Reducing Noninfectious Risks of Blood Transfusion

Blood transfusion is an accepted standard of care in a variety of clinical scenarios and is likely to remain so, despite the absence of randomized controlled trials (RCTs) demonstrating improved outcomes after transfusion. Instead of designing studies to answer the question “should we ever transfuse?” investigators have attempted to answer the question “when should we transfuse?” The question is of principal importance, because several studies have suggested that the use of human blood products may place patients at increased risk of death.6,7Thus, any discussion of strategies for reducing transfusion-related morbidity would be incomplete without emphasizing the importance of evidence-based practice because the safest transfusion is no transfusion.

The primary indication for transfusion of erythrocytes is hemodynamic instability caused by hemorrhagic shock. However, less than 20% of erythrocyte units are transfused for this purpose.8Most are transfused for the routine treatment of anemia in hemodynamically stable critically ill patients.9The Transfusion Requirements in Critical Care trial demonstrated that a conservative transfusion threshold may be equivalent to a liberal threshold in the most critically ill patients and may be beneficial in those less critically ill.9Use of a more liberal threshold may be justified in patients with active ischemic cardiovascular disease10or sepsis, when transfusion may be titrated to the mixed venous oxygen saturation rather than to hematocrit.11 

The American Association of Blood Blanks recently convened a panel of experts to comment on several controversial practices involving plasma transfusion.12The panel recommended the inclusion of plasma during massive transfusion (defined as greater than 10 units per day). A plasma-to-erythrocyte ratio greater than 1:3 is associated with reduced mortality in trauma patients; however, the optimal ratio remains to be determined.13,–,15During routine surgery, in the absence of massive transfusion, transfusion of plasma typically is not indicated. Plasma is used commonly in reversal of warfarin anticoagulation; however, the evidence supporting this practice is limited. It is recommended that plasma be administered during active intracranial hemorrhage, but whether reversal is beneficial during other life-threatening forms of bleeding, such as gastrointestinal bleeding, remains unknown. Finally, transfusion in the absence of coagulopathy, severe anemia, or active bleeding may increase mortality and is rarely indicated.13 

Platelet transfusion typically is indicated for bleeding prophylaxis or therapy. Prophylactic transfusion in thrombocytopenic patients or those with dysfunctional platelets is common, and appropriate thresholds are being established. Thresholds for prophylaxis before surgical procedures are established largely by empiricism.15Thresholds are set to match the risk and consequence of bleeding: high for neurosurgery or ocular surgery, lower for insertion of a central line.15In addition to infectious and noninfectious complications, platelet transfusion may result in refractoriness to subsequent platelet transfusion.16 

Finally, procoagulant products, such as prothrombin complex concentrates, cryoprecipitate, recombinant factor VII, aminocaproic acid, or tranexamic acid, and others, may be indicated in specific clinical situations, although a discussion of these products is beyond the scope of this review. Ultimately, minimizing the use of blood products may be the best way to reduce transfusion-associated morbidity. This end may be achieved in part by minimizing unnecessary phlebotomy and using smaller collection tubes,8limited appropriate use of pharmacologic agents such as erythropoietin (as in renal failure), or substitution of synthetic blood products or hemoglobin-based oxygen carriers.8Of note, there are no hemoglobin-based oxygen carriers available in the United States, given the concern that they increase mortality and myocardial infarction.17 

A select group of noninfectious complications of transfusion are reviewed here, with the most commonly occurring complications discussed first.

Febrile transfusion reactions typically are defined as a 1°C increase in temperature during or within 3 h of transfusion18that cannot be explained by sepsis or a hemolytic reaction. The reported incidence varies widely,19,20but convincing evidence suggests that the number of febrile reactions is significantly reduced by leukoreduction of erythrocyte units.21The average rate is approximately 1 in 330 for erythrocyte transfusions and 1 in 20 for platelet transfusions.18,19Febrile transfusion reaction may be accompanied by chills, rigors, and discomfort. Approximately 50% of transfusions in the United States are administered with acetaminophen and diphenhydramine premedication,18yet little evidence exists to justify this practice, and the few prospective randomized studies available have generated conflicting results.22,–,24A recent Cochrane review concluded (based on low-quality data) that premedication does not reduce the risk of febrile or allergic nonhemolytic transfusion reaction.25Treatment of febrile reactions entails discontinuation of the transfusion and supportive care and may include antipyretic therapy.

Transfusion of blood products may result in circulatory overload presenting as hydrostatic pulmonary edema that can be indistinguishable from the increased lung vascular permeability that is present in TRALI. Patients present with dyspnea, tachypnea, jugular venous distension, and increased systolic blood pressure. The incidence of TACO typically is cited at 1–10% but varies by patient population, and recognition may be heavily provider-dependent. In addition, there is no consensus definition of TACO, which has hampered clinical investigation. Many cases of transfusion-associated pulmonary edema may represent a combination of noncardiogenic pulmonary edema, as in TRALI, and pulmonary edema, as in TACO. Distinguishing between them can be challenging, but algorithms have been published to facilitate diagnosis26(fig. 1). Echocardiography, B-type natriuretic peptide concentration, right-side heart catheterization, and alveolar fluid protein analysis may all be used diagnostically. TACO frequently is a post hoc  diagnosis made evident by the rapid improvement of pulmonary edema with simple measures such as diuresis. The use of slow transfusion rates, diuretics, and identification of at-risk patients, such as those with critical illness, cardiac disease, renal disease, or infants, may reduce the incidence of TACO.27 

Transfusion-related acute lung injury is defined as noncardiogenic pulmonary edema occurring within 6 h of transfusion28,29(table 1). Reports of the incidence of clinically recognized TRALI vary but typically are accepted as roughly 1 in 5,000 transfusions28; however, recent studies have highlighted the presence of previously unappreciated subclinical effects of transfusion, which may be quite common.30,–,33The pathophysiology of TRALI is incompletely understood but may be explained by a “two-hit” hypothesis, in which a “primed” patient (first hit) is transfused with antihuman leukocyte antigen antibodies, antineutrophil antibodies, or other biologic response modifiers (second hit), which precipitates acute lung injury33(fig. 2). Recent data suggest that neutrophils and platelets play significant roles in producing lung injury.34The priming event could be any condition that leads to subthreshold immune activation, including surgery, infection, and possibly trauma. TRALI has emerged as a leading cause of transfusion-related morbidity and mortality, and in 2009, 30% (13 of 44) of transfusion-related mortalities in the United States were attributed to TRALI or suspected TRALI.^bTreatment of TRALI is largely supportive, and efforts have centered on prevention. Plasma mitigation (collection of plasma from males only or never-pregnant females) and limiting unnecessary transfusion may reduce the incidence of TRALI.35 

Urticarial reactions and generalized pruritus are common, occurring during approximately 1–3% of all transfusions, and are thought to result from the presence of soluble antigens in the donor plasma that produce a dose-dependent clinical response. Allergic reactions usually are associated with mild symptoms, such as localized erythema, pruritus, or hives, and typically respond to parenteral antihistamines.

Severe allergic reactions, characterized by bronchospasm, stridor, hypotension, and gastrointestinal symptoms, are referred to as anaphylactic or anaphylactoid transfusion reactions. These reactions occur in 1 in 50,000 transfusions and can be life threatening.19An anaphylactic reaction refers specifically to classically described, immunoglobulin E-mediated reaction to foreign protein, whereas the term anaphylactoid  is used to describe other reactions that produce the same clinical syndrome. Immunoglobulin E-mediated reaction against protein-hapten conjugates and complement-mediated generation of endogenous anaphylotoxins are two proposed mechanisms for anaphylactoid reactions. The latter mechanism is thought to explain anaphylactoid reactions in individuals with IgA deficiency. High-titer antiimmunoglobulin A antibodies in these individuals provoke complement activation and anaphylaxis. Thus, IgA deficiency should be considered when an anaphylactoid reaction occurs. Treatment may require administration of epinephrine in severe cases.

Hemolytic transfusion reactions typically are classified as acute or delayed. Acute hemolytic reactions are defined as those occurring within 24 h of blood transfusion. They are thought to result from the presence of preexisting recipient alloantibodies against donor erythrocytes. Hemolytic transfusion reactions (associated with ABO or non-ABO alloantibodies) are relatively uncommon. However, these reactions were the second leading cause of transfusion-associated death in the United States from 2005 to 2009, accounting for 37% (68 of 267 deaths) because of the very high mortality associated with transfusion of ABO-incompatible blood.‖Most events result from transfusion of incorrectly typed units because of a clerical error. Acute reactions may present as sudden onset of fever or chills, facial flushing, pain, hypotension, dyspnea, renal failure, or disseminated intravascular coagulation. Prevention is based on systems-based efforts to improve blood bank safety, which are the focus of a large industry and beyond the scope of this review. If acute hemolytic reaction is suspected, the transfusion should be stopped, large-bore intravenous access established, and the patient monitored in the intensive care unit (ICU).

Delayed hemolytic reactions typically occur between 24 h and 1 week after transfusion and are thought to occur because of antierythrocyte antibodies acquired from previous transfusions. Delayed hemolytic reactions occur commonly (1 in 1,900 transfusions) and typically are less severe than acute hemolytic reactions.36Such reactions may present as fever or reduced urine output, but most commonly they are associated with no symptoms and are discovered as an unexplained decrease in hemoglobin concentration. Supportive care is appropriate in most cases, including transfusion of appropriately typed erythrocytes. Intravenous immunoglobulin and steroid therapy have been used to treat severe reactions.

Transfusion-related immunomodulation has been the subject of intensive investigation yet remains a subject of controversy in the transfusion medicine community. The idea that an allogeneic blood transfusion could produce immunosuppressive effects first gained wide recognition when Opelz et al.  37noted improved outcomes among recipients of cadaveric renal transplants who had received blood transfusions. This effect has been attributed to the immunomodulatory effect of transfused donor leukocytes, and alterations in circulating lymphocytes, T-cell helper/suppressor ratio, B-cell function, and number of circulating antigen-presenting cells in recipients of allogeneic blood.38Although it has been suggested that transfusion-mediated effects on the survival of renal allografts have disappeared in the era of potent immunosuppressive drugs, prospective studies in the modern era have demonstrated a continued survival advantage for grafts transplanted into transfused patients.39 

Subsequently, the effects of transfusion on bone marrow transplantation, recurrence of malignancy, and susceptibility to infection were proposed. Although the data regarding transplant outcomes are consistent, thus validating TRIM as a real phenomenon, the data describing the effect of transfusion on recurrence of malignancy and infection are mixed and remain controversial.20,40Prestorage leukoreduction of erythrocytes has been proposed as a method of reducing cancer recurrence and postoperative infection. The increased risk of nosocomial infections is discussed in more detail in the section on universal leukoreduction.

Increased risk of cancer progression after transfusion was first proposed in the early 1980s by Gantt.41Since then, many retrospective trials have demonstrated an association between transfusion and cancer progression, which is thought to be attributable to suppression of the host immune system.42Particularly convincing is the association between transfusion and lymphoma. A recent meta-analysis by Castillo et al.  43included 12 observational studies and demonstrated a significantly increased risk of lymphoma, particularly chronic lymphocytic leukemia after erythrocyte transfusions. The common critique of such studies is that transfusion may simply be a marker for worse disease. Three RCTs have been performed to examine the effect of TRIM on cancer recurrence in patients with colorectal cancer; however, none of them had detectable differences in cancer recurrence with reduced exposure to allogeneic leukocytes.44,– 46 

Transfusion-related microchimerism refers to the consistent presence of a population of donor cells in the recipient. The incidence may be as high as 10% in patients who receive massive transfusion after trauma and can last for many years. Foreign cells may represent as much as 5% of circulating leukocytes.39,47The theoretical risks of microchimerism include graft-versus -host disease or autoimmune and inflammatory disorders, but the true clinical implications of this condition are not known.

Posttransfusion purpura is a rare complication characterized by purpura, epistaxis, gastrointestinal bleeding, and thrombocytopenia, typically observed 5–10 days after transfusion.19The reaction is thought to result from antiplatelet antibodies (antihuman platelet alloantigen 1a is the most common) that react with transfused or autologous platelets. Intravenous immunoglobulin is the recommended therapy.19Avoidance of blood product units that are positive for the antigen in patients with a history of posttransfusion purpura is recommended.19 

Hypotensive transfusion reactions may occur during transfusion protocols that activate the intrinsic “contact activation” pathway of the coagulation cascade and increase production of bradykinin, as in bedside leukoreduction through filters with negatively charged filtration surfaces,48infusion of plasma protein fraction and albumin, and therapeutic apheresis. Patients taking angiotensin-converting enzyme inhibitors are at increased risk because of the normal physiologic role of angiotensin-converting enzyme in bradykinin catabolism.49,50 

Transfusion-related graft-versus -host disease is an extremely rare complication in which viable donor leukocytes attack recipient cells. It is typically observed in severely immunocompromised hosts, although it has been reported in normal recipients when the donor is homozygous for one of the recipient's human leukocyte antigen types. In both cases, the donor leukocytes are not recognized as foreign and are not eliminated by the recipient immune system. Transfusion-related graft-versus -host disease is characterized by fever, liver dysfunction, rash, diarrhea, and pancytopenia and is fatal in 84% of cases but can be effectively prevented by irradiation of units for at-risk patients and through the use of leukocyte reduction.51 

Several recent trials have generated data that suggest transfusion may be independently associated with increased risk of renal injury. Habib et al.  52conducted a retrospective review of patients undergoing coronary revascularization procedures and discovered that those with a nadir hematocrit less than 24% were at increased risk of kidney injury. However, transfusing patients with similarly low hematocrit did not decrease the risk of kidney injury. Instead, those transfused have higher postoperative creatinine, a greater percentage increase in creatinine, and longer length of hospital stay. These findings have been replicated in other postoperative coronary artery bypass patients53and in patients after lower extremity revascularization.54It has been suggested that transfusion may worsen, rather than improve, tissue oxygen delivery, and that effect may explain these data.55However, the retrospective nature of the studies raises concern that increased risk of acute kidney injury was worse in the transfused than in nontransfused patients as a result of selection bias.

Iron overload, metabolic toxicities, such as citrate toxicity, hypocalcemia, and hyperkalemia, and complications of massive transfusion,13, 14such as hypothermia and coagulopathy, are sources of significant morbidity but are not discussed in this review. Incidence, etiology, therapeutic measures, and preventative techniques for the reviewed complications are reviewed in table 2.

Several proposed interventions for reducing the noninfectious complications of transfusion have generated great interest and provoked extensive research. We have reviewed three strategies and the relevant evidence but stress that the most effective means for avoiding transfusion-related morbidity is to use blood products in an evidenced-based way.

Universal leukoreduction refers to the process of removing leukocytes from a unit of packed erythrocytes or platelets to a standardized degree of purity.20Traditionally, this has been done through removal of the buffy coat (the fraction of blood that contains leukocytes and platelets) after centrifugation or by pre- or poststorage filtration.56,57The consensus is that leukoreduction helps to prevent three complications of blood transfusion: febrile nonhemolytic transfusion reactions, platelet refractoriness caused by human leukocyte antigen alloimmunization, and transmission of cytomegalovirus.56,57Patients at risk for these complications traditionally have been provided with leukoreduced blood, and this precaution is clinically effective and cost-effective. Other proposed benefits, such as reduction of TRIM, effects on cancer progression, and rates of infection, remain matters of controversy.

In the late 1990s, accumulating evidence of leukocyte-mediated TRIM and the suggestion that leukoreduction might reduce the transmission of Creutzfeldt-Jakob disease provoked an international debate over the appropriateness of universal leukoreduction. Proponents favored universal leukoreduction as a necessary safety measure that would reduce recognized and unrecognized complications related to transfused leukocytes and argued that leukoreduction would save money over time.58Opponents of universal leukoreduction argued that, although there were no apparent clinical risks to universal leukoreduction, a rigorous interpretation of the evidence did not demonstrate a benefit beyond those traditionally described, and that it would not be cost-effective. Most European nations adopted universal leukoreduction in the late 1990s, including Great Britain, Austria, Germany, Portugal, Switzerland, and Ireland.

The American Red Cross, which supplies approximately 50% of all erythrocyte units in the United States, adopted universal leukoreduction in 2000. However, the FDA regulates the blood supply through codes communicated in the Code of Federal Regulations and did not mandate universal leukoreduction. During a meeting on January 26, 2001, the FDA Advisory Committee on Blood Safety voted to recommend that universal leukoreduction be implemented as soon as feasible and also addressed concerns that the adequacy of the blood supply be maintained and that sufficient funding be provided for this transition.#Thus, individual blood banks were confronted with the dilemma of following the FDA recommendation at increased cost or the FDA requirement at increased liability. Currently, approximately 70% of the U.S. blood supply is leukoreduced,§and that percentage is increasing gradually with time.

As the United States continues a transition toward universal leukoreduction, new evidence supporting and discouraging the practice of universal leukoreduction has emerged. Retrospective “before and after” studies performed during transition to universal leukoreduction have provided valuable data. In 2003, Hébert et al.  21published a before-and-after report describing the health of Canadian patients in the 12 months before and after the transition to universal leukoreduction. These investigators noted a 1% reduction in mortality, reduced posttransfusion fever, and reduced antibiotic use after the transition to universal leukoreduction (fig. 3). However, the increased incidence of severe lung disease in the preleukoreduction cohort and increased use of aspirin, β-blockers, and angiotensin-converting enzyme inhibitors in the postleukoreduction group cast doubt on causality of the reported reduction in mortality. In addition, it cannot be ruled out that the decreased incidence of fever was directly responsible for decreased antibiotic use.

Other retrospective studies have been published reinforcing or refuting the relationship between universal leukoreduction and infection. In 2005, Blumberg et al.  demonstrated a 35% reduction of indwelling-catheter infections that could not be explained by any change in hospital policy other than implementation of universal leukoreduction.59The same authors recently published a report demonstrating reductions in TRALI and TACO after the transition to universal leukoreduction.60In contrast, Englehart et al.  61performed a retrospective study comparing outcomes in 495 trauma patients receiving nonleukoreduced, leukoreduced, and mixed transfusions. They found no differences in number of ICU days, hospital days, ventilator days, incidence of acute respiratory distress syndrome, multiple organ dysfunction scores, mortality, or infection rates. In 2004, Llewelyn et al.  62conducted a before-and-after study documenting the effect of universal leukoreduction in approximately 2,100 cardiac and orthopedic surgery patients in 11 hospitals in the United Kingdom and reported no difference in mortality or infection rates.

Since 1998, many new RCTs investigating the association between leukoreduction and postoperative infection, length of hospitalization or mechanical ventilation, and mortality have been published.63Only one RCT published since 1998 has shown a beneficial effect of universal leukoreduction.64Bilgin et al.  observed a reduced infection rate after cardiac valve surgery in patients randomized to prestorage leukoreduced blood compared with buffy-coat–depleted blood. Other studies demonstrate no effect. Several meta-analyses have been published.65,–,67However, discordant results from these meta-analyses have led to an active debate regarding the cause for the discordance. Heterogeneity of studies, use of intention-to-treat as opposed to as-treated analysis, and exclusion of recent studies have all been suggested as causes of the discordant results.63,68,69 

Although the volume of evidence has increased, clarity on the role of universal leukoreduction has not. The mid-1990s may have been a time when this question was susceptible to publication bias, but there has been ample demand for consenting and dissenting opinion regarding universal leukoreduction since 1998, yet consensus has not emerged. If universal leukoreduction does diminish transfusion-related immunomodulatory effects, the clinical effects may be small and difficult to capture in clinical studies, despite a collection of randomized clinical trials enrolling more than 6,000 patients.65Unfortunately, this controversy may never be resolved because universal leukoreduction has become the standard of care in most European nations and the United States.

Generation of a safe blood supply involves screening of potential donors for characteristics that may increase infectious or noninfectious risks for recipients. Certain characteristics prompt temporary or permanent donor deferral, such as extremes of age, low hemoglobin, history of high-risk behaviors, or use of certain prescription drugs.

Donor restriction by gender has emerged as a strategy for reducing the incidence of TRALI due to the accumulation of epidemiologic evidence that women are higher-risk donors because of the alloimmunization that occurs with pregnancy. An early case series demonstrated that the majority of TRALI cases were associated with multiparous female donors,70and several case series demonstrated that multiple reactions could be traced to a single donor, commonly multiparous females.71,72Furthermore, populations with increased alloantigen exposures, such as multiparous women and previous transfusion recipients, have been implicated as high-risk donors. The Serious Hazards of Transfusion program in the United Kingdom reported that all donors between 1996 and 2002 found to have antileukocyte antibodies recognizing recipient antigens have been female.73The observation that alloimmunized individuals were more frequently implicated in TRALI reactions led to the theory that a significant percentage of TRALI cases resulted from transfusion of donor-derived alloantibodies. In fact, fresh frozen plasma and platelets, so-called high-plasma components that contain donor-derived antibodies, are associated with a 6-fold increased risk of TRALI.73 

In a retrospective case-control study by Gajic et al. ,31ICU patients transfused with three or more high-plasma components from females were compared with matched cases transfused with three or more units from males. Patients receiving female-only plasma showed diminished oxygenation, fewer ventilator-free days, and a trend toward increased in-hospital mortality. In a retrospective before-and-after study by Wright et al. ,32patients undergoing repair of ruptured abdominal aortic aneurysms in the United Kingdom were reported to have decreased rates of acute lung injury and hypoxia after surgery when male-only plasma was used.

Clinically evident TRALI occurs in approximately 1 in 5,000 transfusions,28so it has been difficult to conduct RCTs on the effect of female plasma. However, data have emerged to suggest that the incidence of more subtle effects of transfusion on lung function may be significant. In 2001, a prospective, double-blind, randomized crossover study demonstrated diminished FIO²/PaO²in the absence of increased blood pressure after transfusion of plasma from multiparous females in 100 ICU patients.30In contrast, Welsby et al. , from the Duke-CARE group, recently published a retrospective case-control study of 390 matched pairs who received male- or female-only plasma during aortocoronary bypass surgery.74They reported significantly fewer adverse events in those receiving female-only plasma. Their primary outcomes included hybrid measures of pulmonary dysfunction (including pneumonia, acute respiratory distress syndrome, and pulmonary edema) and prolonged hospital stay or death within 30 days. Additional RCTs are likely to be published on the subtle impact of transfusion on lung function, and we are likely to learn more about the pathogenesis of severe TRALI reactions from work describing these physiologic effects.

Based on observations in the Serious Hazards of Transfusion program, the National Blood Service in the United Kingdom instituted a policy in 2003 that all high plasma components be derived from male donors, and by 2005 the United Kingdom achieved a rate of male plasma approaching 90%. The impact of this transition has been well-documented. Although the overall rate of adverse event reporting to the Serious Hazards of Transfusion program increased during this time, the number of cases of highly likely or probable TRALI decreased**(fig. 4a). Although the number of cases associated with plasma or platelets has decreased, the number of cases associated with erythrocyte units has remained relatively constant (fig. 4B).

In the United States, the American Association of Blood Banks workgroup on TRALI recommended on November 3, 2006, that the United States also transition to male-only high-plasma components.††In 2006, the American Red Cross began a pilot program implementing distribution of male-only plasma for transfusion in 13 of their 35 regional blood centers. In April 2010, Eder et al.  35published a report of their observations as the percentage of male plasma increased during 2006–2008. Among voluntarily reported cases of transfusion reaction in which plasma was the only component transfused, the number of TRALI cases decreased each year: 32 in 2006, 17 in 2007, and 7 in 2008. The calculated rate of probable TRALI reactions per 10⁶plasma distributions decreased approximately 5-fold, to a rate equivalent to that observed with transfusion of erythrocyte units. Restriction to male-only plasma has resulted in increased use of plasma frozen within 24 h of phlebotomy (FP24) over that frozen within 8 h (fresh frozen plasma).35However, the transition has not imposed increased cost or supply burden. The 2007, National Blood Collection and Utilization Survey did not report the percentage of plasma or platelets that were derived from male donors overall in the United States. At the time of manuscript submission, final results for the 2009 version are not publicly available, and it is not known by the authors if the report will contain this information.

Although the data in support of using male-only plasma is mostly observational in nature, it is nonetheless convincing, and it is likely that the use of male-only plasma will expand. Additional RCTs describing the subtle effects of female plasma on lung function may further inform practice, and the data from the Duke-CARE group represents an important reminder that changing transfusion practice may have broad implications and should not be made only to optimize TRALI outcomes, but rather to optimize overall outcome for the entire population of transfusion recipients. Implementing policies that have subtle effects on common complications may influence outcomes more than those that have dramatic effects on rare, catastrophic complications.

The clinical impact of erythrocyte storage practices is a subject of great concern for the blood banking community. Historically, successful processing and storage of erythrocyte units has been judged by a nonclinical standard; erythrocytes were transfused into a healthy subject and the percentage remaining in circulation 24 h later was measured. Recovery of 75% of the cells was considered adequate. Methodologic improvements have extended the maximum storage age for erythrocyte units to 42 days, but changes occur in erythrocytes as they age, including potassium leak; loss of 2, 3-diphosphoglycerate; loss of membrane; release of toxic lipids; and rapid decline in S-nitrosohemoglobin, resulting in loss of hypoxic vasodilation75,76(fig. 5).

Recently, focus has shifted to the clinical impact of erythrocyte storage. In March 2008, the New England Journal of Medicine  published a retrospective study by Koch et al.  77from the Cleveland Clinic, who reported outcomes after transfusion of old versus  new erythrocyte units in 6,002 cardiac surgery patients. In the report, outcomes for patients who were transfused exclusively with units stored less than 14 days were compared with those transfused exclusively with units stored more than 14 days. There was reduced in-hospital mortality, intubation time, renal failure, and sepsis in recipients of the erythrocytes stored for less time. Most impressive, the group reported a 3.6% absolute risk reduction of 1-yr mortality among those who received blood stored less than 14 days.

The association of erythrocyte storage age with clinical outcomes had been demonstrated previously in several smaller retrospective studies. In 2003, Leal-Noval et al.  78published a retrospective report of outcomes among 897 cardiac surgery patients and reported that for each 1-day increase in age of the oldest erythrocyte unit transfused, the risk of pneumonia increased by 6%. In 1999 and 2000, Vamvakas and Carven published two companion studies. The first was a retrospective study in 416 patients79undergoing coronary artery bypass grafting, in which the authors reported that each 1-day increase in mean storage age of blood transfused was associated with a 1% increased risk of postoperative pneumonia. However, the second80was a retrospective study of postoperative cardiac surgery patients and reported no association between erythrocyte age and length of stay, ICU days, or days of mechanical ventilation. In 1997, Purdy et al.  81conducted a retrospective analysis of 31 patients admitted to the ICU with severe sepsis and who received transfusions. The individuals who survived or died were similar in age, sex, duration of ICU stay, duration of sepsis, incidence of shock, Acute Physiology and Chronic Health Evaluation (APACHE) score, and number of erythrocyte units transfused. However, the average age of erythrocyte units transfused into the survivors was 17 days, compared with 25 days for the units transfused to those who died. Zallen et al.  82retrospectively compared 23 patients who experienced multiple organ failure after trauma with 40 patients who did not, and demonstrated that transfusion of older erythrocyte units was independently associated with organ failure.

The Koch report provoked great interest and a series of editorials regarding the implications of this work for blood banking practices and transfusion policy. Many critiques were offered. The baseline characteristics of the two groups were not equivalent in ABO type, leukocyte reduction of blood, mitral regurgitation, New York Heart Association class, body surface area, left ventricular function, and peripheral vascular disease. The question of generalizability was also raised because the subjects were an average of 70 yr old and undergoing cardiac bypass. Furthermore, the study raised ethical questions. Given that most studies on the clinical impact of erythrocyte storage age have been conducted in cardiac surgery patients, should such patients be given newer erythrocyte units preferentially over patients undergoing other surgical procedures? In addition, if there is a continuum of benefit to receiving newer blood, who receives the freshest units?

Casting further doubt on the clarity of the conclusions from the Koch article are studies that contradict its findings. In 2008, Yap et al.  83published a retrospective report of the effects of erythrocyte age in a cardiac surgery population. The authors enrolled 670 patients but found no association of erythrocyte age and hours of ventilation, duration of ICU stay, renal failure, or mortality. Van de Watering et al.  84conducted a study of 2,732 repeat cardiac surgery patients and reported that erythrocyte age was not an independent risk factor for survival or days in the ICU. In a retrospective study of 740 Danish patients undergoing resection of colon cancer with curative intent, Mynster et al.  85reported that receiving transfusion of erythrocytes stored less than 21 days was an independent risk factor for cancer recurrence.

The evidence suggesting that increased erythrocyte age may lead to worse outcomes is clearly mixed, and the studies in question have been performed using a variety of outcomes and patient populations. Vamvakas and Carven79and Leal-Noval et al.  78provide strong evidence that increased erythrocyte age increases risk for pneumonia, but this finding was not statistically significant in the Koch report. Furthermore, large studies by van de Watering et al.  84and Yap et al.  83provide convincing contradictory evidence. On July 11, 2008, the Advisory Committee on Blood Safety and Availability released a statement indicating that a change in policy regarding erythrocyte storage age would be premature, and called for increased support to address this question through clinical research.‡‡Prospective randomized studies to address the question of risk associated with older units of erythrocytes are under way. The National Heart, Lung and Blood Institute-funded Red Cell Storage Age Study (RECESS, NCT00991341) began in November 2009. The SCANDAT database study,86which will use large databases in Sweden and Denmark to probe the effects of erythrocyte storage on the short- and long-term outcomes after erythrocyte transfusion, may provide valuable information.§§The ABLE (Age of Blood Evaluation) trial is a double-blind, multicenter randomized trial examining the effects of erythrocyte storage age on 90-day, all-cause mortality in patients requiring positive pressure respiratory support in the ICU. The anticipated end date of the trial is January 2013.

Should these studies confirm that older erythrocytes lead to worse outcome, blood banking policy may change in ways that have huge implications for transfusion services. The current limit set by the FDA for erythrocyte storage of 42 days provides great flexibility in the management of the blood supply, permitting shifts of stock in response to regional shortage or natural disaster and limiting the impact of week-to-week or month-to-month variation in supply. Currently, 2.4% of erythrocyte units are outdated, totaling 401,000 units.§A reduction in the maximum allowable erythrocyte age could result in dramatic changes in the cost and logistics of blood banking. Of course, the degree of impact would depend on the new age limit. The storage age of transfused erythrocyte units at most hospitals in the United States typically is 15–20 days, and the distribution is skewed toward newer units and varies by blood type. Elimination of the oldest units might be accomplished without resulting in critical blood shortages.

The common theme in each of these debates is that there is insufficient evidence to make informed policy decisions. In the case of universal leukoreduction and male-only plasma, standard practice changed without definitive evidence. In the United States, the approach of the FDA has been to provide guidelines, rather than mandates, when the data are unclear, leading to diversity of practice.

Improved hemovigilance infrastructure and funding for large-scale RCTs likely would improve blood banking and transfusion practice. The United Kingdom Serious Hazards of Transfusion program has led the way by developing a detailed database of adverse events dating to 1996. The European Hemovigilance Network, now 6 yr old, boasts 25 member nations. One of the many challenges during implementation of international standards set forth by the European Union Blood Safety Directive of 2003 has been establishing a universal system for reporting of adverse events and reactions. During the May 2009 meeting of the authorities from each member state, this issue was raised again, and a draft of a common form for hemovigilance reporting was presented. The great promise of this organization as a means for collecting useful epidemiologic data has yet to be fully realized.‖‖In the United States, the absence of a national healthcare system and a patchwork of reporting systems from hospitals and blood banks have slowed progress toward development of such a network. However, in 2009, the U.S. Biovigilance Network initiated a nationwide system of reporting and data collection.

The high-profile nature of studies on the effect of liberal versus  conservative thresholds for erythrocyte transfusion7has led to widespread adoption of evidence-based erythrocyte transfusion, but high rates of inappropriate transfusion of fresh frozen plasma and platelets remain87because of insufficient evidence to produce well-informed guidelines. Implementation of transfusion algorithms can change physician practice88and may improve outcomes and reduce costs,9,88,89but doing so requires a systemwide effort90and cannot be implemented without adequate evidence. Funding for high-quality prospective clinical trials is a necessity for generating the necessary evidence. Ultimately, evidenced-based physician practice with avoidance of unnecessary transfusions will be the most effective way to reduce complications of transfusion.