Early Goal-Directed Therapy in the Treatment of Severe Sepsis

(N Engl J Med 2001;345:1368-77.)

This was a trial of sepsis management in the Emergency Department. The study population was a sick group of patients meeting SIRS criteria and either SBP < 90 after 20-30cc/kg fluids OR lactate > 4.0. This population was randomized to protocolized care involving SvO2-based interventions and protocolized care without SvO2-based interventions. There was very impressive mortality: 30% in treatment arm and 46.5% in control arm.

Equally impressive was the benefits to the group receiving SvO2-based interventions - blood transfusion and dobutamine for goal SvO2 > 70. Unfortunately, the study design and implementation lacked the internal validity to conclude that the mortality benefit stems from SvO2-based interventions.

Only the EGDT group was forced to stay in the ED for 6h. The standard therapy group get to be dispo’ed to the ICU whenever they were “stable enough”. In other words, perhaps being treated in the ED for 6h drove the mortality benefit.

Furthermore, the patient’s allocation was not blinded. After all, one group had a clear SvO2 number stamped on the monitor and the other group did not. Dr. River's was frequently at the bedside of patients in the study group directing resuscitation. In other words, perhaps having closer attention by an experienced physician drove the mortality benefit.

So despite the fact that both groups had the same resuscitation goals of CVP, MAP, and urine output, they did not actually receive the same initial resuscitation. Only 86% of the control group reached said goals compared to 99.2% of the EGDT. This explains for the average difference of 1.5L fluid given in the first 6 hours.

After 6 hours of resus – you can already start to see that the control group did not react CVP goals as an average from fluid resus and were on unequal footing for MAP avgs with pressors.

Control EGDT

CVP 11.8 13.8 p < 0.007

MAP 81 95 p < 0.001

Simply put, the treatment arm received more attention, time, tinkering, and fluid resuscitation in the ED than the control arm. This severely confounds whether SvO2 monitoring with dobutamine and transfusion therapies had independent effects on mortality. And this could explain why PROCESS, ARISe, PROMISE trials made the conclusions they did.