Abstract
In 2001, Rivers and colleagues published a randomized controlled trial of early goal-directed therapy (EGDT) for the treatment of sepsis. More than a decade later, it remains a landmark achievement. The study proved the benefits of early aggressive treatment of sepsis. However, many questions remain about specific aspects of the complex EGDT algorithm. Recently, 3 large trials attempted to replicate these results. None of the studies demonstrated a benefit of an EGDT protocol for sepsis. This review explores the physiologic basis of goal-directed therapy, including the hemodynamic targets and the therapeutic interventions. An understanding of the physiologic basis of EGDT helps reconcile the results of the clinical trials.
In 2001, Rivers and colleagues published a landmark trial of early goal-directed therapy (EGDT) for the treatment of sepsis. 1 The central paradigm of EGDT is that invasive monitoring can identify hemodynamic targets that clinicians need to normalize with aggressive resuscitation.
Many components of the Rivers EGDT protocol have been codified in the International Surviving Sepsis Campaign Guidelines. In subsequent years, there has been widespread recognition of the importance of early detection and aggressive fluid resuscitation for patients with sepsis. 2 The adoption of these guidelines has been associated with decreased mortality, despite the lack of new medications and increased bacterial resistance to antimicrobials. 3,4 The literature contains more than 8000 citations of the Rivers EGDT trial. However, questions remain about specific components of the EGDT algorithm. Recently, 3 large randomized controlled trials evaluated the effectiveness of the individual components of EGDT. The trials called Protocolized Care for Early Septic Shock (ProCESS), The Australiasian Resuscitation in Sepsis Evaluation (ARISE), and Protocolised Management in Sepsis (ProMISe) were published in major journals. The studies enrolled patients early in the course of sepsis in the emergency department. All patients in the control and intervention arms were given early antibiotics and fluid resuscitation. The 3 trials unanimously concluded that strict adherence to EGDT protocolized management—specifically the monitoring of central venous pressure (CVP) and oxygen saturation to direct fluid management—does not improve outcomes. 5 –7 This review explores the physiologic basis of EGDT to reconcile the different results between the Rivers EGDT and the subsequent 3 trials.
The Pathophysiology of Goal-Directed Therapy
Prior to EGDT, clinicians used end points such as the normalization of blood pressure, urine output, and mental status to monitor the adequacy of tissue perfusion. However, poor outcomes in patients with shock led to the idea that clinicians should monitor additional hemodynamic variables to identify occult hypoperfusion. Cardiac output is an obvious choice. Cardiac output is the rate of blood flow through the circulation, which can vary from 1 to 25 L/min. 8 However, it is difficult to define a normal cardiac output for patients with sepsis. Therefore, one potential method of estimating the adequacy of blood flow to tissues is to measure the concentration of oxygen exiting from capillary beds. The use of oxygen saturation of hemoglobin in venous blood for EGDT requires an understanding of the physiology of the flow of oxygen.
The resting oxygen saturation of hemoglobin in venous blood depends on the balance between the rates of oxygen flowing in arterial blood and consumed by the tissues. The tissues remove oxygen from systemic capillary blood and consume it in the reactions of metabolism. At rest, adults must consume at least 3 mL of O2·kg−1 per minute to maintain homeostasis. 9 Without this minimum oxygen consumption, cells have energetic failure leading to organ dysfunction. The rate of oxygen flow in the systemic arterial system is called the oxygen delivery. This rate depends on the hemoglobin concentration, cardiac output, and the oxygen saturation of hemoglobin. The tissues consume a fraction of oxygen delivered by the systemic arteries. Therefore, the oxygen saturation of hemoglobin falls to an average of 65% to 75% when the blood enters the systemic veins (Table 1).
The Flow and Content of Oxygen in Systemic Blood in a Healthy Resting Adult.a
aEach gram of fully saturated hemoglobin carries 1.36 mL of oxygen. 10 The resting normal oxygen consumption is approximately 250 mL O2·min−1, and the resting cardiac output (blood flow rate) is 5 L/min.
Venous blood that has passed through more metabolically active tissue (eg, the brain) will have lower oxygen saturation than venous blood that passes through tissues that consume less oxygen (eg, the kidneys). Veins carrying blood of varying oxygen content empty into the right heart. Therefore, the term “mixed venous blood” refers to blood in the pulmonary artery. Clinicians can sample mixed venous blood from a pulmonary artery catheter. The oxygen saturation of hemoglobin in mixed venous blood (SvO2) reflects the global venous oxygen content. 10 As blood travels through the lungs, oxygen diffuses from the alveolar airspaces down its concentration gradient. This process replenishes the oxygen lost to tissues in the capillaries. At steady state, the rate of oxygen uptake from the lungs equals the rate of oxygen consumption by the tissues (Figure 1A).
Venous oxygen saturation depends on the balance between oxygen delivery and consumption. A, Normal resting conditions and the mixed venous oxygen saturation is 74%. B, Decrease in venous oxygen saturation to 49% due to an increase in oxygen consumption from sepsis. C, Decrease in venous oxygen saturation to 49% due to a decrease in oxygen delivery from anemia, hypoxemia, and reduced cardiac output.
Normally, the rate of oxygen delivery greatly exceeds the rate of oxygen consumption by tissues. This is necessary because of the inherent inefficiency of the diffusion of oxygen from capillary blood to the tissue’s mitochondria. The excess oxygen delivery also serves as a reserve supply of oxygen in blood that is readily available to tissues when their metabolic requirements increase. Sepsis and exercise both increase the rate of oxygen consumption. In these situations, the increased consumption reduces the oxygen content in blood flowing from capillaries into veins (Figure 1B). When oxygen consumption increases relative to delivery, SvO2 will decrease. For example, SvO2 often decreases to less than 50% during exercise. 11
The excess oxygen flowing in the systemic arteries also safeguards against reductions in hemoglobin concentration, cardiac output, and oxygen saturation. Because of the excess, if the rate of oxygen delivery decreases, tissues may still continue to consume oxygen at the same rate. In this scenario, SvO2 will decrease (Figure 1C). Oxygen delivery must, however, always exceed a minimum threshold in order to meet the demand for oxygen consumption. This minimum value is called the critical rate of oxygen delivery. In resting adults, the critical rate is less than 7 mL O2/kg·min and may be as low as 5 mL O2/kg·min. 12,13 If oxygen delivery drops below this threshold, oxygen consumption will be inadequate to sustain aerobic metabolism. This problem is amplified when metabolic requirements increase and demand a proportional increase in the critical rate of oxygen delivery. For example, metabolic requirements may be elevated during the inflammatory response of early sepsis, yet oxygen delivery may be inadequate to match this high demand because of reductions in the cardiac output caused by relative intravascular volume depletion and reduced stroke volume. 14 Ideally, clinicians would continuously calculate oxygen delivery to ensure that it exceeded the rate of oxygen consumption by a wide margin. These measurements, however, are impractical in critically ill patients. Since venous oxygen content falls when there is a reduction in oxygen delivery, a low SvO2 can serve as a surrogate marker that oxygen delivery is at or near the critical value.
Trials of goal-directed therapy in the 1990s attempted to increase the arterial oxygen saturation, cardiac output, and hemoglobin to achieve this target. 14 In support of the goal-directed therapy concept, better sepsis outcomes were associated with higher values of oxygen delivery, oxygen consumption, and SvO2. 15 Moreover, multiple trials documented the benefit of prophylactic goal-directed therapy for high-risk perioperative patients. 16 Surprisingly, clinical trials of goal-directed therapy in patients with sepsis failed to demonstrate a benefit. 17,18 However, these trials were performed in intensive care unit patients after their initial resuscitation. At this later stage of sepsis, many patients have a normal or elevated SvO2 despite ongoing hypoperfusion. This phenomenon is due to tissue dysoxia—an inability of cells to utilize the oxygen. In late sepsis, the rate of oxygen delivery does not limit oxygen consumption. 19,20
The paradox of a compelling physiologic rationale for goal-directed therapy coupled with the disappointing clinical trials performed in intensive care units prompted Rivers and colleagues to evaluate a protocol of EGDT. This trial enrolled patients with sepsis when they first presented to emergency departments. Theoretically, patients with unresuscitated sepsis would benefit from restoration of perfusion before the onset of multi-organ dysfunction. This trial revealed a dramatic mortality benefit in the group randomized to correction of the ScvO2 to 70%. 1
A novel feature of the Rivers EGDT algorithm was its use of the oxygen saturation of central venous blood (ScvO2) as a surrogate for SvO2. This innovation provided an alternative to the placement of a pulmonary artery catheter. The obvious limitation of ScvO2 as a resuscitation target is that the blood sampled by a central venous catheter is incompletely mixed. Although there is a variable correlation between SvO2 and ScvO2, they have similar prognostic value in sepsis. 10,21,22
The Components of EGDT
Fluid Therapy
The purpose of fluid therapy is to increase venous pressure in order to augment the rate of venous return to the heart. The heart normally responds to increased venous return with an increase in heart rate and the force of contraction (the Frank-Starling mechanism). In this way, the cardiac output rises to match venous return. 23 The initial step in EGDT was fluid resuscitation in an attempt to increase the cardiac output. Aggressive fluid resuscitation for patients with sepsis at the earliest stage is likely the key benefit of the original Rivers EGDT trial. There is, however, a limit to the extent that fluid therapy can recruit cardiac output. Fluid resuscitation only increases cardiac output if it increases the gradient for venous return, and the heart is able to respond to increased preload. 24 In fact, more than half of fluid boluses in critically ill patients do not increase cardiac output. 25 Furthermore, excessive fluid resuscitation is associated with worse outcomes in sepsis. 26
Central Venous Pressure Monitoring
Much of the criticism of the EGDT protocols surrounds the use of CVP to ensure adequate fluid resuscitation. Central venous pressure is a surrogate for right atrial pressure. However, CVP does not correlate with circulating blood volume 27 or predict whether a fluid bolus will increase cardiac output. 25,28,29 These observations are predicted by physiology: cardiac preload is a volume not a pressure. Moreover, it is the pressure gradient between the systemic veins and the right atrium that determines the rate of venous return.
Central venous pressure is the pressure that limits venous return to the right heart; an elevated CVP may therefore actually decrease cardiac output. 8,23 Likewise, a high CVP may decrease the perfusion gradient across the systemic capillary beds and decrease organ perfusion. 30 Research in the years after the Rivers EGDT has demonstrated that CVP should not be used as a primary guide for fluid resuscitation.
Red Blood Cell Transfusions
There is no method available to directly measure oxygen delivery. Instead, we calculate oxygen delivery by multiplying cardiac output by the oxygen content of arterial blood. Therefore, raising the hemoglobin concentration, by definition, will increase the rate of oxygen delivery as long as there is not a concomitant decrease in cardiac output. Red blood cell transfusion may improve oxygen delivery both by expanding intravascular volume to recruit stroke volume as well as by increasing the oxygen carrying capacity of blood. Although red cell transfusion clearly improves perfusion during severe anemia, its role in septic shock remains controversial for several reasons. Transfusion of stored red blood cells may not increase oxygen delivery because of an associated decrease in cardiac output due to increasing blood viscosity. Moreover, banked allogeneic red cells have suboptimal oxygen-carrying capacity. 31 Because of these effects, studies of heterogeneous patients with sepsis do not show an increase in oxygen uptake after red blood cell transfusion. 32 This lack of physiologic benefit amplifies the potential risks of transfusion, including increased pulmonary vascular resistance, nosocomial infections, and decreased perfusion of splanchnic organs. 32 –35 The Rivers EGDT used a transfusion target of 10 g/dL to improve oxygen delivery. In the past decade, however, a lower transfusion target of 7 g/dL has gained acceptance in critically ill patients, including patients with sepsis. 36 –38 The optimal transfusion target, however, remains uncertain since patients in the Rivers EGDT had more hemodynamic instability than patients in subsequent trials of transfusion triggers. Transfusion targets in patients with significant coronary arterial disease are also poorly defined.
Vasoactive Agents
The Rivers EGDT trial used vasopressors to maintain mean arterial pressures (MAPs) of 65 to 90 mm Hg. Theoretically, vasopressors can counteract the pathologic vasodilation of sepsis and provide an adequate pressure gradient to allow autoregulation of systemic blood flow by tissues. Importantly, vasopressors may restore the coronary perfusion gradient when ventricular diastolic pressures are elevated from cardiac pathology. The benefits of vasoconstriction, however, come with the cost of increased ventricular afterload. Additionally, vasoconstriction may reduce perfusion by impeding microcirculatory flow. 39
Multiple randomized controlled trials have established norepinephrine as the vasopressor of choice in septic shock. 40 Norepinephrine is a balanced vasoconstrictor, raises tone in both the arterial and venous systems, and can both increase cardiac output and improve tissue perfusion. 41 –43 There is, however, a wide variation among patients in the effect of norepinephrine on oxygen consumption and microcirculatory function. 44 Recent data confirm that the MAP targets in sepsis should not be higher than 65 mm Hg (unless there is uncontrolled hypertension at baseline). 45
Inotropic Support
The purpose of inotropic support in sepsis is to increase cardiac output by improving myocardial contractility and lusitropy or ventricular relaxation. However, inotropes also decrease vascular resistance leading to hypotension, promote tachyarrhythmias, and increase myocardial oxygen demand. In normal hearts, inotropes have limited impact unless there is a concomitant increase in venous return. 8 The Rivers EGDT, therefore, restricted inotropes until fluid resuscitation had raised CVP. 1 Despite their physiologic potential, there are limited data demonstrating that inotropes increase cardiac output in sepsis already treated with norepinephrine. 46,47 Furthermore, for reasons already discussed (mitochondrial dysfunction), increases in cardiac output do not reliably correlate with enhanced global oxygen consumption during sepsis. 17
Mechanical Ventilation
The Rivers EGDT protocol prescribed endotracheal intubation and sedation if ScvO2 remained below target despite the aforementioned hemodynamic therapy. This intervention was designed to decrease the rate of oxygen consumed by minimizing the work of breathing during sepsis. Mechanical ventilation, however, has multiple effects on the balance between oxygen supply and consumption. First, a substantial oxygen cost from the work of breathing may persist during assist control ventilation. 48 Moreover, although positive pressure ventilation may improve stroke volume by reducing left ventricular afterload, the more common effect is a reduction in venous return and cardiac output. 49,50 Importantly, the Rivers EGDT protocol did not change the utilization of mechanical ventilation or sedation.
Clinical Trials of EGDT
The intervention arm of the original Rivers EGDT provided aggressive resuscitation in the emergency department for 6 hours after randomization. The resuscitation goals were endotracheal intubation and mechanical ventilation, central line and arterial line placement; crystalloid resuscitation until CVP reached 8 to 12 mm Hg; vasoactive medications to achieve MAP of 65 to 90 mm Hg, followed by transfusion of red cells; and/or addition of inotropes for central venous oxygen saturation (ScvO2) of less than 70%. This intervention arm was compared to a control arm in which physicians used a “protocol for hemodynamic support,” which included the same CVP and MAP goals. The Rivers EGDT study showed that a substantial proportion of patients who present to emergency departments with sepsis have significant volume depletion and low ScvO2. It also established a new standard of care for sepsis: screening for hypoperfusion followed by early aggressive fluid resuscitation. It was, however, difficult to determine the relative of benefits of each of the following aspects of EGDT: trained resuscitation teams, CVP or ScvO2 monitoring, hemoglobin transfusion targets, inotropes, and protocolized of care.
Three new studies attempted to determine which of aspects of EGDT was essential (Table 2). All of the new trials included early screening for sepsis in the emergency department and aggressive fluid resuscitation for all patients in the control and intervention arms. In ProCESS, EGDT was compared with 2 arms—protocolized standard therapy and usual care. In the protocolized standard therapy arm, crystalloid was bolused until clinically volume replete and vasopressors added if still in shock (systolic blood pressure <100 mm Hg or shock index less than or equal to 0.8). Red cells were transfused if hemoglobin concentration fell to less than 7.5 g/dL. Therapy in the usual care arm was up to the discretion of the treating physician. 6 In ARISE, and in ProMISe, the EGDT therapy arms were compared to usual care arms where therapy was at the discretion of the treating physician.
Comparison of the 4 Major Trials of Early Goal-Directed Therapy for Sepsis.
Abbreviations: APACHE II, Acute Physiology and Chronic Health Evaluation II; EGDT, early goal-directed therapy; ScvO2, the mean oxygen saturation of central venous blood from patients at the time of enrollment.
There are a number of reasons why EGDT may not have been effective in the newer trials. First, it is possible that the EGDT interventions (aggressive fluid transfusions, inotropes) may not effectively raise oxygen delivery in many patients. Also, dysoxia, mitochondrial dysfunction, and microvascular dysfunction may be more important factors than inadequate oxygen delivery in sepsis. Additionally, it is possible that the initial fluid resuscitation adequately restores oxygen delivery in the majority of patients, which obviates the need for additional resuscitation. The baseline mean ScvO2 in the EGDT trials supports this possibility. In Rivers EGDT, the baseline ScvO2 was 49%, suggesting the need for significant resuscitation. However, the mean ScvO2 in the ProCESS, ARISE, and ProMISe was greater than 70%. 1,5,7,6 It should not be surprising that protocolized interventions designed to raise ScvO2 to greater than 70% do not help cohorts of patients starting with an ScvO2 of 70%. The higher ScvO2 in the newer trials may have been due to more aggressive resuscitation prior to enrollment or different patient characteristics than in the Rivers EGDT trial. It is possible that an EGDT protocol would still be successful in a different study population, such as one with lower initial ScvO2.
Summary Recommendations: Integrating EGDT Into Clinical Practice
An appreciation of the basic physiology of goal-directed therapy allows the clinician to reconcile the results of the ProCESS, ARISE, and ProMISe trials. These newer trials do not refute EGDT—they refine our understanding of it. The Rivers EGDT study made early identification and aggressive resuscitation of patients with sepsis, the standard of care. These principles are incorporated into the control arms of these recent trials. When considered together, these studies suggest that if clinicians rapidly identify and resuscitate patients with sepsis, they do not need to routinely screen patients for low ScvO2.
As a result of these recent trials, consensus opinion is that there is no longer a need for the routine or protocolized placement of central venous catheters and measurement of CVP in all patients with sepsis. If clinical evidence of hypoperfusion persists despite adequate fluid resuscitation, clinicians should consider additional hemodynamic monitoring to help guide management. The specific type of monitoring should be based on local skill and resources and may include ScvO2 and CVP measurement in select cases. Current guidelines recommend a restrictive approach to allogeneic red blood cell transfusion, excluding those with active bleeding or significant coronary arterial disease. The careful addition of inotropic support is recommended if there is evidence of poor cardiac performance based on bedside transthoracic echocardiography and low ScvO2. Finally, the use of intubation and mechanical ventilation is recommended for respiratory failure and to decrease the work of breathing associated with septic shock, although ScvO2 should not play a prominent role in the decision to intubate.
The Rivers EGDT trial has transformed the treatment of sepsis and led to earlier and more aggressive care. Because of the improvements in sepsis management, as well as the difference in initial ScvO2 in the study populations, newer trials were unable to show a benefit to protocolized care. The modern care of patients with sepsis should not entail blind adherence to a protocol or universally abandoning the principles of EGDT. Rather, clinicians should consider the physiology of each unique patient in considering when to use the elements of EGDT.
Footnotes
Authors’ Note
This work does not involve research on animals or humans.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.