The Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3) recently defined sepsis as a “life-threatening organ dysfunction caused by a dysregulated host response to infection.” Sepsis is a medical emergency with mortality rates of 18% to 29% in developed countries and as high as 80% in low-income countries. In the United States alone there are reportedly 750,000 cases per year.
The oldest reports of sepsis date back to 1600 BCE from a papyrus discovered in Luxor, Egypt. This is probably a copy of another dated 3000 BCE and is best described as a surgical treatise detailing 48 cases of traumatic lesions from wounds, fractures, and dislocations exploring symptoms, signs, follow-up, prognosis, and treatment. The actual term “sepsis” is derived from the ancient Greek word “σηφις” meaning decomposition of animal or vegetable matter. This term was used in Homer’s Iliad to describe Hector’s body as it lay in Achilles’ ship, “σηπω” (translates as “I rot”).
The first theories of sepsis originate from Hippocrates (460–370 BCE) who described the body humors: blood, phlegm, yellow bile, and black bile. He theorized that disturbance of these humors resulted in disease, with blood and yellow bile causing acute disease. This Hippocratic theory remained the mainstay of sepsis understanding until the 19th century, despite Marcus Terentius Varro postulating in 100 BCE the presence of “small creatures invisible to the eye, fill the atmosphere, and breathed through the nose cause dangerous disease.” Microorganisms were not recognized until 1674 when Anthony van Leeuwenhoek built his own compound microscope and made drawings of these animalcules in the shape of “Spheres, Rods and Spirals” ( Fig. 38.1 ). However, their role in disease was not identified until 1897 when Louis Pasteur discovered streptococci in the bloodstream of a patient with puerperal sepsis. Joseph Lister (1827–1912) applied this concept of germ theory to the use of carbolic acid on wounds, significantly reducing rates of wound sepsis and death.
The concept that morbidity relates more to a dysregulated host response than the infection itself was first described by Lewis Thomas. In 1904, Sir William Osler noted: “Except on few occasions, the patient appears to die from the body’s response to infection rather than from it.”
The first formal definition of sepsis was published by Roger Bone and colleagues in 1992 to assist suitable patient enrolment into drug trials. They considered sepsis to be a combination of infection with two or more systemic inflammatory response syndrome (SIRS) criteria; namely temperature > 38°C or < 36°C, heart rate > 90 bpm, respiratory rate > 20 bpm or PaCO 2 < 32 mmHg, and an abnormal white blood count (> 12 or < 4 × 10 /L or > 10% immature forms). “Severe sepsis” represented sepsis plus organ dysfunction while “septic shock” was characterized by severe sepsis plus hypotension or hypoperfusion. However, these authors did not provide formal clinical criteria for dysfunction, hypotension, or hypoperfusion. A second task force reporting in 2003 recognized the limitations but lacked the evidence to recommend change. The recent Sepsis-3 definitions, published in 2016, utilized large hospital databases to identify hospitalized patients with likely infection (850,000 episodes where cultures were taken and antibiotics commenced) to characterize organ dysfunction as a change in “Sequential [sepsis-related] Organ Failure Assessment (SOFA) score” ≥ 2 related to an episode of infection. Patients fulfilling these criteria carried a ≥ 10% risk of hospital mortality. Septic shock is defined as “underlying circulatory and cellular/metabolic abnormalities profound enough to substantially increase mortality.” It is characterized clinically by a requirement for vasopressors to maintain a mean arterial pressure (MAP) ≥ 65 mmHg and hyperlactatemia (> 2 mmol/L) despite adequate volume resuscitation. This subgroup has an in-hospital mortality greater than 40%. SIRS was dropped from the sepsis criteria because it occurs in response to any inflammatory insult and usually represents an appropriate rather than pathologic response to infection.
Sepsis remains a significant cause of mortality, morbidity, and economic burden. Precise numbers are uncertain because of prior inconsistencies in defining sepsis. Fleischmann et al. recently suggested a global annual burden of 31.5 million cases of sepsis with 5.3 million deaths. Recent UK data identified sepsis as the cause of a quarter of all ICU admissions and approximately 11,000 deaths per annum ; mortality depended on the location of infection (e.g., 19.1% for genitourinary sepsis but 43% for respiratory sepsis). Mortality increased with the number of organ failures (18.5% for one to 69.9% for five affected organs). Although various studies suggest an increasing incidence but falling mortality, this probably relates in large part to coding changes, reimbursement incentives, and increased recognition. Nonetheless, mortality appears to be improving, albeit at the same rate as other noninfective critical illnesses. Unfortunately, long-term morbidity and mortality remain high. The reported economic burden of sepsis in the United States was $24.3 billion in 2007, costing $50,000 per patient treated for sepsis. In the UK, the average cost was £25,000 with an annual cost of £2.5 billion.
In a comparison between the SOAP study conducted in 2002 and the ICON audit in 2012 a number of important changes to the epidemiology of intensive care were identified over that 10-year period. Patients admitted to the intensive care unit are now older (60.6 ± 17.4 years in SOAP 2002 vs 62.5 ± 17.0 years in ICON 2012) and tend to be sicker with an increased incidence of shock on admission (SAPS II Score [without age] 26.0 ± 16.1 in SOAP vs 30.8 ± 17.0 in ICON) but have an improved survival rate. More patients admitted to intensive care have significant comorbidities such as COPD (SOAP 10.8% vs ICON 15.3%) and were receiving chemotherapy (SOAP 0.8% vs ICON 2.5%). Other important differences include a reduced use of mechanical ventilation (SOAP 58.8% vs ICON 53.0%) suggesting a higher reliance on noninvasive ventilation (NIV).
Sepsis is more common and serious in the elderly, but other underlying patient risk factors include immunosuppression, which can be either endogenous (e.g., diabetes, malignancy) or iatrogenic, related to corticosteroid, chemotherapy, or other immunosuppressive treatments. A prior insult such as trauma or surgery can also predispose the patient to secondary infection and sepsis. Apart from the presence of various tubes (endotracheal, urethral) and skin-breaching wounds, catheters and drains, a serious noninfectious insult will also cause immunosuppression as will the use of drugs such as sedatives, analgesics, and catecholamines.
By definition, sepsis needs an infective source. The SOAP study, conducted in multicountry intensive care units, identified respiratory (68%) and abdominal (22%) as the most frequent sites of infection. Laboratory identification of an infecting microorganism was positive in 60% of cases; 30% had Staphylococcus aureus , 14% Pseudomonas species, and 13% Escherichia coli.
The infecting microorganisms contain pathogen-associated molecular patterns (PAMPs), which are recognized by the innate immune system. Examples include lipopolysaccharide (endotoxin) found exclusively in gram-negative bacterial cell walls, lipoteichoic acid from gram-positive bacteria, mannan from fungi, and single- or double-stranded RNA from viruses. These are detected by various extracellular and intracellular pattern recognition receptors (PRRs), of which toll-like receptors are the best characterized. PRRs can also detect “self” molecules originating from the host, which can also trigger a similar host response. These molecules, termed damage-associated molecular patterns (DAMPs) include heat shock proteins, high mobility group (HMGB1) ( Box 38.1 ), histones, mitochondria, and DNA.
Procalcitonin was discovered in 1975 as a precursor to Calcitonin. In healthy people, it is produced almost solely by thyroid C cells after transcription of the CALC-1 gene on chromosome 11. During these healthy states, almost none of the protein is released into the systemic circulation with very low levels only < 0.05ng/mL. During inflammation procalcitonin is synthesized via stimulation by lipopolysaccharides or inflammatory cytokines and expressed in more cell types including white bloods cells, spleen, kidney, adipocytes, pancreas, colon, and brain, all of which have mRNA expression in hamsters.
Levels rise within 2–6 hours after infection and peak at 6–24 hours. It also reduces to normal levels much faster than C-reactive protein, giving a good indication of resolution. There are now guidelines regarding which levels of procalcitonin correlate with bacterial infection with levels < 0.1 ng/mL being actively discouraged from using antibiotics to levels > 0.25 ng/mL, which are probably bacterial infections warranting antibiotic use.
Some studies have indicated that levels > 2 ng/mL suggest the patient is more likely to develop septic shock than those with levels < 0.5 ng/mL. Levels of PCT < 0.1 ng/mL can be used to discontinue antibiotics early.
Detection of PAMPs and/or DAMPs by PRRs activates intracellular signal transduction pathways leading to an inflammatory response. This usually involves activation of NF-κB with increased production of proinflammatory cytokines (such as TNF-α, IL-1, and IL-6) and multiple other mediators including nitric oxide, eicosanoids, chemokines, reactive species, and adhesion molecules. Some PPRs assemble into intracellular molecular complexes on activation, forming inflammasomes that secrete potent proinflammatory cytokines such as IL-1β and IL-18. Inflammasomes can trigger programmed cell death pathways via caspase-mediated rupture of the plasma membrane (pyroptosis). Anti-inflammatory mediators such as IL-10 are also released simultaneously to control the degree of inflammation. Although proinflammatory mediators predominate initially, they are subsequently superseded by anti-inflammatory mediators to enable resolution of the inflammation. For reasons still unknown, the normal checks and balances controlling inflammation are overwhelmed, leading to the dysregulated host response that results in organ dysfunction.
An important host defense mechanism is to prevent spread of infection from the source. This is achieved by local endothelial activation and formation of clots. Neutrophils release extracellular traps (NETs) and platelets, endothelial cells, and leukocytes release microparticles. The net result is immunothrombosis whereby microbes are trapped within thrombi, which then attract and further activate leukocytes and prevent hematogenous spread. Consolidation within the lung is also considered to be another means of preventing the spread of microorganisms beyond the affected segments.
Although the inflammatory response is evolutionarily conserved, systemic activation and injury can result in dysfunction affecting organs distant from the site of infection. There is widespread activation of the endothelium with increased production of vasoactive mediators—both constricting and vasodilating—that result in marked heterogeneity of the microcirculation. Degradation of the glycocalyx layer covering the endothelium, with shedding of endothelial cells and loss of tight junctions between the cells, results in increased capillary leak and the development of tissue edema. Although some microregions may be hypoxic, there is little visible cell death. Widespread immunothrombosis can result in disseminated intravascular coagulation (DIC); however, frank DIC is now rarely seen. Notwithstanding a lack of widespread intravascular clots, there is an increased prothrombotic milieu with marked thrombin generation. Thrombin itself is a potent proinflammatory stimulant.
The cardiovascular system is affected at all levels. The autonomic nervous system is activated, with an imbalance between sympathetic and parasympathetic limbs that usually results in tachycardia and an elevated cardiac output. In sepsis there appears to be an uncoupling between cholinergic signaling and the sinoatrial node, which may be related to vagal activation producing an anti-inflammatory response. Nevertheless, myocardial depression is commonplace with systolic and/or diastolic dysfunction being frequently recognized. Multiple mechanisms are involved in cardiac depression including down-regulation of β-adrenergic receptors, decreased activity of post-receptor signaling pathways, mitochondrial dysfunction, abnormal actin–myosin cross-bridge formation, and perturbations of intracellular calcium. Circulating cardiodepressants such as TNF-α, IL-1β, and nitric oxide are contributory. Some data also suggest a role for complement activation, IL-6 and TNF-α inhibiting myocardial fiber shortening, while TNF-α also produces a dose-dependent reduction in left ventricular ejection fraction. Sepsis-induced cardiomyopathy has been associated with a reduced calcium current across cardiomyocytes, reduced calcium sensitivity especially to phospholamban and the ryanodine receptor, and reduced density of L-type calcium channels, all of which can result in reduced cardiac contractility. Nitric oxide (NO) is also implicated with inducible NO synthase (iNOS) indirectly inhibiting myocardial contractility. The interaction of NO and superoxide radicals can exert inhibitory effects on the mitochondrial respiratory chain and depress cardiac contractility as a result. An increase in NO production with low levels of the antioxidant glutathione is indicative of oxidative and nitrosative stress. This has been associated with reduced ATP levels in patients with septic shock. Reduced cardiac contractility has been hypothesized to be akin to cardiac hibernation, protecting cardiomyocytes from excessive stress by limiting ATP production. Indeed, this could be part of a wider cell hibernation and be responsible for multiple organ failure, particularly as histological evidence of cell death is remarkably limited.
The vasculature is also affected by excess production of mediators including eicosanoids, prostacyclins, and NO that can either cause local vasoconstriction or vasodilation (see later for microcirculatory heterogeneity). The net overall result is usually vasodilatation with loss of vascular tone and often profound hyporeactivity to catecholamines. This, in combination with reduced cardiac contractility, is responsible for the profound hypotension that can be seen in septic shock. In such patients, normal compensatory mechanisms have been overwhelmed despite high concentrations of circulating catecholamines and activation of the renin–angiotensin system. The pathology generally lies downstream, with decreased receptor density and impaired signaling. Although vasopressin levels are elevated early in the shock phase, a relative deficiency of vasopressin ensues. Other hormones have also been associated with the reduction in vascular tone seen in septic shock including adrenomedullin, which has both pleiotropic vasodilating and cardiac depressant effects.
The microcirculation is also abnormal in septic shock with increased heterogeneity, potentially creating pathologic shunts. However, even though microregions of hypoxia may exist within tissues during sepsis, there is minimal evidence of cell death in affected organs, even at postmortem examination. The direct contribution of the abnormal microcirculation towards the causation of organ dysfunction remains moot. Arguably, as discussed later, the increased heterogeneity may be a response to decreased cellular utilization of oxygen, perhaps protecting these “dormant” cells from oxygen toxicity by decreasing local oxygen supply. Although the degree of microcirculatory abnormality and failure to improve after resuscitation are associated with poor outcomes, no intervention specifically targeting the microcirculation has yet shown outcome benefit.
Immunoactivation and Paresis
Despite the initial widespread activation of immune cells and an exaggerated inflammatory response to sepsis, immune cells taken from septic patients are hyporeactive when stimulated ex vivo. Various mechanisms are postulated, such as metabolic reprogramming. There is also a depletion in the number of both T and B lymphocytes, with often decreased levels of immunoglobulins, related to both decreased production and increased destruction through apoptosis. Neutrophils, on the other hand, undergo delayed apoptosis. The resulting immune anergy places the critically ill patient at increased risk of secondary or superimposed infection.
Apart from autonomic effects on the circulation, cytokines within the systemic circulation can access the constitutively permeable areas of the carotid body chemoreceptors, vagal afferents, and certain cerebral regions. Vagal cholinergic efferents are activated and these signal to the innate immune system in the spleen and gut to inhibit inflammatory cytokine production.
Hormonal and Metabolic Changes
The initial inflammatory insult generates a stress hormonal response with increased production of catecholamines, cortisol, and glucagon. This response stimulates lipolysis and fat utilization at the expense of glucose, with insulin resistance and hyperglycemia. These stress hormones are catabolic, decreasing turnover of muscle protein yet increasing release of muscle amino acids and lactate into the circulation to be utilized by other organs as energy substrates. Although hyperlactatemia may be related in part to poor tissue perfusion and tissue oxygen insufficiency (anaerobic respiration), increased aerobic respiration related to catecholamine-driven acceleration of glycolysis is likely to be more relevant when the rate of carbohydrate metabolism from gluconeogenesis and glycolysis exceeds the oxidative capacity of the mitochondria. Pyruvate is produced from glucose faster than it can be converted into acetyl CoA by pyruvate dehydrogenase (PDH). Pyruvate also accumulates through increased protein catabolism with metabolism of amino acids to pyruvate. With increased levels of pyruvate in the cytosol, lactate dehydrogenase (LDH) converts it to lactate, recycling NADH to NAD + . Adrenergic mediators activate β receptors on the cell membrane, which increases cyclic AMP in the cytosol. This, in turn, stimulates gluconeogenesis and glycolysis with subsequent activation of the Na + /K + ATPase pump. This further activates glycolysis and production of lactate. Muscle appears to a major producer of lactate in septic shock.
Circulating levels of many other hormones also change in sepsis, mainly to subnormal ranges during established sepsis. This includes thyroid and sex hormones, vasopressin, and gut hormones. Downstream effects at receptor and post-receptor levels for most of these hormones are poorly described, other than for adrenergic vascular hyporesponsiveness and insulin resistance.
The initial increase in total body oxygen consumption and metabolic rate in response to infection is subsequently followed by a decrease toward resting healthy baseline levels during the established phase of sepsis and organ dysfunction, followed by a 60% increase in oxygen consumption during the recovery phase. Although most of the body’s oxygen utilization is directed toward “coupled” respiration with aerobic production of ATP by mitochondria, a greater proportion may possibly be diverted towards heat production (“uncoupled” respiration).
The findings in metabolic rate during the septic process mentioned previously, in conjunction with the minimal levels of cell death in most organs affected by sepsis imply a functional shutdown rather than structural damage as an underlying cause of organ dysfunction. While changes in metabolism may be directly affected by hormonal changes occurring in sepsis, such as the low T3 (triiodothyronine) syndrome, there may also be direct negative feedback from a reduced production of ATP by mitochondria. Multiple factors are recognized in sepsis including tissue hypoxia, particularly before resuscitation; excess production of NO, NO metabolites such as peroxynitrite, and other reactive species that can overwhelm antioxidant defenses and directly inhibit and/or damage the electron transport chain and other mitochondrial structures; transcriptional changes with decreased expression of genes encoding mitochondrial proteins resulting in decreased mitochondrial biogenesis; the impact of hormonal and metabolic alterations; alterations in substrate utilization, and increased uncoupling. The net result is a decrease in ATP turnover with less energy substrate available for utilization by normal cellular processes. Continued cell functioning would deplete ATP levels and trigger cell death pathways with the result that a reduction in metabolic rate, targeting pathways that are not essential for cell survival, will potentially balance out the reduced levels of ATP production. This will result in maintained cell integrity and the potential to recover. However, the normal physiological and biochemical roles of the cell cannot be fulfilled, which manifests clinically as organ dysfunction or failure. Of note, many of the treatments routinely used in sepsis such as antibiotics, catecholamines, and sedatives can impact upon mitochondrial functionality and biogenesis; as a result there may be an unwitting iatrogenic contribution to delayed restoration of cell function and organ recovery.
Mitochondrial dysfunction, microvascular and epithelial abnormalities, and local inflammatory mediators are all thought to contribute to the organ dysfunction witnessed in sepsis. Altered renal and hepatic metabolism can affect metabolism and excretion of drugs and this may further contribute to organ dysfunction. Every organ system can be affected, including the nervous system with both central (encephalopathy) and peripheral (motor and/or sensory neuropathy) manifestations; muscle dysfunction with loss of muscle bulk and power; gut dysfunction with ileus or diarrhea and decreased gut hormone activity (e.g., gastrin, ghrelin); liver and renal dysfunction with decreases in synthetic capability, metabolism, and excretory function. The hematopoietic system is also affected with decreased production of circulating cell constituents and increased consumption of platelets and clotting factors.
Sepsis can present in protean ways. There may be clinical features pointing to the likely source of infection such as dyspnea, hypoxemia, and productive cough for a respiratory source, or peritonism for abdominal sepsis. However, not infrequently, and especially in the early stages, there may not be a clear pointer to the site of infection. The septic patient may present nonspecifically unwell with evidence of organ dysfunction. The infected patient will often present with two or more SIRS criteria such as pyrexia, tachycardia, and neutrophilia, as described earlier, but this is not a prerequisite. One in eight patients in Australasian ICUs with infection-related organ failure did not have the requisite two or more SIRS criteria on admission. The converse is often true: 20% to 40% of patients initially diagnosed and admitted to ICUs as “septic” subsequently turn out to have a noninfective critical illness.
Various bedside scores such as NEWS (National Early Warning Score), which uses seven criteria, and qSOFA ( Box 38.2 ), which uses three criteria, identify clinical markers of unwellness such as abnormal levels of blood pressure, respiratory rate, and conscious level. A high score indicates those patients at increased risk of doing badly; however, this should not be relied upon as a “rule-out.” There is no perfect screening tool at present that offers both high specificity and sensitivity for detecting the septic patient at risk of poor outcome.
There are no specific early diagnostics for sepsis. Laboratory confirmation of infection is found in approximately 60% of cases with blood culture positivity rates ranging between 15% and 30%. Newer molecular diagnostics for infection will identify microorganisms far sooner than traditional culture techniques, but issues surrounding cost, over-sensitivity, and false positive results resulting from sample contamination need to be overcome.
Biomarkers such as neutrophilia and neutropenia, C-reactive protein, and procalcitonin ( Box 38.1 ) are all nonspecific for infection. Although useful in supporting a clinical suspicion of infection, these biomarkers can both give false negatives (especially if the patient cannot mount a host response) and false positives (related to other noninfectious inflammatory insults). Multiple efforts are being made to find superior biomarkers of the host response to infection for both “rule in” and “rule out,” including transcriptional changes, various circulating mediators (or combinations thereof), and proteomic and metabolomic analyses. None has yet become established commercially.
Sepsis and septic shock are medical emergencies and are still associated with significant mortality and morbidity. Consequently, prompt treatment is warranted with early, adequate but not excessive resuscitation, plus administration of appropriate antibiotics and early source control.
Multiple management challenges and questions remain. For example, ongoing debate surrounds what constitutes adequate but not excessive resuscitation, what optimal blood pressure should be targeted for an individual patient, and the true impact of early antibiotics.
Antibiotics and Infection Control
Intravenous antibiotics are recommended by international guidelines to be given within 1 hour of admission to hospital and identification of sepsis, immediately after blood cultures and other relevant microbiological samples have been taken. It is claimed that each hour’s delay in giving antibiotics is associated with significant increases in mortality and worse outcomes. However, this particular dogma is based on retrospective analyses, usually with complex statistical adjustments of preexisting databases, and often without confirmation of infection or knowledge of antibiotic sensitivity patterns. All prospective studies to date, including a recent large randomized study investigating the utility of pre-hospital antibiotics, have failed to show such an association. Although the absolute importance of each hour’s delay can be challenged, especially if “time zero” when infection or sepsis commences is unknown, it is nonetheless rational to give antibiotics without considerable delay.
Depending on local resistance patterns, it may be prudent to commence empiric broad spectrum antibiotics that are likely to be effective until the organism and its antimicrobial sensitivities are identified, with subsequent switch and/or de-escalation to a narrow-spectrum antibiotic. An empiric antifungal and/or antiviral agent may be needed based on risk factors (e.g., in a neutropenic patient following chemotherapy). The evidence base for combination therapy, using two or more antibiotics aimed at the same organisms, is currently inconclusive and may even be harmful. Similarly, while using an initial appropriate antibiotic also appears rational, the evidence base for survival benefit is also conflicting. Other outstanding questions that remain unanswered are the optimal means of administering antibiotics (e.g., intermittent bolus vs continuous infusion ) not just to achieve therapeutic plasma levels but also to improve outcomes, and the optimal course duration both to ensure adequate microbial eradication and yet to avoid the downsides of excessive antibiotic use such as development of resistance, or fungal and Clostridium difficile overgrowth. Although a 7–10-day course is currently recommended, the evidence base for this is surprisingly weak and shorter courses appear adequate for many conditions.
The one area of infection management for which there is clear consensus is the imperative for early and adequate source control. This is particularly pertinent in the postoperative patient who may suffer, for example, from anastomotic leaks or wound infections with abscess formation or necrotizing fasciitis. If suspected, appropriate investigations and imaging should be conducted promptly followed by urgent drainage or debridement.
In terms of a specific early resuscitation regimen, the previously endorsed Early Goal-Directed Therapy (EGDT) strategy proposed by Rivers et al. was recently shown not to offer any benefit over standard-of-care in three separate multicenter studies (ARISE, ProCESS, and ProMISe). EGDT comprised fluid resuscitation ± dobutamine ± blood transfusion targeting a central venous pressure of 8 to 12 cm H 2 O, a mean arterial pressure ≥ 65 mmHg, and a central venous oxygen saturation ≥ 70%. More recent guidelines suggest targeting a mean blood pressure ≥ 65 mmHg using at least 30 mL/kg of intravenous crystalloid within the first 3 hours of admission. More fluid may be given as required or vasopressors if the patient is unresponsive to fluid. However, the optimal means of assessing volume resuscitation remains uncertain. While dynamic assessment (e.g., monitoring the response to a fluid challenge or a straight-leg raise) is superior to a static measurement, this gauges fluid responsiveness but not whether this is necessarily benefiting organ perfusion. Nonetheless, failure to respond positively to a dynamic challenge does suggest that further fluid loading is probably inadvisable. Various measures are used to assess fluid responsiveness including changes in stroke volume, blood pressure, pulse pressure variation, or stroke volume variation. All have limitations and potential methodological pitfalls, although stroke volume monitoring is likely to be the most accurate.
A higher MAP may be needed, particularly if the patient is chronically hypertensive. Persisting hyperlactatemia may be indicative of continuing tissue hypoperfusion. Although such patients who remain hyperlactatemic are more likely to fare badly, there is no strong evidence that normalizing lactate as a primary therapeutic target offers outcome benefit.
The currently recommended first-line vasopressor for fluid-resistant hypotension is norepinephrine ( Box 38.3 ), with vasopressin recommended as a second-line vasopressor. Epinephrine was, however, equally effective compared with norepinephrine ± dobutamine in two multicenter studies assessing outcomes in shock patients. Dopamine has, however, been shown to be inferior to norepinephrine. Although a post hoc analysis of the VASST trial suggested superiority of vasopressin over norepinephrine in patients requiring low-dose vasopressors, the recent VANISH study found no outcome difference but did note a reduction in the need for renal replacement therapy.