Chapter 68. Anemia and Blood Transfusion

• The current practice of transfusion in the ICU is conservative, reflecting a balance between the benefits and adverse consequences of transfused blood.
• Anemia developing in the ICU should not be attributed solely to phlebotomy; red cell loss, destruction, or underproduction should also be considered.
• Evaluation of the basis for anemia, beginning with analysis of the blood smear, reticulocyte count, and red cell indices, may lead to easier management of the anemia and may prove useful in overall treatment.
• For most critically ill patients, a transfusion target for hemoglobin of 7 to 9 g/dL is at least as safe as a higher threshold of 10 to 12 g/dL; the lower value may be superior.
• Transfusion is rarely associated with life-threatening complications such as transfusion-related acute lung injury, major hemolysis, or lethal infection.
• Erythropoietin and oxygen-carrying blood substitutes remain unproven for critically ill patients.
• Massively transfused patients are at risk for hypothermia and dilutional coagulopathy.

Red blood cells are the most efficient vehicle by which oxygen is delivered to the tissues of whole organisms. The evolution of red cells and the circulatory system allowed for the development of larger, more complex organisms with higher levels of activity. Deficiency in the number and in the local delivery of red blood cells causes relative oxygen deficiency in tissues. Oxygen deficit causes metabolic insufficiency, with inadequate energy production to sustain activity, or even life. Hemoglobin, the complex molecule contained within human red blood cells, has unique properties in and of itself, allowing for the efficient uptake of oxygen in an oxygen-rich environment, and the efficient divestiture of oxygen in an oxygen-poor environment. The whole organism also has evolved to attempt to regularly recycle and reuse the important portions of the hemoglobin molecule, thus reducing the need for ongoing sources of iron. Anemia, the lack of red blood cells in normal volumes, is therefore a disorder that threatens the homeostasis of the entire organism, and when it occurs in the setting of already life-threatening illness, anemia is a potential contributor to mortality.

Blood transfusion has long been recognized as a potential remedy for blood loss, although the first efforts to transfuse blood from animals to human beings or from one human to another were marred by the lack of knowledge of the unique antigens on the surface of red blood cells and the immune response to them. Once a systematic approach to the collection, storage, cross-matching, and transfusion of blood was developed, enthusiasm for transfusion reached its peak. Tempered by the recognition of new transmissible illnesses, by the increasing limitations of the blood supply,1 and by the awareness of the limitations of aggressive transfusion, modern critical care practice has evolved to a more conservative approach to the use of transfusion as a remedy for anemia in critically ill patients.

The frequency of anemia in critically ill patients has been studied in a variety of settings, all with the conclusion that it is a common problem that tends to worsen with severity of illness and duration of stay in the ICU. Up to 40% of patients admitted to the ICU in one large sequential cohort study were bleeding on admission,2 and 40% to 50% require blood transfusion during their ICU stay.3 By day two in the ICU, nearly 95% of patients are anemic.4 Trauma patients are subject to higher rates of transfusion, 55%, than medical patients in one comparative study, even though these patients had lower rates of comorbid illnesses.5

In these studies, anemia has been defined in various ways, mostly as any deviation from the normal hemoglobin content of blood for men and women. The severity of anemia, however, may vary, and the data on the effects of anemia show that severity is an important marker of both concomitant complications and outcome. Hemodilution in laboratory animals is a relatively well tolerated phenomenon; acute hemodilution results in evidence of myocardial dysfunction and tissue ischemia demonstrable only at hemoglobin levels of 3 to 5 g/dL. Normal human volunteers also appear to tolerate hemodilution to levels of 5 g/dL.6 However, in animal models of coronary ischemia, and in acute hemodilution in humans with coronary disease, hemoglobin levels of 7 to 10 g/dL may produce evidence of further ischemia and of depressed ventricular function.7 Conversely, the normalization of hematocrits with transfusion and erythropoietin in patients with chronic anemia and coronary disease in one randomized trial did not result in an outcome benefit.8

Measurement of the effects of severe anemia in the critically ill is not straightforward. The assessment of the consequences of anemia is complicated in severely ill patients who cannot report the usual symptoms of anemia—dyspnea, fatigue, lightheadedness, angina, or claudication. The usual signs of end-organ problems due to poor oxygen delivery may also be difficult to interpret: tachycardia and tachypnea have many other causes, cardiac ischemia may be multifactorial, and physical signs of poor organ perfusion may result from a variety of other conditions.

Guidelines for the evaluation of the physiologic impact of anemia are based on what is known of the physiology of anemia. In the absence of adequate delivery of red cells over a given volume of perfused tissue, oxygen extraction rises. Thus the delivered oxygen content may remain constant, but the mixed venous oxygen content will fall compared to normal. Low mixed venous oxygen saturation, usually below 70%, is often employed as a marker of anemic physiology. Under normal circumstances, peripheral oxygen extraction is about 25%, so the ratio of oxygen saturation in mixed venous blood to that delivered (SvO2/SaO2), and the oxygen extraction ratio (SaO2 minus SvO2 divided by SaO2) all have been shown to provide indirect clues to the presence of tissue oxygen delivery problems.9

To make matters more complicated, there are numerous case reports of patients tolerating severe anemia in the postoperative setting in Jehovah's Witnesses, in patients with renal failure, and in situations in which matched blood could not be obtained. These reports were compiled in a formative review by Hébert and colleagues for the Canadian Guidelines on Blood Transfusion.10 This qualitative review concluded that there was little evidence of perioperative mortality when the preoperative hemoglobin concentration was above 8 g/dL and the estimated blood loss at surgery was less than 500 mL. However, patients having the same hemoglobin level and either older age (>50 years) or greater blood loss were subject to greater mortality.11 Hemoglobin levels below 8 g/dL are accompanied by a rise in mortality.12

Risk factors for adverse outcomes from anemia have been examined in a number of recent publications, and these risk factors are more prevalent in patients in the ICU.13–15 Most of the evidence for these adverse risk factors is empiric, retrospective, and cohort based. ICU admission alone is a risk factor for poor outcome, as are advanced age and underlying cardiac, cerebrovascular, or respiratory insufficiency.16,17 In particular, the Carson series of Jehovah's Witnesses allows for a calculation of the relative risk according to perioperative hemoglobin levels: The odds ratio for mortality in patients having a hemoglobin of 6 g/dL preoperatively is 2.5 times the risk of a nonanemic patient, but the mortality rises from 2.5- to 16-fold in the presence of cardiac disease. A retrospective review of Medicare data for patients having myocardial infarction showed that transfusion was associated with lower 30-day mortality in patients having hematocrits of 27% or lower, and with higher mortality in patients having admission hematocrits of 36% or more.18 Perioperative transfusions after hip fracture were associated with no difference in outcome for patients with hemoglobin levels of 8 g/dL or greater. While overall severity of illness appears to be a risk factor, most of the other studies involving patients in the ICU are too small to examine the independent effects of age, cardiac and cerebrovascular disease, and degree of blood loss.

Causes of Anemia in the ICU

Red Cell Loss or Destruction

Hemorrhage is the most common cause of anemia in the ICU; those losses arise from surgical intervention or from medical causes of bleeding, and are the sole contributor to anemia in up to 20% of anemic patients in the ICU (Table 68-1).4 The severity of red cell loss, and the resultant need for transfusion, is also dependent on the degree of underlying illness in a particular patient, as well as the type of surgery being performed, and the exact medical cause of bleeding. Underlying conditions such as renal failure, nutritional failure, liver disease, and drug effects have been shown to increase the risk of bleeding in critically ill patients.2

Red cell losses attributable to phlebotomy can be considerable; when originally reported, they probably amounted to the equivalent of two units of red blood cells in an average ICU patient.19 Phlebotomy losses are greatest in the first few days of ICU admission. Following the initial publications regarding the extent of red cell loss, a variety of conservation measures have been employed, including microanalytic techniques, continuous arterial sampling, the use of external oxygen monitoring, and increasing nursing precautions for blood loss at phlebotomy or other patient procedures.20 Early closure of burns reduces red cell loss dramatically. Recently published prospective series suggest that the fraction of red cell loss attributable to “medical vampires” has been reduced to less than 20% of all calculated red cell loss.21

Red cell destruction also occurs in critically ill patients, although is it a less common cause of red cell loss than is blood loss. Intravascular events that cause mechanical shearing and red cell destruction include disseminated intravascular coagulation (DIC), thrombotic thrombocytopenic purpura (TTP), and endovascular devices, especially newly placed endovascular grafts and older mechanical cardiac prosthetic valves. Thermal destruction of red cells may occur in particular situations, including malfunctioning blood warmers, patients with thermal injuries and with heat exposure, but not in patients with high fever only.

Immune-mediated red cell destruction is also less common than blood loss, but often unrecognized. Autoimmune hemolytic anemia is a very rare event in a critically ill patient, but may be seen in patients with the acquired immunodeficiency syndrome (AIDS), autoimmune diseases, and lymphomas. More common is alloimmunization due to transfusion; delayed red cell hemolysis due to delayed transfusion reactions is often unrecognized and may occur in one in 4000 red cell transfusions.22 The estimates of red cell sensitization for patients undergoing coronary bypass surgery, for example, are a risk of approximately 1% per unit of red cells transfused. While critically ill patients are less likely to experience primary immunization to homologous red cell antigens than are either healthy volunteers or less sick patients,23 nonetheless, prior exposures to homologous red cells may predispose to delayed transfusion reactions even though the patient is at the time quite severely compromised.

Drug-induced red cell destruction may occur by several mechanisms: glucose-6-phosphate dehydrogenase (G-6-PD) deficiency is the most common cause of metabolic hemolysis. In the presence of oxidative stress from drugs such as sulfas, primaquine, nitrofurantoin, phenazopyridine, and dapsone, G-6-PD-deficient patients may experience intravascular hemolysis, occasionally on a large scale with accompanying hyperbilirubinemia and hemoglobinuria. But other minor metabolic abnormalities, unstable hemoglobin molecules, and red cell membrane defects may also be associated with hemolysis under the stress of critical illness. More commonly, drugs bound to the red cell membrane act as antigens and stimulate an immune response or form an immune complex which then binds to the red cell surface as a by-stander site. The result is immune-mediated, usually extravascular, hemolysis and anemia. Because the antibodies only bind when the drug is present, this particular form of immune-mediated hemolysis is particularly reversible with the cessation of the drug. Other drugs such as methyldopa are capable of inducing autoantibodies; still others cause hemolysis by unknown mechanisms. Common causes of drug-induced immune hemolysis include penicillins, cephalosporins, quinine and related drugs, sulfas and sulfonylureas, and procainamide.

Another rare cause of red cell destruction in critically ill patients is sepsis due to a variety of infectious agents that can cause hemolysis. Malaria is usually recognizable by history and blood film examination. Babesiosis is a tick-borne intracellular parasite that may also be recognized on peripheral blood smear. Bartonella bacilliformis adheres to red cell membranes and induces their destruction in the spleen and reticuloendothelial system. Massive intravascular hemolysis may occur in patients with Clostridium perfringens sepsis, due to the formation by the organisms of lysolecithins and resulting red cell membrane proteolysis. Overwhelming infections with the DNA viruses Epstein-Barr, herpes, and cytomegalovirus are associated with hemolysis, probably due to immune complexes. Hemolysis has also been observed in varicella, influenza, measles, and coxsackievirus infections. Mycoplasma infections are commonly associated with the formation of cold agglutinins; occasionally these produce overt cold-mediated hemolysis. Infections with enterotoxin-inducing species of Escherichia coli and other enteric organisms trigger hemolytic uremic syndrome.

Red Cell Underproduction

The humoral response to stress and infection includes the production of a wide variety of inflammatory cytokines that have an adverse effect on red cell production, erythropoietin production and responsiveness, iron metabolism, and the reticuloendothelial system. The end result of this adaptation to stress and infection is the characteristic anemia of chronic disease (ACD). The laboratory phenomena associated with the anemia of chronic disease have all been observed in critically ill patients, and evolve rapidly at the onset of severe illness such as sepsis and multiorgan failure.24 The anemia of chronic disease may also be present prior to ICU admission, as it is a common manifestation of inflammatory and neoplastic disease.

Inflammatory cytokines cause suppression of marrow production of red cells. This phenomenon is seen in a wide variety of illnesses, including minor viral infections. Ordinarily it is brief in duration and produces only a small impact on red cell mass. In critical illness, however, marrow suppression is combined with other disorders of red cell synthesis to produce more profound and durable anemia. One phenomenon that has been documented in the anemia of chronic disease is shortening of red cell lifespan. There is no evidence that this is directly immune-mediated, nor that the red cells themselves are abnormal. The current theory is that the 20% to 30% reduction in lifespan is a result of exaggerated surveillance by an activated reticuloendothelial system.25,26

Underproduction of erythropoietin has also been documented in critically ill patients. This may be a direct consequence of hepatic or renal failure, but is more often related to suppression of production of erythropoietin and perhaps to relative refractoriness of the marrow to its effects. Serum erythropoietin levels below those expected for the degree of anemia have been demonstrated in critically ill patients.27,28 What is more controversial is whether pharmacologic erythropoietin treatment has a significant impact on overall outcomes (see below).

Another phenomenon observed in critically ill patients and in patients with chronic disease is abnormal iron metabolism. Iron has two sources: dietary iron is absorbed in the small intestine, transferred across the epithelial cells and into the bloodstream to the carrier molecule transferrin, where it is made available to the marrow storage cells. Iron is also recycled from senescent red cells by macrophages. Transferrin and ferritin are the molecules that deliver stored iron to developing erythrocytes. In chronic illness, ferritin levels are usually normal or elevated, but iron transport via ferritin into developing erythrocytes is suppressed, probably by responses to inflammation within the macrophages themselves. Transferrin production and transfer of iron from macrophages to transferrin is also often suppressed, and serum iron and iron binding capacity levels are low as a consequence. The result is that although iron absorption is normal, iron reuse from storage is suppressed, and overall production of red cells declines.

While the suppressive effects of the anemia of chronic disease are likely to be the most important elements causing prolonged anemia in the ICU, a variety of other causes of red cell underproduction should be assessed and treated in critically ill patients. Nutritional factors are particularly important, as they lend themselves to obvious solutions. Underlying nutritional deficiency may antedate entry into the ICU, particularly in the elderly, in alcoholics, and in patients with underlying gastrointestinal or other conditions that predispose to undernutrition. Folic acid, vitamin B12, micronutrients (copper, selenium, and zinc) and calorie malnutrition all may cause anemia or perpetuate it in a critically ill patient.29 Bone marrow suppression from drugs is important in cancer patients, but should be evaluated in patients taking suppressive therapies for autoimmune diseases, and in patients on chronic nonphenothiazine antipsychotic medications and phenytoin. The presence of pancytopenia should prompt an evaluation of the peripheral blood smear for evidence of disruption of marrow architecture, and perhaps bone marrow biopsy to evaluate the anemia.

Evaluation of Anemia in the ICU

Patients with anemia upon admission to the ICU should have the same considerations of cause as any patient admitted to the hospital. A medical history may disclose the underlying conditions, the duration of the anemia, and prior evaluation. Patients having unknown causes or new onset of anemia upon ICU admission should be thoroughly evaluated at the onset of their stay, since the underlying cause may both prove useful in the overall treatment of the patient, and will certainly assist in the management of the anemia during the ICU stay. Patients who acquire their anemia during the ICU stay have a slightly different set of conditions to consider.

The work-up of anemic patients should begin with the evaluation of the peripheral blood smear and the reticulocyte count. Normal to low reticulocyte counts should be expected in patients sick enough to merit ICU admission, and thus a high reticulocyte count should prompt an evaluation for chronic hemolysis or other red cell destruction. Patients' mean corpuscular volume (MCV) should also be evaluated; frankly megaloblastic red cells have a limited differential diagnosis, as do frankly microcytic red blood cells. Most patients, however, will have normal red cell volumes. The evaluation of both the peripheral blood smear and the MCV are complicated by the addition of transfused red blood cells. Pancytopenia should prompt a still different set of diagnoses, including bone marrow disorders and marrow toxins.

The evaluation of hemolysis also begins with the peripheral blood smear. Red cell fragmentation and distortion, producing schistocytes and teardrop cells, are indications of intravascular events or intramedullary events that mechanically fragment red cells, and suggest the diagnosis of DIC, TTP, or marrow infiltration.

Most patients will prove to have normochromic, normocytic anemia with a low or normal reticulocyte count and no obvious alterations in red cell morphology on peripheral blood smear. Such patients are likely to have the anemia of chronic disease. This may antedate ICU admission and be due to underlying chronic illness, or may appear in the ICU as a consequence of acute illness and marrow suppression, combined with steady blood loss. Whether such a patient requires some other confirmatory test to prove the presence of ACD is controversial. For example, critically ill patients (and many other categories of chronically ill patients without renal failure) display erythropoietin levels that are below the range considered normal for their degree of anemia. However, measuring erythropoietin levels is expensive, often not readily available, and is not considered a standard for the evaluation of ACD in the ICU. Similarly, the iron transport protein ferritin should be normal or elevated in ACD, and low when ACD is accompanied by iron deficiency. However, serum iron levels themselves are often low, and transferrin levels low in ACD in the ICU. These common indices of the suppression of red cell production are in themselves only suggestive of ACD and are not widely used to make a diagnosis. Hence, the diagnosis of ACD is usually one of exclusion. The extent to which testing for other causes of anemia is undertaken depends on the clinical situation.

Transfusion Therapy in the ICU

Indications for Transfusion

In spite of a long tradition of transfusion of red blood cells in critically ill patients, the precise indications for transfusion remain a source of controversy, and specific transfusion practices may vary widely among clinicians. Prior to the major randomized studies of transfusion policies, survey studies of transfusion practice showed that about half of ICU patients were receiving red cell transfusions,30 and another showed that if the ICU stay was longer than a week, the rate of transfusion was 85%.31 Total numbers of transfusions were high, and ICU practice was characterized by high rates of transfusions.32

There have been only ten randomized trials of transfusion policy in the ICU, and only one of these was large enough to draw specific, statistically significant conclusions.33 The Canadian Critical Care Trials Group compared a liberal (target hemoglobin 10 to 12 g/dL) with a restrictive (target hemoglobin 7 to 9 g/dL) red cell transfusion policy in patients stratified for disease severity. At 30 days from randomization, the restrictive strategy was at least as good as, if not better than (p = 0.11) the liberal strategy, and overall hospital mortality was significantly lower for the restrictive strategy group (p = 0.05). For patients under 55 years of age and for patients with lower (≤20) Acute Physiology, Age, and Chronic Health Evaluation (APACHE) II scores, the restrictive strategy was clearly superior. In addition, liberal transfusion was not associated with shorter ICU stays, less organ failure, or shorter hospital stays; longer mechanical ventilation times and cardiac events were more frequent in the liberal strategy group. A later subgroup analysis of patients having cardiovascular disease, while small enough to have statistical doubt, suggested that for patients with severe ischemic coronary disease a more liberal transfusion strategy was probably appropriate.34 The Canadian study has highlighted the many and complex issues involved in transfusion decision making in the ICU.

Since the publication of the Canadian study, several large reports have examined utilization of red cell transfusions in critical care units. Vincent and colleagues3 surveyed European ICUs, and found that transfusion rates for 3534 patients were 37% during the ICU stay and 12.7% after the ICU stay. The mean pretransfusion hemoglobin level was 8.4 g/dL. Corwin and colleagues35 studied 284 ICUs in the United States a year later and found great similarity: nearly 50% of patients received transfusions, and the mean threshold hemoglobin level was 8.6 g/dL. A single large Scottish teaching hospital reported a more parsimonious practice; the rate of transfusion was still 52% in its ICU patients, but the total volume of blood used was slightly smaller and the mean pretransfusion hemoglobin was only 7.8 g/dL.36 All of these authors have concluded that ICU practice has not fully embraced the guidelines of the Canadian clinical trial. In contrast, eighteen hospitals in Australia and New Zealand have reported on transfusion in 1808 consecutive ICU admissions, and although the authors found a median pretransfusion hemoglobin of 8.2 g/dL, the rate of transfusion was lower, at only 19.7% of patients, 60% of whom were bleeding.37 The “inappropriate” transfusion rate was 3%. The authors speculate that the practitioners may have been influenced by the publication of the Canadian study and their own regional survey of transfusion practices. Nonetheless, they agree that full implementation of the Canadian guidelines in their clinical setting might be controversial.

Components Available from the Blood Bank

Whole Blood and Red Cell Products

Whole blood, approximately 450 to 500 mL obtained by phlebotomy, is generally stored in an acid-citrate-dextrose medium, and is most often available only for replacement during acute hemorrhage, where transfusion requirements are likely to be large. Because most donated units of blood are separated into components rather than remaining as whole blood, whole blood has become less available.

Fresh whole blood, which has been removed from donors within a few hours of its use, has not been shown to be more effective than stored whole blood, and is only very rarely available, in part because complete infection testing may take several days. Whole blood that has been stored less than 10 days does exhibit longer red cell survival than blood that has been subjected to longer storage, and there is some evidence that the red cells in whole blood have slightly longer survival than red cells that are stored as packed red blood cells. Patients who are at high risk of clinical problems from the transfusion of hemolyzed or senescent red cells may be candidates for blood that has been stored for a shorter time period. Such patients include those with anticipated difficulty clearing bilirubin, such as neonates.

Packed red blood cells (PRBCs) consist of essentially all of the red cells in a unit of whole blood, with platelets and plasma largely removed, so that the resulting cells are at a hematocrit of about 70%. Their higher viscosity may result in longer transfusion times.

Red blood cells store well when frozen, usually in glycerol. Their shelf life is vastly extended, and they can be shipped with little risk of loss. Rare donor types may be stored in one institution and made available nationwide to other blood banks. Frozen red cells are no less likely to transmit most diseases than refrigerated stored red cells. Preparation for transfusion of frozen red cells consists of careful thawing and dilution and recentrifugation in order to remove the glycerol. Opening and manipulating the stored red cells results in a short shelf life; they must be used within 24 hours in order to avoid the risk of bacterial contamination.

“Washed” red cells are produced by recentrifugation in saline solution in order to remove plasma and contaminating white blood cells. They must also be used rapidly after washing, and because of this and the cost of the washing procedure, most institutions use white blood cell filters to prevent febrile transfusion reactions prior to considering the washing procedure. Washing does permit the administration of red cell transfusions to patients with a history of major allergic responses to plasma transfusion (see below).

Irradiation of blood products renders the lymphocytes nonviable, and is used to prevent graft-versus-host disease (see below). Irradiated red cells have a normal life span, do not require expedited handling, and may perhaps be less likely to transmit viral disease.

One blood conservation strategy that is probably underused, although it is currently estimated to be the source of approximately 7.5% of all transfusions, is autologous predeposit of whole blood. Designated donor blood, which has not been shown to be safer than volunteer donor blood, comprises 1% to 2% of all transfusions. Whole blood from autologous predeposit or from designated donors is tested and stored in the same fashion as donor blood. Institutional guidelines vary with respect to the handling of autologous units that test positive for transmissible diseases. While federal regulations allow the storage and use of such units of blood, many institutions either label it with biohazard labels or do not allow its use at all in order to attempt to avoid the potential for clerical error and transfusion to another patient (see section on risks, below). Unfortunately, wastage of both autologous and designated donor blood is approximately 50%, higher than that for volunteer donor blood (about 15%). This observation suggests that overcollection may be occurring.

Red blood cells are customarily stored for 72 to 90 days; during that time they gradually become deficient in 2,3-diphosphoglycerate (2,3-DPG), and a small fraction of stored cells also gradually lyse, releasing potassium, hemoglobin, and red cell stroma into the pack. Red cell senescence is accelerated in storage, and the survival of red cells held in storage is shorter than normal. Altered conditions of storage, such as the addition of alanine, that improve red cell survival unfortunately diminish 2,3-DPG levels, and vice versa. Individual blood bank policies and practices with respect to preservation solutions vary. Usage of red cells beyond the storage limit can result in higher risks of bacterial transmission.

Heat and mechanical fragmentation both can disrupt stored red cells. Blood warmers must be checked for temperature accuracy, since exposure to temperatures over 42°C can result in significant hemolysis of red cells. Attempts to restore the 2,3-DPG levels of stored red cells are not necessary, as the cells rapidly replete when transfused. However, citrate toxicity, which results in hypocalcemia, may be seen when large volumes of blood are administered over a very short time (more than a liter in an hour or two). The intravenous administration of calcium gluconate is recommended for treatment of this problem.

Platelets, Plasma, and Other Products

Platelets do not survive refrigeration, and therefore platelets are removed from fresh whole blood and stored at 22°C under continuous agitation. They may also be obtained by pheresis of a single donor, which harvests about the same amount as five to six units of platelets. Stored platelets are viable for up to a week, and transfused platelets last up to 72 hours in normal recipients.

Platelets pheresed from a single donor may be used to reduce the risk of immunization of recipients who are repeatedly exposed to platelet transfusions. Platelets may be matched to specific platelet antigens or to human leukocyte antigen (HLA) haplotypes; both techniques may be used to successfully give platelet transfusions to recipients who have already become sensitized to random donor platelets.

Plasma may be stored whole, usually as fresh frozen whole plasma. Each unit of plasma from the blood bank is about 200 to 250 mL. Fresh frozen plasma must be thawed carefully, and thawed plasma is rarely kept in storage in blood banks. Cryoprecipitate is prepared from fresh plasma by freezing, then slow thawing to 60°F. The resulting flocculant precipitate contains factor VIII C/von Willebrand factor (vWF), fibrinogen (about one-third of that in the unit of blood), and fibronectin. Commercial cryoprecipitate is lyophilized after the precipitate is harvested. Both of these plasma products have long shelf lives, though shorter half lives once transfused than their native counterparts. When made into cryoprecipitate, a unit of plasma yields about 80 to 100 units of factor VIIIC/vWF and about 200 to 300 mg of fibrinogen. Cryoprecipitate can be reconstituted in extremely small volumes, making it useful when volume overload is a concern.

Specific coagulation factors may be manufactured from pooled plasma or as recombinant DNA products. The latter are more expensive, but appear to have fewer complications, including sensitization and disease transmission. These include human recombinant factor VIII, factor IX, activated protein C, and activated factor VII.

Commercial preparations of plasma also include plasma products that are free of immunoglobulins, that consist of activated or non-activated liver-dependent clotting factors (prothrombin complex), and fibrin “glue” or fibrin sheets.

Risks of Transfusion Therapy

Cross-Matching and Sensitization

Red cells are cross-matched by mixing recipient serum with type-specific (ABO and Rh) donor cells at two different temperatures (major cross-match) and donor serum with recipient red cells (minor cross-match). This takes time, and only under desperate circumstances should this step be omitted. Cross-matching may be complicated by a variety of problems. Nonspecific cold agglutinins are seen in a host of medical conditions, and they cause difficulty matching at room temperature and sometimes at 37°C. Cold agglutinins are usually not associated with serious hemolytic reactions in vivo, but they make cross-matching difficult and they may mask the presence of other antibodies. The warm antibodies seen in autoimmune hemolytic anemia are usually IgG, are associated with autoimmune disease and certain malignancies, and make conventional cross-matching of red cells very difficult because they react with almost all red cells. The procedure for cross-matching in patients having warm antibodies involves eluting the warm antibodies from the patient's plasma so that a cross-match may be performed. If antibody elution is not available, the consensus is that blood banks should provide the best possible match of type-specific blood, knowing that hemolysis may be no greater than it is endogenously. Surface antigens are weakly expressed in neonates, the extremely elderly, and massively transfused patients, making cross-matching difficult. Larger volumes of blood will be needed to cross-match these patients.

In spite of cross-matching, sensitization to homologous blood transfusion occurs, and may become a clinical problem for repetitively transfused individuals, especially those who have rare blood types. The risk of sensitization to red cells is estimated at 1% per transfusion episode; in contrast, the risk of platelet sensitization is 5% to 10% per pooled random donor transfusion episode. Recipient sensitization to leukocyte antigens also occurs during red cell transfusion and may be the basis for some febrile reactions, but is otherwise usually of little clinical significance. On the other hand, the presence of leukocyte antibodies in the donor plasma, the presumed cause of transfusion-associated lung injury, can be prevented only by much more complex leukocyte typing.

Hemolytic Transfusion Reactions

Intravascular hemolysis of transfused red blood cells is caused by complement fixation and rapid intravascular destruction of ABO incompatible red cells, or by the same process in a previously sensitized patient with a high-titer antibody such as anti-Kell or antibodies to Rh system components. The syndrome that results from massive intravascular hemolysis is due to the release of red cell stroma, which causes cytokine, bradykinin, and granulocyte mediator release. Hypotension, capillary leak, and oliguria are manifestations that follow very rapidly after the onset of intense intravascular hemolysis. The most feared complications are disseminated intravascular coagulation, acute renal failure, and the acute respiratory distress syndrome (ARDS). The majority of acute hemolytic transfusion reactions are caused by clerical or nursing error, and fortunately, deaths are rare.

The treatment of hemolytic transfusion reactions is supportive. Early recognition of the syndrome and stopping the offending transfusion are critical. Recognition may be difficult during general anesthesia, and the symptoms at the bedside may also be initially confusing. The decision to stop a transfusion is often difficult unless the findings are very extreme. Most clinicians believe that supporting high urine output is important during the resuscitation of these patients, but that diuretics should be used only if the patient shows intravascular volume overload. The use of mannitol is controversial. There is no role for steroids, antihistamines, heparin, colloid, or other specific pharmacologic interventions. Treatment of the complications should be vigorous, as the majority of patients survive the event.

Delayed Transfusion Reactions

Red cell sensitization may result in an antibody response that is initially weak and does not persist. However, the patient's next exposure to that red cell antigen will arouse an anamnestic immune response, and higher titers of antibody may occur fairly rapidly after transfusion. If an extremely brisk anamnestic response produces antibodies capable of fixing complement, patients may have intravascular hemolysis of the transfused cells within a few days of transfusion. These delayed transfusion reactions usually produce jaundice, and elevated lactate dehydrogenase (LDH) and low haptoglobin levels, but may occasionally be severe and result in oliguria. If the antibody does not fix complement, extravascular hemolysis of the transfused red cells will occur. This type of delayed transfusion reaction is often subtle, but results in elevations in LDH and indirect bilirubin, occasionally the appearance of spherocytes in the peripheral blood smear, a positive direct antiglobulin (Coombs) test, and falling hemoglobin levels. Documentation of such a reaction is important so that future transfusions are more carefully screened.

Other Types of Transfusion Reactions

Febrile transfusion reactions are most often attributed to the presence of granulocytes in transfused blood, although they may also be caused by plasma factors such as exogenous immunoglobulins or rarely by bacterial pyrogens. Benign febrile transfusion reactions may be distinguished from the fever associated with intravascular hemolytic transfusion reactions by the lack of hypotension, chills, or other signs of hemolysis, and by the generally delayed onset of fever compared to hemolytic transfusion reactions. Febrile reactions are usually treated with acetaminophen or aspirin; use of these drugs and occasionally antihistamines for prophylaxis is also common. Patients who have repeated febrile reactions should receive leukocyte-filtered blood.

Urticarial transfusion reactions are thought to be caused by plasma components. One rare subset of patients with congenital IgA deficiency (about one in 650 to 900 persons) may have anaphylaxis due to plasma exposure, but in general urticarial reactions are not accompanied by a risk of anaphylaxis. Urticarial reactions are usually treated with antihistamines and may be treated prophylactically. IgA-deficient patients may be transfused with washed packed red blood cells, or failing that, with IgA-deficient red cells. Products such as platelets must be from IgA-deficient donors. Patients who have repetitive urticarial reactions in spite of prophylaxis may need to receive related-donor blood products.

Transfusion-related acute lung injury (TRALI) is an uncommon syndrome that is due to the presence of leukocyte antibodies in transfused plasma. TRALI is believed to occur in approximately one in every 5000 transfusions. Leukoagglutination and pooling of granulocytes in the recipient's lungs may occur, with release of the contents of leukocyte granules, and resulting injury to cellular membranes, endothelial surfaces, and potentially to lung parenchyma. In most cases leukoagglutination results in mild dyspnea and pulmonary infiltrates within about 6 hours of transfusion, and spontaneously resolves; occasionally more severe lung injury occurs as a result of this phenomenon and ARDS results.38 Leukocyte filters may prevent TRALI for those patients whose lung injury is due to leukoagglutination of the donor white blood cells, but because most TRALI is due to donor antibodies to leukocytes, filters are not helpful in TRALI prevention. Transfused plasma (from any component source) may also contain antibodies that cross-react with platelets in the recipient, producing usually mild forms of posttransfusion purpura or platelet aggregation after transfusion.

Another nonspecific form of immunologic transfusion complication is mild to moderate immunosuppression consequent to transfusion. This effect of transfusion is not completely understood, but appears to be more common with cellular transfusion and may result in both desirable and undesirable effects. Mild immunosuppression may benefit organ transplant recipients and patients with autoimmune diseases; however, neonates and other already immunosuppressed hosts may be more vulnerable to infection, and cancer patients may possibly have worse outcomes postoperatively.39

Graft-versus-host disease (GVHD) occurs when immunocompetent lymphocytes are transfused into an immunoincompetent recipient. The lymphocytes proliferate and respond to recipient HLA antigens. The syndrome presents as fever, skin rash, and liver function abnormalities occurring 2 to 6 weeks after transfusion. Graft-versus-host disease is highly lethal in reported cases outside of the transplantation situation; it is very morbid in transplant settings as well. GVHD primarily occurs in organ transplantation, leukemia, and other severely immunosuppressed patients, but it has been reported following cardiac surgery and in neonates. GVHD has been reported only once in a patient with the acquired immunodeficiency syndrome (AIDS). The correct diagnosis is made by recognizing the syndrome and obtaining a skin biopsy. Prevention in the non–bone marrow transplant setting is to irradiate every cellular blood product before use. Treatment of active graft-versus-host disease involves further immunosuppression with cyclosporine, antithymocyte globulin, steroids, and cytotoxic drugs. Critically ill adult patients who are receiving blood transfusions from family members should have those related donor products irradiated.40

In order to reduce the risks of immunosuppression and GVHD, cytomegalovirus (CMV) transmission, and perhaps of alloimmunization, leukocyte reduction through the routine use of white blood cell filters has been undertaken. Several randomized trials of leukoreduced transfusions in critically ill patients have been published, but none of the trials has been large enough to demonstrate benefits unequivocally, and most are negative. However, the largest prospective study performed to date was a Canadian sequential cohort study that has shown some potential benefits to routine leukoreduction.41 The study closely followed 6982 patients using routine practice, followed by the institution of universal leukoreduction in all blood banks and follow-up of 7804 patients' outcomes. The main outcome measure was all-cause in-hospital mortality, which was reduced during the leukoreduction period compared to the previous period (odds ratio [OR] 0.87; 95% confidence interval 0.75 to 0.99). The second outcome measure, nosocomial infections, was not changed by leukoreduction, although the frequency of fevers and the use of antibiotics were reduced. The cost effectiveness of this strategy has yet to be analyzed.

Risks of Disease Transmission

Hepatitis transmission from blood transfusion is rapidly becoming a rare event. Nucleic acid testing for hepatitis B is universal. The current estimate of the risk of hepatitis B is less than 1 in 200,000.42 Nucleic acid–based screening tests for hepatitis C provide excellent protection. It detects more potentially infectious units of blood than the enzyme-linked immunosorbent assay (ELISA) test, and is thought to have the potential to reduce the rate of posttransfusion hepatitis to less than the current 5% nationwide.43 Individuals who have an acute hepatitis-like illness after transfusion should be tested for hepatitis C; however, since the antibody screen is not 100% sensitive. The vast majority of individuals infected do develop antibodies within the first 6 months after transmission. Many of those people may lose their antibodies later on, hence the less-than-perfect screening test. Posttransfusion hepatitis may be caused by other as yet poorly characterized hepatitis viruses, but it is an uncommon event when donor screening by questionnaire and serology is performed.

AIDS transmission is also an unusual event. Both nucleic acid–based tests and HIV antibody tests are routinely used, and the most recent estimates of risk nationwide are less than one in two million.44 Human T-lymphocyte virus (HTLV) I/II is also tested using ELISA antibody detection, and the total number of cases related to transmission of this virus in the U.S. is fewer than two dozen annually. Transfusion-associated HIV is more rapidly lethal than most community-acquired cases. Acute HIV infection is characterized by fever, adenopathy, and pancytopenia. For all disease testing, repetitively tested volunteer donor blood has been shown to have a higher safety profile than family donors or designated donors.

Cytomegalovirus transmission remains a clinical problem. Forty to sixty percent of adults in the U.S. are antibody-positive for CMV; most, however, are not infectious. Clinical studies have estimated transmission to occur in 3 to 12 recipients/100 units of blood transfused. Rises in antibody titers or seroconversion have been shown in up to 30% of transfused surgical patients. More than 90% of seroconverters/titer increases are asymptomatic, but in vulnerable recipients, CMV produces acute illness. Neonates, premature, or low birth weight infants are at highest risk; 13% to 37% acquire CMV, and up to one in four dies of it.44 Bone marrow transplant recipients have a high rate of early death due to CMV pneumonitis, although most of this is thought to represent endogenous reactivation of virus replication. Other vulnerable patients include cancer patients, AIDS patients, seronegative pregnant women, and splenectomy patients. Symptomatic acute cytomegalovirus infection results in a clinical syndrome that resembles infectious mononucleosis. Treatment of cytomegalovirus disease is marginally effective; drugs used in infected patients are mostly suppressive unless remission of the immunosuppression occurs. White blood cell filters, chemoprophylaxis for seropositive patients, and use of seronegative blood products for others are important and useful strategies in the bone marrow transplant setting, and transmission is prevented in neonates by the use of seronegative donor blood products.45

Blood may be contaminated by bacteria at the time of collection. This is a rare event, at a rate of 0.21 per million units of red cells transfused, but it is fatal in about one third of recipients, especially if the contaminating organism is a gram-negative rod.46 Contamination is more than twice as likely from platelets or plasma. Other blood-borne diseases that may be transmitted by transfusion include parvovirus B19, Epstein-Barr virus, human herpesvirus-8 (HHV-8), malaria, brucellosis, trypanosomiasis, syphilis, toxoplasmosis, West Nile virus, the severe acute respiratory syndrome, and possibly variant Creutzfeldt-Jakob disease (or bovine spongiform encephalopathy).

The Massively Transfused Patient

Patients suffering rapid blood loss require massive transfusion, generally defined as replacement of the entire blood volume within a 24-hour period, or equaling 50% of the blood volume in 3 hours. Typical conditions leading to massive transfusion include polytrauma, peripartum hemorrhage, gastrointestinal blood loss, aortic aneurysmal rupture, and perioperative bleeding. These patients are considered separately because the large volume of blood required leads to complications not generally associated with blood transfusion, most notably a severe coagulopathy. Moreover, management may benefit from a style that relies on prevention and empiricism as well as on laboratory testing.

Coagulopathy accompanying massive transfusion is multifactorial, including contributions from direct loss, consumption, and dilution of clotting factors and platelets; hypothermia as cool fluids are infused; and shock-related hepatic dysfunction leading to impaired clearance of activated coagulation factors and their breakdown products. It is generally believed that early detection and prevention of coagulopathy is more effective than later treatment by interrupting a vicious cycle that, if not avoided, culminates in a hypothermic, acidemic patient who responds very poorly to aggressive resuscitation. Early studies implicated thrombocytopenia as being the predominant early consequence of massive transfusion, but more recent studies show that a coagulation factor deficit is the earlier derangement. Hypocalcemia can contribute to coagulopathy during truly massive transfusion (greater than 100 mL/min), but this is not often seen because citrate can be rapidly metabolized by the liver.

Initial priorities in the massively bleeding patient include achieving hemostasis when this is possible; obtaining large-bore intravenous access (number and caliber are more important than location); assessing the need for intubation and mechanical ventilation; and infusing sufficient quantities of fluid (generally normal saline). Fluid warmers should be used early in anticipation of heat losses, not only when core temperature falls to abnormal levels. Packed red blood cells should be infused rapidly, dictated by the clinical assessments of blood loss, intravascular volume, and measured hemoglobin concentration. Laboratory tests of hemoglobin, coagulation parameters, platelets, and fibrinogen are indicated, but given the delay in obtaining results and preparing and transporting blood products, cannot provide the sole basis for plasma and platelet transfusion in all patients. Instead some centers have empiric guidelines recommending transfusion of fresh frozen plasma and platelets in some proportion to the amount of blood transfused once the patient is deemed to be “massively transfused.” Such a formulaic approach has not been shown to improve outcomes and is strongly discouraged by many, but others attribute their unusually good outcomes in massively bleeding patients to such protocols. The roles for cryoprecipitate, prothrombin complex concentrates, and recombinant activated factor VII remain to be determined.

Some patients transfused massively will suffer uncommon complications such as abdominal compartment syndrome (see Chap. 42); electrolyte derangements such as hypocalcemia, hyperkalemia, or hypomagnesemia (see Chap. 76); and transfusion-related acute lung injury (see Chap. 38), although this is difficult to diagnose in a patient also at risk of fluid overload and aspiration.

Alternatives to Use of Blood Bank Products

Presurgical storage, or autologous blood donation, makes up approximately 7.5% of all transfusions, although recent studies suggest that autologous donation is underused. Generally, practice has been to have patients donate a unit every 2 to 3 weeks prior to surgery, while taking oral iron and perhaps erythropoietin. There is some evidence that more units may be predeposited with the concomitant use of erythropoietin, although the ultimate clinical impact of such use is unknown. Patients do present for surgery slightly anemic. The FDA permits the use of autologous units that are hepatitis- or HIV-infected; individual institutions may not wish to assume the risk of handling such units and may reject such autologous donors. If used, this infected donated blood should be labeled as a biohazard and handled with extreme caution.

Designated donors, usually members of a patient's family, are often requested by physicians and patients as sources of blood products, usually with the idea that designated donors are less likely to have infectious diseases. Unfortunately, the published data demonstrate that in comparison to regular volunteer donors who have been screened, designated donor blood is more likely to test positive for infectious diseases. Designated donor blood is not necessarily safer than volunteer donor banked blood.

With the availability of recombinant human erythropoietin and its success in increasing hemoglobin concentration in patients with renal failure and bone marrow failure due to HIV infection or chemotherapy, the use of erythropoietin to reduce blood transfusions in critically ill patients seemed obvious. In the past decade, six randomized studies have evaluated exogenous administration of recombinant human erythropoietin (rHuEPO) to decrease red cell transfusions in the ICU.47 They have varied with respect to dosage regimens, the dose of concurrently administered iron, patient characteristics (trauma, medical, and surgical), and transfusion thresholds. Although the use of rHuEPO does increase hemoglobin levels compared to concurrent or historical controls, the rate at which hemoglobin rises is slow, measured over a period of weeks, not days. The largest study conducted to date using the best methodology administered weekly rHuEPO and found a 19% decrease in transfusion requirements.48 Reduced ICU length of stay was shown in only one study of surgical/trauma patients. Reduced hospital stay after ICU discharge was found in another study of severely ill patients (APACHE II score >22), a potentially important non-ICU outcome. A variety of other outcomes were measured in these studies, but none was significantly altered by rHuEPO use. No adverse events were associated with rHuEPO use, although these studies were not designed to evaluate safety, and the target hemoglobin was higher than that used in the Canadian transfusion trial. There are to date no published cost-effectiveness analyses.

Artificial blood substitutes are still in development. Perfluorocarbon products are chemically inert polyfluorinated hydrocarbons that are insoluble in plasma, but in which oxygen is soluble. They are produced as emulsions with surfactants and added hydroxyethyl starch and they need to be stored frozen. The oxygen delivery capability of these products is good, but the use of perfluorocarbons is limited by less efficient oxygen uptake, which requires 100% inspired oxygen delivery. In addition, the total dose and carrying capacity are limited, so that at the maximum dose, the contribution to the total hematocrit by perfluorocarbon, or “fluorocrit,” is very low. Finally, these drugs are short-acting, with a half-life of only about 24 hours.49,50

Hemoglobin conjugated to polysaccharide is available in a variety of forms, although all have drawbacks with respect to oxygen-carrying capacity, reactions, vasoconstriction, and shorter half-lives. None is currently marketed for use in critical care.

Hemodilution and intraoperative autologous transfusion are other surgical techniques used to avoid transfusion. Hemodilution lowers blood viscosity, thought to be an advantage by some. While there are no formal randomized trials of hemodilution, there are ample nonrandomized studies in cardiac bypass surgery, orthopedic surgery, and Jehovah's Witness patients to suggest that hemodilution techniques are safe and can result in saving blood transfusions.51 Intraoperative scavenging and reuse of red cells using cell saver technology has become widespread practice in major surgeries such as cardiac and vascular procedures. It remains controversial in trauma or “dirty” operative fields, however. A large number of published nonrandomized series attest to the relative safety of most “washing” cell savers now in use, although the research methodology does not allow for unequivocal demonstration of benefit for their use.52 One systematic review of the available trials in abdominal aortic aneurysm repair does not clearly demonstrate an advantage with respect to the use of blood products,53 while an overall review of cell salvage use supports safety and the potential for decreasing red cell use. The use of heparin, antibiotics, filters, and other adjuncts to cell salvage is not standardized. Additional savings in blood loss have been made with the use of continuous intra-arterial monitoring devices, laboratory microtechniques, and concerted nursing efforts in intensive care units to reduce the number and volume of phlebotomies.