The Final Nail in Early Goal Directed Therapy’s Coffin?

Posted by • March 24th, 2015

You are called to evaluate Ms. Smith urgently.  She reports several days of progressive malaise, weakness and a new cough. On your exam she is tachycardic, tachypneic and febrile.  You diagnose Ms. Smith with sepsis and explain the importance of quickly treating this life-threatening condition. 

In the United States, situations like this—presentations leading to the diagnosis of sepsis –play out over 1.6 million times per year and are associated with a 16% mortality rate.  Since 2001 many healthcare professionals have pointed to early-goal directed therapy—a strict protocol aimed at quickly optimizing tissue perfusion and oxygen transport—as the key to beating sepsis.  However, a trial in this week’s NEJM (the last in a series of three) reinforces concerns that in the management of sepsis early goal directed therapy (EGDT) may not be more effective than non-protocolized usual care.

The Protocolized Management in Sepsis (ProMISe) study examined the potential benefits of EGDT, much like the recently published Protocolized Care for Early Septic Shock (ProCESS) and Australasian Resuscitation in Sepsis Evaluation (ARISE) trials.  ProCESS and ARISE did not detect a significant reduction in sepsis mortality with the use of EGDT compared with usual care, but were noted to have overall low mortality rates (less than 20%, compared with 45% seen in the initial 2001 trial).  ProMISe aimed to target a more critically ill septic population in order to examine the same question: can EGDT improve patient mortality in sepsis compared with usual care?

The authors conducted ProMISe at National Health Service hospitals in England where EGDT was not routinely being used.  Patients with (1) a known or presumed infection, (2) two or more SIRS criteria, and (3) elevated lactate or hypotension refractory to a 1 liter intravenous fluid bolus were randomized to receive usual care versus EGDT. The EGDT protocol required placement of a central venous catheter capable of continuous central venous oxygen saturation monitoring and adherence to a 6-hour resuscitation protocol.  This protocol called for the use of IV fluids, vasopressors and blood transfusions triggered by specific physiological measurements. Usual care was determined by on-site treating physicians. The primary outcome measured was 90-day mortality.

In an intention to treat analysis, ProMISe found that there was no significant difference in the 90-day mortality between the EGDT and usual care groups (29.5% vs 29.2%, p=0.90).  This was despite the EGDT group having a higher utilization rate of vasopressors, red blood cell transfusions, and intravenous fluids during their 6-hours following trial randomization.

What does ProMISe add to our knowledge about sepsis management? Simply put, ProMISe demonstrated that in patients who are recognized early as having sepsis and treated with intravenous fluids and antibiotics, EGDT did not provide significant improvement over patients treated with usual care.  However, ProMISe does more than just echo the findings of ProCESS and ARISE.

First of all, the ProMISe study aimed to enroll a more critically ill study population than those examined by the two preceding studies.  They were successful in this respect, with an overall trial mortality of 29%.  However, the observed mortality rate remained significantly below that of the original EGDT trial.  While the results from ProMISe allow an expansion of conclusions from ProCESS and ARISE to include more sick populations, some might argue that they fail to allow for translation to those who carry even higher expected mortality rates, such as those in the original EGDT trial.

Secondly the authors of ProMISe, acknowledging that valuable outcomes in sepsis may be more than just the most immediate and health-related ones, also defined a set of interesting long-term outcomes relating to quality of life and costs.  The study did not find a significant difference between EGDT and usual care groups with respect to reported quality of life and healthcare costs at 90 days and 1 year.  Additionally, ProMISe calculated a low probability that EGDT was cost effective in comparison to usual care. At a time when the conversations about healthcare value and the patient experience are of increasing relevance to actual bedside care, these results add weight to the arguments against EGDT.

So is ProMISe the end for early goal directed therapy? The results of ProMISe imply that one can stray from the strict EGDT protocol and still achieve equivalent results in sepsis care.  However, it would be misleading to depict the results of ProMISe as pushing us towards some kind of post-EGDT world where sepsis interventions are neither early nor goal-directed.  As the study authors note, by the time of their randomization into the study every patient in ProMISe had received antibiotics and an average of 2 liters of intravenous fluids.  Patients in the usual care group were still given transfusions, vasopressors and intravenous fluids, suggesting physicians must have been following some goals (no doubt they were aware of the results of the hallmark 2001 trial), albeit ones less formalized than those promoted by EGDT.  

In the aftermath of the ProCESS-ARISE-ProMISe trilogy, it should come as no surprise that the Surviving Sepsis Campaign—a champion of EGDT—recently announced that they would be re-evaluating the sepsis literature and their EGDT protocols.  Instead of ending EGDT, ProMISe likely marks the beginning of a new chapter in sepsis care, where interventions remain early and goals remain present, but the rules governing the clinical approach put more emphasis on clinician intuition than strict targets.

Have questions or comments about the ProMISe trial? Please join the NEJM Group Open Forum from March 24 to April 2 for a fascinating online discussion with authors, experts and your peers. 

2 Responses to “The Final Nail in Early Goal Directed Therapy’s Coffin?”

  1. First, I want to congratulate Dr. Allen-Dicker for not hopping aboard the “dump EGDT” or “dump protocols” bandwagon, as so many have done. I, for one, appreciate an analysis that views the studies for what they can objectively tell us.

    There is a very important point to note about all three of the recent EGDT vs. “usual care” trials, that point being that on average the patients in the EGDT group of all three trials had a ScvO2 of 70% upon initial line insertion and measurement. This means, of course that a substantial proportion of patients in the protocol group (just exactly how many is not clear) had met the protocol’s goals before the protocol began. One has to assume that these patients received “usual care” from that point forward. Assuming that randomization was effective in all three trials, the same would apply to a substantial proportion of patients in the “usual care” arms – they would not have received any particular protocolized care even if they had been randomized to the EGDT arm of the study. Under these circumstances, and given the power analyses supplied by the authors, it would have been impossible to find a significant difference between the two groups with the number of patients studied. I have not actually done the calculation, but it seems well nigh impossible that any reasonable number of patients in the groups could find a difference when a large number of patients studied didn’t actually require the therapy under study.

    At the time the studies were initiated it seemed entirely appropriate that lactate > 4 and/or septic shock should have been the determining features for group allocation. These were the criteria used by Rivers, et al. in their seminal trial of EGDT in severe sepsis and septic shock. However, in the Rivers trial the average ScvO2 was 49% in both of the study arms. In other words, most patients in the Rivers trial did, indeed, qualify for the study therapy. In spite of the best intentions of the authors of three hugely important trials, and through no fault on their part, the studies may well have missed analyzing the most important question, which is “What is the effect of EGDT in patients with severe sepsis/septic shock and low tissue oxygenation, as evidenced by low ScvO2?” In other words, what is the effect of EGDT in the patients for whom it is actually designed? Unfortunately, ARISE, ProCESS, and ProMISe can’t answer that question and don’t really give us any further useful information on the utility of EGDT, except that it likely doesn’t help in people who don’t need it.

    I do not mean my comments as criticisms, other than to say that this is the nature of clinical science, indeed of science in general. It sometimes takes the experience of having done the experiment to understand how to make it better answer the question. It has happened to anyone who has ever undertaken scientific investigation. There are some good lessons to take from these studies. Why, one might ask, were the study mortalities all three so much lower than in the Rivers study? As Dr. Allen-Dicker points out, nearly all patients in all three groups received early antibiotics and 20 – 30 mL/kg of IV fluids by the time of randomization, and all patients were aggressively recruited and entered into the studies as early as possible in the ED. This was not standard practice at the time of the Rivers study, and one cannot determine from the published data just how it compares on that account. I don’t think we will ever find out whether this is the exact factor that substantially lowered mortality, though. Can anyone imagine designing a prospective study of early vs. delayed antibiotics?

    As for what I will do, myself, with this patient. It still seems prudent to understand that both lactate > 4 mmol/L and presence of septic shock are markers that tissue oxygenation may be inadequate. I am inclined to use those as triggers to investigate the ScvO2 and inclined to address low ScvO2 aggressively, likely following the current EGDT protocol or a good facsimile that perhaps uses dynamic measurements of stroke volume, rather than CVP, as the guide to volume administration. In spite of its single center nature and its relatively low patient numbers, the Rivers study remains the only randomized trial that has actually approached these low ScvO2 patients with a standardized treatment strategy.

  2. Tetsuo Kimura says:

    Sort of Deja-vu sensations.
    We saw something similar to this before, regarding pulmonary artery catheterization.