Guest post: Sepsis 2.5

by Howard Rodenberg, MD, MPH, CCDS

Sepsis-3 is celebrating its fourth birthday this year, and while it can be hard to judge the success or failure of a child until it’s forty, one can at least take an early look to make sure the kid isn’t tearing wings off flies. So, what do we make of this pre-k toddler?

The first thing we find is that contrary to the belief of its parents (and every third party payer who wants to deny payment), Sepsis-3 is not the cutest kid in class. Four years on, it is still not universally accepted as the gold standard for the definition of sepsis. If the key physicians involved in the care of the septic patient are critical care specialists, emergency physicians, and infectious disease experts, it’s important to note that specialty societies for the latter two groups have still not endorsed Sepsis-3. Knowing that two-thirds of the physicians who would potentially use your definition are non-believers doesn’t exactly support the claim that Sepsis-3 is the unquestioned diagnostic authority. In addition, CMS has yet to jump on the Sepsis-3 bandwagon, and the State of New York has enshrined Sepsis-2 as their definition for use in sepsis campaigns and payment plans. Outside of the First World, there are also concerns about this “universal” definition from countries lacking easy access to the laboratory testing required to establish the diagnosis through Sequential Organ Failure Assessment (SOFA) scoring.

One of the more overlooked aspects of the Sepsis-3 discussion has been its inherently subjective nature, and how the SOFA score has therefore been misunderstood. The authors of the Sepsis-3 document are clear in describing sepsis as a “life-threatening organ dysfunction.” There are two problems here. The first is that my definition of “life-threating” may be very different than yours. (Mine tends to focus around ex-wives.)

The second issue—the most important, but also the one that gets lost in the shuffle—is that this definition betrays the author’s bias. If we believe that sepsis is life-threatening, then our search for diagnostic criteria is going to be focused on markers of mortality rather than the simple presence of disease.

Here’s an example of what I mean. If we look at clinical markers in patients with heart failure (ejection fraction, BNP, et al), we recognize that different values suggest increased risks of mortality. However, the diagnosis of heart failure is not dependent upon any particular value that suggests death is imminent. By suggesting up front that sepsis is life-threatening, we’ve already telegraphed that we’re not looking for markers of disease but of mortality.

This is not to say the authors are necessarily wrong. I think most of us would agree that sepsis represents a severe systemic response infection with high rates of morbidity and mortality. But it’s important to keep this bias in mind, especially as we discuss the SOFA score.

Why has SOFA become the keystone of Sepsis-3? The Sepsis-3 team notes that that they took data sets of patients diagnosed with sepsis, looked at the patient’s scores in measures of systemic inflammation (SIRS) or organ dysfunction (SOFA; Logistic Organ Dysfunction System), and correlated these scores with outcomes. Given that sepsis is considered “life-threatening,” the understandable outcome of interest would be death. A two-point change in the SOFA score turned out to be the best predictor of mortality. But, as we’ve previously described, predictors of mortality are different than markers of illness, unless you’ve already established a high risk of mortality as an inherent part of your definition of disease.

Again, it bears repeating that there’s nothing inherently wrong with describing sepsis as a life-threatening event, and nothing wrong with the analysis used to determine correlations between sepsis mortality and the SOFA score. But it seems clear that when you build certain assumptions into your search for diagnostic criteria, you risk taking previously known scores out of context.

If we want to come up with diagnostic criteria for sepsis, perhaps we would be better served by looking at a cohort of patients diagnosed with sepsis under Sepsis-2 and finding the most common elements of their presentation; if we want our definition to focus on those at higher risk, we might even take the patients currently diagnosed with Sepsis-3 using the SOFA score and identify the most common elements in their clinical course. These exercises seem to me more likely to yield a disease profile for sepsis based on easily identified clinical markers rather than forcing a mortality predictor to do double duty.

In fairness to the Sepsis-3 team, they never say that SOFA represents the definition of sepsis. Their definition of sepsis, as noted, is that of a life-threatening organ dysfunction caused by a dysregulated host response to infection. A change in the SOFA score is suggested not as the definition of sepsis, but as a way to operationalize the definition. However, some clinicians and most private payers overlook that distinction in favor of tight adherence to changes in the SOFA as the gold standard measure of sepsis. I would hope that the Sepsis-3 group would be dismayed.

All this being said, what’s been most interesting to me in following the discussion is how the tenor of the conversation has changed over the past few years. While the clinical validity of the Sepsis-3 definition had been questioned early on, more and more literature suggests that the lack of operability may be its fatal flaw.

The term “operability” deserves some additional thought. The book definition of the word means “capable of being able to put into use, operation, or practice.” Taken more broadly and in our clinical context, “operability” means the ability to use a definition or criteria as part of the normal clinical workflow, and to make decisions or institute processes based on that information; a diagnosis of pneumonia becomes “operable” when we use it to institute antibiotic care. Operability encompasses the elements of resources, timing, and goals of care as well. A post-mortem diagnosis made at autopsy may be accurate, but is clearly non-operable in this broader sense.

How does this apply to Sepsis-3? In patients with severe illness, the best “operable” definition would allow us to quickly identify patients at risk early in their clinical course and initiate aggressive care as soon as possible. Sepsis-3 gains discrimination at the cost of sensitivity; this can be seen by comparing mortality rates for patients meeting Sepsis-2 criteria with those diagnosed under Sepsis-3 (12% vs 28% in our own in-house work).

The patient diagnosed under Sepsis-3 is clearly sicker, and farther down the road to mortality, than the one diagnosed via Sepsis-2. Since the criteria to meet the Sepsis-3 definitions are more stringent than those of Sepsis-2 and are generally seen later in the clinical course, Sepsis-3 becomes less “operable” in the context of trying to meet our clinical goals of early patient identification and intervention.

(Those more familiar with Sepsis-3 will recall that the authors proposed a “qSOFA” score as a triage tool for early identification of patients with sepsis. However, while the qSOFA has been found to be an improved predictor of mortality than SIRS, the qSOFA has not been found to be consistently better at the early identification of patients with sepsis than the prior Sepsis-2 criteria.)

 I’ve made no secret of the fact that given the choice, I much prefer Sepsis-2 over Sepsis-3. I think the opportunities for aggressive intervention to head off morbidity and mortality far outweigh the increased specificity of Sepsis-3. But I am beginning (in fits and starts, gritting my teeth the whole time) to give more credence to the fact that the Sepsis-2 criteria are non-specific, and that using SIRS plus infection can’t really sort out those who are truly sick with the systemic illness of sepsis from those who can down some Tylenol, grab a Z-pack, and go home to nurse their ills.

As an ED physician, I over-triage by nature, but also recognize that I use up time, resources, and my one remaining ounce of compassion when patients don’t need the level of care their initial complaints might suggest. (“What emergency medical condition brings you to our Level II Trauma Center at 3 a.m.?”) At the same time, I don’t want to waste time waiting for a patient to fit a more restrictive criteria before I can intervene. I want specificity without losing sensitivity; I want my cake and to eat it, too.

I think there are two possible ways to work through this problem. The first is to use what my colleague Dr. Douglas Campbell, our Pediatric CDI Physician Advisor, calls “Sepsis Cubed.” Patients with sepsis must have SIRS, an infectious source, and look sick—the “Three S’s of Sepsis.” (Interestingly, neither Sepsis-2 nor Sepsis-3 criteria require the patient to appear ill.) This scheme preserves the ability of Sepsis-2 to drive early treatment, while insuring that routine infections in patients described as in “no acute distress” are not considered in the same bucket as those who are truly ill.

The second option would be to take a step backwards, into what I’ll call Sepsis 2.5. This idea accepts that organ dysfunction differentiates sepsis from a localized infection but uses a different set of parameters to establish evidence of that dysfunction.

In this model, the key is to identify the proper measures of organ dysfunction to complement the presence of infection. We’ve already discussed why the SOFA score is really a mortality predictor and not an actual measure of sepsis. Fortunately, Levy, Fink and colleagues (in the original Sepsis-2 document) have given us a table of possible signs and symptoms of systemic inflammation and organ dysfunction, encompassing a wide range of general, inflammatory, hemodynamic, and tissue perfusion parameters. Given that a subset of patients may not manifest classic SIRS criteria despite the presence of infection (such as in the immunocompromised), this scheme allows for a more expansive picture.

Using the Sepsis 2.5 model allows us to expand signs of organ dysfunction into presentations such as altered mental status and acute kidney injury, without requiring that a certain score be met or multiple organ systems fail concurrently before sepsis is recognized and aggressive treatment begun. Sepsis 2.5 moves us away from using the flawed and misappropriated SOFA score into a paradigm that can be used in virtually any setting, can function as an ED triage tool, and ensures that the right patients get the right care. The Sepsis-3 division between sepsis and septic shock is maintained, but as the Sepsis-2 category of “severe sepsis” necessarily includes some measure of organ dysfunction (hypotension or lactic acidemia), both “sepsis” and “severe sepsis” are encompassed within the former term as in Sepsis-3.

There’s clearly a need to reconcile the competing definitions of sepsis in a way that makes clinical and operational sense. Maybe someone will take advantage of this birthday and come up with Sepsis-4. Now that would be something worth candles and cake.

Selected references

Braun D. A Retrospective Review of the Sepsis Definition After Publication of Sepsis-3. American Journal of Medicine 2019; 132(3):382-4.

Deutschman, CS. Imprecise Medicine: The Limitations of Sepsis-3. Critical Care Medicine 2016; 44(5):857-8.

Fernando S, Rochwerg B, Seely AJE. Clinical Implications of the Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). CMAJ 2018; 190 (36):E10058-9.

Freund Y, Lemachatti N, Krastinova E, et al. Prognostic Accuracy of Sepsis-3 criteria for In-Hospital Mortality Among Patients with Suspected Infection Presenting to the Emergency Department. JAMA 2017;317(3):301-8.

Machdo FR, Nsutebu E, AbDilaziz S. et al. Sepsis-3 from the Perspective of Clinicians and Quality Improvement Initiatives. Journal of Critical Care 2017; 40:315-17.

Marshall JC. Sepsis-3: What is the Meaning of a Definition? Critical Care Medicine 2016; 4(8):1459-60.

Raith E, Udy A, Bailey M, et al. Prognostic Accuracy of the SOFA Score, SIRS Criteria, and qSOFA Score for In-Hospital Mortality Among Adults with Suspected Infection Admitted to the Intensive Care Unit. JAMA 2017; 317(3):290-300.

Sartelli M, Kluger Y, Cateno F. Raising Concerns About the Sepsis-3 Definition. World Journal of Emergency Surgery 2018; 13(6): Open Access at https://wjes.biomedcentral.comn/articles/10.1186/s13017-018-0165-6.

Seymour CW, Liu VX, Iwashyna TJ, et al. Assessment of Clinical Criteria for Sepsis from the Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). JAMA 2016; 315(8):762-74.

Verboom DM, Frencken JF, Ong DSY, et al. Robustness of Sepsis-3 Criteria in Critically Ill Patients. Journal of Intensive Care 2019: 7(46); Open Access https://jintentsive care.biomedcentral.com/articles/10.1186/s40560-019-0400-6.

Editor’s note: Rodenberg is the adult physician advisor for CDI at Baptist Health in Jacksonville, Florida. Contact him at howard.rodenberg@bmcjax.com or follow his personal blog at writingwithscissors.blogspot.com. Advice given is general. Readers should consult professional counsel for specific legal, ethical, clinical, or coding questions. Opinions expressed are that of the author and do not represent HCPro or ACDIS.