Chapter 151: Sepsis

Sepsis is a heterogeneous syndrome characterized by widespread inflammation and potential organ harm initiated by a microorganism. Although gram-positive and gram-negative bacteria account for the majority of cases, fungi, viruses, mycobacteria, rickettsiae, and protozoans can trigger sepsis. Invasion of the blood is not necessary to develop or identify sepsis, which is determined by the host response to the infectious insult. As sepsis severity increases, a multifactorial series of events lead to impairments in oxygen delivery, secondary to macro- and microvascular malperfusion, as well as direct cellular damage secondary to inflammation. Eventually, multisystem organ failure occurs, and mortality is high.

The varying clinical presentation, differences in coding, and methodologic differences between studies lead to a wide range of estimates of the annual incidence of severe sepsis, ranging from 300 to 1000 cases/100,000 persons per year.¹ Over 500,000 patients each year present to the ED with suspected severe sepsis, the largest group of all sepsis patients hospitalized. The incidence is increasing and is multifactorial in origin; a key component of this increase relates to an aging patient population, which is not surprising given the fact that sepsis incidence increases >100-fold with age (0.2 per 1000 in children age 10 to 14 years to 26.2 per 1000 in those >85 years of age).²

The mean ED length of stay of a patient with sepsis is 5 hours. Once admitted, more than half of patients with severe sepsis will require care in an intermediate or intensive care unit,² where it represents the leading cause of death. Despite advances in care, mortality rates from severe sepsis remain high, with approximately 20% dying³ during hospitalization in optimal clinical trial scenarios^4,5; this rate approaches 50% when considering the sicker subset of those with septic shock.⁶ These mortality rates exceed many other high-visibility acute care conditions such as acute myocardial infarction,⁷ massive pulmonary embolism,⁸ and cerebrovascular accident.⁹ Morbidity is high and prolonged, and a long-term deficit in cognition and functioning is common.¹⁰ Finally, sepsis care is costly; estimates from 10 years ago suggest a mean case cost of $22,100 with annual national costs reaching $16.7 billion,² figures that have certainly increased in the last decade.

Since 1987, gram-positive bacteria outside of the surgical setting are the predominant pathogens of sepsis.¹¹ With the rise of antimicrobial resistance, methicillin-resistant Staphylococcus aureus, vancomycin-resistant Enterococcus, and other multidrug-resistant organisms are more common.¹² Similarly, the incidence of fungi as the source has risen, particularly in immunosuppressed patients. The most likely causative microorganism varies based on the likelihood of exposure to drug-resistant microorganisms (due to recent healthcare exposures) and the anatomic site of infection, with pneumonia, intra-abdominal, urinary, and skin/soft tissue being the most common locations.

Sepsis syndromes are a continuum with specific definitions^13,14 (Table 151–1). Although the definitions provide a conceptual and practical framework for recognition of the systemic inflammatory response to infection, definitions often are not sensitive or specific in the clinical setting. Classically, sepsis was defined as suspected or confirmed infection with evidence of systemic inflammation (demonstrated either through evidence of the systemic immune response syndrome or laboratory abnormalities). Severe sepsis was sepsis plus evidence of new organ dysfunction thought to be secondary to tissue hypoperfusion. Septic shock exists when cardiovascular failure occurs, evidenced as elevated serum lactate, persistent hypotension, or need for vasopressors despite adequate fluid resuscitation. It has a poor prognosis.

The Third International Conference of Sepsis (Sepsis-3) in 2016 altered these definitions, removing severe sepsis as a term.¹⁵ In Sepsis-3, septic shock identifies those requiring vasopressors after adequate resuscitation and with an elevated lactate; all others have forms of sepsis. These new definitions are simpler and better identify the highest risk group. However, the new definitions may diminish the sensitivity of detecting those in the middle ground (previously called severe sepsis) where death and morbidity rates are elevated compared to those simply infected.

An overlapping and potentially confusing recent mandate is the Centers for Medicare Services quality measure called SEP-1.¹⁶ Enacted prior to the Third International Conference definitions, it identifies sepsis populations and actions associated with better care. Core measures and definitions of sepsis and septic shock, as defined by Centers for Medicare Services, include the following:

• Sepsis: two Systemic Inflammatory Response Syndrome (SIRS) criteria plus suspected infection

• Severe sepsis: above plus lactate > 2 or organ dysfunction

• Septic shock: Severe sepsis with hypoperfusion despite adequate fluid resuscitation or lactate > 4

• Between initial recognition and 3 hours, lactate measurement (if not done), blood and other cultures, antibiotic therapy directed at specific source or broadly, and initial fluid therapy with 30 cc/kg of crystalloid is targeted.

• Reassessment occurs between 3-6 hours of meeting sepsis criteria and being treated, using clinical or hemodynamic/oxygenation measurements and a repeat lactate (if initially elevated), to judge adequacy of volume status and tissue perfusion interventions.

qSOFA

The “qSOFA” tool (for “quick Sequential Organ Failure Assessment”) was developed in conjunction with Sepsis-3. It is a simple screening tool derived to identify patients at higher risk of death. Criteria include altered mental status, respiratory rate ≥ 22, and systolic BP ≤100, with a score ≥2 indicating a high risk for poor outcome. Previous tools that helped with assessment—classic SOFA, MEDS, SIRS—all have flaws that limit usefulness. However, qSOFA requires validation, and it requires rather than replaces judgment about potential/actual infection and > one system involvement—that is, sepsis.^17,15,18 One particular concern with qSOFA is that like the newer definitions, it enhances specificity for short term mortality assessment at the expense of lower sensitivity; it may not identify those where disease progresses later, and so would fail to allow for aggressive early treatment. Whether qSOFA aids early ED care overall requires further study.

SYSTEMIC INFLAMMATORY RESPONSE SYNDROME PROGNOSIS

Having systemic inflammatory response syndrome criteria does not confirm the presence of infection or sepsis because these features are shared by many other noninfectious conditions such as trauma, pancreatitis, and burns (Figure 151–1). The systemic inflammatory response syndrome response is not a diagnosis or a good indicator of outcome; it is a crude means of stratification of patients with systemic inflammation. In a prospective study of the epidemiology of patients demonstrating systemic inflammatory response syndrome (infectious and noninfectious), mortality rates were 3% in patients without systemic inflammatory response syndrome, 6% in those meeting two criteria, 10% in those meeting three criteria, and 17% in those meeting all four criteria.¹⁴ Death rates were similar whether or not patients had positive blood cultures.

Complementary methods of classification of sepsis based primarily on physiologic abnormalities include the third Acute Physiology and Chronic Health Evaluation acuity system,¹⁹ the Sequential Organ Failure Assessment score, and the Mortality in Emergency Department Score²⁰; each allows for prognostication. These scoring systems are typically limited to research and are not for guiding patient care. Any evidence of septic shock confers the highest mortality risk assessment.

Serum lactate across a wide spectrum²¹ provides excellent prognostic data in those with sepsis, but is not a singular test to diagnose or exclude sepsis. Lactate has traditionally been attributed to anaerobic metabolism secondary to tissue hypoperfusion; it also accumulates from changes in aerobic metabolism that occur in response to widespread inflammation.²² As lactate increases, the risk of mortality rises, with the degree of lactate elevation and hypotension being independent predictors of mortality.^23,24 Specifically, 28-day mortality rate associated with modest elevation of serum lactate to 2 to 4 mmol/L approaches 15% even in those patients without hypotension,²⁵ and the rate continues to rise with increasing lactate levels. In general, venous or arterial lactate levels on either point-of-care bedside devices or measured in a central laboratory are useful if done with good clinical technique²⁶; even better are serial levels using the same method of sampling (arterial or venous) to identify improvement that mirrors clinical responses.²⁶ This is called lactate clearance and is defined as a drop over time; lactate clearance provides an independent prognostic value in addition to vital signs and other measurements. Patients with sepsis who do not clear lactate (adequate is at least a 10% drop with therapy over the first few hours) have a higher mortality rate than patients who do clear the lactate.^27,28 The major limitation is that lactate is nonspecific and can be elevated in a number of conditions, and it must be used together with clinical judgment.

In sepsis, the host immune response fails to control and/or overreacts to invasive pathogens, leading to two critical events.^29,30 The first event involves marked abnormalities in the inflammatory response in the host. The host response typically varies from a hyperinflammatory response in the early stages of sepsis, to a blunted inflammatory response in the later stages of sepsis, which leads to an increased risk of secondary hospital-acquired infections. The blunted inflammatory response results in programmed death of key immune, epithelial, and endothelial cells, leading to tissue injury and perpetuating multiorgan dysfunction.

The second event is an imbalance in procoagulant and anticoagulant functioning; in the most extreme of situations, this results in the clinical syndrome of disseminated intravascular coagulation. Disseminated intravascular coagulation results in micro- and macrovascular clot formation, impaired microvascular tissue perfusion, and thrombosis of small vessels. Notably, subclinical disseminated intravascular coagulation is often lurking despite relatively normal basic laboratory findings, with impaired microvascular perfusion (detectable in research settings) that is associated with a worse prognosis.³¹ Continued microvascular ischemia likely contributes to organ failure, as well as to the release of proinflammatory intracellular contents, which further stimulates the innate immune response and perpetuates the underlying pathology.³² As these events progress, they intensify the inflammatory response and a destructive cycle ensues, as illustrated in Figure 151–2. At this time, there are no specific therapies to remedy microvascular dysfunction.

While some presentations of severe sepsis are immediately clinically apparent, sepsis can present in a subtle or occult manner, particularly early in the course. Given that systemic inflammation produces physiologic changes, vital sign abnormalities—notably fever, hypotension, and/or tachycardia—are a hint to sepsis, recognizing that many patients with these findings may have another cause. In ED patients with undifferentiated hypotension, 40% will ultimately have an infectious cause of symptoms.³³

Although traditionally sepsis is categorized as an example of distributive shock (peripheral vasodilation evidenced by warm extremities with a compensatory increased cardiac output), this presentation does not accurately describe all patients with sepsis. ED patients with sepsis are often volume depleted from decreased intake and increased fluid losses (from emesis, diarrhea, or insensible losses associated with fever and tachypnea). Intravascular volume depletion directly affects preload, cardiac output, and ultimately peripheral perfusion. Further complicating matters is septic cardiomyopathy, a reversible process characterized by impaired systolic function and diastolic relaxation.³⁴ Finally, the combination of intravascular volume depletion and septic cardiomyopathy may manifest as "cold shock," impaired peripheral perfusion and cool extremities.

PULMONARY INJURY

Widespread inflammation secondary to sepsis commonly affects pulmonary function even in the absence of pneumonia. Acute lung injury is common and may result in acute respiratory distress syndrome, which is characterized by new lung edema from increased alveolar and capillary permeability. Classification of acute respiratory distress syndrome is based primarily on the degree of hypoxemia.³⁵ The three mutually exclusive categories are mild (Pa_O2 divided by fraction of inspired oxygen [FIO2] of 200 to 300), moderate (Pa_O2/FIO2 of 100 to 200), and severe (Pa_O2/FIO2 <100), within 1 week of a clinical insult or change in respiratory status coupled with new bilateral pulmonary infiltrates on chest radiographs not explained by effusions, lung collapse, or nodules and not fully explained by heart failure or volume overload. Clinically, severe refractory hypoxemia, noncompliant lungs noted on mechanical ventilation, and a chest radiograph showing bilateral pulmonary alveolar infiltrates suggest the diagnosis. Mortality increases from 27% to 45% with increasing severity of acute respiratory distress syndrome.

RENAL INJURY

The kidney is another common sepsis target; acute kidney injury can present with azotemia, oliguria, or anuria. Factors increasing acute kidney injury risk include pre-existing kidney dysfunction or vasculopathy, depth and duration of hypotension, dehydration, and use of nephrotoxic substances (e.g., aminoglycoside antibiotics, nonionic intravenous contrast). Renal ischemic injury from hypoperfusion is a major factor in the pathogenesis of acute kidney injury in sepsis, although toxic products resulting from neutrophil-endothelial interactions, endothelial damage by various mediators, reperfusion injury, and microvascular thrombosis also contribute.

HEPATIC INJURY

The most frequent hepatic abnormality is cholestatic jaundice, although it occurs infrequently. Increased concentrations of transaminase, alkaline phosphatase (one to three times the normal level), and bilirubin (usually not >10 milligrams/dL) may be observed. Marked elevations of transaminases or bilirubin are less common unless septic shock is present; if seen, consider a biliary source of infection. Smaller elevations in liver function tests can result from intermittent or prolonged macro- or microvascular hypoperfusion and ischemia or can be secondary to direct endotoxin, cytokines, or immune complex damage. Red blood cell hemolysis from microvascular coagulation can also rarely cause jaundice.

GI CHANGES

The most common GI manifestation of sepsis is ileus, which may persist for days after shock resolves. Major blood loss secondary to upper GI bleeding is rare in septic patients. Minor GI blood loss within 24 hours of developing severe sepsis can result from painless erosions in the mucosal layer of the stomach or duodenum.

HEMATOLOGIC CHANGES

Multiple abnormalities are possible in the hematologic system in the setting of sepsis, including neutropenia or neutrophilia, thrombocytopenia, or disseminated intravascular coagulation. Neutrophilic leukocytosis with a "left shift" results from demargination and release of newer granulocytes from the marrow storage pools. However, the presence of excessive bands (the immature neutrophil) is neither sensitive nor specific for infection. Neutropenia occurs rarely and is associated with an increase in mortality; it results from increased peripheral use of neutrophils, damage to neutrophils by bacterial by-products, or depression of marrow granulocyte production by inflammatory mediators. Functional neutropenia also creates a relative immunosuppression, particularly later in patients' hospital course. Both red cell production and survival decrease during sepsis, but anemia is not expected unless it existed prior to the infection, the infection is extremely prolonged, or there is concomitant bleeding.

Thrombocytopenia may arise as a consequence of disseminated intravascular coagulation but is present in >30% of cases of sepsis even in the absence of overt disseminated intravascular coagulation, and lower platelet levels are associated with worse outcomes.^36,37 Proposed mechanisms of thrombocytopenia include inhibition of thrombopoiesis, increased platelet turnover (due to a consumptive coagulopathy), increased endothelial adherence, and increased destruction secondary to immunologic mechanisms.

Finally, fulminant disseminated intravascular coagulation where clinical clotting and bleeding exist is rare but associated with a very poor prognosis.³⁰ Occult disseminated intravascular coagulation, with microvascular flow abnormalities, subclinical increased activation of the coagulation system, and evidence of increased fibrinogen split products in the absence of overt bleeding and clotting, is much more common. The activation of the hemostatic (clotting) system is due primarily to the activation of the extrinsic pathway of clotting. The fibrinolytic system is also activated in sepsis and plays an important role in limiting fibrin deposition in the microcirculation. The release of tissue plasminogen activator activates the fibrinolytic system, at least initially in sepsis. As sepsis progresses, there is an increased release of plasminogen activator inhibitor 1, which blocks plasmin generation and thus contributes to fibrin deposition in the microcirculation and subsequent multiple-organ failure. Laboratory studies suggesting the presence of disseminated intravascular coagulation include thrombocytopenia, prolonged prothrombin and activated partial thromboplastin values, decreased fibrinogen and antithrombin III levels, and increased fibrin monomer, fibrin split products, and D-dimer levels.

METABOLIC CHANGES

Sepsis induces multiple metabolic changes. Abnormalities in lactate metabolism due to tissue hypoperfusion with resultant anaerobic metabolism as well as increased aerobic production of lactate have been discussed previously (see "Systemic Inflammatory Response Syndrome Prognosis"). Hyperglycemia is seen even in patients without a history of diabetes; in this latter group, it is associated with a worse prognosis, in contrast to less impact of glucose elevations in those with known diabetes.³⁸ Hypoglycemia with glucose levels as low as 10 to 20 milligrams/dL is reported but uncommon, and may result due to depletion of hepatic glycogen and inhibition of gluconeogenesis and/or adrenal insufficiency. Adrenal insufficiency can occur,³⁹ caused by hypoperfusion of the adrenal glands, adrenal or pituitary hemorrhage, cytokine dysfunction of the adrenals, drug-induced hypermetabolism, inhibition of steroidogenesis by chemotherapeutics (e.g., ketoconazole), and desensitization of glucocorticoid responsiveness at the cellular level.

SKIN

There are five potential cutaneous manifestations of sepsis: direct bacterial involvement of the skin and underlying soft tissues (cellulitis, erysipelas, and fasciitis); lesions from hematogenous seeding of the skin or the underlying tissue (petechiae, pustules, cellulitis, ecthyma gangrenosum); lesions from hypotension and/or disseminated intravascular coagulation (acrocyanosis and necrosis of peripheral tissues); lesions from intravascular infections (microemboli and/or immune complex vasculitis); and lesions caused by toxins (toxic shock syndrome). Look for any necrotizing or surgical source of sepsis, and recall that the toxic shock syndromes present in an overlapping fashion (see chapter 150, "Toxic Shock Syndromes").

In the ED, sepsis is a clinical diagnosis predicated on suspicion or confirmation of infection, systemic inflammation, and evidence of new organ dysfunction and/or tissue hypoperfusion.⁴⁰ Septic shock is present or likely in any individual with the preceding signs (Table 151–1) and a systolic blood pressure of <90 mm Hg (or 2 standard deviations below normal for age in children) after an initial fluid bolus (often cited as 20 to 30 mL/kg crystalloid, generally 1.5 to 3 L in adults), or who displays evidence of inadequate organ perfusion.

Many patients with minor infections can meet consensus criteria for sepsis due to evidence of systemic inflammation. If there is new organ dysfunction or tissue hypoperfusion secondary to sepsis, consider more aggressive assessments and interventions. However, in the absence of new organ failure, the diagnosis of sepsis does not necessarily imply a poor prognosis.

The differential diagnosis of severe sepsis and septic shock includes other causes of shock. This includes cardiogenic, hypovolemic, anaphylactic, neurogenic, or obstructive shock (pulmonary embolism, cardiac tamponade), or endocrine disorders (adrenal insufficiency, thyroid storm) that mimic sepsis. Given the high frequency of sepsis in ED patients with undifferentiated shock (40%), think of this cause when uncertain. Children may not have hypotension until very late in the course of the disease process due to an ability to upregulate their heart rate as a compensatory response to tissue hypoperfusion. This means considering compensated shock and occult sepsis in infected children who do not appear well, even with normotension.

Once sepsis is diagnosed, seek the source, but do not withhold interventions such as resuscitative measures and antimicrobials. Often the source is clear, as the patient may present with signs and symptoms attributable to a pulmonary, genitourinary, skin and soft tissue, or intra-abdominal source. However, a source of infection may not be readily apparent or may be difficult to detect because of altered mental status.

The most common sepsis trigger is acute bacterial pneumonia, although viral and bacterial etiologies are difficult to differentiate initially. The most frequent causative organisms are Streptococcus pneumoniae, S. aureus, gram-negative bacilli, and Legionella pneumophila. A chest radiograph may identify air space changes and pneumonia, although radiograph findings can lag behind the clinical presentation (see also chapter 65, "Pneumonia and Pulmonary Infiltrates").

Acute pyelonephritis, typically due to gram-negative enteric bacteria or enterococci, is another frequent cause of severe sepsis. Other abdominal triggers include cholecystitis and cholangitis, each rare but particularly devastating sources of septic shock, and both require immediate surgical assessment. If there is abdominal tenderness or peritonitis, the possibilities include perforated viscous, appendicitis, diffuse colitis, or intra-abdominal abscesses. Acute pancreatitis can result in a presentation identical to that of septic shock due to widespread inflammation.⁴¹ If peritonitis or organ ischemia/necrosis exists, early surgical consultation is needed. In women of childbearing age, septic abortion and postpartum endometritis or myometritis are unusual etiologies of septic shock.

The most common skin and soft tissue infection triggering a sepsis syndrome is cellulitis due to S. aureus or Streptococcus pyogenes. Soft tissue infections caused by gram-negative organisms are indistinguishable from those caused by staphylococci or streptococci. Shock associated with a generalized erythematous macular rash may be a toxic shock syndrome. Necrotizing soft tissue infections are usually seen in immunocompromised patients, diabetic patients, or patients with a history of poor vascular circulation (see chapter 152, "Soft Tissue Infections"). Pay close attention to skin and soft tissue infections, particularly in elderly or obtunded patients. Include an assessment for decubitus ulcers and a genitourinary assessment to look for a necrotizing infection such as Fournier's gangrene.

Individuals without an obvious source of septic shock may have a primary bacteremia or endocarditis. The most prevalent causes of primary bacteremia in outpatients are S. aureus, S. pneumoniae, and Neisseria meningitidis. Pseudomonas aeruginosa and other gram-negative bacteria are occasional causes of bacteremia and endocarditis in injection drug users. Secondary bacteremia from indwelling medical devices, including intraperitoneal or intravascular dialysis catheters, chemotherapy ports, peripherally inserted central catheters, ventriculoperitoneal shunts, and pacemaker/defibrillators, is possible. Although bacteremia can be suspected, it takes time for a blood culture to be fully assessed, making ED diagnosis difficult.

Acute bacterial meningitis is a devastating but rare cause of septic shock. Community-acquired meningitis with shock is usually caused by S. pneumoniae or N. meningitidis (see chapter 174, "Central Nervous System and Spinal Infections"). Brain and spinal abscesses, subdural or epidural empyemas, and viral CNS infections are seldom associated with shock on the initial presentation. Shock is also unusual in neurosurgical patients with S. aureus or enteric gram-negative meningitis secondary to neurosurgical procedure or skull fracture.

LABORATORY TESTING AND IMAGING

Tests are useful to assess the general hematologic and metabolic state of the patient and aid in detecting occult bacterial infection, uncovering a specific microbial cause of infection, and identifying occult severe sepsis. Common tests include a CBC with platelet count; serum electrolytes (including calcium and glucose); renal function panel (BUN and creatinine levels); lactic acid level; liver function panel (bilirubin, alkaline phosphate, and aspartate and alanine aminotransferase levels); and urinalysis. An arterial blood gas will aid in assessing the metabolic, ventilatory, and oxygenation adequacy in select patients. Type and screening of red blood cells is wise with any hypotension or anemia. In the setting of active bleeding or suspected disseminated intravascular coagulation, consider prothrombin time, activated partial thromboplastin time, fibrinogen, and D-dimer.

Given the frequency of pneumonia or concomitant acute respiratory distress syndrome, obtain a chest radiograph. Abdominal plain radiographs or CT scans are needed in patients if perforation or diffuse inflammation is suspected as a potential source of intra-abdominal sepsis, and US can identify a biliary source. Soft tissue CT scans to assess for the presence of free air in the tissue or deep space abscesses may assist in making the diagnosis of necrotizing skin and soft tissue infections. Perform head CT and lumbar puncture in any patient with potential meningitis, but do not delay antibiotics pending these tests (see chapter 174). If considering a spinal epidural abscess, an MRI is obtained after resuscitation.

In adults with sepsis, obtain at least two separate sets of specimens for blood culture from different venipuncture sites when possible. Order Gram staining and culture of secretions from any potential site of infection, if possible.

Other tests of interest in the assessment of sepsis include C-reactive protein and procalcitonin, both of which reflect systemic inflammation; neither of these tests can exclude sepsis, and they are not routinely needed in the ED.^41,42 Both of these tests may be better suited to guide the necessity of initiating or continuing antimicrobial therapy outside of sepsis or after a treatment course.⁴³

The cornerstones of the initial treatment and stabilization of severe sepsis are early recognition, early reversal of hemodynamic compromise, and early infection control.⁴⁰ Base the resuscitation on administering fluids, frequently assessing response, and adding adjunct therapies including vasopressors based on the conditions. The specific method of titrating resuscitation is less important than treating early and aggressively.

The goals of resuscitation are to improve preload, tissue perfusion, and oxygen delivery. There is no "set amount of fluid," although most patients will require a total (bolus plus infusion) of 2 to 5 L of crystalloid in the first 6 hours to achieve optimal outcomes. Similarly, do not delay vasopressors when blood pressure does not respond to volume or if volume overload seems likely. Administer appropriate antibiotics, and remove any nidus of infection. Each of these interventions will be discussed in detail below. Once the patient is stabilized, other interventions such as appropriate management of oxygenation and ventilation, fever control to reduce metabolic demand, and control of hyperglycemia may be needed to improve patient outcomes.

EARLY QUANTITATIVE RESUSCITATION

The primary deficit leading to poor outcomes in shock states (typically studied in hemorrhagic shock) is a mismatch between oxygen supply and demand,⁴⁴ and after a certain time point, processes initiated by this mismatch are irreversible. In 2001, Rivers et al⁴⁵ described the use of an early, structured hemodynamic resuscitation protocol driven by a central venous catheter with continuous oximetric capability to titrate fluids to central venous pressure, vasopressors to mean arterial pressure, and transfusion of blood or inotropes to central venous oxygen saturation (S_CVO2). This protocol, termed "early goal-directed therapy," decreased mortality when compared with standard care, although all patients had central venous catheters placed early. Follow-up uncontrolled observational studies confirmed that even partial use of this approach lowered mortality in settings compared with previous times without aggressive early sepsis detection or care.^46-50

In a second large trial, Jones et al⁴ randomized patients to the Rivers early goal-directed therapy protocol or to a protocol that measured lactate clearance of ≥10% rather than S_CVO2. The lactate clearance protocol was noninferior to continuous S_CVO2 monitoring in the setting of ED-based resuscitation of septic shock. Lactate clearance–guided therapy (titrating fluids and vasopressors) became an alternative to invasive S_CVO2 monitoring and early goal-directed therapy. Although the ideal lactate clearance target remains unclear, a minimum of 10% relative lactate clearance^27,28 and normalization of lactate to <2 mmol/L emerged as goals.^40,51 Still, about one third of patients with severe sepsis have a normal lactate,⁵¹ and up to one half or more have a normal S_CVO2,⁵² limiting either tool in guiding therapy.

The newly released ProCESS trial⁵³ compared three ED treatment groups after early identification and antibiotics: protocol-based resuscitation using invasive central venous monitoring (the Rivers early goal-directed therapy protocol), protocol-based care using clinical parameters such as blood pressure and shock index (but no catheter-driven care), and nonprotocol usual care chosen by the bedside physician(s). These resuscitative approaches all delivered different care, but all groups received aggressive fluids and pressors in addition to early recognition and antibiotic use; however, no one approach was superior in terms of morbidity or mortality. Thus, the ProCESS trial emphasizes that the early recognition of sepsis, administration of antimicrobials, adequate volume resuscitation, and assessment of the adequacy of circulation are the important elements that improve outcomes, not any specific path of resuscitation. These observations allow providers or sites the flexibility of crafting the best approach to care but show that mandatory central venous catheterization and monitoring of central venous pressure or S_CVO2 are not necessary for all patients with sepsis.

VOLUME RESUSCITATION

When resuscitating an ED patient with sepsis, first assess and replenish circulating volume, typically with an initial 20 to 30 milligram/kg crystalloid bolus, preferably through large-bore IVs. This amounts to an approximate 1- to 2-L bolus of saline or lactated Ringer's solution in a 70-kg adult, although some patients require more or less. Saline can produce a hyperchloremic metabolic acidosis if used in large-volume resuscitation, causing some to switch to lactated Ringer's if large volumes are planned. Colloids are not needed in early sepsis care, and hydroxyethyl starch can worsen acute kidney injury.

The most important variable process in volume replacement is determining if a patient is volume responsive. There are several more sophisticated measurements that can be used to make this determination such as cardiac output and stroke volume variation. The latter entails lifting the legs of a supine patient for 60 seconds; improved blood pressure or less variation in the peak blood pressure wave (if an arterial line is present) identifies a volume-responsive patient who will likely benefit from more fluid. If these tools are unavailable, assess the need for further volume expansion with empiric crystalloid boluses until the patient fails to demonstrate a physiologic or hemodynamic response (e.g., rise in systolic or mean pressure, decrease in heart rate, improvement in peripheral pulses or extremity perfusion, increase in urinary output).

VASOPRESSORS

Once the patient fails to respond to further intravenous volume expansion, provide adequate perfusion pressure with vasoactive agents. In general, a mean arterial pressure goal of 65 mm Hg is sought; routine targeting of a higher mean arterial pressure does not aid outcomes, although in some patients, it may be necessary.⁵⁴ A mean arterial pressure is preferable to a specific systolic blood pressure goal; however, use of systolic pressure is often preferred by clinicians, who often seek a goal of 90 mm Hg or higher.

Ideally, deliver vasoactive agents through a central venous line to limit extravasation and resultant tissue necrosis. If a central line is unavailable, use a large, secured peripheral IV temporarily for vasopressor administration. In septic shock, norepinephrine at a dose of 0.5 to 30 micrograms/min is the best first choice since the dual α- and β-adrenergic effects result in peripheral vasoconstriction and cardiac inotropy. Dopamine has a higher rate of complications, most notably dysrhythmias and failure, compared with norepinephrine and is no longer routinely recommended.⁵⁵ Vasopressin is a second-line agent and may allow for the down-titration of the norepinephrine dose.⁵⁶ Give vasopressin as a constant infusion at a rate of 0.03 or 0.04 U/min. Do not titrate the dose, because higher rates are associated with vasospasm and high morbidity. Epinephrine at a dose of 1 to 20 micrograms/min appears safe and equivalent to norepinephrine when dosed appropriately,⁵⁷ although the risk of medication dosing errors related to epinephrine concentration may make norepinephrine a superior choice. If tachydysrhythmias are a problem, phenylephrine is an option as a pure α-adrenergic agonist.

CENTRAL VENOUS OXYGENATION

After volume repletion and perfusion pressure optimization, if tissue perfusion appears to continue to be compromised (cool extremities, poor pulses, worsening organ function), one may assess oxygen balance. Although not needed routinely, insertion of a continuous central venous oxygen saturation (S_CVO2) measuring catheter can aid additional therapy as outlined in the early goal-directed therapy approach.⁴⁵ If S_CVO2 is less than 70%, it implies a relative oxygen supply and demand mismatch.

LACTATE CLEARANCE

An alternative is to guide therapy through the use of lactate clearance, because a relative decrease in lactate over time suggests a restoration of adequate tissue perfusion. Measure lactate using the same method 1 to 2 hours apart; improvement of 10% or more is associated with improved clinical outcomes.

TREAT INFECTION

Give broad-spectrum antibiotics as early as possible for severe sepsis.⁴⁰ Combination antibiotic therapy as opposed to monotherapy leads to improved outcomes, potentially due to higher rates of bactericidal activity.⁵⁸ Recommendations for antibiotic regimens are in Table 151–2. Therapy can later be tailored based on clinical response and microbiologic data. Given the rising rates of community-acquired methicillin-resistant S. aureus (CA-MRSA), ensure this is covered along with gram-negative pathogens and anaerobic organisms. Vancomycin is often underdosed in clinical practice, which can impair recovery⁵⁸; guidelines suggest an initial vancomycin dose of 25 to 30 milligrams/kg in critically ill patients.⁵⁹ In immunosuppressed patients, consider antifungal or antiviral agents.

The Surviving Sepsis Campaign recommends giving antibiotics as soon as feasible, ideally within 1 hour of the recognition of severe sepsis and/or within 3 hours of triage. However, these are suggestions, not a standard of care,⁴⁰ and these recommendations are often not achievable.⁶⁰ Whether antimicrobials are best given initially or only after hemodynamic stabilization of the patient is unclear; given the poor prognosis associated with delays in either resuscitation or infection therapies, we recommend delivering both concurrently. Ideally, sample any organ infectious source and obtain blood cultures prior to the initiation of antibiotics, but do not delay the initiation of antibiotics to obtain cultures.⁴⁰

Infection control is not confined to antimicrobial administration; it also includes addressing potential surgical sources of infection. Drain any source and remove indwelling vascular lines and other medical devices suspected as infected once the patient is stabilized.

OTHER THERAPIES: VENTILATION, GLYCEMIC CONTROL, ACTIVATED PROTEIN C, AND STEROIDS

Sepsis is the leading cause of acute respiratory distress syndrome. Data from a large study of patients with acute respiratory distress syndrome found that low tidal volume ventilation (6 mL/kg ideal body weight) leads to superior outcomes compared with large (12 mL/kg) volume ventilation,⁶¹ with the latter likely creating barotrauma from overdistension of the alveoli. Thus, low tidal volume ventilation with a positive end expiratory pressure adequate to prevent alveolar collapse is recommended. Although starting at a low tidal volume is recommended,⁴⁰ it is unclear whether a fixed volume of 6 mL/kg is best for all; some titration is permissible. To accommodate low tidal volumes, patients may be allowed some hypercapnia as long as the accompanying acidosis does not threaten hemodynamic deterioration.

Multiple trials note a link between hyperglycemia and worsened outcomes in the setting of sepsis, notably in patients without pre-existing diabetes. Although there is evidence of improved outcomes in patients who achieve tight glucose control (target glucose level of 80 to 110 milligrams/dL), the high rate of hypoglycemia occurring with attempts at this targeted glycemia may explain increased harm seen.⁶² More modest efforts (seeking blood glucose <180 mg/dL) allow for similar outcomes with less variation.⁶² Glucose control is most important after initial care and is best if protocolized, especially for patients receiving extended sepsis care in the ED.

Therapies to manipulate the coagulation cascade are unproven; the most promising candidate to date, activated protein C, failed to improve outcomes and is not used.⁶³ Systemic corticosteroids are the only anti-inflammatory agent in use in clinical practice but are controversial. Despite early literature suggesting a mortality benefit,⁶⁴ recent data failed to confirm these findings.⁶⁵ Corticosteroids shorten the time to shock reversal and are adjunctive agents in patients with refractory hemodynamic shock (i.e., requiring more than one vasopressor or active upward titration of vasopressors after adequate volume restoration). Stimulation testing using adrenocorticotropic hormone is not clinically helpful. Although even a single dose of etomidate may alter adrenal activity as measured by laboratory tests,⁶⁶ there is no clear evidence that the use of etomidate to facilitate endotracheal intubation alters any clinically important outcome.

PATIENTS WITH POSITIVE BLOOD CULTURES

There is no consensus for interpretation or management of positive ED blood cultures. Before assessing positive blood culture reports and choosing a plan, you should know local typical pathogens and contaminants along with patient data, including reason for seeking care, past medical history, immune status, and current condition at the time the positive culture was called to the clinician. Corynebacterium spp., Bacillus spp., and Propionibacterium acnes rarely represent true bacteremia or require action. Other pathogens such as viridans group streptococci, enterococci, and coagulase-negative Staphylococcus (the most common organism reported on blood culture notifications) may represent a pathogen or require new care in a given patient. In both ambulatory and ED patients, blood cultures are positive in less than 10% of cases, with only one third to two thirds of those cultures representing true bacteremia.⁶⁷

An important predictor of true-positive first blood culture is persistence of the organism on subsequent culture if the latter was done separately (time and location); in a discharged patient, this requires re-evaluation, either in the ED or by another provider. Decisions about antibiotics or admission are guided by the pathogen and clinical grounds: All highly lethal pathogens should receive therapy and admission, whereas some pathogens in blood culture may clear with or without therapy, with a plan made after all current data are assimilated.

POSTSPLENECTOMY PATIENTS

Patients without a spleen are at increased risk for infection with encapsulated species such as Salmonella or Haemophilus influenzae. Table 151–2 details treatment recommendations of these patients. Although not relevant at the time of initial sepsis care, these patients should have appropriate bacterial immunization following a splenectomy.

NEUTROPENIC FEVER

Due to an impaired immune system, neutropenic patients are at increased risk for developing sepsis and having worse outcomes. Any fever in a neutropenic patient is suspicious for infection, and all patients should have source organ and blood cultures drawn followed by empiric antibiotics. After conferring with an oncologist, some patients may be treated at home with oral broad-spectrum agents (often including a quinolone) if they look well and are closely followed; the rest are admitted.