When the Third International Consensus Definitions for Sepsis and Septic Shock, known as Sepsis-3, were released in February, some in the field felt some shock themselves.
Among other changes, the expert authors recommended that severe sepsis no longer be used as a diagnostic term. They also suggested replacing systemic inflammatory response syndrome (SIRS) criteria with other diagnostic tools to identify sepsis—the Sequential Organ Failure Assessment (SOFA) score and an entirely new concept, the quick or qSOFA.
To learn more about these changes and the reasoning behind them, ACP Hospitalist recently spoke with Clifford S. Deutschman, MS, MD, co-chair of Sepsis-3 and a professor of pediatrics and molecular medicine at Hofstra-Northwell School of Medicine in Hempstead, N.Y.
Q: Why were new sepsis criteria needed?
A: A little bit of history is required. Prior to 1991, no 2 people meant the same thing when they said sepsis....So a group of experts headed by Roger Bone [MD] got together and came up with the first set of definitions, what we call Sepsis-1. The key things were that sepsis was defined as suspected, or documented, or presumed infection plus 2 out of 4 SIRS criteria.... They then said that if you had some kind of organ dysfunction on top of meeting the SIRS criteria and having suspected infection, you had severe sepsis, a brand-new term that no one had ever used before.
On the one hand, it was fairly well accepted and it really helped with standardization; on the other hand, there were people who voiced concerns about it from the word go. In response to those concerns, a second conference was convened in 2001, a paper published in 2002....Basically what came out of that was a statement that despite concerns, there really wasn't any good way to improve what was already out other than to add a whole lot of other things to the SIRS criteria.
[Since then] we've learned a lot more about the biology, the pathobiology behind sepsis, [and] we've had a bunch of clinical interventions that have improved management....It looks for all intents and purposes like outcome has improved, but there are huge discrepancies between sites. For example, the mortality for septic shock in the United States is about 20%, in Australia it's about 20%, in Great Britain it's about 35%, in France and Italy and Germany, it's 55 to 60%. Our care for these patients is not 3 times better than the care is in Germany, so something is wrong.
Probably the most important thing of all, the SIRS criteria are incredibly sensitive. They get everybody that we think has sepsis, but they get everybody who has a bad cold, too. That's probably part of what accounts for the big discrepancy in mortality. We have more ICU bed capacity than they do in Europe, so we put less sick patients in the ICU and it makes our outcomes look better.
Q: What was the process for developing Sepsis-3?
A: It was a joint project of the Society of Critical Care Medicine and the European Society of Intensive Care Medicine. The 2 societies put up some of the infrastructure and some of the money, but basically, other than appointing myself and Mervyn Singer [MD] from University College London as co-chairs, they had absolutely no input into the process at all....Mervyn and I selected people based on a track record of investigation into sepsis.
In the first 2 [sepsis conferences], these were definitions based on expert opinion. In the first meeting we had, we decided that wasn't good enough, that whatever we did, we had to produce something that was data-driven. We now have these huge electronic databases, things that weren't available then. We agreed right off the bat that severe sepsis and sepsis were used interchangeably and there was no point in having 2 terms for the same thing, and so we recommended that severe sepsis be dropped.
Q: How did you come to the new definitions and criteria?
A: If you want to absolutely 100% without fail identify a patient who has sepsis, you can't....It's like pornography, I can't define it, but I know it when I see it. We took the best biological evidence we could, from patient studies, from animal studies, from cell studies, from everything, and synthesized it into something that we thought was the best description of what sepsis is [that is] possible at this time “Sepsis is life-threatening organ dysfunction caused by a dysregulated host response to infection.” Relatively straightforward and clinically pretty much useless.
We have to use something else instead to identify the patients who are most likely to have sepsis. So we turned to Chris Seymour, MD, a critical care physician at the University of Pittsburgh Medical Center, who is an expert in examining large databases...What we decided was if you have somebody with suspected infection, and they have a bad outcome, it probably was sepsis. Chris defined a bad outcome as spending 3 or more days in the ICU or dying, and used these parameters to examine several large databases to find clinical criteria that could identify these patients....It turned out that SIRS was pretty good, better than we expected. But was there something better? It turns out a SOFA score ≥2 was better....But Chris also found that if you eliminated patients from the ICU, and just looked at patients on the hospital ward or in emergency departments, the presence of 2 out of 3 easily measured clinical criteria—systolic blood pressure less than 100 [mm Hg], a respiratory rate greater than 22 breaths per minute, and an altered mental status—identified patients better than SIRS and almost as well, probably as well, as the SOFA ≥2...We called this new composite quick SOFA or qSOFA.
The 1 thing that absolutely separates a patient with septic shock from a patient with sepsis is a higher mortality rate, so we looked again at the literature and came out with a definition that said septic shock is a subset of sepsis, where circulatory and cellular and metabolic abnormalities result in a markedly elevated mortality. It turned out...that the only combination that identified a group with statistically elevated mortality was when 3 things were present—when you had a lactate that was more than 2 [mmol/L], when you were requiring vasopressor medication, and when you had a systolic blood pressure that was low.
Q: What are your thoughts on the criticisms of Sepsis-3?
A: Change scares people....And despite the best of intentions, none of the critical editorials is based on data. The concerns we've heard are as follows.
“How can you leave lactate out?” Well, we're not saying leave lactate out. We're saying that, in the model that we constructed, lactate didn't add anything to qSOFA. If, for example, you have somebody who is breathing fast and has an elevated lactate, that might be somebody else to worry about. We don't have data to say yes or no to that, but we're certainly not going to say don't draw lactate. The flip side is when we have lactate—we require it in [the criteria for] septic shock—we're getting people from resource-limited environments, both in and outside the United States, saying, “We can't measure lactate.”
The second criticism is that none of these data were derived prospectively, which is absolutely 100% correct. It's absolutely essential that moving forward, both SOFA ≥2 and qSOFA be tested prospectively. There was a recent paper in Critical Care looking at applying qSOFA to critically ill emergency patients with suspected pneumonia, and that's the first time I've seen somebody actually applying this in print and it worked. So that's a good thing.
The other big criticism is that we didn't include patients from smaller hospitals, from nonacademic medical centers—absolutely not true. The data we assessed did indeed include those patients.
We've been told, “The SIRS criteria—We're finally getting people to use them. How can we abandon them?” The answer to that is we're not abandoning them. They are things you're going to measure routinely anyway....They just aren't very good at identifying patients who go on to have poor outcomes. They're very good at identifying patients with infection. That includes a lot of patients who aren't all that sick....We're not saying don't use them.
Q: Another concern has been the impact of Sepsis-3 on coding and documentation. What are your thoughts on that?
A: The Centers for Medicare and Medicaid Services have basically come out with their own set of criteria, SEP-1, and they're not budging on it. They want to keep using severe sepsis. This is the way they're going to do coding and that means it's the way that everybody in the U.S. will need to follow. We hope that over time this would change. The CDC is sponsoring a couple of data-gathering projects that we hope will serve to catalyze that change.
So we're in the very uncomfortable position of telling people that you shouldn't call it severe sepsis, but you have to call it severe sepsis if you want to get paid for it in this country....We derived [Sepsis-3] based on scientific evidence and data-driven results. It wasn't in any way, shape, or form geared to meet billing requirements or coding requirements, which only apply in the United States anyway....Probably to our detriment, we do have recommendations in the [Sepsis-3] paper on how you should code for this stuff that may need to be reconsidered based on SEP-1, which wasn't part of our discussion in any major way. It's a valid criticism. I also think it's a criticism of the way we pay for medicine in this country and not of the science and data-driven results that are represented in these 3 papers.
Q: Do you expect the Surviving Sepsis Campaign to make changes based on Sepsis-3?
A: Mitchell Levy [MD, executive committee member of the Surviving Sepsis Campaign] was part of the task force that generated the new definitions and the clinical criteria. Importantly, Surviving Sepsis is about intervention. It's about management. The definitions are not about management. We would hope that the Surviving Sepsis Campaign would change the term severe sepsis to just sepsis and stop talking about severe sepsis. I would hope that in the future...in acquiring data, they would apply the new clinical criteria and perhaps compare them to the old ones, because that is a really good way to see what works better in a prospective manner. I'm hopeful of that; there's no guarantee of it.
Q: Your editorial in Critical Care Medicine asked clinicians to try out the new criteria and offer feedback on its value. How should they do that?
A: Other than hoping they'll tabulate large sets of data and [publish] their results from accumulated patient care, I don't know that we've put together a mechanism. Maybe that's something I should talk to the Surviving Sepsis people about, see if that would be a good repository for people to tell each other how they think things are working.
[For example] at our institution here they have something that they've been using for a couple of years that they call Super SIRS, which it turns out is very similar to things that were in our paper. We're going to try to sit down with the Super SIRS database and apply the SOFA and qSOFA, [to] see if the same patients are identified, see what the comparison looks like.
Q: What are the biggest messages you'd like hospitalists to remember?
A: 1) Don't call SOFA ≥2 or qSOFA definitions anymore. Call them clinical criteria. Definitions are about what something is, whether or not that actually helps you identify it. Clinical criteria can be used in the real world to help identify the most likely suspects. So let's try and be correct and separate the real biology from what we're using to help us identify patients. 2) Keep track of what you're doing. Tabulate your data. Collect data moving forward. 3) If the little alarm starts dinging in the back of your brain or you get this funny feeling in the pit of your stomach that says to you, ‘This ain't right. This patient is not going to do well,’ pay attention. Because nothing we're saying supersedes the judgment of a good clinician.