Sepsis leads to significant patient morbidity and mortality affecting 750 000 hospitalized patients yearly, with nearly 200 000 annual deaths in the United States related to sepsis.1 The annual cost of sepsis in the United States is estimated at approximately $16.7 billion.2 Many critically ill and septic patients will require care in the intensive care unit (ICU). It is reported that as many as 95% of ICU patients will have a lower than normal hemoglobin (Hb) level by their third day in the ICU.3 As many as one-third to one-half of ICU patients will receive transfusions during their ICU stay.4 The decreased Hb level may be due to the patient's underlying disease process, impaired production of red blood cells (RBCs), or iatrogenic blood loss due to phlebotomy. The magnitude of iatrogenic blood loss can be quite significant. A study in 1986 reported that ICU patients lost approximately 65 mL of blood daily due to phlebotomy, with a mean total blood loss of 762 mL per ICU stay. Researchers observed a blood loss of 944 mL in patients with arterial catheters.5 A subsequent study in 1999 reported that, even with blood conservation strategies in the ICU, the amount of daily blood loss due to phlebotomy can still be as high as 41 mL.6 It can be surmised that the sicker the patient, the higher the likelihood of laboratory testing requiring phlebotomy and, therefore, higher iatrogenic blood loss due to monitoring.

It is now relatively accepted in the medical community that RBC transfusion can lead to increased patient morbidity and mortality. Studies have shown a relationship between RBC transfusion and an increase in duration of hospital stay, ICU stay, duration of mechanical ventilation, risk of postoperative infection, and risk of multiple organ failure. Observational studies have demonstrated a dose-dependent increase in morbidity and mortality when comparing patients who received RBC transfusion with patients who were not transfused or who received fewer RBC transfusions.7-10 Increased morbidity and mortality in transfused patients, of course, leads to an increase in the total cost of health care.

If a critically ill or septic patient does need RBC transfusion, then an obvious question is simply, "How much is enough?" To discuss this point, recent information regarding RBC transfusion to critically ill and septic patients is discussed below, including data comparing a liberal versus restrictive transfusion strategy in this patient population.

The pathophysiology of sepsis is complicated and shares overlapping features with some of the changes that occur to RBCs during storage. In addition, there are new and emerging data indicating that storage of RBCs causes changes that may potentially decrease oxygen delivery to the microcirculation. With all of these factors combined, some propose that transfusion of RBCs to a septic patient may actually be "throwing fuel on the fire." The following information will also look at the pathophysiology of sepsis and the similarities to the changes that occur during RBC storage. These changes may contribute to the increased patient mortality and morbidity observed with RBC transfusion.

TRANSFUSION IN CRITICALLY ILL AND SEPTIC PATIENTS

RBC transfusion is a common and important life-saving procedure, with approximately 30 million blood components being transfused each year in the United States.11 In previous decades, liberal and early-goal transfusion therapy was the standard. Increasing emphasis, however, has been placed on evidence-based transfusion practice. To assess appropriateness of RBC transfusion, an international committee of 15 experts reviewed 494 publications and found that transfusion was appropriate in 81% of transfusions to patients with Hb less than 7.9 g/dL and 71.3% in patients with Hb levels of 8 to 9.9 g/dL. However, all transfusions with Hb greater than 10 g/dL were felt to be inappropriate.12

The Transfusion Requirements in Critical Care (TRICC) trial is a randomized, controlled trial comparing use of liberal versus restrictive transfusion strategy. Within the liberal arm, the goal was to maintain Hb of 10 to 12 mg/dL, with a lower RBC transfusion trigger of 10 g/dL. The restrictive strategy had a goal of maintaining Hb between 7 and 9 g/dL with a lower RBC transfusion trigger of 7 g/dL. The trial included 838 patients. Although the overall 30-day mortality rates were not significantly different within the 2 groups, the researchers observed that the mortality rate during hospitalization (in-hospital rate) was 22.2% in the restrictive strategy group compared with 28.1% in the liberal strategy group (P = .05). They also noted that in the restrictive strategy group, patients who were clinically less ill (as determined by an Acute Physiology and Chronic Health Evaluation [APACHE] score less than 20) or younger (less than 55 years old) had significantly fewer 30-day mortality events than the more severely ill (an APACHE score greater then 20) or older patients (older than 55 years) when compared with similar patients in the liberal transfusion strategy group. The patients who were less ill had a 30-day mortality rate of 8.7% in the restrictive strategy group compared with 16.1% in the liberal strategy group (P = .03). The younger patients had a 30-day mortality rate of 22.2% in the restrictive strategy group compared with 28.1% in the liberal strategy group (P = .05).13,14 Additional analysis of the data (n = 713) found that there was no advantage with liberal over restrictive transfusion strategy in weaning patients from mechanical ventilation.15 This finding supports those reported by Fernandes et al16 in 2001, who reviewed Hb level and hemodynamic and oxygen use effect in 15 critically ill septic patients on mechanical ventilation and with Hb less than 10 g/dL. Fernandes et al reported no improvement in global or regional oxygen use in anemic septic patients by increasing Hb. In addition, Silverman and Tuma17 in 1992 compared efficacy of dobutamine infusion and RBC transfusion. Their findings suggest that dobutamine is more effective than RBC transfusion in increasing oxygen delivery to tissue. Another review of the TRICC data in 2004 observed that patients with severe ischemic heart disease showed a better trend with liberal transfusion strategy. Conclusions from these TRICC data report that restrictive transfusion strategy is at least equivalent to liberal transfusion strategy in all groups except patients with severe ischemic heart disease. In addition, restrictive transfusion strategy is potentially better in patients who are less ill and younger.13

The Anemia and Blood Transfusion in Critical Care trial reviewed 3534 patients from 146 Western European ICUs. The researchers reported that the 28-day mortality rate was 22.7% in transfused versus 17.1% in nontransfused patients (P = .02). They also noted that patients with a similar degree of organ dysfunction had a higher mortality rate if transfused. If a patient in the ICU received a transfusion, the patient's odds of dying were increased by a factor of 1.37.4,13

Another study evaluating anemia and blood transfusion in the critically ill is the CRIT study ("Anemia and blood transfusion in the critically ill-Current clinical practice in the United States"). This study evaluated 4892 ICU patients in 213 US hospitals and observed that the number of RBC transfusions was an independent predictor of length of ICU stay, overall hospital length of stay, and mortality. In addition, the association with mortality was particularly pronounced when more than 2 RBC units were transfused.18

The Surviving Sepsis Campaign (SSC), led by an international panel of experts, aimed to establish a series of international guidelines to reduce morbidity and mortality in severe sepsis and septic shock. The guidelines were initially reported in 2004 and updated in 2008 to cover a wide spectrum of treatment parameters, including transfusion. The SSC used the Grades of Recommendation Assessment, Development and Evaluation system to guide assessment of quality of evidence from high to very low and determination of strength of recommendation from strongly recommended (desirable effects clearly outweigh its undesirable effect) to weak. The SSC recommends transfusion if Hb is less than 7 g/dL with posttransfusion targets in adults of 7 to 9 g/dL. Higher Hb may be indicated in special circumstances such as myocardial infarction, severe hypoxemia, acute hemorrhage, cyanotic heart disease, or lactic acidosis. They also report that Hb 10 to 12 g/dL has not been shown to be associated with improved outcome.19,20

STORAGE OF RBCs

The main objective in transfusion of RBCs is to increase oxygen-carrying capacity. Recent data suggest, however, that changes resulting from RBC storage may actually decrease the ability of transfused RBCs to deliver oxygen to the microcirculation. Blood is a biologic substance, and with all biologic substances, the changes that occur with storage and the interactions with the recipient are complex and still not fully understood. There are, however, well-recognized changes that occur with red cell storage that are collectively referred to as the red cell storage lesion. It has been known for some time that, with storage, glucose is consumed by RBCs, levels of adenosine triphosphate decrease, extracellular potassium levels increase, pH decreases, levels of 2,3-diphosphoglycerate (2,3-DPG) decrease, and RBC shape changes occur.21 All of these may have potential effects in the recipient; however, RBC shape changes are of particular interest in this discussion. RBCs are normally a deformable biconcave disc. This deformability allows RBCs to enter more easily into the microvasculature of the patient. With time, however, RBCs are deformed into an irreversible form that is more rigid than the biconcave shape. This rigid form decreases the ability of the RBCs to enter the microcirculation, resulting in decreased oxygen delivery to those areas that the RBCs cannot reach in the microcirculation.21 RBC shape changes are partially due to the loss of fragments of the cell membrane during cellular degeneration. These fragments are often referred to as red cell microparticles (RMPs). The number of RMPs increases with duration of storage, have proinflammatory-promoting activities, and may affect nitric oxide (NO) levels.22

Recent attention has focused on the role of NO, a relaxing factor derived from endothelial cells (ECs) normally lining blood vessels. NO acts to regulate local blood flow in tissue by recognizing hypoxia and subsequently causing vasodilation (hypoxic vasodilation). NO activity allows more blood to enter the hypoxic area, resulting in increased delivery of oxygen. Simply stated, more RBCs in an area equate to more oxygen delivered. In addition to producing NO, ECs produce other regulators involved with vascular contraction. Patients who are critically ill have endothelial dysfunction. It is proposed that the endothelial dysfunction acts in synergy with the effects of blood storage to produce a relative vasoconstriction.23 In addition to oxygen, Hb within RBCs can bind and deliver NO to tissue via the circulation. With red cell storage, the ability of Hb to bind and deliver NO is reduced. With decreased NO vasodilation activity, there are fewer RBCs within the hypoxic region and, therefore, decreased oxygen delivery to hypoxic tissue. This theory is further supported by evidence linking decreased NO levels with numerous disease states that are associated with tissue hypoxemia, including pulmonary hypertension, sickle cell disease, diabetes, congestive heart failure, and sepsis.24

The relationship to ECs and transfusion is also complicated by the adherence of transfused RBCs to vascular ECs. Normally, RBCs do not readily adhere to blood vessel walls, allowing smooth flow within the blood vessels. With storage, however, RBCs undergo cellular changes that allow them to adhere to ECs more readily. Adhesion of RBCs to the blood vessel walls disrupts the normal, smooth flow, resulting in disturbances of the local blood flow and decreased oxygen delivery to tissues, and possibly leading to occlusion of the microcirculation.25 The importance of RBC adhesion is also supported by significantly increased adhesion seen in RBC-related diseases such as malaria, sickle cell disease, thalassemia, and diabetes mellitus.26 Some forms of infection, including sepsis, have clinical effects resulting from endotoxin, which is produced by some forms of bacteria. Models have shown that pretreatment with endotoxin activates ECs and increases RBC adhesion.27 There are many proposed mechanisms for this increased adhesion that results from storage; these include changes in the expression of cell adhesion molecules on the RBC, changes in the RBC membrane, and possible interaction with RMPs containing shed lipids.26

Hb is normally contained within RBCs. With storage, RBCs will eventually break down and release free Hb, which is then transfused with the blood product. Free Hb is associated with thrombosis and is reported to alter EC and vascular function.8 NO normally adheres to RBCs to circulate through the body. Free Hb is known to be a potent scavenger of NO, with NO binding 1000 times more rapidly to free Hb than to RBCs. The free Hb rapidly destroys the NO by oxidation, thereby decreasing NO bioavailability.13 Iron is also released during RBC breakdown, resulting in free iron (ie, iron that is not bound to transferrin, the physiologic iron transport protein). The amount of extracellular, free iron increases with the number of days in storage. Free iron is taken up by the monocyte-macrophage system (MMS) and subsequently induces a pro-inflammatory response. Healthy adults at steady state have approximately 20 mg of iron cleared by the MMS daily. Transfusion of just one unit of older, stored RBCs will acutely deliver a 60-fold increase in the hourly amount of iron delivered to the MMS. Although iron is an essential element for life, iron overload may also have adverse effects, including risk of certain bacterial infections such as gram-negative sepsis.28

PATHOLOGY OF SEPSIS

The administration of blood products to the septic patient is a controversial topic requiring an understanding of the pathophysiology of sepsis and a detailed understanding of the events from the initial infectious insult through the body's immune response and, ultimately, the dysregulation of the body's coagulation system. An infection is defined as the inflammatory response to an invasion of microorganisms into sterile tissue. Sepsis is an infection with evidence of systemic inflammatory response syndrome (SIRS). Septic shock is sepsis with hypotension that persists despite intravenous fluid resuscitation. SIRS is defined as clinical evidence of infection including 2 or more of the following: a temperature greater than 38[degrees]C or less than 36[degrees]C, a heart rate greater than 90 bpm, hyperventilation (respiratory rate greater than 20 breaths per minute or PaCO₂ less than 32 mm Hg), and a white blood cell count greater than 12 000 cells per microliter or less than 4000 cells per microliter.29 Until recently, gram-negative rods were the most frequent type of bacteria implicated in sepsis. The proportion of cases of sepsis from gram-positive cocci rose in the mid-1980s, however, and is now the most common cause of sepsis. Of ICU-acquired infections, the most frequent are infections of the respiratory tract followed by the bloodstream and urinary system. Older patients, however, are more likely to have gram-negative infections than younger patients because of increased genitourinary infections. Escherichia coli is the most common cause of bloodstream infection in adults over 65 years of age, whereas Staphylococcus aureus is the most common cause of bloodstream infection in younger adults. The Sepsis Occurrence in Acutely Ill Patients study has suggested that 50% of episodes of Staphylococcus aureus sepsis are resistant organisms (eg, methicillin-resistant Staphylococcus aureus). Fungemia is also attributed in 17% of the septic population.4

The absolute mortality of sepsis is 65.5 per 100 000 persons in the United States and is highest within a cluster in the Southeast and Mid-Atlantic states.1 Sepsis results in 570 000 emergency room visits affecting 750 000 hospitalized patients yearly. There are nearly 200 000 US sepsis-related deaths annually, and there is greater than 60% mortality with septic shock. It is estimated that the cost of septic episodes amounts to an annual cost in the United States of $16.7 billion.2

Ultimately, the outcome of a septic episode is multifactorial with host physiologic defense mechanisms and the specific infectious agent paramount. An increased mortality is associated with Candida and Pseudomonas infections of the bloodstream.30 Common morbidities associated with sepsis are described in all organ systems. In the lungs, the large microvascular surface area and widespread endothelial injury combined with disrupted capillary blood flow produce interstitial and pulmonary alveolar edema. The lungs then show acute lung injury and acute respiratory distress syndrome.31 Central nervous system (CNS) complications often occur even before complications develop in other organ systems. Dysfunction of the blood-brain barrier is found in septic patients. Toxic mediators and leukocytes facilitate inflammation within the CNS, resulting in encephalopathy and peripheral neuropathy.32 Decreased hepatocellular synthetic function in the septic patient is a key dysfunction. Liver damage aids the progression of sepsis because the reticuloendothelial system helps to clear bacteria and bacterial toxins. Damage to the liver can decrease clearance of these bacteria-derived toxic substances. Hepatocellular damage also reduces synthesis of clotting factors and contributes to decreased hemostasis in the septic patient by deregulation of the clotting cascade.31 Acute renal failure is a frequent occurrence in sepsis, and acute tubular necrosis is a common morbidity. The mechanism whereby circulating endotoxins and sepsis produce renal injury is, however, uncertain.31

Circulatory system dysregulation often results in a mismatch of oxygen demand and supply. NO released by ECs results in vasodilation and increased vascular permeability at the site of infection. Induction of NO synthesis may produce excess NO with sufficient vasodilation to contribute to septic shock. Induction of NO synthesis may also contribute to injury in the CNS, further affecting autonomic regulation of the circulatory system. Hypotension is the most severe type of circulatory dysfunction in sepsis, and it compounds tissue injury by decreasing oxygenation in the microcirculation.31,33

The cellular physiology of sepsis is a result of a complex immune response that includes activation of leukocytes, activation of ECs, release of inflammatory mediators, and, ultimately, activation of the coagulation cascade.33 The overwhelming host immune response (producing SIRS) is triggered by bacterial cell wall components, including lipopolysaccharides, lipoteichoic acid, and other exotoxins. These bacterial factors stimulate the MMS to initiate an excessive proinflammatory response causing diffuse capillary injury. The wave of inflammatory mediators stimulates coagulation and suppresses fibrinolysis, leading to diffuse thrombosis. In severe sepsis, disseminated intravascular coagulopathy consumes coagulation factors and platelets, which results in bleeding.34

The hyperinflammatory response may be followed by a secondary anti-inflammatory response. The subsequent immune suppression in this case may be profound, leading to secondary infection. Ongoing investigations into immune system signaling have suggested that individual cell necrosis resulting in apoptosis and inappropriate removal of lymphocytes may contribute to further immune suppression.35,36 The ultimate pathology of mortality in severe sepsis is likely a result of imbalance between systemic inflammatory responses and compensatory anti-inflammatory responses.37

DISCUSSION

Sepsis is a complicated disease process in which the systemic reaction can lead to magnification of morbidity and mortality. The addition of stored RBCs may potentially accelerate the adverse effect of the septic process. With RBC storage, changes in the RBC shape, decreased delivery of NO, formation of RMPs, and inactivation of NO by free Hb all potentially decrease the ability of RBCs to deliver oxygen to tissue in the microcirculation.21,22 In addition, RMPs and free iron are associated with activation of a proinflammatory response that may exacerbate the inflammatory state of the patient.22 There is also overlap with regard to ECs. Stored RBCs result in EC activation with increased RBC adhesion to blood vessel walls. Many disease states, including sepsis, are associated with disruption of vascular wall integrity, and addition of transfused RBCs may potentially act in synergy, resulting in increased EC activation.23-25

RBC transfusion is common in treatment of critically ill and septic patients. The TRICC study has found that RBC transfusion in critically ill and septic patients using a restrictive transfusion strategy is at least equivalent to outcomes using a liberal transfusion strategy in most clinical situations. This study also suggests that restrictive transfusion strategy is associated with lower 30-day mortality in less ill and younger patients than liberal transfusion strategy.14,15 Restrictive transfusion strategy is also the recommendation of the SSC.19,20 Because there is no documented advantage of liberal transfusion strategy and there are potential adverse effects associated with RBC transfusion, it is reasonable to conclude that restrictive transfusion strategy is the preferred, best practice. Is RBC transfusion to a critically ill or septic patient adding to the domino effect and throwing the proverbial fuel on the fire? Additional research and clinical trials are necessary to fully answer this question, but evidence suggests a possible synergistic effect. As stated in the essence of the Hippocratic oath, first do no harm.

REFERENCES