Are traditional protocols for goal directed therapy for sepsis dead? (ARISE trial)
Update: As astute commenters have mentioned below, and as I stated in our post on the ProCESS trial, protocols of some kind could still have a place in the care of sepsis if they ensure more rapid recognition and thorough treatment. Accordingly, I changed the headline to clarify that it's only traditional sepsis protocols to which I refer, not the existence of protocols for sepsis treatment in general (which I support).
Also, the ProCESS and ARISE investigators did not pronounce sepsis protocols "dead," rather, they urged caution and reinforced the absolute need for vigilance for early recognition and treatment of sepsis to ensure good outcomes. Sepsis protocols might help achieve those goals.
In other words, rumors of sepsis protocols' death have been greatly exaggerated. This is a controversial and important public policy issue and you should rely on national experts and the original literautre, not just a blog, to guide management of your patients.
Okay, disclaimers complete. Read on.
Protocols for goal directed therapy for sepsis and septic shock had their day in the sun. But as standard care for sepsis has steadily improved, the sun is setting on their usefulness, at least at the best centers, and as they are currently written.
You remember that in May 2014, the large U.S. multicenter ProCESS trial showed that protocolized sepsis care did not improve outcomes in patients as compared to individualized physician judgment. The even larger ARISE trial, just published online in the New England Journal of Medicine, came to the same conclusion outside the U.S. Like ProCESS, it's a must-read.
At 45 centers in Australia and New Zealand (and 6 centers in other countries), investigators randomized 1,600 adults presenting to emergency departments with severe sepsis or septic shock to receive either protocolized early goal-directed therapy or usual care for 6 hours.
The protocol used for early goal-directed therapy was the "sepsis bundle" that has been promoted in the U.S. since the 2001 publication of a single-center trial by Emanuel Rivers et al.: continuous monitoring of blood pressure and central venous oxygen saturation (ScvO2), with IV fluids, vasopressors, dobutamine and/or red blood cell transfusions administered to attain goals of central venous pressure, mean arterial pressure, and ScVO2.
Usual care was whatever the physician felt appropriate, but could include any of the above interventions except monitoring of central venous oxygen saturation, which was forbidden for the first 6 hours. At enrollment, most patients (70%) had full-on septic shock (hypotension unresponsive to fluid resuscitation), in both groups. Also:
Almost all (90%) of patients in the goal-directed group got central venous catheters with continuous ScvO2 monitoring; about 62% of patients in the usual care group got ordinary central lines but their central venous oxygen saturation was not monitored at all; almost 40% of patients had peripheral IVs only.
Those in the goal-directed therapy arm got significantly more of most everything else too: arterial lines, fluids in the first 6 hours, vasopressors, red blood cell transfusions, and dobutamine. There was very high compliance (>90%) with goal-directed therapy in the intervention group.
What happened? After 90 days, mortality was equal in both groups: 18.6% vs 18.8%. There were no meaningful differences in any other outcome measured.
The bigger and better-conducted a randomized trial is, the less there is to say about it (besides Yep, that's probably true). This was a large, well-conducted study by the currently preeminent critical care research consortium worldwide. ARISE had a lot of power to detect a difference between groups -- far more than the original Rivers study (single-center, n=263), and even more than ProCESS with its three groups (n=1,341 total). As a multicenter trial, its results are more valid than a single-center study.
Both ARISE and ProCESS showed no benefit of protocolized treatment for sepsis over physicians using clinical judgment on individual patients. The only reasonable conclusion is that at centers with experience and effective systems in place for treating sepsis, using protocols does not improve outcomes compared to conscientious, individualized care. Central venous catheters and arterial lines are often helpful, but appear to be unnecessary at least some of the time. Red cell transfusions and dobutamine never had a strong basis for use in sepsis, and there is nothing to suggest they helped in ARISE.
It's essential to mention that care in both ProCESS and ARISE was started in the emergency department in almost all cases and included prompt resuscitation and antibiotics -- which does not always happen outside of clinical trials.
"But the mortality was so much lower than expected! This is probably not valid."
19% mortality is low compared to the old days of sepsis treatment, but outcomes have improved; this mortality rate likely reflects the new normal, at least in Australia and New Zealand where sepsis mortality has fallen by half over the past 10 years, from 35% to 18%.
There were plenty of patients (70%) in both groups with true septic shock in ARISE. Adherence with EGDT was high, and more patients got the protocolized interventions. They just didn't help at these centers.
"This doesn't prove protocols don't work -- 'usual care' included the sepsis bundle elements too."
This is half true but misses the point. It's not that protocols don't work, it's that they don't seem to provide a benefit over an attentive physician using her clinical judgment. Most destructive to this argument is the fact (see above) that almost 40% of patients in the usual care arm did not get central lines, and none had measurements of their central venous oxygen saturation -- a variable whose manipulation was previously believed to be essential to optimal sepsis care.
There's another serious problem with existing sepsis bundles as written. Most bundles mandate red blood cell transfusion for critically ill patients with hemoglobin below 10 g/dL -- a liberal approach which has now been shown to have no benefit in patients with septic shock in a separate randomized trial. Red cell transfusion's inclusion in sepsis protocols (at hemoglobin levels above 7 g/dL) is no longer defensible. There has never been a good case for dobutamine either, as it has never been shown to improve outcomes and in fact does not improve microvascular perfusion in septic shock.
Clinical Takeaway: Protocols for treatment of severe sepsis and septic shock served a good purpose, improving the standard of care by drilling the importance of continual evaluation and intervention of patients with early severe sepsis to ensure end-organ perfusion. At many centers, they may continue to serve this purpose. But with abundant proof that good doctors can treat sepsis perfectly well without a protocol, that red cell transfusion in most septic patients is inadvisable, and with concerns that the original Rivers trial had potentially serious methodological problems, it's time to consider the sepsis bundle optional at centers where the systems and experience are in place to manage sepsis well.