase at any time during the postoperative period because of pain, fever, shivering, or physical activity.
Aside from the more obvious potential benefits of RBC transfusion in improving oxygen-carrying capacity, other unsubstantiated claims of benefit have been made, including effects on wound-healing. However, in healing tissue, collagen deposition is dependent on oxygen tension and perfusion and not on blood hemoglobin concentration. [76].
Effectiveness of Red blood cell Transfusion
The transfusion of one unit of whole blood or RBCs increases the hematocrit by approximately 3%, or the hemoglobin concentration by 1 g/dL, in a 70-kg non-bleeding adult. [77]Controlled studies have not been performed to determine the hemoglobin concentration at which RBC transfusion improves clinical outcome. Indirect evidence suggests that many RBC transfusions are unnecessary, because mild to moderate blood loss does not appear to be associated with increased perioperative morbidity or mortality [78–84]and reduced transfusions brought about by concerns over transfusion-related infections have not been associated with poorer perioperative outcomes. [85–87]However, most of these studies were uncontrolled and lacked long-term follow-up.
A relationship between perioperative anemia and myocardial ischemia or infarction has been proposed. A controlled observational study of 27 high-risk patients undergoing infrainguinal arterial bypass surgery found that the incidence of postoperative myocardial ischemia and morbid cardiac events was significantly higher among the 14 patients with hematocrits less than 28% than among patients with higher hematocrits. [88]However, the anemic group was significantly older and underwent longer procedures, the results were not adjusted for confounding variables that increase ischemic risk, and the study did not examine the effectiveness of RBC transfusions. A study of 30 postoperative critical care patients with hemoglobin concentrations less than 10 g/dL found that RBC transfusions had little impact on oxygen consumption, [89]a finding consistent with other studies of critical care patients. [90–93].
Strength of Evidence: Category II-2 and II-3 (Controlled and Uncontrolled Observational Studies)
Expert Opinion.
Recommendations of Other Groups. In 1988, the National Institutes of Health Consensus Conference on Perioperative Red Blood Cell Transfusion concluded that evidence did not support the use of a single criterion for transfusion, such as a hemoglobin concentration less than 10 g/dL, nor was there evidence that mild to moderate anemia contributed to perioperative morbidity. [4]In 1992, the ACP recommended distinguishing between stable and unstable vital signs in determining whether to transfuse anesthetized patients. The ACP concluded that patients with stable vital signs and no risk of myocardial or cerebral ischemia do not require RBC transfusion, independent of hemoglobin level, and recommended transfusing patients with unstable vital signs only if risks of myocardial or cerebral ischemia were present. [7]A consensus conference convened by the Royal College of Physicians of Edinburgh concluded that RBC transfusion is indicated only to increase oxygen-carrying capacity, the transfusion decision should be made by a qualified practitioner as part of the overall management of the illness, patients should be offered information about RBC transfusion and available alternatives, and the indication for transfusion should be documented in the medical record. [94].
Expert Opinion of Task Force. The task force believes that any uniform "transfusion trigger" (e.g., an absolute hemoglobin or hematocrit value) provides an inappropriate basis for determining the need for perioperative or peripartum RBC transfusion. Hemoglobin concentration alone is an inadequate measure of oxygen delivery. The ACP recommendation to rely solely on vital signs [7]is inappropriate for anesthetized patients. The decision to transfuse often is affected by the dynamic nature of surgical hemorrhage. [95]Patients with hypovolemia and anemia may be transfused more aggressively when rapid blood loss is anticipated (e.g., aortic unclamping). Changes in vital signs often are masked by anesthetics and other drugs and are frequently a late sign of cardiovascular decompensation. Moreover, silent ischemia of the myocardium, cerebrum, liver, kidney, and other tissues can occur in the presence of stable vital signs. Intraoperative myocardial ischemia, a predictor of cardiac morbidity and mortality, [74]is associated with tachycardia in only 26% of patients and with blood pressure changes in less than 10% of patients. [75].
Factors affecting the surgical patient's response to decreased hemoglobin concentration, and thus the factors that should influence the physician's decision to transfuse, include the patient's cardiopulmonary reserve (determined by the presence or absence of cardiac and/or pulmonary disease and hemodynamic indexes, and affected by drugs and anesthetics), the rate and magnitude of blood loss (actual and anticipated), oxygen consumption (affected by body temperature, drugs/anesthetics, sepsis, muscular activity), and atherosclerotic disease (cerebrovascular, cardiovascular, peripheral, renal).
Recommendations. The task force bases its recommendations on available category II-2 and II-3 evidence and expert opinion. The task force concludes that (1) transfusion is rarely indicated when the hemoglobin concentration is greater than 10 g/dL and is almost always indicated when it is less than 6 g/dL, especially when the anemia is acute;(2) the determination of whether intermediate hemoglobin concentrations (6–10 g/dL) justify or require RBC transfusion should be based on the patient's risk for complications of inadequate oxygenation;(3) the use of a single hemoglobin "trigger" for all patients and other approaches that fail to consider all important physiologic and surgical factors affecting oxygenation are not recommended;(4) when appropriate, preoperative autologous blood donation, intraoperative and postoperative blood recovery, acute normovolemic hemodilution, and measures to decrease blood loss (deliberate hypotension and pharmacologic agents) may be beneficial; and (5) the indications for transfusion of autologous RBCs may be more liberal than for allogeneic RBCs because of the lower (but still significant) risks associated with the former.
Platelets
More than 7,000,000 units of platelets are transfused each year in the United States. [1]Platelets are used in the perioperative and peripartum setting when a quantitative or qualitative platelet defect is the suspected cause of bleeding. The scientific rationale rests on two principal arguments:(1) surgical patients experience adverse outcomes as a result of thrombocytopenia and/or platelet dysfunction, and (2) platelet transfusion can correct platelet defects and thereby reduce, minimize, or prevent bleeding. Evidence to support these arguments follows.
Adverse Effects of Thrombocytopenia and Platelet Dysfunction
Patients with thrombocytopenia or platelet dysfunction may experience morbidity and mortality from severe surgical hemorrhage. The platelet count at which surgical and obstetric patients are likely to experience increased bleeding is unknown. In nonsurgical patients, spontaneous bleeding is uncommon with platelet counts greater than 20 x 109/l, [96]and some studies suggest low complication rates in surgical patients with thrombocytopenia. [97]Performance of paracentesis and thoracentesis was not associated with increased bleeding in patients with platelet counts of 50–99 x 109/l. [98].
Coexisting clinical conditions influence the value of platelet counts in predicting the occurrence of bleeding in surgical and obstetric patients, but the probability of clinically significant thrombocytopenia increases in proportion to the number of units of blood transfused. [99–102]In a study of 39 massively transfused patients, platelet counts less than 50 x 109/l were found in 75% of patients who received 20 or more units of RBCs and in no patients who received less than 20 units. [103]Consumption of platelets, as well as simple dilution, can lead to microvascular bleeding (e.g., diffuse bleeding from wound edges, mucous membranes, insertion sites of vascular cannulae). [102].
Gestational thrombocytopenia is usually a benign phenomenon during pregnancy. [104]There is little evidence of bleeding complications from anesthetic procedures (e.g., epidural placement) performed in pregnant women with thrombocytopenia [105]Because hemostasis after placental separation is largely mechanical, thrombocytopenia has virtually no effect on the incidence of postpartum uterine bleeding. [106]Mild thrombocytopenia is detected in approximately 15% of women with preeclampsia. [107]With the HELLP (hemolysis, elevated liver enzymes, and low platelet count) syndrome associated with preeclampsia, the thrombocytopenia is more severe, but spontaneous resolution usually occurs by the 4th postpartum day. [108,109].
In some circumstances, platelet dysfunction (e.g., secondary to preoperative aspirin therapy) may be more important than platelet count in explaining a bleeding disorder. Although preoperative platelet aggregation studies may be helpful in predicting bleeding in some surgical patients, [110]the bleeding time is a poor predictor. [111–113]Neither test is useful in the operating room. The bleeding time, which assesses both platelet function and the vascular component of hemostasis, lacks specificity, is affected by technique and body temperature, and is subject to individual interpretation.
Effectiveness of Platelet Transfusion
It is well established that platelet transfusion can increase platelet counts. The magnitude of effect is variable and is influenced by the release of stored platelets from the spleen and peripheral platelet destruction. Transfusion of one platelet concentrate will increase the platelet count by approximately 5–10 x 109/l in the average adult. The usual therapeutic dose is one platelet concentrate per 10 kg body weight. Single-donor platelets obtained by apheresis are the equivalent of approximately six platelet concentrates. Patients repeatedly transfused over a prolonged period may become alloimmunized and refractory to platelet transfusion. In such patients, human leukocyte antigen-matched or crossmatched platelets may be required. [114].
There is indirect evidence from nonsurgical settings regarding the effectiveness of platelets in controlling bleeding. Studies of leukemic patients with platelet counts of 30 x 109/l or less have suggested that the incidence of spontaneous bleeding can be decreased by platelet transfusions. [115,116]However, similar studies in surgical patients are lacking. Controlled trials of prophylactic platelet transfusion have not demonstrated benefit for patients undergoing cardiopulmonary bypassed [117]or massive transfusion. [101].
Strength of Evidence: Category II-2 and II-3 (Controlled and Uncontrolled Observational Studies)
Expert Opinion.
Recommendations of Other Groups. In 1987, the National Institutes of Health Consensus Conference on Platelet Transfusion Therapy recommended prophylactic platelet transfusion for patients with platelet counts less than 10–20 x 109/l, noting that patients with platelet counts above 50 x 109/l were unlikely to benefit. It added that platelet transfusions at higher platelet counts may be indicated for patients with systemic bleeding or those at increased risk of bleeding because of coagulation defects, sepsis, or platelet dysfunction. [3]In 1994, the CAP recommended platelet transfusion in patients with decreased platelet production and platelet counts below 5 x 109/l. The CAP also recommended considering prophylactic platelet transfusions in patients with platelet counts between 5 x 109/l and 30 x 109/l. For major surgery with life-threatening bleeding, they concluded that transfusions may be indicated at higher platelet counts to maintain a concentration greater than 50 x 109/l. The CAP recommended transfusing patients with enhanced platelet destruction and platelet counts below 50 x 109/l in the presence of microvascular bleeding. [8]It indicated that some experts recommend platelet transfusion after cardiopulmonary bypass in patients with normal coagulation values and platelet counts below 100 x 109/l when major unexplained bleeding occurs. Similar guidelines may be appropriate for patients undergoing neurosurgical procedures, in part because of concerns about local tissue damage from extravasation. The ACOG recommends platelet transfusion in certain patients with hereditary or acquired thrombocytopenia. [5]A 1992 survey of 630 hospitals caring for hematology and oncology patients reported that prophylactic platelets transfusions usually were ordered for patients with platelet counts of 20 x 109/l or less; for patients undergoing minor invasive procedures (e.g., biopsy, central line placement, lumbar puncture), the most commonly cited criteria was 50 x 10 sup 9 /l or less. [118].
Expert Opinion of Task Force. The task force believes that the need for platelet transfusion is dependent on multiple risk factors and not a single laboratory value (e.g., platelet count, bleeding time). The risk in surgical and obstetric patients is defined by the type and extent of surgery, the ability to control bleeding, the consequences of uncontrolled bleeding, the actual and anticipated rate of bleeding, and the presence of factors that adversely affect platelet function (e.g., extracorporeal circulation, renal failure, medications).
There is inadequate scientific evidence to determine the platelet count below which the risk of surgical bleeding is increased. Some recommendations of other groups (e.g., CAP) regarding safe platelet counts are based on evidence that may not be applicable to all surgical patients. In the absence of evidence, the opinion of the task force is that platelet transfusion is justified in bleeding patients at higher platelet counts than recommended for nonbleeding patients because of the increased risk of complications due to bleeding in the surgical patient. The task force believes that intraoperative platelet counts are beneficial in bleeding patients who are massively transfused. Under normal circumstances, platelet counts should be obtained to determine the need for platelet transfusion. In unusual situations, massively transfused patients with microvascular bleeding suspected to be secondary to platelet deficiency may benefit from empirical platelet therapy.
Recommendations. The task force bases its recommendations on available category II-2 and II-3 evidence and expert opinion. The task force concludes that (1) prophylactic platelet transfusion is ineffective and rarely indicated when thrombocytopenia is due to increased platelet destruction (e.g., idiopathic thrombocytopenic purpura);(2) prophylactic platelet transfusion is rarely indicated in surgical patients with thrombocytopenia due to decreased platelet production when the platelet count is greater than 100 x 109/l and is usually indicated when the count is below 50 x 109/l. The determination of whether patients with intermediate platelet counts (50–100 x 109/l) require therapy should be based on the risk of bleeding;(3) surgical and obstetric patients with microvascular bleeding usually require platelet transfusion if the platelet count is less than 50 x 109/l and rarely require therapy if it is greater than 100 x 109/l. With intermediate platelet counts (50–100 x 109/l), the determination should be based on the patient's risk for more significant bleeding;(4) vaginal deliveries or operative procedures ordinarily associated with insignificant blood loss may be undertaken in patients with platelet counts less than 50 x 109/l; and (5) platelet transfusion may be indicated despite an apparently adequate platelet count if there is known platelet dysfunction and microvascular bleeding.
Fresh-frozen Plasma
Approximately 2,000,000 units of FFP are transfused each year in the United States. [1]A significant portion of FFP is transfused inappropriately. [119,120]The scientific rationale for administering FFP rests on the assumptions that (1) patients are at risk of adverse effects from inadequate coagulation factors, and (2) FFP transfusions can decrease those risks. Evidence to support these assumptions is reviewed below.
Adverse Effects of Inadequate Plasma Coagulation Factors
Evidence that coagulation factors can be depleted sufficiently to produce perioperative bleeding due to dilutional coagulopathy is limited. Blood usually coagulates appropriately when coagulation factor concentrations are at least 20–30% of normal and when fibrinogen levels are greater than 75 mg/dL. [101,102,121]Replacement of an entire blood volume leaves the patient with approximately one-third of the original concentration of coagulation factors. [122,123]Although laboratory values such as prothrombin time (PT) and partial thromboplastin time (PTT) may be abnormal, clinical coagulopathy from dilution does not usually occur until replacement exceeds one blood volume or when the PT and PTT exceed 1.5–1.8 times control values. [99–102,121,124–127]In a study of 39 massively transfused patients, Leslie and Toy reported PT values greater than 1.5 times normal in all patients who received 12 or more units of RBCs and in 36% of patients who received less than 12 units. [103]In a series of 12 patients who received RBCs for massive blood loss, Murray et al. reported that PT and PTT were increased in nine patients before replacement of one blood volume but that there was no clinical evidence of abnormal bleeding. Abnormal bleeding occurred primarily in patients with PT or PTT values greater than 1.5 times normal. [121]In a subsequent study of 32 patients who lost more than 50% of their blood volume during elective spinal surgery, Murray et al. found once again that patients with PT or PTT values greater than 1.5 times normal were more likely to have evidence of abnormal hemostasis during surgery. Abnormal PT or PTT, which occurred in 30 of the 32 patients, was not predictive of intraoperative bleeding due to inadequate hemostasis, which occurred in only 17 patients. [127].
Shock, independent of blood loss, may be associated with a consumptive coagulopathy, leading to microvascular bleeding. A study of 36 massively transfused patients found that approximately 150 min of shock was required before significant prolongation of PTT or decreases in factor V activity were noted. [128]A retrospective review of 64 massively transfused patients found that prolongation of PTT during the first 3–4 h correlated with the volume of electrolyte solution administered. Thereafter, the prolongation of PTT corresponded with the duration of the preceding hypotension. [129]A retrospective study of 44 trauma patients found that 33% of those with blunt trauma and brain injuries and 55% of those with penetrating trauma and no brain injuries had a PT greater than 18 s and a PTT greater than 55 s on arrival in the emergency department, after having received only electrolyte solution as prehospital therapy. [130].
In the preoperative patient with no history of bleeding, retrospective studies show that abnormal PT and PTT are poor predictors of surgical bleeding. [131,132]Friedman and Sussman reported no cases of significant bleeding in 71 invasive procedures performed on patients with chronic liver disease and a PT greater than 15 s. [133]McVay and Toy reported that paracentesis and thoracentesis were not associated with increased bleeding in patients with a PT or PTT of up to 2 times control values, [98]Ewe found no association between clotting indexes and bleeding in 200 patients undergoing percutaneous liver biopsy, [134]and Foster et al. reported no complications from coagulopathies in 259 central venous catheterizations. [135]However, outcomes from these noninvasive procedures may not be relevant to the surgical setting.
Prolongation of the PT and/or PTT rarely is seen in association with preeclampsia. However, evidence of increased fibrinolysis without clinically significant hypofibrinogenemia is found in approximately one-third of patients when a D-dimer assay is used [136]; these abnormalities generally do not require treatment.
Effectiveness of Fresh-frozen Plasma Transfusions
Few studies have been performed to determine whether perioperative administration of FFP improves clinical outcome. Although Miller et al. reported correction of PT and PTT with FFP administration in patients massively transfused with whole blood, no change in bleeding occurred until thrombocytopenia was corrected. [124]Spector et al. reported that 600–1,800 ml of FFP was required to reduce the PT to within 3 s of control values in patients with liver disease and that the responses were generally transient (a finding possibly related to hepatic dysfunction and not to normal surgical conditions). [137]A retrospective review of 100 sequential patients having coronary artery bypass surgery and given either albumin or an average of six units of FFP did not demonstrate any differences in blood loss or transfusion requirements. [138]In 17 patients with abnormal intraoperative bleeding due to dilutional coagulopathy, Murray et al. reported that hemostasis improved after FFP administration in 14 patients. Coagulation studies in this group were comparable after FFP transfusion to those of massively transfused patients undergoing the same procedure with no evidence of abnormal hemostasis. [127].
Strength of Evidence: Category II-2 and II-3 (Controlled and Uncontrolled Observational Studies)
Expert Opinion.
Recommendations of Other Groups. In 1985, the National Institutes of Health Consensus Conference on Fresh-Frozen Plasma concluded that FFP is indicated for the following conditions that may occur in the perioperative or peripartum setting: certain coagulation factor deficiencies, selected cases of massive transfusion, and multiple coagulation defects (e.g., liver disease). [2]In 1994, the CAP recommended FFP transfusions for the following indications: massive blood transfusion (more than one blood volume) with active bleeding, urgent reversal of warfarin therapy, and a history or clinical course suggestive of an inherited or acquired coagulopathy (with active bleeding or before an operative procedure). The CAP indicated that the use of FFP as a volume expander or for wound healing was contraindicated. [8]Similar guidelines for FFP transfusion have been issued by the ACOG [5]and the British Committee for Standards in Hematology; the Committee noted that at least four units of FFP usually will promote coagulation in adults. [139].
Expert Opinion of Task Force. The task force believes that few clinical circumstances in the perioperative or peripartum setting result in coagulopathies that require replacement of coagulation factors with FFP. The special clinical circumstances that might warrant FFP in the nonbleeding patient include the urgent reversal of warfarin therapy and the treatment of known coagulation factor deficiencies for which specific factor concentrates are unavailable.
In the patient with microvascular bleeding, there is adequate scientific evidence to suggest that coagulation studies (PT/PTT) obtained in the operating room are useful in detecting a coagulopathy that may respond to FFP transfusion. Although massive blood replacement can produce prolongation of PT and/or PTT, the task force believes that a true dilutional coagulopathy does not ordinarily occur until more than 100% of the patient's blood volume has been replaced. The task force believes that FFP is beneficial in patients with microvascular bleeding or hemorrhage who are massively transfused if the PT/PTT values exceed 1.5 times the laboratory's normal values. If PT and PTT cannot be obtained in a timely fashion, the task force believes that massively transfused patients with microvascular bleeding that is believed to be secondary to coagulation factor deficiency may benefit from empirical FFP therapy.
Recommendations. The task force bases its recommendations on available category II-2 and II-3 evidence and expert opinion. The task force recommends the administration of FFP with the following guidelines:(1) for urgent reversal of warfarin therapy;(2) for correction of known coagulation factor deficiencies for which specific concentrates are unavailable;(3) for correction of microvascular bleeding in the presence of elevated (> 1.5 times normal) PT or PTT;(4) for correction of microvascular bleeding secondary to coagulation factor deficiency in patients transfused with more than one blood volume and when PT and PTT cannot be obtained in a timely fashion;(5) FFP should be given in doses calculated to achieve a minimum of 30% of plasma factor concentration (usually achieved with administration of 10–15 ml/kg of FFP), except for urgent reversal of warfarin anticoagulation, for which 5–8 ml/kg of FFP usually will suffice. Four to five platelet concentrates, one unit of single-donor apheresis platelets, or one unit of whole blood provide a quantity of coagulation factors similar to that contained in one unit of FFP (except for decreased, but still hemostatic, concentrations of factors V and VIII in whole blood); and (6) FFP is contraindicated for augmentation of plasma volume or albumin concentration.
Cryoprecipitate
Almost 1,000,000 units of cryoprecipitate are transfused each year in the United States. [1]Cryoprecipitate, which contains factor VIII, fibrinogen, fibronectin, von Willebrand's factor, and factor XIII, is used for the correction of inherited and acquired coagulopathies. Its use in the operative setting is based on the assumptions that (1) patients with these coagulation factor deficiencies are at increased risk of hemorrhagic complications, and (2) replacement of coagulation factors is effective in decreasing these risks. Evidence to support these assumptions is reviewed below.
Adverse Effects of Coagulation Factor Deficiencies
There is limited evidence from observational studies that patients with certain inherited or acquired coagulopathies (e.g., hemophilia A, von Willebrand's disease, hypofibrinogenemia, disseminated intravascular coagulation, hepatic insufficiency) are at increased risk of perioperative or peripartum bleeding. [99,140].
Effectiveness of Cryoprecipitate Transfusion
One unit of cryoprecipitate per 10 kg body weight raises plasma fibrinogen concentration by approximately 50 mg/dL in the absence of continued consumption or massive bleeding. No studies have been performed to determine whether perioperative or peripartum administration of cryoprecipitate improves clinical outcome. Indirect observational evidence suggests a beneficial effect for patients with factor VIII deficiency and certain subtypes of von Willebrand's disease. [141,142]However, most patients with factor VIII deficiency are treated with factor VIII concentrates, and patients with some subtypes of von Willebrand's disease respond to administration of desmopressin acetate (DDAVP). [143–146]Similarly, coagulopathy associated with uremia can be treated with cryoprecipitate, but DDAVP is usually the first-line therapy.
Severe placental abruption frequently is associated with disseminated intravascular coagulation, in which thrombocytopenia, hypofibrinogenemia, and increased fibrinolytic activity are the most consistent findings. Indirect evidence suggests that administration of cryoprecipitate in these settings increases the plasma fibrinogen concentration. The hypofibrinogenemia responds well to treatment with cryoprecipitate once delivery has been accomplished. [140,147].
Strength of Evidence: Category II-3 and III (Uncontrolled Observational and Descriptive Studies)
Expert Opinion.
Recommendations of Other Groups. In 1994, the CAP recommended cryoprecipitate transfusions in bleeding patients with hypofibrinogenemia, von Willebrand's disease, and patients with hemophilia A (when factor VIII concentrate is not available). [8]Similar recommendations have been issued by the ACOG. [5]The British Committee for Standards in Haematology recommended the administration of cryoprecipitate for massively transfused patients with microvascular bleeding when the fibrinogen level is less than 80 mg/dL. [114].
Expert Opinion of Task Force. There is little scientific evidence regarding the effectiveness of cryoprecipitate in improving clinical outcomes, and therefore, the task force believes that its perioperative and peripartum use should be limited to selected indications. Based on clinical experience, the task force believes that cryoprecipitate is likely to be effective in patients with von Willebrand's disease unresponsive to DDAVP, congenital fibrinogen deficiencies, and consumptive coagulopathies when fibrinogen levels are below 80–100 mg/dL.
Recommendations. The task force bases its recommendations on available category II-3 evidence and expert opinion. The task force recommends considering the administration of cryoprecipitate for (1) prophylaxis in nonbleeding perioperative or peripartum patients with congenital fibrinogen deficiencies or von Willebrand's disease unresponsive to DDAVP (whenever possible, these decisions should be made in consultation with the patient's hematologist), (2) bleeding patients with von Willebrand's disease, and (3) correction of microvascular bleeding in massively transfused patients with fibrinogen concentrations less than 80–100 mg/dL (or when fibrinogen concentrations cannot be measured in a timely fashion).
Conclusions
Adherence to proper indications for blood component therapy is essential because of the potential adverse effects and costs of transfusion. These risks can be reduced further by other effective measures, the most important being efforts to minimize exposure to allogeneic blood through use of autologous transfusion and other blood conservation techniques, but the cost-effectiveness of these practices requires further study. [148]The unnecessary complications of incompatible blood transfusions can be minimized by careful specimen, unit, and patient identification before blood sampling and transfusion and by maintaining a high index of suspicion for transfusion reactions. Most importantly, transfusion decisions should be based on sound physiologic principles and a comprehensive assessment of the patient's risk factors.
A variety of comprehensive intervention strategies and transfusion algorithms have been associated with reductions in inappropriate blood component therapy. [9,41,87,149–153]Inappropriate RBC use has been decreased by audits, discussions with ordering physicians, ward rounds, computer-based decision-support systems, and comprehensive educational outreach programs. [47,154–156]Studies have suggested that FFP transfusions can be reduced by programs that include chart audits and the review of results with ordering physicians, dissemination of practice guidelines, case presentations, house staff education, and review by pathologists of transfusion orders. [157–160]Similar results have been achieved for platelet use. [161,162].
All available evidence regarding the effectiveness of blood component therapy qualifies as category II-2, II-3, or III evidence. The lack of data from prospective, randomized studies with adequate sample size, control groups, clinical outcome measurements, and other features of well designed clinical effectiveness research impedes development of evidence-based clinical practice guidelines for blood component therapy. To establish a stronger scientific basis for transfusion practice, future research is essential to provide meaningful evidence regarding the indications and effectiveness of blood component therapy. As these data become available, guidelines for blood component therapy are likely to change.