Chapter 46. Sepsis with Acute Organ Dysfunction

• Sepsis is the combination of a known or suspected infection and an accompanying systemic inflammatory response.
• Severe sepsis is sepsis with acute dysfunction of one or more organ systems; septic shock is a subset of severe sepsis.
• Severe sepsis is common, frequently fatal, and expensive. More than 750,000 cases occur annually in the United States.
• Effective management of patients with severe sepsis requires early identification, cardiopulmonary support, antibiotics, source control, and general supportive care.
• The prognosis of the patient with severe sepsis is related to the number of dysfunctional organs.
• Cardiopulmonary support consists of early and aggressive fluid resuscitation, maintenance of mean arterial pressure at 65 mm Hg or higher, and measures to maximize and maintain tissue oxygenation; judging the adequacy of tissue oxygenation is informed by measurement of the mixed venous oxygen saturation, and supportive therapy is most effective and beneficial when provided in the first hours of presentation.
• Patients should receive early intravenous empirical antibiotics directed at all possible sources of infection. Appropriate antibiotics decrease the mortality rate by 10% to 15% in patients with severe sepsis.
• Source control can be surgical or nonsurgical and is intended to remove or lessen the burden from the primary focus of infection.
• Despite appropriate antibiotics, source control, and organ support, the mortality rate in patients with severe sepsis remains at 28% to 50%.
• Specific antisepsis interventions have recently been introduced that target multiple pathophysiologic aspects of the sepsis cascade and can improve outcomes.
• To maximize outcomes, supportive measures must be introduced to ensure proper nutrition; maintain fluid, glucose, and electrolyte homeostasis; promote tissue oxygenation; and prevent complications.

Sepsis with acute organ dysfunction (severe sepsis) is common and frequently fatal and represents a significant health care burden. The incidence and associated mortality and morbidity rates of severe sepsis are commonly underestimated. This is a function of a number of factors. Severe sepsis is not generally reported as a primary diagnosis. For example, although steps are underway to address this issue, the most recent edition of the International Classification of Diseases, Ninth Revision, Clinical Modification lacks a diagnostic code for severe sepsis. Instead, severe sepsis is often coded as a complication of another disorder (e.g., cancer or pneumonia). Several recent publications have evaluated the epidemiology of severe sepsis in the United States.1,2 They have estimated that the annual incidence of severe sepsis in the United States is in the range of 240 to 300 cases per 100,000 population. Further, in 2003, there were approximately 750,000 cases of sepsis in the United States. In Europe, the incidence of severe sepsis exceeds 200,000 annually.3 Reported mortality rates in patients with severe sepsis range from 28% to 50% or higher.1,4 Thus, in the United States and Europe, at least 700 to 1300 patients die daily from severe sepsis. Patients with severe sepsis account for annual health care expenditures in excess of $16 billion in the United States and £5.2 billion in Europe.1,3

The incidence of severe sepsis peaks in children younger than 12 months, remains low until midlife, and then progressively increases.1 In a study by Angus et al1 of the patients who developed severe sepsis, nearly 66% were older than 65 years. This population also accounted for more than 75% of the overall health care costs of the disease. The incidence of severe sepsis is anticipated to increase approximately 1.5% per year until at least 2050.1 This increase is due to a number of factors, including age shifts in the population, prevalence of more critically ill patients (e.g., transplant recipients), and increases in the numbers of invasive diagnostic procedures and monitoring techniques. This predicted increase has significant implications for the critical care community because it has been estimated that by 2020 there will be a 22% shortfall of available intensivists' hours to meet this demand.5

In 1992, members of the American College of Chest Physicians (ACCP) and the Society of Critical Care Medicine (SCCM) developed a set of consensus definitions for sepsis and related disorders (Table 46-1).6 The consensus committee believed that standardized terminology would improve the ability of clinicians to make an early diagnosis of sepsis, provide for more reliable reporting of the incidence and severity of sepsis, and facilitate early therapeutic interventions. In addition, they hoped that acceptance of the definitions would help to standardize research protocols and improve the dissemination and application of clinical information from subsequent studies. The consensus committee acknowledged that the clinical presentation of sepsis spanned a continuum of severity; however, they recognized certain common phases and provided specific definitions for them (Fig. 46-1). Sepsis was defined as the presence of at least two of the four systemic inflammatory response syndrome (SIRS) criteria developing in response to a documented or suspected infection. Severe sepsis was sepsis plus acute organ dysfunction (Table 46-2). Although these definitions have provided an important framework, they have been criticized as being too sensitive and not providing enough information for clinicians attempting to make a diagnosis.7 In addition, the definitions lack a clear pathologic basis and fail to incorporate the hemostatic component of severe sepsis.6

However, the ACCP/SCCM definitions have endured the test of time. For example, the critical importance of organ dysfunction and outcomes first emphasized by the conferees was recently reaffirmed by a study of risk factors for hospital mortality rate in 3608 intensive care unit (ICU) patients included in the European Sepsis Study (Fig. 46-2).8 In December 2001, representatives from the SCCM, the ACCP, the European Society of Intensive Care Medicine, the American Thoracic Society, and the Surgical Infection Society met to revisit the definitions of sepsis and related disorders.9 After an intensive review, the group concluded that there was no evidence that would justify altering the definitions first proposed by the ACCP/SCCM. As part of the discussion, however, they expanded the lists of signs and symptoms of sepsis that are reflected by experience gained at the bedside that prompts a clinician to say that the patient “looks septic.” Among these criteria, they included additional general (e.g., altered mental status, significant edema or positive fluid balance, or hyperglycemia), inflammatory (increased C-reactive protein [CRP] or pro-calcitonin [PCT]), hemodynamic (O2 saturation in venous blood >70%, cardiac index >3.5 L/min per square meter), organ dysfunction (coagulation abnormalities, ileus, or hyperbilirubinemia), and tissue perfusion (decreased capillary refill or mottling) abnormalities. They also emphasized the need for a precise method to characterize and stage patients with sepsis. Creation and validation of such a system would bring an increased degree of precision to clinical trials and facilitate the ability of clinicians to more fully characterize the disease and to select appropriate therapies. As a subject for debate, they proposed the Predisposition, Insult/Infection, Response, Organ Dysfunction (PIRO) model, a staging process that uses concepts from the Tumor Nodes Metastases (TNM) system of clinical oncology.

The pathogenesis of the septic response involves interaction of the host with a microbial invader. The outcome of the process depends on the capability of the immune system, endothelium and hemostatic mechanisms to contain and then eliminate the process, and the ability of the patient to restore homeostasis over time.

Microbial Factors

Microbes possess several factors that facilitate their growth in a normally sterile environment (Fig. 46-3). These include properties of their capsule or envelope, cell wall, and metabolic factors such as the production of exotoxins. Some such factors are cited by way of example.

Pili of strains of Escherichia coli can enable the coliform bacillus to adhere to Tamm-Horsfall protein coating the epithelium of the lower urinary tract or P-blood group antigens expressed on the epithelium of the renal pelvis. Capsular polysaccharides of certain strains of Streptococcus pneumoniae render the organism resistant to phagocytosis. The cell wall of gram-negative bacteria consists of an inner phospholipid bilayer with embedded transport proteins and an outer layer composed of lipoproteins, lipopolysaccharides (LPS), outer membrane proteins, and capsular polysaccharides. Bacterial LPS is generally considered to be the principal element in the initiation of the septic response in patients with severe gram-negative infections. Whereas components of the cell wall of gram-positive bacteria may be able to elicit some of the effects of endotoxin, much of the pathogenicity of gram-positive organisms is the result of exotoxins. Some of these exotoxins can function as superantigens: chemical substances that simultaneously bind to common sequences on the major histocompatibility complex and the T-cell receptor. Superantigen binding and T-cell activation can produce a flood of cytokines and is independent of the antigenic specificity normally required for activation of various T-cell clones and the need for costimulatory molecules. Streptococcal and staphylococcal pyrogenic exotoxins acting as superantigens are believed to play a critical role in the pathogenesis of toxic shock syndrome. Hyaluronidase secreted by some bacteria can facilitate their spread along tissue planes.

Host Factors

The host has a spectrum of mechanisms that supports its ability to resist the invasive properties of some microorganisms (Table 46-3). The epithelium is the first line of defense against infection. It not only provides a mechanical barrier but also contributes other protective functions in a site-specific fashion (e.g., mucociliary flow in the respiratory tract and gastric pH). The immune status of the host is a function of inherited and acquired components. Genetic polymorphisms can be responsible for dissimilar responses to infection. For example, polymorphisms in toll-like receptors (TLRs) may govern an individual's response to endotoxin, whereas variants in the promoter region of the tumor necrosis factor (TNF) gene can help determine the risk for sepsis after trauma.10 Age, disease, exposures, and interventions are acquired risk factors for sepsis. Young children and the elderly have an increased incidence of sepsis. Various morbidities can increase the risk of sepsis. Patients with diabetes mellitus are at increased risk of infection for a variety of reasons. Hyperglycemia appears to facilitate colonization and growth of organisms such as Staphylococcus aureus and Candida species. In addition, they have a variety of defects in cell-mediated immunity and phagocytosis. Genetic and acquired factors often interact to determine susceptibility to disease. For example, patients positive for the human immunodeficiency virus (HIV) who are heterozygous for a deletion in the gene for the macrophage CCR5 receptor or in the promoter region of the gene have slow progression of HIV disease compared with patients with the wild-type genotype.

Innate Immune System

The human immune system is composed of adaptive and innate components. The adaptive elements are specialized B and T cells. Each of these cells has a structurally unique B- or T-cell antigen receptor that is generated randomly (i.e., not genomically encoded). When a lymphocyte bearing a useful receptor encounters a pathogen, it is selected for clonal expansion. Although adaptive immunity contributes significantly to the overall immune response and provides protection against future encounters with the organism, generation of a significant primary humoral or cellular immune response takes days to weeks, time that is unavailable to a patient with severe sepsis. Unlike the adaptive immune system, elements of the innate immune system can respond instantly to a challenge. This property helps explain the overwhelming systemic inflammatory response generated in patients with severe sepsis. Therefore, we focus the remainder of this section on innate immunity.

Cellular Components of Innate Immunity

In humans, cellular effectors of innate immunity are monocytes/macrophages, neutrophils, natural killer cells, and platelets. The monocyte/macrophage is the human analogue of the amebocyte, an innate immune cell present in invertebrates. Monocytes originate in the bone marrow and then transit the peripheral blood to lodge in tissues as macrophages, Kupffer cells, and other reticuloendothelial elements. Pathogen-associated molecular pattern receptors comprise a key component of the innate immune system.11 Unlike the antigen receptors of B and T cells, these receptors are encoded in the germline, are functionally divided into three classes (secreted, endocytic, and signaling), and trigger an immediate effector function. Secreted pathogen-associated molecular pattern identification is discussed in the section on fluid-phase elements of innate immunity (below). Endocytic pattern recognition receptors such as the macrophage mannose or scavenger receptor recognize highly conserved molecular components of microbial cell walls and mediate their phagocytosis by monocytes/macrophages. Signaling receptors on monocytes/macrophages identify microbial molecular patterns and activate signal transduction pathways that upregulate a spectrum of immune response genes responsible for a variety of cellular functions, including cytokine release.

The toll gene, a transmembrane signaling protein present in organisms as primitive as the fruit fly, is believed to be very important to humans in the septic state. Humans possess at least 10 toll gene homologues known as TLRs. The TLR genes code for cell membrane molecular pattern receptors that play an integral role in the generation of local and systemic inflammatory responses. Natural ligands of the various TLRs include LPS (TLR4), peptidoglycan (TLR2), and flagellin (TLR5). The role of TLR proteins can be illustrated by the response of the innate immune system to LPS. LPS is an outer leaflet compound molecule that is unique to gram-negative bacteria; after infusion into humans, LPS reproduces many of the features of sepsis, including activation of the coagulation and complement systems.12,13 After entering the plasma, LPS is recognized, bound, and transported to innate immune effector cells by LPS-binding protein. The cell membrane CD14 receptor on monocytes and neutrophils accepts the LPS from LPS-binding protein and forms a trimeric complex with TLR4 and MD-2, another membrane protein. Once formed, the complex activates a signal transduction pathway that phosphorylates IκB. As a result, IκB undergoes degradation and releases bound nuclear factor κB. The latter then migrates into the nucleus, where it induces transcription of a wide variety of immune- and inflammatory-response genes.11 Transcription of these genes is responsible for the upregulation of tissue factor (TF) on monocytes/macrophages and endothelium and for the cytokine cascade that creates many of the elements of SIRS in patients exposed to LPS. These key elements of the inflammatory response are therefore linked to the coagulation cascade, with these two responses combining to form the major facets of the pathophysiologic derangements in septic humans. Surfactant appears to play an important role in maintaining innate immunity within the lung. This may be due to the protein's ability to block LPS signaling by inhibiting mitogen-activated protein and IκB kinases of alveolar macrophages.14

TF is responsible for initiating coagulation through the previously named extrinsic pathway, now called the TF pathway. TF is expressed constitutively on a limited number of cells and can be induced in a wide variety of cells after exposure to endotoxin or cytokines. Cells normally expressing TF include extravascular monocytes/macrophages and fibroblasts.15 Sequestered from circulating factor VIIa, these cells are poised to activate coagulation immediately after disruption of a blood vessel wall. TF expression can be upregulated in a variety of other cells that do not express the cell membrane protein constitutively. As a result, after exposure to TNFα, CRP, CD40 ligand, or other substances, TF expression can be identified on subsets of cells including circulating monocytes, vascular smooth muscle cells, a subset of endothelial cells, and possibly alveolar epithelial cells.16 During acute inflammation, activation of the cytokine cascade is followed by rapid upregulation of TF on monocytes and a shift of components of the hemostatic system from an antithrombotic to a prothrombotic state. Circulating TF antigen and microparticles shed from inflammatory cells and platelets contribute to the process.

Fluid-Phase Elements: Cytokines and Complement

Once monocytes/macrophages have been activated, there is a response by fluid-phase elements. These include cytokines and activated components of the complement system. TNFα is a central cytokine in this process. Once released, it functions in an autocrine, paracrine, and endocrine fashion in a positive feedback loop to stimulate additional TNFα production and production of other cytokines. Other proinflammatory cytokines are released, including interleukin (IL)–1, IL-2, IL-6, IL-8, IL-10, platelet activating factor (PAF), and interferon γ. The cytokine cascade produces a vast array of biological effects recognized as the systemic inflammatory reaction.

However, as sepsis persists, there may be a shift in the cytokine profile from an inflammatory to a predominantly anti-inflammatory one.17 Loss of T cells, B cells, and follicular dendritic cells through apoptosis may contribute to the immunosuppression through two processes: loss of immunocytes and the direct immunosuppressive effect of apoptotic cells.18,19 As the anti-inflammatory reaction progresses, the septic patient may become anergic. Although it may be helpful to think that the response of the immune system to a septic insult progresses in discrete and identifiable stages, this may not be true. Some have proposed that, distant from the original source or inflammatory focus, the body's reaction is predominantly anti-inflammatory from the onset.20 The complex nature of this process may partly explain why clinical trials of various biologic response modifiers such as TNFα antagonists have not improved overall survival rate and have even been detrimental in some cases.

Plasma components of the complement system circulate in an inactive state and develop enzymatic activity only after proteolysis or a conformational change. The system works by depositing components of complement on foreign substances or organisms requiring opsonization and by promoting inflammation. Complement can be activated through three pathways: (1) classic pathway, (2) alternate pathway, and (3) mannan-binding lectin pathway. The classic pathway is activated by immunoglobulin (Ig) M or IgG antibodies bound to a particular pathogen or by CRP released during the acute phase reaction. The classic pathway serves as a link between the complement and hemostatic systems. Components generated during activity of the classic pathway increase the exposure of phosphatidylserine on endothelial cell surfaces.21 This action provides the phospholipid that is essential for various phases of clot formation. Activation of the alternate pathway results when complement binds directly to pathogen-associated molecular patterns on the surfaces of bacteria or fungi. Mannan-binding lectin is an example of a secreted pathogen-associated molecular pattern receptor. It is synthesized in the liver and secreted into the plasma during the acute-phase reaction. It binds to carbohydrate structures common to gram-positive and gram-negative bacteria, yeasts, and some viruses and parasites. Once binding occurs, mannan-binding lectin-associated proteases are activated and help to destroy the organism. After complement activation, some of the molecular elements of the system are cleaved, thus producing fragments with biological activity. For example, N-terminal proteolysis of C3, C4, and C5 release small peptides (e.g., C5a) that bind to G protein–coupled receptors on a variety of cells to produce chemotaxis, increase vascular permeability, and so forth. It has been proposed that C5a in excess can paralyze neutrophil function.22 In an experimental model, antibody-mediated neutralization of excess C5a protected animals against septic death.22

Endothelial Cell Dysfunction

With approximately 1011 endothelial cells and a surface area estimated to exceed 1000 m2, the endothelium surpasses the skin as the largest organ in the body.23,24 The great majority of the endothelium is located in the microvasculature; in that region, the endothelial surface area per unit of blood volume is 2000 to 3000 times greater than in the larger blood vessels.25 The endothelium serves as the interface between inflammation and coagulation.26 Situated at the junction of the flowing blood and extracellular space, the endothelium influences a variety of inflammatory and hemostatic processes (e.g., cell-trafficking, vasoregulation, thrombosis, and antithrombosis) and plays a key role in the inflammatory, prothrombotic, and impaired fibrinolytic components of sepsis.27

Endothelial dysfunction is a key element in the pathogenesis of severe sepsis. This may be secondary to the effects of endotoxin or proinflammatory cytokines on endothelial cells. For example, a single injection of endotoxin in rabbits produces desquamation of approximately 25% of the aortic endothelial surface area within 5 days.28 Alternatively, organisms such as Rickettsiae can directly infect endothelial cells. Severe endothelial injury can produce a microvascular coagulopathy and acute organ dysfunction.29 Importantly, however, endothelial cells are differentially regulated in space and time. That is, the endothelium of the lung may respond differently to a septic insult than similar cells in the spleen.30 The phenotype of endothelial cells also changes over time. For example, an endothelial cell population with an antithrombotic phenotype may transiently switch to predominantly prothrombotic activity for some period after exposure to TNFα.

Thrombomodulin (TM), intercellular adhesion molecule 1, E-selectin, and von Willebrand factor (VWF) are normal membrane-bound and intracellular endothelial cell components. Therefore, elevated circulating levels of the soluble forms of these proteins are markers of endothelial dysfunction and injury.31 In children with septic shock, soluble TM (sTM) levels are increased, and the levels correlate with survival status and the extent of organ dysfunction.32 Immunohistochemical studies of skin biopsy samples from patients with meningococcal septicemia have demonstrated decreased stainable TM and endothelial cell protein C receptor (EPCR).33 This appears to result from a combination of shedding of cell surface TM and EPCR plus downregulation of endothelial cell synthesis and/or expression of these essential molecules. These changes suggest that many patients with sepsis may have insufficient TM and EPCR to convert protein C (PC) to activated PC (APC), although this is not universally true.34 In a population of patients with sepsis, plasma levels of VWF, intercellular adhesion molecule 1, and sE-selectin were significantly increased within 8 hours of the development of the acute respiratory distress syndrome (ARDS).35 Damaged endothelium and activated leukocytes and platelets also release microparticles with procoagulant properties (e.g., microparticle-associated TF).36

Apoptosis differs from necrosis because it is a highly regulated physiologic process that plays an important role in cell physiology. Apoptosis can be identified through morphologic and molecular techniques. Cytokines and inflammatory cells can damage the endothelium and induce apoptosis.24 In patients with severe sepsis, the apoptotic process also changes in space and time. That is, the apoptotic process may not occur uniformly, in all organs, at the same time, or with all types of infections.19,37 This may help to explain the commonality of ARDS in patients with severe sepsis. Endothelial cell apoptosis may produce abnormalities in cell trafficking, vasoregulation, and antithrombosis/thrombosis that may contribute to the pathogenesis of septic manifestations, such as multiple organ dysfunction, impaired microvascular blood flow, and inability of the body to restore homeostasis.

Cardiovascular System

Severe sepsis is accompanied by significant alterations in cardiovascular physiology. Sepsis-induced hypotension is multifactorial in origin. Contributory factors include redistribution of blood flow, impaired metabolic autoregulation, release of vasoactive mediators, third-spacing of fluids, dehydration, increased insensible fluid losses, vomiting, diarrhea, etc. Early in the process, systemic vascular resistance may be high and cardiac output (Q̇t) may be decreased. However, the SVR typically decreases and Q̇t increases, thus producing a hyperdynamic circulatory state.

Myocardium

Reversible myocardial dysfunction of variable severity may occur in as many as 40% of patients with severe sepsis.38 Myocardial depression often occurs in survivors very early in the septic process, progresses over the first 3 days, and then resolves after 7 to 10 days.39 Dysfunction may be systolic, diastolic, left ventricular, or biventricular. Potential explanations include myocardial injury mediated by cytokines such as TNF, release of noncytokine mediators such as prostanoids and nitric oxide (NO), catecholamine-induced myofibrillar damage, and myocardial ischemia with reperfusion injury.40 An aggregate of clinical studies suggests that a circulating myocardial depressant factor, not global myocardial ischemia, is a major contributor to myocardial dysfunction in patients with severe sepsis. In a study of coronary hemodynamics in patients with severe sepsis and septic shock, Dhainaut et al reported that coronary blood flow and net myocardial lactate consumption are increased in this disorder.41

Cardiovascular dysfunction may also manifest as ventricular dilation, abnormalities of wall motion, increased levels of cardiac troponin, and electrocardiographic changes. The explanation for the paradoxical association of a hyperdynamic circulation and myocardial dysfunction involves an interaction of compensatory mechanisms that may maintain Q̇t in the face of impaired myocardial contractility. Fluid resuscitation raises the mean systemic pressure while sepsis acts on the systemic vessels to reduce the resistance to venous return. These factors combine to maintain venous return (i.e., cardiac output) in the face of pump dysfunction (see Chapters 20 and 21). It is important to note that, by the time initial resuscitation measures are conducted, most patients with severe sepsis will exhibit high-output hypotension; when this circulatory abnormality is encountered in the critically ill, sepsis should be considered to be present if it has not already been contemplated.

The introduction of pulmonary artery catheters with thermodilution Q̇t capacity and portable radionuclide cineangiographic techniques has allowed a greater understanding of cardiovascular physiology and outcomes in patients with severe sepsis. Parker et al reported that patients who survived severe sepsis experienced an acute decrease in left ventricular EF that was accompanied by left ventricular dilatation, pathophysiologic abnormalities that resolved over a 7- to 10-day period.42 This was associated with a negative correlation between EF and left ventricular end-diastolic volume index.43 Conversely, nonsurvivors did not undergo left ventricular dilation, had preserved EF, exhibited a lower SVR, and had a positive relation between EF and left ventricular end-diastolic volume index. There are several potential explanations for this apparently paradoxical phenomenon. The lower SVR in nonsurvivors may allow them to maintain their Q̇t without the need for ventricular dilation. Alternatively, the endothelial damage and microvascular dysfunction may be more severe in nonsurvivors; this could lead to a significant microcapillary leak, severe interstitial edema of the myocardium, and impaired cardiac compliance. The inability of the heart to dilate may be a surrogate for significant multiorgan dysfunction and inability of the host to respond to the multiple alterations associated with severe sepsis.

Vascular System

Vasoconstriction is the physiologic response to low blood pressure (BP) in patients with hemorrhagic or cardiogenic shock. In contrast, shock in patients with severe sepsis is vasodilatory in nature. That is, their peripheral blood vessels not only dilate but all too often fail to show an adequate physiologic or pharmacologic contractile response to vasopressors. Severe sepsis is the most common cause of vasodilatory shock.44 However, vasodilatory shock is also a component of disorders with impaired tissue oxygenation, such as carbon monoxide intoxication, and can serve as a final common pathway of any form of profound and long-lasting shock. At the vascular level, the pathophysiology of shock in patients with severe sepsis involves three major mechanisms: (1) activation of adenosine triphosphate–sensitive potassium (KATP) channels in vascular smooth muscle cells, (2) synthesis of increased amounts of NO through induction of the inducible form of NO synthase (iNOS), and (3) vasopressin deficiency.44

Vascular Smooth Muscle Cell Membrane Hyperpolarization in Vasodilatory Shock

When the membrane potential of vascular smooth muscle cells is within the physiologic range of −30 to −60 mV, vasopressors such as norepinephrine and angiotensin II open voltage-gated calcium channels and allow calcium to enter cells; through a complex process involving calmodulin, myosin phosphorylation, and myosin ATPase, cytosolic hypercalcemia produces muscle contraction. Under conditions of tissue hypoxia or increased tissue metabolic activity, lactic acidosis activates the KATP channels. The resultant entry of potassium into the cell hyperpolarizes the vascular cell membrane, closes the calcium channels, decreases intracellular calcium concentrations, relaxes smooth muscle, and dilates the blood vessels. In patients with severe sepsis, elevated circulating levels of atrial natriuretic peptide, calcitonin gene-related peptide, NO, and adenosine can also open KATP channels and contribute to membrane hyperpolarization.45–47

Role of Nitric Oxide in Vasodilatory Shock

NO is a potent endogenous vasodilator. In patients with severe sepsis and shock, cytokines upregulate iNOS in a number of cells including vascular smooth muscle and endothelium.48 NO then presumably exerts its vasodilatory activity through activation of myosin light-chain phosphatase and perhaps by activating vascular smooth muscle cell potassium channels such as the calcium-sensitive channel. The latter would then augment the KATP-induced hyperpolarization of the membrane of vascular smooth muscle cells.

Role of Vasopressin Deficiency in Vasodilatory Shock

In addition to its role in water conservation, vasopressin secreted under baroreceptor control constricts vascular smooth muscle. During the initial phases of shock in patients with severe sepsis, vasopressin helps to maintain arterial BP. However, over time, neurohypophyseal stores of the hormone become depleted, and plasma vasopressin concentrations decline.44 Administration of vasopressin in doses large enough to produce plasma concentrations comparable to those seen in acute hypotension will raise arterial BP by 25 to 50 mm Hg.49 This marked hypersensitivity to exogenous vasopressin in patients with vasodilatory shock appears to result from a number of factors, including: (1) availability of receptors for occupancy by the exogenous hormone, (2) autonomic dysfunction, (3) ability of vasopressin to potentiate the activity of high levels of circulating endogenous catecholamines, (4) direct and indirect blunting of the contributions of NO to vasodilation, and (5) inactivation of KATP channels with restoration of vascular smooth muscle cell membrane potential toward normal. The role of vasopressin in inhibiting the effects of NO and blocking vascular KATP channels has been used clinically in patients with severe sepsis and shock who are refractory to other maneuvers to raise mean arterial BP (MAP). The efficacy of vasopressin in septic shock is currently being evaluated within the context of a multicenter, randomized controlled trial in Canada.

Respiratory System

Approximately 40% of cases of acute lung injury (ALI) are ascribed to sepsis (see Chap. 38).50 Presumably, this is a consequence of the organ's large microvascular surface area. In patients with severe sepsis of extrapulmonary origin, ALI initially occurs through an indirect process; however, as the disease progresses, secondary pneumonia, mechanical ventilation, and its associated complications may contribute to ALI and, through direct and indirect mechanisms, produce ARDS.

Systemic Components of Acute Lung Injury

The alveoli and pulmonary capillary bed are separated by two layers of cells: pneumocytes (types I and II) and endothelial cells. In patients with severe sepsis, ALI appears to begin with endothelial injury from circulating bacterial products plus cytokines released into the circulation during the initial stages of SIRS. Endotoxin has been shown to be a potent proinflammatory molecule that can induce a variety of endothelial cell inflammatory responses and produce endothelial cell dysfunction or apoptosis.51 Under the influence of IL-1, TNFα, components of complement and other endogenous mediators, the endothelial phenotype is altered. Activated endothelial cells become prothrombotic, upregulate adhesion molecules, and secrete different inflammatory mediators including chemoattractants.24 Neutrophilic leukocytes attracted to the pulmonary microvasculature become adherent and activated and migrate from the capillary lumen into the alveolar space, a process accompanied by the release of different oxidants, proteases, leukotrienes, and inflammatory mediators such as PAF. In response to alveolar wall damage, pulmonary macrophages also secrete cytokines that amplify the process. Damage to endothelium and the easily injured type I pneumocytes allows flooding of alveoli by proteinaceous edema fluid. Injury to type II pneumocytes decreases the production and turnover of surfactant, whereas the protein-rich alveolar fluid inactivates it.52,53 All of these factors contribute to the dyspnea, tachypnea, pulmonary infiltrates, and decreased ratio of arterial partial pressure of O2 (PaO2) to fractional inspired O2 (FiO2) of ALI.

The coagulopathy of sepsis is also seen at the individual organ level and can contribute to the development and progression of ALI. Compared with controls, the bronchoalveolar lavage fluid of patients with ARDS showed significant increases in the concentrations of TF and activated coagulation factor VII (p <0.001).54 Activation of the TF pathway of coagulation in the presence of large amounts of fibrinogen-rich intra-alveolar exudate can lead to extravascular fibrin deposition, hyaline membrane formation, inflammation, and pulmonary dysfunction.55 With the passage of time, this exudative stage can be replaced by proliferative and fibrotic changes that can lead to pulmonary fibrosis. Deposition and persistence of alveolar fibrin are potentiated by depression of pulmonary fibrinolytic activity.56 The fibrinolytic defect is largely the consequence of local amplification of plasminogen activator inhibitor 1 coupled with inhibition of urokinase plasminogen activator and downregulation of its receptor.57 Failure of urokinase plasminogen activator to bind to urokinase plasminogen activator receptor on pneumocytes and alveolar macrophages can impair the remodeling that is an essential element of recovery from ALI and ARDS. At autopsy, patients with ARDS also showed a predictable gamut of vascular changes that mirror those within the pulmonary parenchyma.55 For example, postmortem angiograms of patients with ARDS demonstrated the presence of serial vascular alterations that included thrombotic, fibroproliferative, and obliterative changes. The clinical significance of these lesions is signified by the development of pulmonary hypertension in patients with severe ARDS.

Mechanical Ventilation and Acute Lung Injury

Mechanical ventilation is a common supportive measure in patients with severe sepsis who develop respiratory failure. In addition to its many life-saving benefits, mechanical ventilation is potentially harmful to patients with severe sepsis and other types of critical illness (see Chaps. 37 and 43). The presence of an endotracheal tube bypasses normal airway defenses. When combined with impaired host defenses and malnutrition, patients with severe sepsis from an extrapulmonary source are at risk of a ventilator-associated pneumonia (VAP). In the absence of pre-existing disease, pneumonia may be the primary cause of ALI and ARDS. Further, high concentrations of inspired oxygen can damage alveolar membranes and worsen ALI. Moreover, barotrauma can mechanically stress alveolar and capillary walls, and shear forces generated during intratidal opening and closing of pulmonary units can damage lung tissue.58 In addition to regional effects, barotrauma can induce release of cytokines into the pulmonary parenchyma and systemic circulation. These cytokines can contribute to the development of the multiple organ dysfunction syndrome (MODS) in mechanically ventilated patients with severe sepsis.59 As an example of how dysfunctional organs may interact in the septic patient, it is thought that ALI (and mechanical ventilation) may contribute to dysfunction of other organs; through its effects on splanchnic blood flow, neurohumoral systems and proinflammatory cytokines, mechanical ventilation has the potential to adversely affect the gastrointestinal (GI) tract and predispose the patient to GI complications.60 In the Acute Respiratory Distress Syndrome Network (ARDSNet) Lower Tidal Volume Study, patients ventilated with the lower tidal volumes (6 mL/kg ideal body weight) had lower levels of IL-6, more days without organ failure, and improved survival rate.61 The complicated therapeutic considerations regarding mechanical ventilatory support in ALI are discussed later in this chapter.

Gastrointestinal System

With the decrease in the effective arterial blood volume, hypoperfusion of the gut becomes an important pathophysiologic component of severe sepsis. Early in the process, experimental evidence indicates that autoregulation of the microcirculatory blood flow is largely intact.62 Before adequate fluid resuscitation, this regulatory process diverts blood from the muscularis mucosa to the more metabolically active mucosa. Redistribution of blood flow within the gut may preserve the mucosal barrier protection and prevent bacterial translocation. However, this redistribution of blood flow may not always be sufficient to meet mucosal oxygen demand. During this time, the intracellular PH of the gut decreases and lactate levels in the portal venous blood rise. Experiments have indicated that mesenteric lymph, rather than portal blood, may be the avenue that disseminates cytokines, endotoxin and other bacterial products, and microorganisms from the compromised gut to other parts of the body. In separate experiments, Magnotti, Sambol, and their associates showed that gut-derived mesenteric lymph, not portal blood, increases endothelial cell permeability and promotes lung injury after hemorrhagic shock and that ligation of the mesenteric lymphatic duct provides long-term organ protection.63,64 However, Alverdy et al cautioned that experimental demonstrations of alterations in the permeability of the gut mucosa to substances such as labeled dextran do not indicate a cause-and-effect relation between the “leaky mucosa” and gut-derived sepsis or multiorgan failure.65 Instead, the two processes may be unrelated, and gut-derived sepsis may be a manifestation of nosocomial pathogens that express “potent virulence traits while competing for scarce resources in the hostile environment of the intestinal tract of a critically ill patient.” Whatever the mechanism of gut-derived sepsis, experimental evidence has shown that early enteral feeding protects against bacterial translocation.66 Whether or not such effects of enteral feeding are beneficial to overall patient outcomes is the point of ongoing study.

Hepatobiliary Tract

The liver plays multiple key roles in patients with severe sepsis.67 As previously described with regard to the lungs, the liver is a source and a target of inflammatory mediators. Its central role in the pathophysiology of severe sepsis is a function of the organ's blood flow and its cellular composition. In the postabsorptive state, the liver receives approximately 25% of Q̇t. Decreased perfusion is the most important event initiating hepatic dysfunction in the first hours after a septic insult.68

The three most important cellular elements participating in the liver's response to severe sepsis are Kupffer cells, hepatocytes, and endothelial cells. Although described separately, the liver's reaction to severe sepsis represents a collective and interactive response of all three cellular elements. The central role of the endothelium in sepsis pathogenesis has been discussed in detail above.

Kupffer Cells

Kupffer cells are the fixed macrophages of the liver. They represent approximately 70% of the organ's total macrophage pool. They clear the portal blood of endotoxin, bacteria, cytokines, toxins, and activated coagulation factors. Once primed, Kupffer cells secrete a variety of cytokines, including TNFα, IL-1, IL-6, IL-8, IL-12, IL-18, granulocyte colony-stimulating factor (GCSF), and granulocyte-macrophage colony-stimulating factor (GM-CSF). In a paracrine fashion, Kupffer cells modulate the response of adjacent hepatocytes to inflammatory signals through synthesis and release of acute-phase proteins and other mediators.

Hepatocytes

Hepatocytes play an integral role in the body's response to severe sepsis.67 They bear receptors for endotoxin, vasoactive substances, various inflammatory mediators, and a number of cytokines. As a result of metabolic alterations during sepsis, hepatocytes re-prioritize their metabolism toward gluconeogenesis, amino acid uptake, and protein synthesis. Alterations in protein synthesis, however, are nonuniform, with levels of albumin, PC and antithrombin (AT) decreasing and concentrations of a variety of acute-phase proteins increasing. The latter include α1-antitrypsin (α1AT), ceruloplasmin, α2-macroglobulin (α2MG), CRP, fibrinogen, C4-binding protein, and thrombin-activated fibrinolysis inhibitor (TAFI). Antiproteinases such as α1AT neutralize elastase and other proteases released by leukocytes. Ceruloplasmin and α2MG serve a scavenger function by inactivating superoxides, hydroxyl radicals, and cytokines such as IL-6. Because many of the aforementioned proteins are involved in the hemostatic response to sepsis, they are discussed in the following section. As the septic process evolves, patients may develop a secondary hepatocellular dysfunction. This is a consequence of hepatocellular inflammation that results from the local release of inflammatory mediators and damage to the hepatic parenchyma by activated neutrophils within the hepatic sinusoids.

Blood and Bone Marrow

Although dispersed in space, the blood and bone marrow comprise an organ system equal in importance to any other. During severe sepsis, the hematologic system attempts to restore homeostasis by eliminating the pathogen and isolating the infected focus. In patients with a localized infection, this process functions to the complete benefit of the host. However, in patients with severe sepsis, an exuberant response of the hematologic system can have detrimental effects. Elements of the hematologic system can be divided into cellular and fluid components.

Although the cellular elements originate in the marrow, the response of the bone marrow to sepsis is reflected in the peripheral blood. Fluid-phase elements of the hematologic system consist of various coagulation proteins. Changes in the hematologic system in patients with severe sepsis are the result of the influences of cytokines and hematopoietic growth factors, such as GM-CSF.

Cellular Response to Sepsis

Cellular elements of the peripheral blood involved in patients with severe sepsis include red blood cells, leukocytes, and platelets.69 During the initial stages of severe sepsis, the hematocrit (HCT) reflects the opposing effects of third spacing of fluids and results of aggressive fluid resuscitation. Over time, the HCT decreases due to changes in the erythron, the red blood cell component of the hematologic system. Although an infection by a hemolytic organism such as Clostridium perfringens can acutely lower the HCT, the principal change in the mass of the erythron is due to a block in the reticuloendothelial transfer of iron to erythroid progenitors, erythroid hypoplasia, and a shortened survival of red blood cells. Changes in iron metabolism sequester this essential element from species of microorganisms that require iron as a growth factor. The white blood cell count may be high, normal, or low in patients with severe sepsis. The initial leukocytosis results from recruitment of neutrophils from the marginating pool and release of the more mature granulocytic elements from the bone marrow. As the process continues, the white cells show a “left shift” as less mature granulocytic elements enter the blood. This is associated with morphologic changes such as toxic granulation, Döhle bodies, and toxic vacuolation. Leukopenia may result from migration of large numbers of granulocytes into the infected focus or failure of the bone marrow to meet the demand; leukopenia resulting from infection itself is a sign of poor prognosis. Severe sepsis is usually accompanied by a variable degree of thrombocytopenia. Thrombocytopenia is a sign of dysfunction of the hematologic system and is due to platelet adhesion and aggregation throughout the microvasculature.

Fluid-Phase Response to Sepsis

Severe sepsis is a prothrombotic state. This is the result of ongoing coagulation and impaired fibrinolysis. In response to acute inflammation, hepatic synthesis of the natural anticoagulants PC and AT decreases. At the same time, other elements of the PC system are compromised. Increased synthesis and release of α1AT and α2MG inhibit PC, whereas elevated levels of C4-binding protein bind protein S and prevent it from serving as a cofactor for activation of PC. In combination with the effects of microbial elements and cytokines, release of CRP by the liver upregulates the expression of TF on various cells. Increased production and release of fibrinogen and TAFI also contribute to the ongoing coagulopathy: the former by providing extra substrate for clotting, and the latter by inhibiting clot dissolution. The ongoing coagulopathy of sepsis is marked by a variable degree of thrombocytopenia; increased levels of D-dimer, a fibrin split product; and an elevation in the ratio of thrombin-antithrombin (TAT) to plasmin-antiplasmin (PAP). The TAT/PAP ratios are higher in patients with severe sepsis than in those who have sepsis without acute organ dysfunction.21 This rise in TAT/PAP may precede the development of organ dysfunction, suggesting that enhanced coagulation and depressed fibrinolysis have a cause-and-effect relationship with the development of the MODS.

Kidney

Different factors present during an episode of severe sepsis have the potential to damage the kidneys. These include systemic hypotension or redistribution of blood flow, renal vasoconstriction, the effects of endotoxin and cytokines on the endothelium of the renal vasculature, and activation of inflammatory cells by LPS and inflammatory mediators.70 In patients with severe sepsis, systemic vasodilation decreases the effective arterial blood volume. As the blood volume decreases, intrarenal vasoconstriction helps to maintain glomerular blood flow. Substances are released locally that promote intrarenal vasoconstriction in patients with severe sepsis and decreased effective arterial blood volume. They include endothelin, thromboxane A2, and leukotrienes.71 Eventually, however, compensatory mechanisms fail, the glomerular filtration rate decreases, prerenal azotemia develops, and acute tubular necrosis can develop. Endothelial injury from proinflammatory mediators, contents of neutrophilic granules, and components of complement can impair autoregulation of renal blood flow and may lead to microvascular thrombosis. Because the hypoperfused kidney is sensitive to nephrotoxic agents that are commonly administered in severe sepsis, these patients may be predisposed to the development of acute renal failure. For example, nonsteroidal anti-inflammatory drugs used to treat fever can inhibit the production of prostaglandins by the afferent arterioles and impair the ability of the kidney to regulate its blood flow. In the patient with severe sepsis and decreased blood volume, this may be enough to precipitate acute tubular necrosis.

Nervous System

Patients with severe sepsis have abnormalities of their central, autonomic, and peripheral nervous systems.

Abnormalities of the Central Nervous System

Septic encephalopathy is common in patients with severe sepsis. In a prospective study of 69 patients with sepsis, Young et al reported that 71% exhibited mild to marked abnormalities of cerebral function.72 The diagnosis of septic encephalopathy in a patient with severe sepsis requires evidence of extracranial infection plus impaired cerebral function. In patients without previous neurologic disease, septic encephalopathy is characterized by symmetrical neurologic findings without the asterixis, tremor, and multifocal myoclonus observed in subjects with metabolic encephalopathies of hepatic, renal, or endocrine origin.73

Delirium is perhaps the most costly and prevalent form of septic encephalopathy. This form of organ dysfunction has been shown to occur in more than 80% of all mechanically ventilated patients and is an independent predictor of poor outcomes.74 Delirium can now be diagnosed by bedside techniques in 1 to 2 minutes with a high degree of accuracy and reliability by using the Confusion Assessment Method for the ICU (CAM-ICU).

The etiology of disordered function of the central nervous system in patients with severe sepsis is multifactorial. It includes abnormalities in the blood-brain barrier, alterations in cerebral blood flow, abnormal cellular physiology, and changes in the composition of neurotransmitters in the reticular activating system. In contrast to hemorrhagic shock, in which there is no change, the blood-brain barrier is disrupted in patients with septic shock.75,76 This appears to be secondary to effects of circulating cytokines on cerebral endothelial cells. Accumulated perivascular edema fluid may impede the diffusion of oxygen and metabolic substrates, with a resultant decrease in cerebral oxygen consumption. The latter can occur despite an increase in cerebral blood flow and may also reflect the presence of mitochondrial dysfunction.77 Although severe sepsis is vasodilatory in nature, defects in the blood-brain barrier in some patients can allow vasopressors with α1-adrenergic activity to constrict cerebral vessels in patients with this disease.78 Glial and neuronal defects are also present in patients with severe sepsis. Damage to astrocytes can further impair the blood-brain barrier, disrupt the autoregulation of cerebral blood flow, disturb the transfer of metabolic substrates to adjacent neurons, and increase the susceptibility of neural elements to toxic oxygen radicals. Elevated concentrations of the breakdown products of aromatic amino acids in the brain tissue of patients with septic encephalopathy may disrupt central noradrenergic pathways.73

Role of the Autonomic Nervous System

In an experimental model of sepsis, the parasympathetic nervous system plays an important role in the body's response to sepsis. Stimulation of afferent vagal nerve fibers produces the release of corticotropin-releasing hormone (CRH), adrenocorticotropic hormone (ACTH), and cortisol.79 Subdiaphragmatic vagotomy blocks cortisol release in this situation. Stimulation of the vagus nerve prevents the onset of shock in a mouse model of endotoxemia.80 The febrile response to IL-1 can be attenuated by experimental vagotomy.81

Role of the Peripheral Nervous System and Skeletal Muscles

Abnormalities of the peripheral nervous system and neuromuscular units are a major cause of morbidity in patients with severe sepsis (see Chap. 66). Two major types of defects can be identified. The primary manifestation in this arena in survivors of severe sepsis and other types of prolonged serious disorders is electromyographic abnormalities. In a study by Fletcher et al, 95% of patients with prolonged critical illness had electromyographic changes indicative of chronic partial denervation, changes that were found up to 5 years after discharge from an ICU in more than 90% of patients who were in the unit for 28 days or longer.82 Another manifestation is myoneuropathy, a critical illness present in almost 66% of patients. Clinical manifestations can be sensory or motor. The exact etiology of this process is unknown; however, it appears to result from a combination of factors including cytokines, impaired blood flow in the vasa nervosum, extended neuromuscular blockade, immobility, compression, and disuse.83

After immediate stabilization, efforts should be directed toward determining the nature of the infection and identifying the presence of organ dysfunction. Special attention should be directed toward any predisposing condition such as trauma, surgery, organ transplantation, or immunosuppression (e.g., HIV infection, chemotherapy, and malignancy). Steps that define the nature of the infection and assess the degree of organ dysfunction include a complete history and physical examination, laboratory testing including selected microbiology procedures, and medical imaging studies.

Clinical Evaluation

Vital signs are routinely recorded and monitored, although the frequency and method of monitoring depend on the severity of the process. As a general rule, patients can be considered febrile when they have a temperature of at least 38.0°C (100.4°F) or hypothermic with a temperature no higher than 35°C (95°F). Although the presence of fever is most accurately evaluated by assessing core temperature with a bladder or intravascular thermistor, electronic probes in body orifices (e.g., ear) are acceptable in appropriate patients. Notably, temperature must be assessed in a fashion that does not facilitate the transfer of nosocomial pathogens. Importantly, fewer than 50% of febrile episodes are infectious in origin, and almost 50% of septic patients are normothermic or hypothermic.84 Variables related to fever may help in the differential diagnosis. Temperatures higher than 41.1°C are most probably noninfectious (e.g., drug fever or thyroid storm). Fever with a relative bradycardia and a rash may be drug induced or indicate Salmonella typhi or S. paratyphi infection. Vital signs can also provide a wealth of information about the patient's prognosis. In a study of community-acquired pneumonia (CAP), a temperature below 35°C or 40°C or higher, a pulse 125 beats/min or faster, a respiratory rate of at least 30 breaths/min, and a systolic BP lower than 90 mm Hg were independently associated with increased mortality rate.85

Altered mental status of the patient is an important clue to the presence of organ dysfunction. Delirium is an acute disorder of attention and cognition. It develops in 60% of older hospitalized patients and in more than 80% of mechanically ventilated patients.76,86 Delirium may be hyperactive (agitated) or hypoactive. Whereas health care workers commonly recognize the former, hypoactive delirium is commonly missed. The presence of delirium is often an important early indication of sepsis, and it can be very hazardous and prolonged in the elderly. Delirium is associated with prolonged hospital stays, institutionalization, and death. Photophobia, nuchal rigidity, papilledema, or cranial nerve palsies should direct attention to a focus of infection, such as meningitis, within the central nervous system. Orbital pain, periorbital erythema, proptosis, or unilateral rhinorrhea may be seen in a patient with bacterial or fungal sinusitis. Although this may be the presenting illness in an immunocompromised patient, it can also be a secondary infection resulting from the obstruction of sinus ostia by nasotracheal or nasogastric tubes. Fetid breath may be detected in some patients with anaerobic oropharyngeal or pulmonary infections. Oral candidiasis may be a clue to fungal septicemia, although it is more commonly a local phenomenon. Immunosuppressed and neutropenic patients may be unable to produce purulent sputum or other exudates and may be unable to vigorously cough. Tachypnea may be the only sign of pneumonia. Careful auscultation of the lung fields may detect localized rales (crackles). Patients with lower lobe pneumonia may sometimes complain of upper abdominal pain. Associated hypoactive bowel sounds in these patients may be mistakenly ascribed to an intra-abdominal focus of infection. Tenderness over the right upper quadrant may indicate the presence of a hepatic or subphrenic abscess or acalculous cholecystitis. Abdominal tenderness and hypoactive bowel sounds may be the only evidence of a localized intra-abdominal infection or peritonitis. Pain or tenderness (direct and/or rebound) localized to the right or left lower quadrant may be evidence of appendicitis or a ruptured sigmoid diverticulum, respectively. Right lower quadrant pain in a neutropenic patient (often leukemic) may indicate the presence of typhlitis, an inflammation of the cecum that may require altered medical management or surgery.

The skin should be carefully examined. Livedo reticularis and poor capillary refill are a signs of impaired cutaneous perfusion and are associated with a poor prognosis in patients with severe sepsis. In neutropenic patients, the nail folds, axillae, perianal region, and groin provide ready entry for systemic pathogens. In this subset of patients, pain and erythema may be more reliable clues to infection than the leukocyte count. The dressings around any implanted device should be removed, and the area should be carefully inspected for purulent discharge, erythema, increased warmth, pain, and crepitus. Patients with vascular catheters also should be carefully checked for evidence of vascular compromise or embolic events originating from an infected device. Sudden hypotension after access of a vascular access device or line is a clue to the presence of an infected line. Operative wound infections are the second most common type of hospital-acquired infection. All surgical dressings should be removed to allow adequate inspection of operative sites. Patients with recent GI or gynecologic surgery, diabetes mellitus, or peripheral vascular disease are at risk for invasive polymicrobial infections such as necrotizing fasciitis or synergistic gangrene. Clues to the diagnosis include swelling around the wound, rapidly advancing border, crepitus, blisters, necrosis, or the radiographic appearance of gas within the soft tissues.

Medical Imaging Studies

Although universally obtained in critically ill patients, technical limitations of plain radiographs from portable machines render them of limited value in characterizing the nature of the infection in patients with severe sepsis. For example, in the critically ill population with pneumonia, portable chest radiographs have a diagnostic accuracy of 0.5, a sensitivity of 0.6, and a specificity of only 0.28.87 Notably, chest radiographs may be essentially normal if the films are taken within the first 24 hours after the onset of symptoms or if the patient is neutropenic. Importantly, portable supine chest radiographs may be of value for other diagnostic purposes, such as estimating the patient's intravascular volume status or the nature of pulmonary edema by using the vascular pedicle width (especially when other invasive procedures are not planned).88 Therefore, imaging studies should be selected based on the presumed nature of the infection, the clinician's need for other data related to the patient's overall clinical status, and the advantages and disadvantages of various approaches.

Plain films and ultrasound have limited utility in establishing the diagnosis of sinusitis. Therefore, a clinical diagnosis of septic sinusitis should be evaluated by computed tomographic (CT) scans of the paranasal sinuses. Although the presence of air or fluid levels is abnormal, the diagnosis is not established without recovery of infected material by sinus puncture, aspirate, and culture. A flat plate of the abdomen may demonstrate the presence of an ileus, a potential explanation for bacteremia in a patient without localizing findings. Abdominal radiographs may also show evidence of a perforated viscus or an intra-abdominal abscess. Ultrasonography of the gallbladder may demonstrate evidence of calculous or acalculous cholecystitis or of biliary tract obstruction in a patient with biliary sepsis. Ultrasound or CT studies of the abdomen may demonstrate localized fluid collections, abscesses, ureteral dilation or a perinephric abscess.

Microbiology

All potential infected foci should be cultured in patients with severe sepsis. For example, the presence of pleural fluid adjacent to an infiltrate on a chest radiograph may provide valuable diagnostic information regarding the cause of a patient's severe sepsis. However, in 20% to 30% of patients, a definite site of infection cannot be identified.89 The search for the responsible pathogen should include all measures necessary to establish the site and nature of infection, including direct examination and culture of material from the lower respiratory tract, fine-needle aspirates, and cultures of infected lines. The blood is also an appropriate material to culture. However, 70% of patients with sepsis will not have organisms recovered from the blood.89 Material obtained for culture should be immediately gram stained. Cultures should also be supplemented with appropriate immunologic procedures such as detection of microbial antigens in body fluids and immunostaining. In patients with intra-abdominal infections, wound infections, and lung abscesses, the material should also be cultured anaerobically.

Laboratory Markers of Sepsis

Several biochemical markers have been used to identify the patient with systemic activation of the cytokine cascade. None of the markers are currently widely used to diagnose sepsis, although some clinicians have argued that they should be incorporated into the entry criteria for patients in clinical trials of severe sepsis. Elevated levels of IL-6, IL-1, IL-8, TNFα, and monocyte chemotactic proteins 1 and 2 are present in septic patients and often correlate with the severity of the disease. Levels of IL-6 are increased in most patients with sepsis and have been shown to correlate with prognosis, but IL-6 is not used in clinical practice for diagnosis or prognostication.90 Although the level of IL-6 may be able to differentiate cases of infectious from noninfectious SIRS, no correlation has been established between the concentration of IL-6 and outcomes.

CRP is an acute-phase reactant synthesized in the liver. After exposure to IL-1, IL-6, or TNFα, CRP levels increase. Levels of CRP have been used for years as a diagnostic marker of systemic inflammation. Whereas levels are normally low, they rise rapidly in patients with sepsis, paralleling the course of infection.91 The complement system is activated in patients with sepsis; this may be mediated in part through CRP. Interactions of the complement system with microorganisms generate increased amounts of C3a and terminal components of the cascade. Elevated levels of C3a have been found to be one of the single and most sensitive and specific laboratory markers that can differentiate between sepsis and noninfectious causes of SIRS.90

PCT is derived from the preprohormone, pre-PCT. Although PCT levels are elevated in patients with sepsis, little is known about its source and function. Interestingly, calcitonin levels are normal in this population. In a porcine model of polymicrobial sepsis, immunoneutralization of calcitonin precursors was found to attenuate the adverse physiologic response to the infection.92 For diagnostic purposes in critically ill patients, PCT levels have a better predictive value for sepsis than do IL-6 or CRP. When PCT concentration was combined with levels of C3a in a “sepsis score,” the combination had a sensitivity of 91% and a specificity of 80% for sepsis versus noninflammatory SIRS.90 As a prognostic marker, PCT levels have been shown to correlate with mortality rate, although the clinical utility of this prognostic marker remains uncertain at this time.93–95

Severe sepsis is a medical emergency. The first priority should be to assess and address abnormalities in the “A,B,Cs”: airway, breathing, and circulation. In many instances, the clinical assessment of tissue perfusion and response to therapy can be aided by monitoring devices and laboratory measurements (Table 46-4). Once immediate stabilization has been accomplished, the source of the infection should be established and controlled, and specific antimicrobial agents should be administered. After stabilization, source control, initiation of appropriate antimicrobial agents, and further support for dysfunctional organs, disease-specific interventions should be considered. The Surviving Sepsis Campaign is a collaborative effort by the SCCM, European Society of Intensive Care Medicine, and the International Sepsis Forum. The group's primary aim is to decrease the relative mortality rate of sepsis by 25% over a 5-year period. To reach this objective, the group has a number of goals. One of these is to improve the therapy of patients with severe sepsis. To achieve this, they reviewed published literature and achieved consensus on a number of management issues for patients with severe sepsis. These are described in the published Surviving Sepsis Campaign Practice Guidelines and listed in Table 46-5 and Figure 46-6.

The initial therapeutic intervention in patients with severe sepsis is to reverse organ hypoperfusion. Patients with severe sepsis often have a relative intravascular hypovolemia. As a consequence, initial therapy often consists of the rapid administration of large amounts of fluids.

Fluid Resuscitation

Large amounts of fluids may be required to restore tissue perfusion and oxygen delivery. In some instances, correction of large fluid deficits may require at least 6 to 10 L of crystalloids.96 Fluid resuscitation is often titrated by the responses of clinical end points of heart rate, BP, and urine output.97 Appropriate amounts of fluid will improve the cardiac index by 25% to 40%. In approximately 50% of patients with severe sepsis who present with hypotension, fluid resuscitation alone will normalize BP and restore hemodynamic stability. Hemodynamic monitoring should be considered for patients who do not respond rapidly to the initial fluid challenge and those with a history of coronary heart disease.

Choice of Fluids

There has been an ongoing debate as to whether fluid resuscitation is best accomplished with crystalloids or colloids. Choi et al conducted an evaluation of randomized clinical trials of crystalloids versus colloids in adult patients requiring fluid resuscitation.98 They reported that there were no apparent differences in the incidences of pulmonary edema, mortality rate, or length of stay in patients who received crystalloid or colloid fluid resuscitation. In contrast, Schierhout et al compared crystalloid with colloid resuscitation in 26 clinical trials using stringent selection criteria.99 They reported that resuscitation with colloids was accompanied by an increased absolute risk of mortality of 4% (95% confidence interval = 0% to 8%). Because colloids were not associated with improved survival rate and are significantly more expensive than crystalloids, they concluded that the evidence does not support the continued use of colloids for volume replacement in the critically ill. The crystalloid versus colloid controversy continues.100 However, in the presence of normal oncotic pressure, most patients are managed with crystalloids; patients requiring osmotic support can also be treated with colloids such as 25% albumin.100

Pyruvate is a potent antioxidant and free radical scavenger.101 Although it is unstable in aqueous solution, its ethyl derivative has greater stability. Ringer ethyl pyruvate has been evaluated in different preclinical models and has been shown to improve survival rate and decrease inflammatory markers in acute endotoxemia, acute bacterial peritonitis, hemorrhagic shock, and mesenteric ischemia/reperfusion injury.102 There is some basic science and clinical evidence that suggests that infusions of small amounts of hypertonic saline may have some future role in the resuscitation of patients with severe sepsis. In 1980, Velasco et al reported that administration of a hypertonic solution of 7.5% saline had beneficial effects in a canine model of hemorrhagic shock.103 Experimental studies in models of severe sepsis confirmed beneficial effects similar to those originally reported. Small clinical studies have also reported beneficial effects on cardiac indices, SVR, and oxygen transport.104,105 Although it was initially proposed that the beneficial effect of this intervention was on cardiac preload, increasing evidence suggests that its effects extend far beyond the osmotic properties of hypertonic saline. Hypertonic saline also appears to improve myocardial contractility, reduce endothelial and interstitial edema, improve microcirculatory blood flow, and exert immunomodulatory activity.106 Its effects on endothelial cell function and microvascular blood flow are particularly interesting in light of the growing evidence of endothelial cell dysfunction in patients with severe sepsis. Endothelial cell edema can serve as a surrogate for endothelial injury and potential dysfunction. Edema of the endothelial cells occurs in the initial phases of hypovolemia. Small infusions of hypertonic saline can reduce endothelial volume by approximately 20%, increase tissue perfusion, and improve cellular function.107 Microcirculatory blood flow is also improved because the hypertonic fluid produces hemodilution and decreased blood viscosity.

Patients with severe sepsis very often have dramatic decreases in hemoglobin (Hb) levels. This is secondary to a number of factors including hemodilution and impaired erythropoiesis. Further, fluid resuscitation may decrease Hb levels by up to 1 to 3 g/dL.97 This degree of anemia is tolerated by most patients because the resultant decrease in blood viscosity reduces afterload and improves venous return. Potential benefits of transfusion in this population include an increase in arterial oxygen content and an improvement in systemic oxygen availability. However, transfusions of packed red blood cells have several potentially detrimental effects. These include rheologic changes that decrease red blood cell deformability and increase blood viscosity, immunosuppression, disease transmission, risk of nonhemolytic transfusion reactions, and the fact that oxygen delivery is not immediately restored after transfusion because the infused red blood cells must restore their depleted 2,3-diphosphoglycerate levels to improve oxygen release.

Hebert et al conducted a multicenter, controlled trial of restrictive versus liberal transfusion strategies in a population of critically ill patients.108 They enrolled 838 euvolemic, critically ill patients who had Hb concentrations below 9.0 g/dL within 72 hours of admission to the ICU. Patients were then randomized to a restrictive transfusion strategy in which Hb concentrations were maintained within the range of 7.0 to 9.0 g/dL and transfusions were administered only if the Hb dropped below 7.0 g/dL or to a strategy in which Hb levels were maintained in the range of 10.0 to 12.0 g/dL and transfusions were given if the Hb dropped below 10.0 g/dL. Even though the overall 30-day mortality rate was similar in the two groups, patients who were less severely ill as stratified by a score no higher than 20 on the Second Acute Physiology and Chronic Health Evaluation (APACHE II) and patients who were younger than 55 years had a significantly lower in-hospital mortality rate (22.2% versus 28.1%; p = 0.05). This benefit did not extend to patients with acute coronary syndromes. However, Rivers et al recently released the results of a clinical trial of early goal-directed therapy (EGDT) that suggest that there may be some benefit of Hb concentrations of 10 g/dL or greater, at least as part of a strategy of early resuscitation of patients with severe sepsis and septic shock.109 Although these results are discussed below, they emphasize the importance of proper tissue oxygenation.

At this time, the optimal Hb level for patients with severe sepsis is unclear. The default transfusion threshold currently used by many ICUs is a conservative one of 7.0 g/dL, unless patients are elderly and have a history of myocardial infarction, in which case a more liberal threshold of 10 g/dL is supported by an 80,000-patient observational study by Wu et al.110

Cardiorespiratory Support

Vasopressors

Patients who remain hypotensive despite aggressive fluid resuscitation are candidates for pressor support; however, vasopressors should not be considered in lieu of adequate fluid resuscitation. BPs determined by a cuff are often inaccurate, and an arterial cannula is usually inserted to monitor intra-arterial BP in patients receiving vasopressor therapy. Many agents with different pharmacodynamic properties are available to treat the hypotensive patient with severe sepsis (Table 46-6). Patients receiving pressor support should be monitored to ensure that treatment restores perfusion to vital organs without compromising stroke volume. Surrogates of improved peripheral perfusion and splanchnic blood flow include an increase in MAP (≥65 mm Hg is often used), warming of the extremities, and resolution of cutaneous mottling and impaired capillary refill. Increased urinary output and improved mentation are other good surrogates of adequate systemic perfusion. Goals should be a urine output greater than 0.5 mL/kg per hour and an alert and oriented cognitive status.

One of the obstacles to determine the “best” pressor in patients with severe sepsis is the lack of large-scale, randomized controlled trials. Most studies of various pressor agents have been observational in nature. Norepinephrine appears to be the preferred agent for initial BP support in patients with severe sepsis who remain hypotensive despite an adequate fluid challenge. Because of its pure α-adrenergic activity, phenylephrine may be useful in patients with coronary heart disease who may be at risk for tachycardia or arrhythmias from agents with β-agonist properties. Meadows et al evaluated the effects of norepinephrine on reversal of shock in 10 patients with severe sepsis.111 Patients with severe sepsis who remained hypotensive and oliguric despite expansion of their plasma volume to a target Paop and infusions of increasing doses of dopamine and dobutamine were then treated with norepinephrine monotherapy. After discontinuation of dopamine and dobutamine, an infusion of norepinephrine significantly elevated the MAP, reversed the hypotension, and increased the SVR, left ventricular stroke work index, and urinary output without significantly affecting heart rate. In a separate study, Martin et al evaluated the effect of norepinephrine on the outcome of patients with septic shock.112 These investigators evaluated the effects of various independent variables that might affect outcome in an observational cohort of 97 adult patients with severe sepsis. Patients who received norepinephrine as a component of their hemodynamic support had a significantly lower hospital mortality rate than did those treated with large-dose dopamine and/or epinephrine (62% versus 82%; 95% confidence interval = 0.68, range = 0.54 to 0.87; p <0.001). Further support for the use of norepinephrine as the pressor of choice was provided by Di Giantomasso et al.113 They evaluated the effects of norepinephrine on vital organ blood flow in a sheep model. They concluded that norepinephrine infusion does not induce ischemia in vital organs in normal mammalian circulation. In addition, infusion of norepinephrine significantly increases coronary and renal blood flow, with associated improvements in urine output and serum creatinine.

Over a short period, tachyphylaxis may appear as vascular smooth muscle becomes resistant to the effects of norepinephrine and other catecholamines. Vasopressin is involved in cardiovascular homeostasis. Normally, it plays only a minor role in BP control. However, in hypotensive patients, vasopressin released under baroreflex receptor control stimulates vascular smooth muscle and helps restore BP. During the initial phases of severe sepsis-induced hypotension, high levels of vasopressin contribute to the effects of endogenous catecholamines in attempting to restore BP. However, as shock worsens, plasma concentrations of vasopressin decline significantly and become inappropriately low for the patient's MAP, presumably because neurohypophyseal stores of the hormone are depleted in the face of constant baroreceptor stimulation. In support of this hypothesis, Landry et al demonstrated by immunohistochemistry in a canine model that, after 1 hour of hemorrhagic shock, the stainable hormone was essentially absent in tissue from the animal's neurohypophysis.44 In patients with severe sepsis and other forms of vasodilatory shock, administration of vasopressin to produce concentrations similar to those identified during acute episodes of hypotension increases arterial BP by 25 to 50 mm Hg. It has been postulated that the marked sensitivity of BP to vasopressin in this situation is multifactorial in origin: (1) in patients with severe sepsis and established hypotension, plasma levels of vasopressin are low and its receptors on vascular smooth muscle are unoccupied; (2) patients with severe sepsis may have dysfunction of the sympathetic nervous system and be more responsive to noncatecholamine pressor agents; (3) plasma concentrations of norepinephrine are markedly elevated in patients with vasodilatory shock, and vasopressin potentiates the effects of the catecholamine; (4) vasopressin directly activates KATP channels in vascular smooth muscle, and; (5) vasopressin directly and indirectly inhibits the vasodilatory activity of NO.44 Studies have supported the role of vasopressin in this situation. Landry et al reported that an infusion of 0.04 U/min of vasopressin in 10 hypotensive patients with severe sepsis increased arterial BP from a mean of 92/52 mm Hg to 146/66 mm Hg.114 Importantly, however, withdrawal of vasopressin resulted in hypotension in six patients in whom it was the only pressor. In a blinded study of 24 patients with septic shock, Patel et al compared the effects of norepinephrine and vasopressin.115 Doses of vasopressin and norepinephrine, starting at 0.01 U/min and 2 μg/min, respectively, were titrated to maintain a clinically acceptable MAP. Treatment with vasopressin produced an important catecholamine-sparing activity (p <0.001), and its use was accompanied by a larger volume of urine output and an improved creatinine clearance (p <0.05). Although these preliminary observations require a large-scale randomized trial to confirm the beneficial effects of vasopressin in patients with vasodilatory shock, critical care specialists are commonly treating patients with severe sepsis who are refractory to catecholamines with vasopressin.

Inotropic Agents

Patients with severe sepsis have a normal to low BP, elevated cardiac index, and decreased SVR. However, myocardial depression is common.97 It is characterized by a decreased EF, ventricular dilation, impaired contractile response to volume loading, and a low ratio of peak systolic pressure to end-systolic volume. In patients who are unable to maintain an adequate cardiac index, MAP, mixed venous oxygen content, and/or urine output, an inotropic agent such as dobutamine may be useful.109 However, because inotropic agents can produce clinically significant tachycardia in patients who have not been adequately resuscitated with fluid, it is important to ensure that the dobutamine candidate has adequate or increased intravascular volume. In addition, the increased myocardial oxygen consumption in patients treated with an inotrope can produce angina or an acute coronary syndrome.

Oxygen and Mechanical Ventilation

Sepsis places significant stress on the respiratory system. At presentation, most patients are tachypneic and hypoxemic. These clinical manifestations result from a combination of factors including increased oxygen requirements, decreased pulmonary compliance, increased airway resistance, and impaired efficiency of the respiratory and other muscles.89 Approximately 88% of patients with severe sepsis will require ventilatory support, generally for 5 to 10 days. Therefore, the patient's arterial oxygen saturation (SaO2) and/or the PaO2 should be evaluated and steps taken to maintain the SaO2 at 88% or greater by increasing the amount of FiO2 or improving respiratory physiology with institution and adjustments of mechanical ventilation. Prone positioning has the potential to improve oxygenation in critically ill patients with severe sepsis. Position-related changes in respiratory physiology are a function of changes in end-expiratory lung volume, ventilation-perfusion matching, and chest wall mechanics. However, although the maneuver improves oxygenation, Gattinoni et al were unable to demonstrate an advantage with respect to mortality rate of ARDS patients when using a protocol of prone positioning for 6 hours/day over 10 days.116 The disconnect between improved oxygenation and mortality rate in this study may be a function of several trial design considerations, including statistical power, low severity of illness of many of the enrolled patients, and the duration of prone positioning employed.117

Most patients with severe sepsis will require an FiO2 greater than 50%. Although high-flow oxygen through a face mask can provide an FiO2 in the 70% range, in general, patients with severe sepsis requiring high levels of oxygen supplementation should be treated with positive pressure ventilation. Importantly, the clinician should be proactive and recognize the signs of impending respiratory failure, so that patients can be intubated early and receive respiratory support before they aspirate, develop hypoxemic encephalopathy, or develop a potentially fatal arrhythmia. During intubation and immediately thereafter, the patient should receive an FiO2 in the 95% to 100% range to maximize SaO2. However, to avoid potential oxygen toxicity through the formation of free radicals and other mechanisms, the FiO2 should be tapered to no greater than 50% as soon as possible, and SaO2 should be maintained above 88%.

Oxygen delivery (QO2) is a function of Q̇t, Hb level, and SaO2 (QO2 = 10 × Q̇t × [1.34 Hb × SaO2 + 0.003 × PaO2]). If filling pressures are low, the Q̇t can be raised by fluids; alternatively, patients who do not respond to fluids and have normal or elevated filling pressures may benefit from an inotrope. Although it is important that QO2 be adequate, the therapy required to achieve supranormal QO2 may be injurious in some patients.118 In a prospective randomized study of 109 critically ill patients, Hayes et al attempted to maximize DO2 by volume expansion.119 Patients failing volume expansion alone were randomized to a control group or a treatment group. The intervention group was treated with intravenous dobutamine (5 to 200 μg/kg per minute). All nine patients who responded to volume replacement alone survived to leave the hospital. The predicted risk of death for the remaining 100 patients was 34%. However, the in-hospital mortality rate was lower in the control group (34%) than in the supranormal QO2 (54%) group, suggesting that the aggressive efforts to increase tissue QO2 were detrimental.

One reasonable ventilatory strategy in patients requiring mechanical ventilation is to employ assist-control ventilation to provide 75% to 100% of the minute ventilation. Positive end-expiratory pressure can be employed to reduce the risk of ventilator-induced injury and reduce the FiO2 required to maintain the SaO2 at 88% or greater. The results of the ARDSNet Tidal Volume Study indicated that a lung protective strategy using tidal volumes of 6 mL/kg (versus 12 mL/kg) can significantly improve outcomes in patients with ALI or ARDS.61 Patients who were randomized to the lower tidal volume had a significantly lower mortality rate (31.0% versus 39.8%; p = 0.007), more days without a ventilator (12 versus 10; p = 0.007), more days without organ failure (15 versus 12; p = 0.006), and lower levels of the proinflammatory cytokine, IL-6. It is important that every effort be made to reduce mechanical ventilation settings to detect readiness for extubation from the ventilator. The spontaneous breathing trial is the best indicator of the patient's ability to breathe without mechanical support.120 Ely et al reported that a spontaneous breathing trial protocol using respiratory therapists and nurses significantly reduced the number of days that patients spent on a ventilator (1 versus 3; p <0.001), decreased the complication rate by half, and cut the costs of ICU care by more than $5000 per patient.121,122

Importance of Early Goal-Directed Therapy

Rivers et al conducted a randomized, nonblinded trial of standard therapy versus early goal-directed therapy (EGDT) in patients with SIRS criteria and systolic BP no higher than 90 mm Hg and who were unresponsive to 20–30 mL/kg of crystalloid or had a lactate concentration of 4 mmol/L or greater.109 Patients were randomized into two groups and treated by protocol. Hemodynamic goals were identical in the two groups: central venous pressure between 8 and 12 mm Hg, vasopressors for a systolic BP of 90 mm Hg or lower, urinary output maintained at 0.5 mL/kg per hour or greater, and a MAP established at 65 mm Hg or higher. Patients in the group randomized to EGDT also received an oximetric central venous catheter capable of measuring the saturation of blood in the superior vena cava or right atrium (ScvO2) and were treated early in the emergency department according to protocol for at least 6 hours before they were transferred to the first available inpatient bed. If the ScvO2 of patients in the EGDT group was less than 70%, they received transfusions of packed red blood cells until their HCT was at least 30%. If the ScvO2 was still less than 70%, patients were treated with dobutamine up to a dose of 20 μg/kg per minute. If necessary, mechanical ventilation and sedation were employed to raise ScvO2 to the goal concentration, and interventions were directed at lowering oxygen consumption and, hence, raising the saturation of blood returning to the right heart.

Patients treated with EGDT received approximately 1.5 L of additional fluid during the first 6 hours (5000 versus 3500 mL); they also received significantly more transfusions of packed red blood cells (PRBCs; 64.1% versus 18.5%; p <0.001). Vasopressor support was not significantly different in the two groups during the first 6 hours; however, the EGDT group received significantly fewer vasopressors after 7 hours than did the standard group (p = 0.03). Whereas fewer than 1% of the standard therapy group received an inotrope, dobutamine was administered to approximately 14% of the EGDT group. Early EGDT significantly affected outcomes. Patients treated in the EGDT arm of the protocol had a significant benefit in mortality rate at 28 days (30.5% versus 46.5%; p = 0.009) and at 60 days (44.3% versus 56.9%; p = 0.03). The number needed to treat (NNT) to prevent one death with EGDT was six to eight patients. The results of this study indicated the importance of early identification of patients with severe sepsis at high risk for cardiovascular collapse and implementation of measures designed to restore equilibrium between oxygen demand and supply. The specific features of the protocol that contributed to the benefit with respect to mortality rate (e.g., the relatively large quantity of PRBCs and dobutamine used in the intervention group and the titration of ScvO2 using oximetric central venous catheters) require further study.

Infection Management

Infection management consists of source control and the administration of anti-infectives that cover the spectrum of microbial organisms suggested by the patient's initial presentation. Importantly, appropriate cultures should be obtained; in some cases, this is accomplished during the process of source control (e.g., removal of an infected intravascular device).

Source Control

The term source control has replaced surgical intervention to describe measures to control the source of infection. This change in terminology is intended to highlight the fact that interventions falling into this category are not limited to surgery. Source control can be divided into three broad categories: (1) drainage of an abscess, (2) débridement of devitalized and infected tissue, and (3) removal of foreign bodies colonized by pathogens and diversion, repair, or excision of an infected focus in a hollow viscus.123 Attention directed to source control should be included in the initial management of all patients with severe sepsis. Whereas septic foci resulting from a ruptured appendix or a perforated duodenal ulcer may be obvious, other presentations may be subtle. For example, a patient with CAP who remains septic may have an associated empyema. The timing of source control measures must take the patient's clinical condition into account. For example, removal of an infected central line should be performed immediately. Alternatively, transferring an unstable patient to the medical imaging suite, where monitoring may not be as intense, for percutaneous drainage might be best deferred until resuscitative measures have restored cardiopulmonary homeostasis. In other instances, surgical intervention should be early in aggressive disease processes in which early intervention clearly leads to improved outcomes such as necrotizing fasciitis.

Anti-Infective Therapy

Several general principles govern antibiotic therapy of the patient with severe sepsis (Table 46-7). Antibiotic therapy is delayed in approximately 10% of cases, a situation responsible for a mortality rate that can be 10% to 15% higher than that in patients treated early.124,125 Iregui et al evaluated the basis and consequences of delayed appropriate antibiotic therapy in a study of 107 consecutive patients with VAP.126 In 75.8% of patients, delays of 24 hours or longer were due to the simple failure to write the order for antibiotics after the diagnosis was made. Microbial resistance could explain only 18.2% of the cases in which therapy was classified as inappropriate. The initial delay of appropriate antibiotic therapy was associated with a significant 41.1% increase in hospital mortality rate from VAP (p <0.01). Kreger et al evaluated the epidemiology and response to therapy in 612 patients with gram-negative bacteremia.127 They reported that appropriate antibiotic therapy reduced the frequency at which shock developed by approximately 50%. In addition, even after shock had developed, appropriate antibiotic therapy significantly decreased mortality rates. As essential as antimicrobial therapy is, the mortality rate of patients with severe sepsis treated with appropriate antibiotics remains unacceptably high. Pittet et al analyzed bedside predictors of mortality rate in 176 ICU patients with bacteremic sepsis and found that appropriate antibiotic therapy reduced mortality rate by 11%.124 Nevertheless, in-hospital mortality rate remained at an unacceptable 43%. Kollef et al reported even stronger evidence for the importance of appropriate antibiotic therapy.128 The infection-related mortality rate for patients receiving inadequate antimicrobial therapy was significantly greater than that seen in patients treated appropriately (42.0% versus 17.7%; p <0.001). Analysis of their data in a logistic regression model indicated that inadequate antimicrobial therapy of infection is the most important independent determinant of hospital mortality rate in critically ill patients admitted to a medical or surgical ICU. Table 46-8 lists some general guidelines for the selection of initial antibiotics for patients with severe sepsis.

Pneumonia

Major potential pathogens in patients with serious CAP include S. pneumoniae, Haemophilus influenzae, and Legionella species (see Chap. 51). Initial therapy can be an extended-spectrum cephalosporin or a β-lactam/β-lactamase inhibitor combination plus a macrolide (azithromycin, clarithromycin, or erythromycin) or an antipseudomonal fluoroquinolone (ciprofloxacin, levofloxacin, or gatifloxacin) alone. If there are concerns about infection by a strain of drug-resistant S. pneumoniae, the β-lactam should be cefotaxime or ceftriaxone. Notably, some researchers have advised against fluoroquinolone monotherapy in this situation because most therapeutic trials studying these drugs excluded seriously ill patients.129 For patients with structural disease of the lung (e.g., bronchiectasis), an antipseudomonal regimen would be an excellent one. If aspiration is suspected, treatment should be clindamycin, metronidazole, or a β-lactam/β-lactamase inhibitor combination, with or without fluoroquinolone. Patients allergic to β-lactams can be treated with a fluoroquinolone with or without the addition of clindamycin. The most probable organisms in a patient with nosocomial pneumonia are S. aureus and aerobic gram-negative rods. A possible initial therapy for these patients might be an antipseudomonal β-lactam (cefepime, imipenem/cilastatin, meropenem, or piperacillin/tazobactam) plus an intravenous antipseudomonal fluoroquinolone such as ciprofloxacin.

Intra-Abdominal Infections

In serious community-acquired intra-abdominal infections, the major pathogens are most likely to be E. coli, Klebsiella pneumoniae, and Bacteroides fragilis (see Chap. 89). Initial treatment in the ICU to cover these organisms may consist of imipenem/cilastatin, ampicillin/sulbactam, or piperacillin/tazobactam with or without an aminoglycoside. The spectrum of potential pathogens increases in patients with nosocomial intra-abdominal infections to include aerobic gram-negative rods, including those that might produce an extended-spectrum β-lactamase, other anaerobes, and Candida species.

Skin and Soft Tissue

In patients with severe sepsis from community-acquired skin and soft tissue infections, the most common pathogens are group A streptococci, S. aureus, Clostridium species, polymicrobial organisms, various enteric gram-negative rods, and Pseudomonas aeruginosa (see Chap. 55). The risk that the infecting organism will be a gram-negative rod increases in patients who are neutropenic. In these patients, initial treatment in the ICU might consist of vancomycin with imipenem/cilastatin or piperacillin/tazobactam. The most probable pathogens in patients with serious nosocomial skin and soft tissue infections are S. aureus (including methicillin-resistant S. aureus) and aerobic gram-negative rods. A reasonable initial antibiotic regimen in these patients might consist of vancomycin plus cefepime. Patients with vancomycin-resistant organisms can be treated with the streptogramin combination, quinupristin/dalfopristin or the first oxazolidinone, linezolid. If microbiological studies demonstrate that a Staphylococcal strain is not resistant to methicillin, nafcillin should be administered instead of vancomycin because of its comparatively superior efficacy and to prevent the emergence of vancomycin-resistant strains. This allows vancomycin to be reserved for infections resistant to other agents.

In patients presenting with severe sepsis from a cutaneous or soft tissue infection with diabetes mellitus, peripheral vascular disease or neutropenia, or a history of recent GI or gynecologic surgery, the index of suspicion should be high for potentially fatal necrotizing fasciitis. These infections are usually polymicrobial and are produced by a combination of anaerobic organisms and aerobic gram-negative rods. This disorder has a high mortality rate unless treated aggressively. Percutaneous aspirates or biopsies should be obtained, gram stained, and cultured aerobically and anaerobically. If crepitation is detected or radiographs show gas in the deep tissues, the wound and surrounding tissues should be immediately débrided surgically.

Urosepsis

The most probable organisms in patients with severe community-acquired urosepsis are E. coli, Klebsiella species, Enterobacter species, and Proteus species (see Chap. 56). Reasonable antibiotic selections for urosepsis include a parenteral fluoroquinolone, an aminoglycoside with or without ampicillin, or an extended-spectrum cephalosporin with or without an aminoglycoside. If a gram stain of the urine demonstrates the presence of gram-positive cocci, a rational choice for initial therapy would be ampicillin/sulbactam with or without an aminoglycoside to cover enterococci. A reasonable initial antibiotic combination in patients presenting with this diagnosis would be ciprofloxacin with or without an aminoglycoside. Although bacteriuria or candiduria is common in catheterized patients in the ICU, an indwelling catheter is uncommonly the cause of urosepsis. Nosocomial urosepsis is most common in patients with a urinary tract obstruction or those who have recently undergone instrumentation of the lower urinary tract. Likely pathogens in patients with serious nosocomial urosepsis are aerobic gram-negative rods and enterococci. Unless the prevalence of vancomycin-resistant enterococci is high in the institution, a reasonable initial choice of antibiotics would be vancomycin plus cefepime.

Meningitis

The most probable organisms in patients with serious community-acquired meningitis are S. pneumoniae, Neisseria meningitidis, Listeria monocytogenes, E. coli, and H. influenzae. Initial treatment of meningitis should be guided by results of the cerebrospinal fluid examination, including examination of a gram stain of an ultracentrifuged specimen of cerebrospinal fluid, plus tests for antigens of S. pneumoniae, type B H. influenzae, and N. meningitidis in patients with purulent meningitis (see Chap. 52). Without a probable pathogen from initial microbiological studies of the cerebrospinal fluid, a reasonable initial treatment regimen for adults 18 to 50 years old would be a broad-spectrum cephalosporin. For patients older than 50 years or those with impaired cell-mediated immunity, ampicillin plus ceftazidime would be appropriate. Organisms that should be suspected in cases of nosocomial meningitis are Pseudomonas aeruginosa, E. coli, Klebsiella species, and Staphylococcal species. A reasonable initial treatment in these cases would consist of cefepime or meropenem plus vancomycin.

Line Sepsis

Readily accessible, stable vascular access is a routine requirement in critically ill patients and chemotherapy recipients. The most common pathogens in patients with an infected line are coagulase-negative staphylococci, S. aureus, aerobic gram-negative bacilli, and Candida albicans130 (see Chap. 50). Infections of these devices are a major cause of morbidity and mortality in the United States. In a meta-analysis of 2573 cases of line sepsis, Byers et al reported an overall case fatality rate of 14%.131 Of the various pathogens, S. aureus and S. epidermidis had the highest and lowest attributable mortality rates, respectively.

Management varies by the type of catheter. In most cases, infected, non-tunneled catheters should be removed and replaced. The decision to replace a tunneled catheter or implanted device is based on documentation that the device is infected, the type of pathogen, and the presence of complications such as septic thrombosis, metastatic infection, or tunnel infection. Until the susceptibility is known, staphylococcal strains should be treated with vancomycin. If the isolate is susceptible to β-lactams, a semi-synthetic penicillin such as nafcillin should be used. Patients who are allergic to penicillin and lack a history of anaphylaxis or angioedema can be treated with a first-generation cephalosporin such as cefazolin; 90% of patients treated in this fashion will not have an allergic response.130 Empirical antibiotic therapy for patients with gram-negative line sepsis should include drugs active against P. aeruginosa. Patients with severe sepsis due to systemic candidiasis have several options, including amphotericin B, fluconazole, voriconazole, and caspofungin. Often patients at high risk of Candida species infection and no positive cultures will be treated with presumptive therapy if they are doing poorly on a well-chosen antibiotic regimen.

Despite early institution of appropriate antibiotics and organ support measures, 28% to 50% or more of patients with severe sepsis die. This has stimulated basic scientists and clinical investigators to seek and develop unique compounds that can serve as specific antisepsis interventions (Table 46-9). As with any new intervention, however, an evidence-based approach should be taken to ascertain whether the data, in aggregate, are sufficient to change clinical practice.132 The analysis should include the quality of the evidence, magnitude of treatment effect, precision of treatment effect, magnitude of hazards, cost-benefit analysis, and various intangibles.

The magnitude of treatment effect can be expressed as absolute risk reduction (ARR), relative risk reduction, or NNT. The NNT represents the number of patients who must be treated with a specific intervention to prevent one adverse outcome or to attain one additional benefit. Clinicians can use the NNT, calculated as the reciprocal of the ARR, or 1/ARR, as a tool to guide them in reaching rational therapeutic decisions. Although there are few large-scale, randomized, placebo-controlled trials of specific antisepsis interventions to use as comparators, information from a large database of interventional trials in clinical cardiology has provided measures of the NNT in various medically accepted interventions. Thrombolytic therapy, for example, provides a survival benefit for patients with acute myocardial infarction. An analysis of nine randomized, placebo-controlled trials, each with 1000 patients or more with suspected acute myocardial infarction, demonstrated that thrombolytic therapy saved one life for every 55 patients treated (NNT = 55).133 In the Thrombolysis in Myocardial Infarction 11B trial of patients with unstable angina pectoris or non–Q-wave myocardial infarction, enoxaparin prevented one composite end point (death, nonfatal myocardial infarction, or severe recurrent ischemia requiring urgent revascularization), compared with unfractionated heparin, for every 40 patients treated (NNT = 40).134 Clinicians can use the NNT as a yardstick to make a relative comparison among the various interventions described below.

Activated Protein C

PC is a vitamin K–dependent protein that circulates in the blood as an inactive precursor. Conversion of PC to APC requires thrombin and TM. Once thrombin is bound to endothelial cell TM, the serine protease undergoes a conformational alteration that shifts the substrate specificity of thrombin from fibrinogen to PC. Activation of PC is accelerated by the presence of the EPCR in concert with its plasma cofactor, protein S. APC modulates endothelial cell function, an effect whose phenotypic manifestations include antithrombotic, profibrinolytic, and anti-inflammatory properties.135 Physiologically, APC limits coagulation and restores endogenous fibrinolytic activity by (1) inactivating coagulation factors Va and VIIIa, (2) inhibiting release of plasminogen activator inhibitor 1, and (3) downregulating activation of TAFI by thrombin. The anti-inflammatory activity of APC results from downregulation of thrombin-induced inflammation, modulation of nuclear factor κB activity, and inhibition of TNFα and macrophage migration inhibitory factor production in monocytes.136,137 Experimentally, APC also blocked the accumulation of leukocytes at inflammatory foci by inhibiting the ability of leukocytes to bind to endothelial cell selectins.138 In another experimental model, the activated molecule also blocked p53-mediated endothelial cell apoptosis, an effect that contributes to its neuroprotective activity in ischemic human brain tissue and may help explain its antisepsis activity.139 In a baboon model, an infusion of E. coli at a concentration of 4.0 × 1010 organisms/kg produced septic shock, decreased concentrations of PC, fibrinogen, and platelets, and induced death within 24 to 36 hours.140 These outcomes can be prevented when plasma-derived APC is added to the E. coli infusion. If activation of PC is prevented by preinfusion administration of antibody to PC, shock develops more rapidly, the coagulopathy is worse, and the animals die more quickly. Importantly, the levels of endogenous PC in the experimental animals remained within normal limits during the experiment. These preclinical data suggested that APC had the potential to protect against the morbidity and mortality of severe sepsis.

Low levels of PC are present in more than 85% of patients with severe sepsis.141 Levels of endogenous PC decrease before the appearance of the clinical signs and symptoms of sepsis with acute organ dysfunction. In a population of cancer patients with chemotherapy-induced neutropenia, Mesters et al determined that the concentrations of PC decreased 16 hours before a diagnosis of severe sepsis could be made.142 In a review of published experimental and clinical studies, Yan et al reported that the lower levels of PC in patients with sepsis occurred in the absence of appreciable conversion to APC. The safety and efficacy of a range of doses of drotrecogin alfa (activated), a recombinant form of human APC, were tested by Bernard et al in a double-blind, randomized, placebo-controlled, multicenter, dose-ranging (sequential), phase II clinical trial of 139 adult patients with severe sepsis.141 The investigators demonstrated that the drug was safe and well tolerated and that it produced a significant dose-dependent reduction in D-dimer and IL-6, markers of coagulopathy and inflammation, respectively. Because the endothelium is the largest organ in the body and a component of all other organ systems, the ability of infused recombinant human APC to protect endothelial cells may explain the drug's ability to improve outcomes in patients with MODS.

In the phase III randomized, double-blind, placebo-controlled, multicenter, international Protein C Worldwide Evaluation in Severe Sepsis (PROWESS) trial of 1690 patients, treatment with drotrecogin alfa (activated) produced significant relative and absolute reductions of 19.4% and 6.1%, respectively, in the risk of death (p = 0.005).143 The NNT with drotrecogin alfa (activated) to save a life in the PROWESS trial was 16. Patients treated with the drug showed a trend toward a higher incidence of serious bleeding (3.5% versus 2.0%; p = 0.06). In patients at higher risk of death, as stratified by APACHE II scores, the ARR of death was 13% and the NNT was eight. In addition, the benefits of drotrecogin alfa (activated) persisted after discharge. In a cross-sectional observational study of the PROWESS cohort followed for up to 3.6 years, Angus et al reported that patients who were treated with the drug had an increased median survival of 9 months compared with placebo; these persistent benefits were greatest in patients with the most severe disease.144 In those who have a reasonable life expectancy if they survive the episode, targeting drotrecogin alfa (activated) to the most severely ill patients with severe sepsis has been found to be cost effective according to societal standards.144,145 Drotrecogin alfa (activated) therapy has recently been approved for inclusion in the hospital prospective payment system and for reimbursement under Section 533 of the Medicare Benefits Improvement and Protection Act of 2000.

Identifying Candidates for Drotrecogin Alfa (Activated) Therapy

As the first approved therapy specifically for patients with severe sepsis, drotrecogin alfa (activated) has been subjected to intensive evaluations to help determine the ideal candidates and to assess the drug's risk:benefit profile. Drotrecogin alfa (activated) is currently indicated for patients with severe sepsis at high risk of death who are felt to be candidates for aggressive care and lack contraindications for anticoagulation (Fig. 46-4).

In the PROWESS trial, mortality rates in nearly all of the prospectively defined subgroups were lower in the group that received drotrecogin alfa (activated) than in the group that received placebo infusion. However, analysis of the data from the PROWESS trial showed that the greater the baseline predicted risk of death (e.g., APACHE II score ≥25 or sepsis-induced organ dysfunction in two or more vital systems), the greater the ARR in mortality rate.146 At this time, patients with the greatest likelihood of benefit are those who fulfill the following three criteria: (1) severe sepsis with vasopressor dependence and/or requiring mechanical ventilation; (2) no active bleeding, platelet count above 30,000/μL, and international normalized ratio (INR) below 3.0, and no identifiable risks of central nervous system bleeding; and (3) the patient/family and medical team approve of an aggressive approach to management of sepsis in the patient, with a reasonable baseline quality of life to return to after the patient's stay in the ICU. A useful treatment algorithm is provided in Figure 46-5.

Vasopressor dependence can be defined as a systolic BP of 90 mm Hg or lower or a MAP 70 mm Hg or lower despite aggressive fluid resuscitation to maintain adequate intravascular volume status. In the PROWESS trial, the mortality rate of patients treated with conventional therapy with an APACHE II score of 25 or higher was 25%. In this same group, the mortality rate of vasopressor-dependent patients was 32%. As a consequence, vasodepressor dependence is used by some as a surrogate for APACHE II scores of at least 25 (i.e., high risk of death). In the PROWESS trial, patients treated with drotrecogin alfa (activated) had significantly higher MAP by day 5 (p = 0.01) and were able to achieve a MAP above 70 mm Hg without vasopressor support over days 1 through 7 more rapidly compared with controls (p = 0.01).147 Likewise, patients with severe sepsis and significantly impaired oxygenation (e.g., respiratory rate >30 breaths/min, PaO2 <55 mm Hg) are at high risk of death. In the PROWESS trial, treatment with drotrecogin alfa (activated) resulted in significantly improved oxygenation by day 7 (p = 0.01) and enabled recipients to achieve PaO2/FiO2 ratios greater than 400 without ventilator support significantly faster over days 1 through 7 (p = 0.009).

In the PROWESS trial, the rate of serious bleeding events attributable to drotrecogin alfa (activated) during drug infusion was 1.5%. Most serious bleeding events in the population treated with drotrecogin alfa (activated) occurred in patients with predisposing factors such as GI ulceration, traumatic injury of a blood vessel or a highly vascular organ, or markedly abnormal tests of coagulation. Notably, the presence of material in a nasogastric aspirate that tests positive for heme (e.g., guaiac test) in the absence of overt bleeding should not be considered a contraindication to treatment with drotrecogin alfa (activated). Drotrecogin alfa (activated) therapy should be used with caution in patients with markedly abnormal hemostasis (e.g., platelet count <30,000/μL). An INR above 3.0 is also used as a relative guide; in patients with INRs above this level, the benefits must be carefully weighed against the risks. The potential for intracranial hemorrhage (ICH) represents the most important complication warranting screening. In a review of approximately 3000 patients, the ICH rate was 0.5%; in an ongoing open-label study, the incidence was 0.2%. This incidence should be viewed from the perspective of the placebo rate of ICH in other large-scale, randomized clinical trials of patients with severe sepsis. For example, in the trial of large-dose AT III in severe sepsis, the incidence of ICH in the control arm was 0.4%.148

In elderly patients, the cost:benefit ratio for drotrecogin alfa (activated) depends on the patient's potential to have a reasonable quality of life after recovery from the episode of severe sepsis. In general, age should not be used to stratify patients for potential life-saving interventions such as drotrecogin alfa (activated) therapy. In the Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatments (SUPPORT), life support measures such as mechanical ventilation were withheld more often in patients older than 70 years, even after adjusting for severity of illness.149 However, analysis of the ARDSNet data indicates that, despite a decreasing survival rate with advancing age, 50% of patients 80 years and older survived an episode of ARDS.150 In the PROWESS trial, the ARR of drotrecogin alfa (activated) in patients 75 years and older was 15%, and the NNT to save a life was six to seven patients.151 Importantly, this benefit was not accompanied by an increase in the risk of bleeding compared with that of the younger cohort. Although the pivotal study was conducted in a broad range of patients with severe sepsis, the data clearly showed that drotrecogin alfa (activated) produces the greater absolute benefit in those patients with high risk of death. Because a mortality benefit was not demonstrated in PROWESS for patients with a lower severity of illness (using APACHE II scores) and potential for harm could exist in those not likely to benefit, these low-risk patients are currently being evaluated in an 11,000-patient randomized controlled trial.

Several trial design issues have been raised and answered about PROWESS that may have limited its early adoption.152 These include concerns regarding mid-trial changes to the study protocol and drug preparation, high acquisition costs, and bleeding.153 The U.S. Food and Drug Administration (FDA) and the principal investigators of the trial have addressed these issues with firm conclusions: (1) large-scale clinical trials are routinely amended, and analysis of PROWESS by the FDA does not support the belief that changes in the protocol or drug were responsible for the outcomes; and (2) the greatest benefits of drotrecogin alfa (activated) are in the most severely ill patients.154,155 In acknowledging the benefits seen in the most severely ill, the FDA indication specifies that the drug should be prescribed to patients with severe sepsis at high risk of death. In addition, the FDA requested that the manufacturer provide additional clinical trial data on the use of drotrecogin alfa (activated) in patients with severe sepsis at lower risk of death; trials in such low-risk patients are ongoing.

Administration and Monitoring of Drotrecogin Alfa (Activated)

Once a decision has been made to treat a patient with drotrecogin alfa (activated), the infusion should be started immediately. Although there was no indication from the clinical trials to indicate that watchful waiting is harmful within the 48-hour window after the development of sepsis-induced organ failure, it would seem prudent to treat a patient with severe sepsis at high risk of death earlier rather than later. This is supported by analysis of data from the ENHANCE trial. Patients treated with Drotrecogin alfa (activated) within the first twenty-four hours had a lower 28-day mortality by degree of organ dysfunction before treatment (P = 0.014) and by time to treatment stratified by APACHE II quartile. Patients treated within the first twenty-four hours also had significantly fewer days on the ventilator and lower ICU and hospital lengths of stay (P <0.004). Treatment consists of an infusion of 24 μg/kg per hour of drotrecogin alfa (activated) for a total infusion duration of 96 hours. The dosage is based on the patient's actual body weight rather than ideal body weight. There is no need to monitor the infusion with tests of hemostasis or levels of PC or APC. In the event of clinically significant bleeding, the infusion should be stopped immediately. Drotrecogin alfa (activated) is rapidly cleared by plasma proteases, including PC inhibitor and α1 antitrypsin: 80% is eliminated within 13 minutes and 98% is cleared within 2 hours. If the nature and severity of bleeding has been clarified and appropriate actions have been successfully taken, the drug infusion may be resumed for the remainder of the total 96-hour duration as long as the management team feels that treatment with an antithrombotic is now safe.

The increased rate of bleeding in patients treated with drotrecogin alfa (activated) is usually related to procedure and predominantly develops during the time of drug infusion. Because of its short half-life, there is no significant residual drug effect if the infusion is temporarily interrupted 2 hours before any surgical procedure. The decision as to when to restart the infusion should be based on the nature of the procedure and assessment of the patient.156 For minor procedures such as tracheostomy or change or placement of an arterial line or a femoral venous catheter, the drotrecogin alfa (activated) infusion can be restarted immediately thereafter if there is no sign of bleeding. In the absence of bleeding or complications, the drug infusion can be restarted 2 hours after procedures such as placement of a Swan-Ganz catheter or placement of a chest tube. The 2-hour postprocedure interval before the infusion is resumed is also reasonable for patients undergoing lumbar puncture, percutaneous drainage, nephrostomy, or sinus puncture. In the absence of bleeding, the infusion can be restarted after 12 hours in patients undergoing major surgical procedures or removal of an epidural catheter. Once the infusion is restarted, the drug should be administered so that the patient continues to receive the calculated amount for the full 96-hour duration. No attempt should be made to accelerate or “catch-up” and bolus infusions are contraindicated.

Hemostatic abnormalities are common in patients with severe sepsis.157 In addition to APC, the body has a number of other natural antithrombotics that have the potential to improve outcomes in patients with severe sepsis. These include AT, TF pathway inhibitor (TFPI), and endothelial cell glycosaminoglycans.

Antithrombin

AT III is a serine protease that is a slow, progressive inhibitor of thrombin and factor Xa; binding of heparin to a lysine site on the AT protein produces a conformational change in the molecule that converts it to a rapid inactivator of thrombin and factors Xa, IXa, XIa, and XIIa.158 Experimentally, AT III supplementation has been shown to decrease inflammatory changes associated with severe sepsis.159 Based on experimental and early clinical studies, Warren et al evaluated the safety and efficacy of AT III in a double-blind, placebo-controlled, multicenter study of 2314 adult patients with severe sepsis.148 At 28 days, overall mortality rates were 38.9% in the treated group and 38.7% in the placebo group (p = 0.94). Study subjects who received AT III had a relative risk of bleeding greater than 1.7. Studies of this natural antithrombotic for the treatment of severe sepsis have been discontinued.

Tissue Factor Pathway Inhibitor

TFPI is a Kunitz-type inhibitor; after binding and inactivating factor Xa, TFPI then inhibits factor VIIa within the factor VIIa/TF complex.160 Circulating TFPI has been shown to modulate the activity of the extrinsic pathway. Experimentally, TFPI inhibits LPS-induced pulmonary edema and tissue damage.161 These features suggest that recombinant human (rh) TFPI (tifacogin) therapy might decrease the coagulopathy and inflammatory changes of severe sepsis in a fashion similar to that of rhAPC. Because the proposed benefits of tifacogin would be anticipated to be through stabilization of coagulation, trials of the drug have focused on patients with a mild coagulopathy (INR ≥1.2). Although a phase II trial showed a trend toward reduction in the 28-day all-cause mortality rate in the rhTFPI group compared with placebo, the phase III OPTIMIST (Optimized Phase III Tifacogin in Multicenter International Sepsis Trial) study failed to show a mortality benefit.162,163 Overall mortality rate at 28 days in the group treated with tifacogin (n = 880) was 34.2% versus 33.9% in the placebo group (n = 874), a difference that did not reach statistical significance (p = 0.88). In addition, there was an increase in serious adverse events for bleeding in the treatment group; this difference was irrespective of the patients' INR. A second phase III study of TFPI is now being conducted to determine if a higher dose of the drug will produce a survival advantage.

Glycosaminoglycans

Endothelial cells have important anticoagulant systems, including the heparan sulfate-AT system. Heparan sulfate is an endothelial cell glycosaminoglycan similar in structure and pharmacodynamic properties to heparin. By interacting with endothelial cell surface heparan sulfate, endogenous AT III stimulates prostacyclin generation and inhibits cytokine and TF production in endothelial cells and monocytes.164 It may also be involved in the AT III–associated desensitization of leukocyte chemoattractant receptors. Ahmed et al demonstrated that heparin also possesses clinically significant anti-inflammatory properties that are independent of AT III.165 They showed that inhaled enoxaparin can produce a dose-dependent, statistically significant inhibition of exercise-induced bronchoconstriction independent of the drug's anti–factor Xa activity. In a hypothesis-generating exercise, Davidson et al reported that their analysis of the PROWESS and AT III trials suggests that administration of small-dose heparin to enrollees in the two trials is associated with a 45% (95% confidence interval = 1.18 to 1.78; p <0.001) improvement in the odds for survival among the placebo recipients compared with those who did not receive heparin.166 Similar observations have been made about a heparin effect in the trials of other natural anticoagulants. In the AT trial, mortality rate was lower in the placebo group who received heparin.148 In the OPTIMIST trial, mortality rate was lower in patients receiving placebo and heparin than in those who did not.163 Heparin's potential anti-inflammatory and antileukoadhesive properties were investigated in a human endotoxemia model by Derhaschnig et al.167 They showed that, although the drug displayed little effect on cytokine production and endothelial cell activation in endotoxemia, it downregulated expression of l-selectin and produced lymphocytopenia. As intriguing as these observations are, experimental studies in a sepsis model reported that heparin does not prevent ALI produced by smoke inhalation.168 Thus, the clinical relevance of heparin's anti-inflammatory properties is unclear and requires further studies.

Other Agents

Several experimental approaches are underway to investigate other hemostasis-related agents for the treatment of patients with severe sepsis. They include soluble TM (sTM), an inhibitor of activated coagulation factor VII (rhFVIIa), and PAF inhibitor. TM is an endothelial cell protein that alters thrombin's substrate specificity from fibrinogen to PC, thereby converting PC to APC. Studies have indicated that endotoxin and proinflammatory cytokines such as TNFα decrease endothelial cell TM expression. This may be a consequence of downregulation of TM expression, increased degradation of the protein, or some combination thereof. When coupled with a glycosaminoglycan moiety, a soluble form of TM can block the fibrin formation and cell activation, convert PC to APC, and promote thrombin inhibition by AT III and heparin cofactor II.169 These properties suggest that sTM might be useful in sepsis. In an experimental model, rh-sTM blocked endotoxin-induced disseminated intravascular coagulation (DIC) and pulmonary vascular injury.170,171 Although rh-sTM has been shown to be safe in normal human volunteers, no clinical trials have been published describing its efficacy in patients with severe sepsis.172

Upregulation of TF on monocytes and perhaps endothelial cells is a significant element in the evolution of severe sepsis. Although clinical trials of TFPI failed to demonstrate a survival benefit, the extrinsic pathway of coagulation remains an attractive target. Active site-inhibited rhFVIIa can act as a competitive inhibitor of factor VIIa generated during activation of coagulation. Welty-Wolf et al and Carraway et al demonstrated in a primate model that rhFVIIa can prevent sepsis-induced pulmonary and renal dysfunction and attenuate organ damage in established sepsis.173,174 However, Jilma et al reported that, although rhFVIIa was an effective inhibitor of TF-induced thrombin generation, it did not alter endotoxin-induced changes in fibrinolysis, cytokine release, or markers of endothelial cell or platelet activation.175 Studies of this drug are ongoing.

During the systemic inflammatory response to infection, phospholipase A2 is released. This enzyme catalyzes conversion of cell membrane phospholipids to PAF and arachidonic acid. It has been hypothesized that a phospholipase A2 inhibitor administered within the first 18 hours of sepsis-induced organ failure may improve survival rate. However, a recent multicenter, double-blind, placebo-controlled trial of 14-kd group IIA secretory phospholipase A2 inhibitor failed to provide evidence that this approach offers a survival benefit.176

PAF acts as a signaling molecule that can trigger and amplify inflammatory and coagulation signals.177 Levels of PAF are increased in patients with severe sepsis and there is evidence that PAF signaling is defective. Human recombinant PAF-acetyl-hydrolase inhibits monocyte/macrophage and neutrophil activation in response to inflammation. In patients at risk for multiple organ dysfunction syndrome, rhPAF-acetyl-hydrolase was safe and improved organ scores.178 However, a phase II clinical trial failed to demonstrate benefits of this drug compared with placebo on survival rate, hemodynamic status, respiratory function, or organ failure scores.179

The results of these studies raise an interesting issue: with the demonstrated efficacy of drotrecogin alfa (activated) and preclinical data that support the use of other natural antithrombotics in patients with severe sepsis, why have the results of the clinical studies of these compounds been so disappointing? The answers to these questions are unknown but several possibilities exist. Recombinant human APC has multiple mechanisms of action that may be additive to its antithrombotic properties. These include several anti-inflammatory, profibrinolytic, and endothelial cell survival properties. Other issues may be related to trial design. In the trials of AT III and TFPI, the lack of overall benefit may have been related to the dose, the ability of thromboprophylactic doses of heparin to block the drug's anti-inflammatory effect, or the concept. For example, the continuous infusion of AT III may have been insufficient to maintain supraphysiologic levels of AT III in severely septic patients. Because AT III must bind to endothelial cell and neutrophil glycosaminoglycans to promote local anticoagulant and anti-inflammatory activities, prophylactic heparin administered to some patients may have prevented the infused drug from significantly binding to its target on endothelial cells.

(See Chap. 79) Cortisol is the predominant glucocorticoid secreted by the adrenal cortex. During acute illness, cortisol production increases as much as sixfold, an increase that is roughly proportional to the severity of the illness.180 Increased cortisol levels in acutely ill patients are a result of a combination of increased production of CRH and ACTH plus a reduction in the degree of cortisol-mediated feedback inhibition. Other factors that may contribute to the adrenal cortical stress response include loss of the normal diurnal secretory pattern, increased levels of free cortisol resulting from a combination of a decrease in the levels of cortisol-binding globulin and cleavage of cortisol-binding globulin by neutrophil elastase, plus inflammatory cytokine-mediated alterations in peripheral cortisol metabolism and changes in the affinity of cellular glucocorticoid receptors for their ligand.

There are several potential mechanisms of action that suggest that glucocorticoid therapy might be beneficial in patients with severe sepsis: (1) correction of a state of relative adrenal insufficiency, (2) inhibition of the synthesis of iNOS, (3) improved hemodynamics secondary to restoration of the sensitivity of vascular catecholamine receptors, and (4) liberation of bound IκB-a with decreased transcription of inflammatory cytokines and increased synthesis and release of the receptor antagonist for IL-1.181,182

Megadose steroid therapy for patients with severe sepsis was abandoned in the late 1980s when it failed to improve mortality rate.183 However, in 1998, Bollaert et al reported the results of a prospective, randomized, double-blind, placebo-controlled study of small doses of hydrocortisone (100 mg intravenously, three times daily for 5 days) on hemodynamics and survival.184 After 28 days, patients treated with glucocorticoids demonstrated a significant ability to reverse the hemodynamic alterations of shock (p = 0.005) and had a 31% reduction in mortality rate (32% versus 63%; 95% confidence interval = 1% to 61%; p = 0.091). Because there was no correlation between the outcomes and the results of a short corticotropin test, the investigators commented that the results do not appear to be related to the presence of adrenal cortical insufficiency. However, in 2000, Annane et al described a three-level prognostic classification for septic shock based on cortisol levels and response to a short corticotropin stimulation test.185

In a follow-up placebo-controlled, randomized, double-blind, parallel-group trial, Annane et al reported the effect of hydrocortisone and fludrocortisone on mortality rate in 300 medical and surgical patients with severe sepsis.186 Other eligibility criteria included mechanical ventilation, hypotension unresponsive to fluids, and dysfunction of other organ systems characteristic of severe sepsis. Patients who fulfilled the usual criteria for septic shock were enrolled after undergoing a short ACTH stimulation test. Steroid therapy consisted of 200 mg of hydrocortisone (50 mg at 6-hour intervals) and a single 50-μg dose of fludrocortisone daily for 7 days. In patients characterized on the basis of the short ACTH stimulation test as having relative adrenal insufficiency (i.e., nonresponders), steroid therapy significantly decreased mortality rate and the duration of vasopressor therapy. Steroid treatment of the nonresponder population produced 16% and 10% relative and absolute reductions, respectively, in the risk of 28-day mortality (p = 0.04). Further, steroid treatment of this subpopulation improved hemodynamics and enabled the investigators to wean patients from vasopressors 3 days earlier than placebo-treated controls. The contribution of the administered mineralocorticoid to the outcomes is unclear. Adverse events potentially associated with steroids such as superinfection, GI bleeding, and psychiatric disorders did not increase in the treatment group.

Identifying Candidates for Low-Dose Steroid Therapy

As opposed to absolute adrenal insufficiency, which is relatively uncommon in patients with severe sepsis, evidence of inadequate adrenal reserves is common. Annane et al reported that evidence of an inability of the adrenal cortex to respond to ACTH-releasing hormone (nonresponders) was present in approximately 66% of patients with severe sepsis in their study.186 This observation coupled with the results of the study and the ready availability of glucocorticoids and mineralocorticoids provide preliminary support for low-dose steroids in appropriate patients with severe sepsis. Notably, however, steroids did not provide uniform benefits in all subgroups: in patients with normal adrenal function (i.e., responders to ACTH stimulation), administration of low doses of hydrocortisone plus fludrocortisone was associated with increased mortality rate (61% versus 53%; p = not significant). Although the results were not statistically significant, when combined with previous studies indicating that steroids can have adverse effects in patients with severe sepsis, the possibility was raised that steroid therapy may be detrimental to patients in this population with adequate adrenal reserve.187 Because the number of patients in this study was small and demonstrated interesting results, an ongoing large European trial is evaluating this intervention.

Administration and Monitoring of Low-Dose Steroid Therapy

In patients with severe sepsis at high risk of death, it is reasonable to administer combination glucocorticoid/mineralocorticoid therapy with the 7-day protocol described by Annane et al.186 Therapy should begin immediately after a short ACTH stimulation test (ACTH 250 μg [one vial of Cosyntropin] intravenously or intramuscularly with a serum cortisol level 45 to 60 minutes later). Once the results become available, steroids can be discontinued in patients whose baseline serum cortisol is increased by 10 μg/dL or greater (i.e., responders). Nonresponders, patients with a change in their serum cortisol levels of 9 μg/dL or less, can continue to receive low-dose steroids for the full 7-day period. The significance of the level of serum total cholesterol in patients with sepsis is unclear since values fall with the development of critical illness-associated hypoproteinemia. Serum free cholesterol may more accurately reflect adrenal function. Consequently, some patients may be selected for low-dose steroid therapy unnecessarily. There is no evidence that steroids and drotrecogin alfa (activated) are competitive therapies. Patients in the PROWESS trial received the best therapy as determined by their treating physicians; in many cases, this included steroids plus drotrecogin alfa (activated) or placebo. In an editorial accompanying the low-dose steroid trial, Abraham et al pointed out that more than 50% of the patients with inadequate adrenal reserve died despite treatment.187 They emphasized that this observation underscores the need for additional therapies that can reverse organ dysfunction and improve outcomes in this population.

Tumor Necrosis Factor Antagonists

TNFα is thought to play a major role in the pathogenesis of the clinical manifestations of patients with severe sepsis. Evidence supporting the cytokine's central role in the process includes: (1) experimentally, TNFα levels increase after an endotoxin or a bacterial challenge; (2) increased levels of TNFα can be demonstrated in preclinical models and patients with sepsis; (3) presence of increased levels in patients with the disease and persistent high concentrations are associated with nonsurvival; (4) TNFα challenge simulates manifestations of sepsis and can lead to organ dysfunction; and (5) neutralization of TNFα frequently leads to a decrease in sepsis-associated symptoms and increases survival rate in experimental models.188 Compounds designed to test the efficacy of TNF antagonists can be divided into two broad groups: (1) monoclonal antibodies against human TNFα (e.g., afelimomab, CDP571) and (2) soluble TNFα receptors (e.g., p55-IgG). Studies of these compounds have indicated that they may produce small survival benefits in some subgroups of patients with severe sepsis.188,189 Future studies of these agents require methods to better indicate patients who might benefit from the drugs and multimodal therapy that also includes TNFα neutralization.

Other Potential Strategies

The autonomic nervous system plays a major role in the pathophysiology of severe sepsis. Studies have shown that electrical stimulation of the vagus nerve attenuates the release of TNF and other cytokines in response to endotoxin and can prevent the development of shock.80 At a molecular level, strategies to neutralize macrophage migration, block apoptosis, or inhibit the activity of poly(adenosine 5′-diphosphate-ribose) synthase can also provide protection against the effects of severe sepsis.190–192 Activation of the classic or alternate pathway of complement generates C5 convertase. This enzyme converts C5 to C5a and C5b. In the presence of C6, C7, and C8, C5b forms the membrane attack complex responsible for complement's lytic activity. In contrast, C5a is a proinflammatory peptide also known as anaphylatoxin. Antibodies to the C5a peptide decrease bacteremia, inhibit apoptosis, and improve survival in experimental models of sepsis.193–195

Nutritional Support and Glycemic Control

Importance of Adequate Nutrition

Severe sepsis is a catabolic state. Metabolic changes associated with the disease include rapid breakdown of proteins, carbohydrates, and lipids, protein diversion, negative nitrogen balance, hyperglycemia with insulin resistance, and altered blood flow to key metabolic sites. As with any other critically ill population, patients with severe sepsis require adequate nutritional support to correct any previous nutritional deficiencies, maintain organ system structure and function, compensate for anticipated or existent inadequate self-feeding, improve wound healing, promote immune function, decrease susceptibility to superinfection, and increase survival rate (see Chap. 11).

Energy needs can be calculated by determining the patient's basal energy expenditure from the patient's weight, height, and age inserted into the Harris-Benedict equation.196 The weight inserted into the equation should be based on clinical evaluation. Although it is reasonable to use the ideal body weight for obese patients, the pre-illness weight can be used for those patients whose weight has been stable before hospital admission. In most instances, the basal energy expenditure will be approximately 25 to 30 kcal/kg per day; however, due to the catabolic nature of severe sepsis, which seems to reach its maximum during the second week, the basal energy expenditure may be multiplied by a stress factor of 0.5 to 2.5.197,198 The normal protein requirements for a well-nourished individual are 0.6 to 1.0 g/kg per day.199 In contrast, patients with severe sepsis may require up to 2.0 g/kg per day. Glutamine and arginine supplementation may improve outcomes.200,201Glucose should consist of 30% to 70% of the total nonprotein calories to maintain blood glucose within an acceptable range (see below). Lipids should comprise 15% to 30% of nonprotein calories. The ω-6 polyunsaturated fatty acid content of nutritional support for a septic patient should be decreased while maintaining a level that will prevent a deficiency of essential fatty acids. The ω-3 fatty acids are important for cell membrane integrity, receptor expression and function, and activation and signal transduction. Enteral preparations supplemented with these important lipids are available.

Enteral versus Parenteral Nutrition

Enteral nutrition offers several advantages including lower cost, preservation of the GI mucosal barrier, buffering of gastric acid, preservation of enteral hormone secretion, provision of unique nutrients, decreased incidence of infections, improved wound healing, and avoidance of parenteral nutritional catheters and their complications.202 Immunomodulatory enteral preparations that are supplemented with arginine, nucleotides, and fish oil have been shown to improve outcomes in critically ill patients, although much work is needed in this area.203,204 For septic patients who cannot tolerate enteral nutrition because of an ileus or other complications, parenteral nutrition is a consideration. Heyland et al conducted a meta-analysis of 26 randomized clinical trials of 2211 patients that compared morbidity and mortality rates in patients receiving enteral or total parenteral nutrition.205 The mortality rate of critically ill patients was higher among those receiving TPN (RR, 1.78; 95% CI, 1.11–2.85). Notably, the hypothesis that parenteral nutrition is superior to no nutrition during the acute episode has not been rigorously tested.

Management of Blood Glucose Levels

Stress-induced hyperglycemia (>200 mg/dL) is common in patients with severe sepsis. If doubt exists, it can be differentiated from chronic diabetic hyperglycemia by an assay of HbA1c concentrations, values that will be within the normal range in a previously healthy patient. The etiology of this metabolic derangement in patients with severe sepsis is insulin resistance coupled with increased gluconeogenesis, abnormalities that can be compounded by excessive parenteral nutritional support. Insulin resistance is multifactorial in origin and can result from a combination of endotoxin, elevated levels of cytokines such as TNFα, IL-1, and increased circulating concentrations of epinephrine, glucocorticoids, and growth hormone. The effects of cytokines and hormones on blood glucose levels result primarily from the development of insulin resistance in skeletal muscles.206,207 Despite hyperglycemia, the liver is also resistant to the effects of insulin and is driven to gluconeogenesis by counterregulatory hormones such as glucagon, epinephrine, and growth hormone. An associated cause that may be overlooked in a critically ill patient is provision of excess parenteral glucose. Rosmarin et al reported that nondiabetic patients who received total parenteral nutrition in which dextrose was infused at rates greater than 4 mg/kg per minute had a 50% chance of developing hyperglycemia.208

The deleterious effects of hyperglycemia in patients with severe sepsis have been considered to be predominantly related to concerns about an increased risk of infection.209 Although it is not completely clear whether secondary infections in critically ill patients with hyperglycemia is directly due to the elevated glucose or is another manifestation of latent or overt diabetes mellitus, evidence suggests that hyperglycemia directly impairs immune function.210,211 Prolonged elevation of blood glucose may potentiate the coagulopathy of severe sepsis. Rao et al demonstrated that prolonged hyperglycemia activates the TF pathway of blood coagulation.212 Activation of the extrinsic pathway of coagulation was independent of insulin levels, hypertriglyceridemia, and hyperosmolality. This was presumably due to activation of monocytes and/or endothelial cells, upregulation of TF expression, and enhanced conversion of factor VII to VIIa.213

Good glycemic control requires regular monitoring of the blood glucose and attention to parenteral nutrition and insulin therapy. However, the target blood glucose level has been unclear. For example, long-term outcomes are improved in patients with diabetes who have had a myocardial infarction if their blood glucose level is maintained at a level below 215 mg/dL or 11.9 mmol/L.214,215 Recently, van den Berghe et al reported the results of a prospective, randomized controlled trial of more than 1500 mechanically ventilated surgical ICU patients treated with intensive versus conventional insulin therapy.216 They reported that patients who were randomly assigned to intensive insulin therapy (target blood glucose = 80 to 110 mg/dL, or 4.4 to 6.1 mmol/L) or standard care (target blood glucose = 180 to 200 mg/dL, or 10 to 11.1 mmol/L) had significantly different outcomes. Tight glucose control produced the following advantages: (1) decreased overall ICU mortality rate (4.6% versus 8.0%); (2) decreased ICU mortality rate in patients admitted longer than 5 days (10.6% versus 20.2%; p = 0.005); and (3) lower in-hospital mortality rate (7.2% versus 10.9%). Other benefits included a 46% relative reduction in bloodstream infections, a 41% decrease in the need for renal replacement therapy (RRT), a 50% decrease in the median number of red blood cell transfusions, and a 44% reduction in the incidence of critical-illness polyneuropathy. More than 97% of the patients in the tight-control group received intravenous insulin, compared with 33% in the control group. Episodes of hypoglycemia (glucose <40 mg/dL, or 2.2 mmol/L) were more common in the intensive-control group (5.0% versus 0.7%); however, these episodes were not accompanied by hemodynamic instability or seizures. Although these findings need to be confirmed, they suggest that patients with severe sepsis should receive intensive insulin therapy to maintain their blood glucose levels between 80 and 110 mg/dL (4.4–6.1 mmol/L). Realistically, accomplishing tight glycemic control is often difficult. Many internists have adopted protocols that have adjusted glucose in the range of 100 to 140 mg/dL (5.6–7.8 mmol/L). A point of interest deserving future research is whether or not the benefits of this approach are due primarily to the lower glucose area under the curve or result from glucose-independent effects of insulin.217 Recently, Finney et al evaluated the relation between glucose control and mortality rate in 523 critically ill patients.218 They reported a positive association between increased insulin administration and death, independent of prevailing glucose level. They suggested that it is the control of glucose rather than absolute levels of exogenous insulin administered for glucose management that accounts for the mortality benefits of glycemic control. Further, regression analysis of their data suggests that the mortality benefit occurs at blood glucose levels below 145 mg/dL. If these data are supported by others, less stringent glycemic management may provide many of the benefits without intense resource utilization.

Renal Replacement Therapy

Transient oliguria is common in patients with severe sepsis and vasodilatory hypotension or shock. Notably, anuria is rare, and fewer than 5% of patients with severe sepsis develop acute renal failure requiring RRT.89 Importantly, when it occurs, acute renal failure and its associated metabolic derangements increase the risk for potentially fatal extrarenal complications.219,220 Therefore, steps should be taken to preserve renal function in the setting of severe sepsis. Important measures include adequate volume replacement, monitoring drug dosage and avoiding nephrotoxins, and excluding lower urinary tract obstruction as the cause for oliguria or anuria. There is no evidence that low-dose dopamine administered by continuous intravenous infusion can protect against renal dysfunction in the critically ill population.221 If potential nephrotoxins such as aminoglycosides must be used, the dosage should be adjusted for the patient's degree of renal impairment, and drug levels should be monitored. In a patient with severe sepsis and acute renal failure, renal replacement therapy should be considered early in the disease course. In addition to providing renal replacement, continuous renal replacement therapy (CRRT) has become a part of the overall management of the critically ill.

In the ICU, peritoneal dialysis and conventional forms of hemodialysis have been superseded by various forms of CRRT (see Chap. 75).222 CRRT is better tolerated than other renal replacement therapies because solutes and fluids are removed in a slower fashion. The choice of a CRRT modality is based on a number of factors, including access to the equipment, experience of the staff, the degree of hemodynamic stability, vascular access, and whether the primary need is for fluid and/or solute removal. Detailed explanations of these interventions are beyond the scope of this chapter and are discussed elsewhere in the textbook.

In addition to removal of metabolic wastes, other indications for CRRT in patients with severe sepsis include establishing hemodynamic stability and providing more precise fluid and electrolyte control. Although cytokines can be found in the dialysate or ultrafiltrate of patients undergoing renal replacement, there is no evidence that cytokine removal by CRRT improves outcomes. CRRT can help to improve outcomes in patients with severe sepsis and ARDS. This potential benefit may be primarily the effect of extravascular lung water. However, the procedure can also be used to stabilize the pH of the blood in patients treated with permissive hypercapnia. Whereas benefits of CRRT are known, until recently the “dose" of CRRT has been unclear. Schiffl et al conducted a prospective study in 160 patients with acute renal failure to compare the effect of intensive (daily) HD with conventional (alternate-day) HD on mortality rate.223 They found that daily HD reduces mortality rate without increasing hemodynamically induced morbidity compared with the alternate-day approach. Patients treated with intensive HD had better control of uremia, fewer hypotensive episodes during the procedure, and more rapid resolution of acute renal failure (mean ± SD = 9 ± 2 days versus 16 ± 6 days; p = 0.001) than did those undergoing conventional HD. In an intention-to-treat analysis, the mortality rates were 28% for intensive HD and 46% for alternate-day dialysis (p = 0.01). In a multiple regression analysis, conventional (versus daily) HD was an independent risk factor for death. Currently, large-scale, multicenter trials are evaluating the effects of CRRT on survival rate in critically ill patients.

Thromboprophylaxis

Patients with severe sepsis have at least one risk factor that places them at increased risk for venous thromboembolism (VTE). In addition to immobilization, they may have risk factors such as advanced age, recent trauma or surgery, respiratory failure, or the presence of a central venous access device. Serial scans with radiolabeled fibrinogen or ultrasonograms have indicated that approximately 30% of critically ill patients have evidence of VTE.224,225 Importantly, VTE can occur even in patients receiving pharmacologic thromboprophylaxis. Thrombi are most common in the lower extremities, and most pulmonary emboli arise from veins in that region. The incidence of embolic events increases as the thrombus extends proximally to the popliteal fossae. Thrombi can also occur in the veins of the upper extremities; the presence of cancer or a central venous catheter increases the risk of VTE originating in this region.

All patients with severe sepsis should be assessed for VTE risk factors, and, in the absence of contraindications, they should receive pharmacologic thromboprophylaxis. Treatment can be with low-dose unfractionated heparin (UFH), adjusted-dose UFH, or low-molecular-weight heparin (LMWH). The most widely employed regimen is 5000 U of UFH administered subcutaneously, twice daily. Although LMWHs are more expensive, they offer several advantages over UFH. These include improved pharmacokinetics and negligible effects on standard tests of coagulation. The thromboprophylactic dose of a specific LMWH should be the one recommended by the manufacturer. In a meta-analysis, LMWH was demonstrated to provide at least equivalent prophylactic activity as UFH for the prevention of postoperative VTEs and was shown to be associated with less bleeding.226Fondaparinux is a chemically synthesized anticoagulant that possesses the critical pentasaccharide sequence responsible for the anti–factor Xa of UFH and LMWH. Its acquisition cost should be balanced with a safety and efficacy profile that is superior to LMWH for the reduction of VTE events demonstrated in patients undergoing major orthopedic surgery.227,228 Treatment with thromboprophylactic doses of UFH or LMWH is not contraindicated during or after a 96-hour infusion of drotrecogin alfa (activated). The incidence of serious bleeding events in those who received drotrecogin alfa (activated) alone and in those who were also treated with heparin was similar (3.7% versus 3.5%).143 In patients with severe sepsis who are judged to be at high risk of serious bleeding or who are actively bleeding, mechanical prophylaxis with graduated compression stockings or intermittent pneumatic compression devices can be employed. However, the efficacy of these devices is significantly less than that of available antithrombotic drugs. Therefore, once the bleeding risk has decreased, pharmacologic thromboprophylaxis can be started or resumed.

Stress Ulcer Prophylaxis

Patients with severe sepsis have an increased risk for GI bleeding. The incidence of GI bleeding is proportional to the number of risk factors, including prolonged mechanical ventilation, coagulopathy, head injury, solid organ transplant, burns, severe sepsis, and a history of peptic ulcer disease.229 In a prospective study of 174 patients admitted to a medical ICU, Schuster et al reported that overt or occult GI bleeding developed in 14%.230 Risk factors for bleeding were mechanical ventilation for longer than 48 hours and/or the presence of a coagulopathy. Patients who developed GI bleeding had a higher mortality rate (64% versus 9%), longer duration of ICU stay (median = 14.2 days versus 4.2 days), increased risk for mechanical ventilation (84% versus 26%), and a requirement for extended duration of mechanical ventilation (9.5 versus 4.2 days) compared with those who did not bleed. The risk of bleeding can be reduced by strategies that decrease acid secretion (H2 blockers), buffer the acid (antacids), or protect the gastric mucosal barrier (sucralfate). In a study of 1200 mechanically ventilated patients, Cook et al compared the efficacy of sucralfate and ranitidine for the prevention of GI bleeding in this population.231 They reported that administration of ranitidine was significantly more effective than sucralfate in reducing the rate of clinically significant GI bleeding. The incidences of clinically important GI bleeding were 1.7% in patients treated with the H2 blocker and 3.8% in those receiving mucosal protection (p = 0.02). The NNT to prevent one clinically significant bleed with ranitidine over sucralfate was 48.

Nevertheless, it has been hypothesized that mucosal protective strategies such as sucralfate might be superior to H2 blockers for raising gastric pH above 4.0 because gastric acidity would prevent proliferation of bacteria in the stomach. That is, the raised pH with H2 blockers might result in an increased risk of nosocomial pneumonia from reflux of bacteria colonizing the stomach into the respiratory tract. However, there is little evidence to support this hypothesis. Prod'hom et al randomized 244 intubated patients to GI bleeding prophylaxis with ranitidine (150 mg as a continuous infusion), sucralfate (1 g every 4 hours), or antacids (20 mL of aluminum and magnesium hydroxide every 2 hours).232 Although the incidence of early onset pneumonia was not significantly different among the three groups, among 213 patients observed longer than 4 days, late onset pneumonia was less common in the mucosal protective group than in those treated with antacid or ranitidine (5%, 16%, and 21%, respectively; p = 0.022). However, there are several problems with this hypothesis, and the studies by Cook et al found no significant differences in the rates of VAP among those treated with sucralfate or ranitidine.231 Therefore, in patients with severe sepsis, stress ulcer prophylaxis with ranitidine or the newer available intravenous proton pump inhibitors may decrease the incidence of serious GI bleeding without a convincing increase in cases of VAP.

Sepsis Coagulopathy

In patients with severe sepsis, coagulation is activated and fibrinolysis is suppressed. A sepsis-associated coagulopathy is a sign of dysfunction of the hematologic organ system. In the Ibuprofen in Sepsis Trial, a coagulopathy with elevated levels of D-dimer and decreased levels of PC was almost universal in patients with severe sepsis.233 The coagulopathy in patients with this disease is linked to endothelial cell dysfunction and fibrin deposition, pathologic processes that contribute to multiple organ system dysfunction and death. Thrombocytopenia, prolongation of the prothrombin time and/or the activated partial thromboplastin time, or any combination thereof are uncommon in patients with severe sepsis. Although there is no single clinical manifestation or laboratory value that is diagnostic of DIC, the following changes are reasonably consistent with the diagnosis: (1) presence of a disorder such as severe sepsis that is known to be linked with DIC, (2) musculocutaneous bleeding and dysfunction of one or more organ systems, and (3) ongoing consumption of coagulation factors. The first two criteria are detected by clinical examination; the last criterion can be recognized by thrombocytopenia and prolongation of one or more global tests of coagulation. Taylor et al proposed a scoring system for the diagnosis of DIC.234 It uses the characteristics described above to provide a DIC score; values of at least 5 are characteristic of DIC.

In the absence of bleeding, patients with severe sepsis-associated coagulopathy can simply be monitored. In most instances, once the infection is controlled, the abnormal laboratory parameters return toward normal. In patients who are bleeding despite active treatment of the infection, a careful examination should be conducted to rule out a discrete and correctable anatomic lesion (e.g., peptic ulcer) as the cause of the bleed. If bleeding is from multiple sites, treatment is generally broken down into three general areas: (1) platelets and coagulation factor replacement, (2) antithrombotics, and (3) fibrinolysis inhibitors. Replacement of platelets and coagulation factors is generally the first step. Platelet transfusions are often used to maintain the platelet count above 20,000/μL, although the risk of spontaneous bleeding in the absence of anticoagulants occurs at platelet counts lower than 10,000/μL. Importantly, to avoid the risk of intracranial hemorrhage in patients who are to receive drotrecogin alfa (activated), platelet counts should be above 30,000/μL. In patients with severe sepsis, platelet turnover is accelerated; therefore, patients receiving platelets should be monitored with regular platelet counts. Although some patients are transfused with platelets to maintain the threshold, drotrecogin alfa (activated) is not usually given to patients who require multiple platelet transfusions or who have bleeding diathesis due to DIC. Coagulation factors can be replaced with cryoprecipitate or fresh frozen plasma. Each unit of these products has the equivalent of 1 U of fresh whole blood. Cryoprecipitate contains fibrinogen, factor VIII, factor XIII, VWF, and fibronectin. Each bag of cryoprecipitate has approximately 250 mg of fibrinogen and 100 U of factors VIII and XIII in a volume of approximately 10 mL. One unit of fresh frozen plasma contains all hemostatic proteins and albumin, globulins, and electrolytes in 1 U of fresh whole blood. It consists of approximately 225 mL of plasma. Cryoprecipitate may be administered to maintain the fibrinogen level above 100 mg/dL and the level of coagulation factors above 0.5 U/mL. Notably, platelets, cryoprecipitate, and fresh frozen plasma can transmit blood-borne infectious agents.

Low-dose heparin has been used in the past to interrupt the vicious cycle of coagulation, thrombosis, and factor depletion in patients with DIC. However, despite 20 years of calls for such studies, its efficacy has not been demonstrated in large-scale randomized clinical trials.235 Aoki et al reported that low-dose rhAPC corrected the clinical and laboratory abnormalities in patients with DIC more effectively than UFH.236 The 28-day all-cause mortality rates in patients treated with APC and heparin were 20.4% and 40%, respectively (p <0.05). In addition, patients treated with APC had significantly less bleeding (p = 0.009). Joyce et al performed a retrospective subgroup analysis to determine the effect of drotrecogin alfa (activated) on patients with severe sepsis and DIC.237 Treatment with drotrecogin alfa (activated) produced a 42% reduction in the relative risk of death compared with placebo (95% confidence interval = 58 to 19). Fibrinolytic inhibitors are not recommended in patients with severe sepsis and DIC because they may convert a hemorrhagic DIC to a thrombotic DIC. These agents are recommended only for patients with primary or excessive secondary fibrinolysis.

Temperature Management

An abnormality in temperature (i.e., ≤36°C or ≥38°C) is one of the four SIRS criteria and a manifestation of sepsis that will resolve once the infection is controlled. In most instances, mild hypothermia does not need to be actively corrected; however, hypothermia that leads to shivering can increase tissue oxygen requirements and usually can be managed by covering the patient with a blanket. If the patient is to receive large amounts of refrigerated blood, significant hypothermia can be prevented by the use of an in-line blood warmer. For every 1°C above 37°C, fever increases basal oxygen consumption 13%. In most instances, mild fever requires no treatment. However, fever in patients can aggravate pre-existing cardiac or pulmonary disease and worsen behavioral changes in patients with organic brain disease. Measures to lower core temperatures are prudent in patients with temperatures above 40°C to 41°C (104°F to 105.8°F). Steps taken in this situation can include antipyretics such as acetaminophen and measures such as a cooling blanket or tepid water sponging. Notably, however, cooling measures may induce shivering with its undesirable metabolic effects. Therefore, to prevent shivering, the patient can be treated with parenteral meperidine (25 to 50 mg). Muscle relaxants such as pancuronium or vecuronium may be required on rare occasions to control shivering and rigors in adequately sedated and mechanically ventilated patients who are hyperpyrexic.

In the past two decades, several scoring systems have been developed to identify and describe patient populations in the ICU. These scoring systems also have been useful for ensuring comparability in clinical trial populations. These outcome prediction models and organ dysfunction scores are discussed in detail in Chapter 6. Therefore, we confine our discussion specifically to points relevant to prognostic assessment in patients with severe sepsis. These models can be subdivided into two broad categories: outcome prediction models and organ dysfunction scoring systems.

Outcome Prediction Models

(See Chap. 6) General outcome prediction models include the APACHE, Mortality Probability Model (MPM), and the Second Simplified Acute Physiology Score (SAPS). The APACHE score has become the most commonly used survival prediction model in the critical care unit. APACHE II is a simplified version with a point score derived from initial values of 12 physiologic measurements, age, and previous health status. The APACHE III model was developed in 1991. It is complex, requires software to determine values, and is proprietary; therefore, most investigators have continued to use APACHE II. The MPM consists of two scores: MPM0, the admission model with 15 variables, and, MPM24, the 24-hour model that uses five of the admission variables and eight additional variables. The SAPS II uses 12 physiologic variables, age, type of admission, and three variables related to the nature of the underlying disease. When attempting to apply outcome prediction models at the bedside, three caveats should be kept in mind. (1) Outcome prediction models are incorporated into clinical trials of severe sepsis interventions to predict and evaluate mortality rate for cohorts of patients; at this time, it is inappropriate to use them to predict outcome or determine therapy for individual septic patients. (2) For the sake of comparability, the models must be used as originally validated. For example, the APACHE II model is validated for the use at the time of admission or within the first 24 hours. However, to obtain consistency in clinical trials, the APACHE II score is often computed from the worst values in the first 24 hours before randomization, not during the first 24 hours after admission. Because many of the physiologic variables used to compute the APACHE II score can be corrected within the first 24 hours in a patient with severe sepsis (e.g., pH, electrolytes), this trial design practice limits comparison to the original APACHE database. (3) There are other inherent limitations of outcome prediction models. For example, they are cumbersome to use outside a clinical trial setting, imprecise diagnoses can invalidate the results, they are poor discriminators, and they have lead-time bias.238,239 Lead-time bias is reflected by the fact that the admission source predicts hospital mortality rate independent of APACHE II. For example, a septic patient who is admitted to the ICU from an inpatient floor has a higher probability of death than one who is admitted from the emergency department.240

Organ Dysfunction Scoring Systems

The risk of death in patients with severe sepsis is a function of the number of dysfunctional organs. Organ dysfunction scoring systems supply a tool to characterize the severity of illness in patients with sepsis. Commonly used organ dysfunction scoring systems in patients with severe sepsis include the Sequential Organ Failure Assessment (SOFA), the Multiple Organ Dysfunction Score, the Logistic Organ Dysfunction System, and the Brussels Score. Vincent et al used the SOFA Score to assess the incidence of organ dysfunction in critically ill patients.241 They reported that mortality rates were lowest in patients without organ dysfunction (9%) and that rates increased progressively in patients with dysfunction in one (22%), two (38%), three (69%), and four or more organs (83%; p <0.00001). Using logistic regression analysis of the results of a simple Multiple System Organ Failure Score, Hebert et al reported that the type of dysfunctional organ system was also linked to mortality rate.242 The odds ratios for covariates most predictive of mortality rate by organ system were hematologic (odds ratio = 6.2), neurologic (odds ratio = 4.4), hepatic (odds ratio = 3.4), and cardiovascular (odds ratio = 2.6). However, like the outcome prediction models, there are caveats when attempting to employ organ dysfunction scores for the management of individual patients with severe sepsis. (1) The nature of the score is a morbidity, not a mortality, score. (2) Organ dysfunction is an analog parameter that varies from mild laboratory abnormalities to overt organ failure. (3) The degree of organ dysfunction is not static; instead, it changes over time. For example, the platelet count can change considerably over time. Therefore, trends are sometimes more important than absolute values. (4) Because the degree of organ dysfunction must be evaluated over time, by different individuals, and in a variety of community and academic settings, any variables employed in organ dysfunction scores should be objective, routinely measured, and easy to assess.

The authors thank George H. Karam, MD, Paul Garvey Manship Professor of Medicine, Louisiana State University School of Medicine, New Orleans for his thoughtful review of the infectious disease and antibiotic issues in this chapter.