Critique of early goal-directed therapy protocol for sepsis

This chapter is dedicated to the dissection of perhaps the most famous randomised controlled trial in the management of sepsis, the 2001 Rivers trial titled "Early Goal-Directed Therapy in the Treatment of Severe Sepsis and Septic Shock".

Why is it important to dedicate an entire page to Rivers-bashing?

Well. Certainly, our illustrious college has already dedicated two past paper questions to interrogating their candidate's understanding of this trial, their ability to recall the "goals", and their familiarity with the controversy which has enveloped this topic. Given the influence of this paper, it is not surprising that even now  it is being waved around in ICU exit exams: Question 16 from the second paper of 2013 and Question 28 from the first paper of 2015 are certainly not where it is going to end.

The most important resources for answering all questions of this nature would have to include the following:

In brief summary:

Classical goals  as per Rivers et al. (2001)

  • CVP  8-12 mmHg
  • MAP 65 – 90 mmHg
  • Urine output >0.5 ml/kg/hr
  • SvO2 >65% / ScvO2 >70%
  • Haematocrit >30%

Disadvantages of EGDT

  • CVP is unrelated to preload or fluid responsiveness
  • No benefit of SvO2 or ScvO2  as compared to lactate
  • No evidence for HCt > 30% as a transfusion goal (TRISS)
  • Rivers' methodology attracted criticism: non-blinded, single-centre, suspiciously high mortality rate, only enrolled ED patients, etc.

Rationale and advantages of EGDT

  • Other EGDT bundles (eg. for MI) improve survival
  • Rivers trial demonstrated improved survival
  • SSG strongly endorse its use (or .. did)
  • MAP goals are evidence-based (SEPSISPAM)

Evidence against EGDT

  • ANZICS did not endorse it.
  • ProCESS,  ARISE and ProMISE studies did not demonstrate any survival benefit from EGDT

The "goals" of goal-directed therapy

When one embarks upon a course of "early goal-directed therapy", if one is faithful to the standards of the original paper, one commits to achieving a series of physiological parameters:

  • CVP  8-12 mmHg
  • MAP 65 – 90 mmHg
  • Urine output >0.5 ml/kg/hr
  • Mixed venous oxygen saturation >65% / ScvO2 >70%
  • Haematocrit >30%

Means of achieving the goals of EGDT

Rivers used a protocolised pathway of interventions to achieve the abovementioned goals. These interventions consisted of:

  • Decrease metabolic demand by means of sedation, paralysis and mechanical ventilation
  • Fluid resuscitation to maintain CVP and MAP
  • Vasopressors to maintain CVP and MAP if fluids have failed
  • Blood transfusion to increase ScvO2 above 70%
  • Inotropes to increase ScvO2 if haematocrit is > 30%

The philosophy of early goal-directed therapy

The rationale for an early aggressive assault on oxygen delivery is grounded in some fairly sensible assertions.

Rationale for early goal-directed therapy

  • Poor oxygen delivery to organs (another words, shock) leads to multi-organ system failure.
  • Duration of shock correlates with severity of organ dysfunction.
  • Severity of organ dysfunction
  • Patients with sepsis frequently have shock- they present with a high lactate, deranged oxygen extraction and suboptimal hemodynamic parameters.
  • Ergo, early normalisation of oxygen delivery should improve survival in septic shock, by decreasing the incidence of multi-organ system failure.

The official answer to the CICM question adds a parallel to EGDT. If ealy goal-directed therapy for myocardial infarction, stroke and trauma works, then why not for sepsis? This is an unusual leap to make, as the abovementioned "early management" strategies all rely on the correction of very different physiological abnormalities, eg. the coronary artery thrombus or catastrophic exsanguination. However, the uniting feature is the suggestion that perhaps critically ill patients should be managed in a rapid, attentive and coordinated manner, rather than being allowed to fester in the corner of the emergency department.

Critique of Rivers' methodology and rationale

  • This was a single centre study. I imagine Rivers himself went down to the ED and managed the trial patients. The trial conclusively demonstrates that the constant presence of a senior intensivist at the bedside of a septic patient will improve their chances of survival.
  • The trial was non-blinded, which introduces a thick layer of bias.
  • The trial only enrolled ED patients, and we have no idea whether the same sort of mortality benefit would be seen in patients who become septic in ICU, or those whose sepsis manifests in the middle of their ward admission.
  • The control group had an above-average mortality rate. This mortality rate - a reduction from 46.5% to 30.5% - was still higher than the average rate in Australian ICUs (which do not practice early goal-directed therapy).
  • The whole protocol was the test intervention, and it is unclear which particular goals should be prioritised. Much has been made of the use of ScvO2, which is not validated as an endpoint of management in sepsis, and it may not be a good surrogate for mixed venous saturation.
  • Haemodynamic end-points used in this trial do not address sepsis-induced microvascular haemodynamic changes. In sepsis, the ScvO2 may appear normal purely because of microvascular shunting; however the lactate will still be raised, as there may be whole islands of tissue which are essentially bypassed by the deranged microciculation.
  • It is unclear why ScvO2 was used, when lactate monitoring is non-inferior to ScvO2 and is not subject to inaccuracies such as the position of the sampling catheter tip in relation to the right atrium. Lactate may be a better surrogate marker of microvascular perfusion.
  • The only independently validated time-critical intervention is early antibiotic administration. It is unclear as to how much contribution to survival is made by the pursuit of a high CVP, or a high haematocrit.
  • Transfusion target to improve DO2 contradicts restrictive transfusion practice and may be associated with increased mortality in the critically ill. Transfusion of blood is not a uniformly benign process, and most people would agree that haematocrit is a weird endpoint to pursue. Surely, if one wished to improve tissue oxygen delivery, one would rather leave the blood at a lower viscosity, and use a microvascular vasodilator like dobutamine.

Support for early goal-directed therapy in the literature

  • A multi-centre study from 2010 duplicated Rivers' results (15.7% improvement in mortality)
  • EGDT has been incorporated into the Surviving Sepsis guidelines as a 1C recommendation.

Defeat of goal-directed therapy by the evidence

Though previously this section was brief and to the point, Question 28 from the first paper for 2015 has generated the need for indepth dissection of the evidence. The ProCESS, ARISE and ProMISE studies were quoted by the examiners. Even though "candidates were not expected to provide specific details of the trials", the college model answer provides these details in abundance, and so shall I.

ProCESS trial (2013): n=1341; multicentre RCT (31 US hospitals)

  • 3 arms: protocol-based therapy (full EGDT), protocol-based standard care, and some sort of  "anything goes" non-protocol therapy
  • The primary end point was 60-day in-hospital mortality.
  • Mean APACHE II score was about 20.8; this closely matches Rivers (whose groups had an APACHE II score around 20-21)
  • Outcome for this were 21.0% mortality (EGDT), 18.2% (standard protocol) and 18.9% (anything goes).
  • There were no significant differences in 90-day mortality, 1-year mortality, or the need for organ support.

ARISE trial (2014): n=1600; multicentre RCT (50 Australian hospitals)

  • 2 arms: protocol-based therapy (full EGDT), and "anything goes" non-protocol therapy.
  • The primary outcome was all-cause mortality within 90 days.
  • Mean APACHE II score was about 15.8
  • Rates of death were 18.6% (EGDT) and 18.8% (standard)
  • Again, there was no significant difference in survival time, in-hospital mortality, duration of organ support, or length of hospital stay.

 ProMISE trial (2015): n=1260; multicentre RCT (56 UK hospitals)

  • 2 arms: protocol-based therapy (full EGDT), and "anything goes" non-protocol therapy.
  • The primary outcome was all-cause mortality at 90 days.
  • Mean APACHE II score was about 18.0
  • Rates of death were  29.5% (EGDT) and 29.2% (standard)

Meta-analysis (2015) - Angus, Barnato, Bell, Bellomo et al...

  • The authors of the ProCESS, ProMISE and ARISE trials scraped the databases for abstracts. Of 2395 eligible abstracts, only five RCTs met the inclusion criteria, totalling 4735 patients. Of these, the ProCESS, ProMISE and ARISE trials contributed 4201. The remaining 532 patients came from the Rivers trial and the lesser-known Jones trial (2010) which actually had more patients than Rivers.
  • In summary, EGDT was not associated with decreased mortality, but it was associated with increased admission to ICU.
  • Did this increase in ICU admissions improve the organ system function of admitted patients? No it did not. The same organ system support requirements were the same, except for the fact that the EGDT group got more vasopressors.
  • The authors of this meta-analysis included the principle investigators of the ProCESS, ProMISE and ARISE trials, but it did not include Emanuel Rivers.

Advantages of using early goal-directed therapy

  • The hemodynamic goals are laudable - at least MAP is a physiological parameter worth pursuing.
  • If one is comfortable with the methodology of the trial, the improvement in survival (16%) is substantial.
  • Individual use of the EGDT strategy is validated by its incorporation into the Surviving Sepsis guidelines.
  • It is possible to embrace the philosophy of early aggressive resuscitation while remaining opposed to the controversial wackyness of venous oxygenation monitoring, and adding one's own style to the process (by using inodilators early, for example).

Disadvantages of using this approach in one's own practice

  • You end up using more fluids and blood products
  • You end up resorting to more vasopressors and inotropes.
  • The use of EGDT was trialled in the ED population only- benefit in the ICU setting is uncertain.
  • The use of the whole package has been criticised, and one may wish to discard certain parameters from it (eg. the abovementioned controversial ScvO2 measurement, or blood trasnfusion)


Rivers, Emanuel, et al. "Early goal-directed therapy in the treatment of severe sepsis and septic shock." New England Journal of Medicine 345.19 (2001): 1368-1377.

Jones, Alan E., et al. "The effect of a quantitative resuscitation strategy on mortality in patients with sepsis: a meta-analysis." Critical care medicine 36.10 (2008): 2734.

Kumar, Anand, et al. "Duration of hypotension before initiation of effective antimicrobial therapy is the critical determinant of survival in human septic shock*." Critical care medicine 34.6 (2006): 1589-1596.

Early Goal-Directed Therapy Collaborative Group of Zhejiang Province. "The effect of early goal-directed therapy on treatment of critical patients with severe sepsis/septic shock: A multi-center, prospective, randomized, controlled study." Zhongguo wei zhong bing ji jiu yi xue= Chinese critical care medicine= Zhongguo weizhongbing jijiuyixue 22.6 (2010): 331.

Yealy, Donald M., et al. "A randomized trial of protocol-based care for early septic shock." The New England journal of medicine 370.18 (2014): 1683-1693.

Power, GSarah, et al. "The Protocolised Management in Sepsis (ProMISe) trial statistical analysis plan." Critical Care and Resuscitation 15.4 (2013): 311.

Delaney, Anthony P., et al. "The Australasian Resuscitation in Sepsis Evaluation (ARISE) trial statistical analysis plan." Critical Care and Resuscitation 15.3 (2013): 162.

Marik, Paul E. "Early Management of Severe Sepsis: Concepts and Controversies." CHEST Journal 145.6 (2014): 1407-1418.

Peake, Sandra L., et al. "Goal-directed resuscitation for patients with early septic shock." The New England journal of medicine 371.16 (2014): 1496.

Yealy, Donald M., et al. "A randomized trial of protocol-based care for early septic shock." The New England journal of medicine 370.18 (2014): 1683-1693.

Mouncey, Paul R., et al. "Trial of early, goal-directed resuscitation for septic shock." New England Journal of Medicine 372.14 (2015): 1301-1311.

Angus, D. C., et al. "A systematic review and meta-analysis of early goal-directed therapy for septic shock: the ARISE, ProCESS and ProMISe Investigators." Intensive care medicine (2015): 1-12.

Jones, Alan E., et al. "Lactate clearance vs central venous oxygen saturation as goals of early sepsis therapy: a randomized clinical trial." Jama 303.8 (2010): 739-746.