We must admit that we are all imperfect beings and, as such, we are all incorrect from time to time. In order to evolve to proverbial ‘higher planes of enlightenment,’ we must accept our cognitive errors – sometimes as individuals and other times as a collective entity. Within this framework of improvement through critical reflection, evidence-based medicine has been born. In this commentary, the authors address an area of smoldering contention and conflicting evidence—the role of EGDT in managing sepsis. If their call for an individualized approach to therapy is ultimately the ideal strategy, it may vindicate us all by simultaneously proving us all wrong.

Lee E. Morrow, MD, FCCP

Critical care – like all of medicine – is evolving at a rapid pace. In the relatively recent past, we moved from an era of consensus-based (if thinking optimistically) or opinion-based (if being less charitable) medicine to an era of evidence-based medicine. Despite the many valid concerns about ubiquitous adoption of evidence-based medicine, there is little doubt that, on average, an aggregate population managed according to the best available literature does better than one managed solely on widely varying physician expertise. At the same time, there is no doubt that one size does not fit all, and in applying evidence-based protocols to all patients equally, we are helping many, having no effect on many, and are harming some. This has led to a still ongoing transition into an era of precision medicine where each patient gets the best care specifically for them. While the intellectual appeal of personalized therapy is obviously immense, the tools with which to do so currently remain relatively limited.

The approach to sepsis resuscitation is emblematic of the challenges and opportunities of the evolution in this transition. There was no standardized approach to early sepsis resuscitation in the 20th century, and mortality from the disease approached 50% in many studies. This changed in 2001 with the publication of the landmark early-goal-directed therapy (EGDT) trial (Rivers et al. N Engl J Med. 2001;345[19]:1368). This single center trial demonstrated a dramatic 16% absolute decrease in mortality secondary to usage of an aggressive protocol for sepsis resuscitation within the first 6 hours after presentation to the ED. In addition to early cultures and antibiotic therapy in patients randomized to both EGDT and “usual care,” EGDT involved a number of mandatory elements, including placing both an arterial catheter and a central venous catheter capable of measuring continuous central venous oxygen saturation (ScvO₂). Patients received crystalloid or colloid until a predetermined central venous pressure was obtained, and if their mean arterial pressure was still below 65 mm Hg, therapy with pressors was initiated. If their ScvO₂ was not 70% or greater, patients were transfused until their hematocrit was greater than 30%, and, if this still did not bring their ScvO₂ up, patients were started on a regimen of dobutamine. Multiple trials of varying design subsequently demonstrated efficacy in this approach, which was rapidly adopted worldwide in many centers managing patients with sepsis.

However, many questions remained. All patients were managed the same in EGDT, with no capacity to individualize care, regardless of clinical situation (comorbidities, age, origin of sepsis). In addition, it was never clear which specific elements of the EGDT protocol were responsible for its success, as a bundled protocol could potentially simultaneously include beneficial, harmful, and neutral components. Further, many of the elements of EGDT have not been demonstrated to be beneficial in isolation. For example, multiple studies demonstrate that patients not receiving transfusions until their hemoglobin value reaches 7 g/dL is at least as effective as receiving transfusions to a hemoglobin value of 10 g/dL. Also, there is a wealth of data suggesting that central venous pressure is not an accurate surrogate for intravascular volume.

To address these issues, three international, multicentered randomized controlled trials were published in the New England Journal of Medicine in 2014 and 2015: ARISE, ProCESS, and ProMISe (ARISE investigators. N Engl J Med. 2014;371[16]:1496; ProCESS investigators. N Engl J Med. 2014;370[18]:1683; Mouncey, et al. N Engl J Med. 2015;372[14]:1301). Each of these studies randomized patients either to EGDT as defined in the original Rivers study or to a “usual care” group with management directed under the guidance of a bedside health-care provider. Across all three trials, the EGDT group received more fluids, inotropes, vasopressors, and transfusions than the” usual care” group. However, there was no mortality benefit detected in any of the trials.

The difference between the original Rivers trial (demonstrating a huge benefit of EGDT) and the three subsequent trials leads (showing no benefit) was striking and leads to the obvious questions of (a) why were the results so disparate and (b) what should we do for our patients moving forward? Perhaps the most obvious difference in the trials is the baseline mortality in the “usual care” groups between the studies. In the original Rivers study, in-hospital mortality was 46.5% for the “usual care” group. For ARISE, ProCESS, and ProMISe, 60- to 90-day mortality ranged from 18.8% to 29.2% in the “usual care” group. This means either that the patients in the original EGDT trial were significantly sicker or that something fundamental has changed over time. A closer review of the papers reveals it is likely the latter as, in actuality, the “usual care” group in the three NEJM trials looked a lot like the EGDT group in the original trial. Most patients received significant volume resuscitation in these studies prior to enrollment, and the original ScvO₂ was 71% in ProCESS (as opposed to 49% in the original Rivers trial). This suggested that increasing awareness of sepsis that occurred during the 15 years between the EGDT trial and the subsequent three trials – likely due to the Surviving Sepsis Campaign, as well as other efforts from both advocacy groups as well as medical organizations – led to better sepsis care on patient presentation. In essence, what was “usual care” in the time of the original EGDT study had become inappropriate care in the modern era, and much of what was protocolized in EGDT had been transformed into “usual care,” even if a specific protocol was not being used. In the setting in which “usual care” had dramatically improved, the original EGDT protocol was not helpful if implemented on all comers. One key reason is that many patients simply improved with volume and antibiotics (which had become “usual care”) and did not need additional interventions. Another reason is that some of the interventions in EGDT (blood transfusion, continuous ScvO₂ monitoring) are likely not beneficial in the majority of cases.

These studies have led to significant changes in recommendations in sepsis management guidelines. The 2016 Surviving Sepsis Campaign guidelines – published after ARISE, ProCESS and ProMISe trials – still recommend antibiotics, cultures, adequate volume resuscitation (without specifying how to do so), targeting an initial mean arterial pressure of 65 mm Hg, and vasopressors if a patient remains hypotensive despite adequate fluids (Rhodes et al. Crit Care Med. 2017; 45). However, no recommendations are made regarding mandatory placement of a central venous catheter, measuring central venous pressure, transfusing to higher hemoglobin, etc.

In many ways, the last 15 years of fluid resuscitation in sepsis represents the triumph of evidence-based medicine over opinion-based medicine and the challenges of moving toward precision medicine. When “usual care” was highly variable without a consistent scientific rationale, EGDT markedly improved outcomes – a clear victory of evidence-based medicine that likely saved thousands of lives. However, when EGDT effectively became “usual care,” each individual element of EGDT bundled together failed to further improve outcome. The new evidence suggested that for all comers, EGDT is no better than the new normal, and, thus, newer guidelines do not recommend most of its components.

Moving forward, what is the best way to resuscitate newly identified patients with sepsis? A big fear in eliminating EGDT in its entirety is that practitioners will not have any guidance on how to manage resuscitation in sepsis and so will revert to less rigorous practice patterns. While we acknowledge that concern, we are optimistic that the future will continue to yield decreases in sepsis mortality. Optimally, volume status will be assessed on an individual basis. Rather than resuscitating every patient with a one-size-fits-all parameter that is fairly crude at best and inaccurate at worst (central venous pressure), bedside caregivers should use whatever tools are most appropriate to their individual patient and expertise. This could include bedside ultrasound, stroke volume variation, esophageal Doppler, passive leg raise, etc, depending on the clinical situation. The concept of appropriate volume resuscitation raised in EGDT continues to be 100% valid, but the implementation is now patient-specific and will vary upon available technology, provider skill, and bedside factors that might make one method superior to the other. Similarly, the failure of EGDT to improve survival in the ARISE, ProCESS, and ProMISe trials does not mean there is never a role for checking venous blood gases and measuring ScvO₂. From our point of view, this would be a gross misinterpretation of the trials, as the finding that all elements of EGDT combined fail to benefit all patients with sepsis on arrival should assuredly not be interpreted as none of the elements of EGDT can ever be beneficial in any patients with sepsis. While we can – and should – learn from the data as they pertain to “all comers,” we equally can – and should – look at each individual patient and determine where they align with what is known (and unknown) in the literature and simultaneously attempt to both personalize and optimize their care utilizing our general knowledge of physiology and individual information that is unique to the patient.

In the future, we hope that sepsis resuscitation will be performed in an analogous fashion to cancer therapy. Understanding a patient’s response at the organ level and cellular and subcellular levels will allow us to individualize initial therapy. For instance, an “omics” evaluation of a patient’s immune system may be helpful for guiding treatment. Distinct patterns of gene and protein expression could potentially demonstrate in advance how different patients will respond differently to the same therapy and, in a dynamic manner, determine whether they are responding according to the expected trajectory. Unfortunately, since this is impractical today, the best we can do is to follow recommendations that are applicable to large populations (the Surviving Sepsis bundles) while simultaneously individualizing therapy when no clear data are available. Further, it is critical to assess and reassess the response at the bedside to optimize outcomes. While it is frustrating that no clear guidance can be given on the best way to measure volume status or fluid responsiveness or when there is utility in measuring ScvO₂, there is comfort in knowing that best practice has evolved over the past 15 years such that the majority of EGDT is now “usual care.” Moving forward, the challenges in transitioning sepsis resuscitation from population-based evidence-based medicine to individualized therapy are real, but the opportunities for improved outcomes in this deadly disease are enormous.

Dr. Head is with the Department of Anesthesiology, and Dr. Coopersmith is with the Department of Surgery, Emory Critical Care Center, Emory University School of Medicine, Atlanta, GA.

Critical Care Commentary: Sepsis resuscitation in a post-EGDT age