A group of three manuscripts published in 2016 describe the deliberations, research efforts, and decisions of a group of individuals who revisited and revised the definitions of sepsis and septic shock (1–3). The results of these efforts should profoundly affect sepsis research. Previous groups had delineated the original definitions in 1991 (sepsis-1) (4) and marginally revised them in 2001 (sepsis-2) (5). Therefore, the JAMA papers represent the first real revision of these definitions since 1991. The authors of this viewpoint, both members of the sepsis-3 task force, will summarize the new definitions and share our opinions on how the changes will impact investigators engaged in basic and translational research.
DEFINITION OF SEPSIS
One of the critical discussions in sepsis-3 surrounded the meaning of the word “definition.” Ultimately, the Task Force consulted the Merriam-Webster's Dictionary that stated that a definition is “something that identifies the essence of a thing” or, more succinctly, “what a thing is.” As such, it should be subject to change only when something fundamentally new is learned to suggest that the previous understanding is incorrect or incomplete. The essence of sepsis—or any clinical entity—should not be altered to expedite identification of patients by including some diagnostic element (e.g., bedside appearance, physiologic measurement, pathologic, laboratory or radiologic finding). Rather, it was the task force's belief that the definitions of sepsis and septic shock should incorporate the most up-to-date understanding of the pathobiology of the disorders. Critically, this involved an uncoupling of our intellectual understanding of sepsis from clinical signs and symptoms seen at the bedside. While the latter may help diagnose the disorder, they are simply visible manifestations of the actual underlying causal process. Based upon this intellectual framework, the new definition of sepsis is “life-threatening organ dysfunction caused by a dysregulated host response to infection.”
Several important considerations impacted on the new definitions as ultimately formulated. First, we recognized that sepsis is a disorder in which infection is a key initiating factor. As such, we limited our discussion solely to processes involving infection. However, we explicitly did not attempt to determine what constitutes an infection, leaving that to specialists in microbiology, infectious diseases, etc. In addition, we recognize that a very similar response can be initiated by noninfectious insults.
Next, we confronted how sepsis differs from simple infection. This is a critical point and one that was highly problematic in prior approaches. Specifically, a major drawback of the Systemic Inflammatory Response Syndrome (SIRS) criteria first enumerated in the 1991 definitions was that they are highly specific but lack the ability to discriminate between a patient with a life-threatening disorder from one with, for example, a bad cold. Whereas the authors of the 1991 paper identified sepsis as suspected infection plus 2 of 4 SIRS criteria, they identified severe sepsis as sepsis accompanied by organ dysfunction. In contrast, the sepsis-3 task force members felt that organ dysfunction was a defining component of sepsis that must be included in any revised definition. Therefore, sepsis was defined in terms of organ dysfunction, rendering the term severe sepsis unnecessary. In essence, what was previously called sepsis (which was not associated with organ failure) is now called infection and what was previously called severe sepsis is now called sepsis. Further, sepsis-associated organ failure is associated with significant mortality. Because sepsis is potentially lethal by its very nature, the term “life threatening” was added to the sepsis-3 definition.
Finally, since the prior sepsis definitions were formulated, there has been a change in the conceptualization of the host response. Sepsis was initially identified solely as an immunologic syndrome, specifically as a disorder of excessive inflammation (6–13). However, a number of studies have subsequently demonstrated early, and frequently concurrent, development of inadequate inflammation, or immunosuppression (13–17). Recent evidence also suggests that very early dysfunction in other systems is also important in the pathogenesis of sepsis (18–20). In essence, the entire host response is dysregulated, not just the immune component.
CLINICAL CRITERIA FOR SEPSIS
While the sepsis definition is scientifically (and entomologically) sound, it is not very useful clinically. Thus, an additional goal of the task force was to provide clinical criteria—simple changes in symptomatology, physical examination, and radiologic or laboratory findings that could help the practitioner identify a septic patient at the bedside. Indeed, what had been called “definitions” in sepsis-1 and sepsis-2 (e.g., infection + 2/4 SIRS criteria) are more appropriately thought of as clinical criteria. A critical component of sepsis-3 was that any newly proposed clinical criteria should differ from the previous iteration by being data-driven. The chosen approach therefore involved interrogation of a number of large electronic health-record databases. It was important to examine data from the largest and most diverse patient sampling possible. Therefore, the five databases assayed evaluated patients in multiple countries, multiple types of healthcare systems (academic, community, federal), and multiple sizes to attempt to maximize the generalizability of our findings.
However, the task of determining clinical criteria for sepsis was not simple. There is no “gold standard” with which to verify the ability of independent variables to identify sepsis. Therefore, it was necessary to apply certain epidemiology techniques. We first relied used the concept of face validity (i.e., each element chosen had to make clinical sense to the practitioners taking care of septic patients) to justify the outcome variables selected. We equally relied upon predictive validity to choose independent variables that best identified our chosen outcomes.
Interrogation of large databases required us to make several key decisions. First, we needed to determine how to identify patients with suspected infection. We assumed that true suspicion that infection was present would lead a provider to send cultures and start antibiotics.
The next assumption was that infected patients with sepsis would be more likely to have a poor outcome. Multiple criteria for poor outcome among infected patients were examined in the electronic health records derived from five databases detailing more than 7 million patient encounters. It was ultimately decided to focus on either death or an intensive care unit (ICU) stay of 3 or more days. Receiver-operator curve (ROC) characteristics were applied to clinical criteria currently in use, including SIRS, the sequential organ failure severity (SOFA) score (21), and the logistic organ dysfunction score (LODS) (22). These studies demonstrated that in patients with suspected infection, a SOFA score of ≥2 identified patients with poor outcome with reliability exceeding that of 2/4 SIRS criteria, especially among patients in the ICU who were presumed to have new infection. The SOFA score also had prognostic reliability similar to LODS despite having fewer variables and being easier to calculate.
Logistic regression and ROC analysis were also applied to the 21 variables listed in Table 1 of the sepsis-2 definition (5). This technique identified a new index—quick SOFA (qSOFA)—comprised of a systolic blood pressure less than 100 mm Hg, a respiratory rate of greater than 22 breaths/min and an altered mental status. When applied to patients in the Emergency Department or on the hospital wards, having two-thirds qSOFA points was nearly as effective as the full SOFA score in identifying patients with suspected infection who would have poor outcomes (although qSOFA performed less well in ICU patients). Importantly, two-thirds qSOFA points were also more accurate in predicting poor outcome than 2/4 SIRS criteria regardless of patient location. This difference is especially notable because it is easier to calculate qSOFA than SIRS since the latter requires a blood test whereas the former does not.
DEFINITION OF SEPTIC SHOCK
The discussion surrounding the revised definition of septic shock was in some ways more challenging than the one addressing sepsis, because some members of the task force questioned the notion that septic shock needed to be distinguished from sepsis. Sepsis-1 and sepsis-2 distinguished septic shock from severe sepsis exclusively by the presence of hemodynamic compromise that could not be restored with fluid resuscitation. Organ dysfunction is a fundamental concept in the revised definition of sepsis. It could thus be reasoned that cardiovascular dysfunction does not differ from any other type of organ dysfunction, and therefore does not need to be designated by a specific term. Therefore, there would be no reason for a unique term to distinguish septic shock from sepsis with other types of organ dysfunction, especially since treatment of sepsis and septic shock do not differ (23). Alternatively, other participants opined that septic shock is indeed a separate entity but that the distinction from sepsis involves more than cardiovascular dysfunction. Ultimately, this latter view prevailed and it was decided that the definition of septic shock should include not only cardiovascular dysfunction but also cellular and metabolic abnormalities. However, cardiovascular, cellular, and metabolic dysfunction can also occur in sepsis, so something additional was needed to distinguish sepsis from septic shock. It was agreed that mortality in septic shock exceeds mortality in sepsis and that this difference in outcome is sufficiently different intellectually to warrant a different definition. In simple terms, if sepsis is “bad infection,” septic shock is “bad sepsis” or “really, really, really bad infection.” Combining these concepts led to the new definition of septic shock as “a subset of sepsis where particularly profound circulatory, cellular, and metabolic abnormalities are associated with a greater risk of mortality than with sepsis alone.”
CLINICAL CRITERIA FOR SEPTIC SHOCK
The identification and verification of clinical criteria for septic shock involved a process that differed slightly from the one used in sepsis. No comprehensive list of potential signs, symptoms, or laboratory/radiologic abnormalities, such as Table 1 from the sepsis-2, exists for septic shock (5). Therefore, clinical criteria to be tested were identified by the task force members using a modified Delphi approach. Three parameters were identified—hypotension despite adequate fluid resuscitation, an ongoing vasopressor requirement, and elevated serum lactate levels. These three parameters were then applied to nearly 30,000 patient records in the Surviving Sepsis Campaign database and subsequently validated in over 3 million patient records in two additional large databases.
Hypotension was defined as a mean arterial blood pressure <65 mm Hg. Although multiple different thresholds of elevated lactate were tested, a lactate level greater than 2 mmol/L was found to have prognostic significance. Note that this data-driven level is lower than the lactate level of 4 mmol/L used in the 6-h Surviving Sepsis bundle for reassessing volume status and tissue perfusion. Ultimately, a combination of three binary possibilities (MAP <65 yes/no, vasopressors yes/no, lactate >2 mmol/L yes/no) placed patients into one of six combinations. Mortality in the group containing all three factors (hypotension, vasopressor support, elevated lactate) was 42%—significantly higher than that in all the other groups and in the database as a whole. Based on the markedly decreased survival in this group, the task force determined that the clinical criteria for septic shock required the combination of hypotension despite adequate fluid resuscitation combined with a lactate level >2 mmol/L. It should be noted that these clinical criteria represent the majority view of the task force; however, this was not a unanimous decision.
THE VALUE OF THE NEW DEFINITIONS OF SEPSIS TO BASIC AND TRANSLATIONAL RESEARCHERS
Defining sepsis as “life threatening organ dysfunction caused by a dysregulated host response to infection” provides a refined focus to a number of areas currently being investigated and identifies potential new investigative targets. To illuminate some of the challenges and opportunities of the definition, it is appropriate to consider each of its component parts.
On the surface, “life threatening” is relatively straightforward to identify in the broadest terms. However, while multiple scoring systems are beneficial on a population basis, individual prognostication of patient outcome is challenging. Studies aimed at improving prediction of individual patients are assuredly needed. Further, determining which elements feed into that prognosis (organism, comorbidities, genetics, age, etc.) will be necessary to bring the concept of precision medicine to sepsis.
Determination of what constitutes organ dysfunction has long been a research goal not only in sepsis but also for other disorders that result in critical illness. Our metrics for determining organ dysfunction are currently based upon abnormalities that become apparent by using simple bedside monitoring or laboratory tests that identify gross perturbations of organ function. In the future, research should seek to identify and refine the elements of organ dysfunction. While the SOFA score has served the clinical critical care community well, a more accurate, more mechanistic score would be a welcome advance for sepsis-4. More importantly, it is incumbent upon the basic/translational research community to extend our understanding of organ dysfunction to the cellular and subcellular levels. For example, one important aspect would be to seek an abnormality that can occur in a variety of cell types that, if present, can lead to malfunction of cells in multiple organs. While the abnormality itself might be the same in cells of all types, the actual manifestation of the defect would reflect the specialized function of the cell/organ in question. Extending this work to the clinical realm highlights the imperative to identify biomarkers capable of clinically delineating sepsis-specific abnormalities in different organ systems. Such markers need to be firmly rooted in an enhanced understanding of the underlying pathobiology of sepsis.
The concept of a “dysregulated host response” is poorly understood and, as such, presents the greatest number of research opportunities. On a practical level, one of the most important clinical conundrums facing today's clinicians is an inability to distinguish between sepsis and infection. Differentiating a “regulated” from a “dysregulated” response might provide the insight required to solve this problem. Most investigations to date have focused on the immunological aspects of sepsis. While our understanding has advanced dramatically over the last 20 years, it is clear that far more insight into the immune response to infection is required. However, we believe that an understanding of the dysregulated host response requires attention to more than just the immune system. Multiple studies have focused on the gut and the liver as the “motor” of MODS, suggesting the need for a better understanding of the regulatory events initiated by these systems (24, 25). Additionally, while many think of vital organs as targets in sepsis rather than drivers, additional understanding of how they potentially can be drivers of a dysregulated host response is indicated, as well as how interorgan crosstalk might be critical in driving and/or propagating the host response. Intriguingly, recent studies have also demonstrated that endocrine and neural abnormalities may be primary manifestations and even the initiators of sepsis (18, 19). Understanding these systems—which interface with every part of the overall host response—may yield novel paradigms of how sepsis induces dysfunction in nearly every component of the body. Opportunities for understanding the pathophysiology of sepsis on a subcellular level—genomic, transcriptomic, proteomic, metabalomic, bioenergetic, molecular, etc.—also represent a rapidly evolving frontier in sepsis research.
The final component of the new sepsis definition is “infection.” The task force felt that it was not appropriately constituted to approach a redefinition of infection. However, the very inclusion of the term in the new definition of sepsis indicates that advances leading to earlier, simpler, and more accurate identification of infection are essential to improve the care of septic patients and thus represent an important goal of sepsis research. In addition, we currently distinguish sepsis from other forms of noninfectious critical illness (trauma, burns, pancreatitis). However, all the elements of the sepsis definition other than “infection”—“life threatening,” “organ dysfunction,” “dysregulated host response”—also characterize the clinical presentation of these other disorders (26–28). It is possible that something uniquely distinguishes the response to infection (insults initiated by pattern recognition receptors) from the response to noninfectious insults (insults initiated by damage recognition receptors)—but perhaps there is not. Thus, it may well be that what we refer to as “sepsis” arises in response to stimuli other than infection. Comparison of the biology induced by different insults will be required to evaluate this further.
THE VALUE OF THE NEW DEFINITIONS OF SEPTIC SHOCK TO BASIC AND TRANSLATIONAL RESEARCHERS
Septic shock presents a different set of research challenges than sepsis. These issues are reflected in the discussions surrounding the derivation of the definitions. Inherent in the problem was difficulty determining why septic shock required a specific, unique definition that differed from the definition of sepsis. The distinction has clear epidemiologic and prognostic significance because the conditions are associated with different mortality rates. However, it is not clear whether septic shock represents a more severe form of the same disease process or if the entity is something that is biologically distinct (or at least partially unique). This issue represents an important question for the future.
However, if one concludes that sepsis and septic shock are at least in part disparate entities, there must be a biological rationale that is ultimately reflected in distinctly different clinical trajectories and outcomes. Viewing septic shock as “sepsis where the dysfunctional organ is the cardiovascular system”—as in sepsis-1 and sepsis-2—may limit our understanding of this entity. Because the need for vasopressors despite adequate fluid resuscitation has historically defined septic shock, it has been appropriate for a basic or translational investigator to focus on the primacy of the cardiovascular system in driving the process.
The alternative view is that septic shock involves a global cellular or metabolic defect that differs from that of sepsis alone and that is not confined to the cardiovascular system. There is a great deal of evidence that both global cellular processes and metabolism are profoundly altered in both sepsis and septic shock. However, it has not yet been possible to identify a change that is characteristic of septic shock but not of sepsis, if such a difference truly exists. Certainly, attempting to elucidate a fundamental defect for either sepsis or septic shock is a valid and enticing goal for any investigator, whether or not that defect is the same in both disorders.
THE VALUE OF THE NEW CLINICAL CRITERIA OF SEPSIS AND SEPTIC SHOCK TO BASIC AND TRANSLATIONAL RESEARCHERS
The new clinical criteria also provide wide-ranging opportunities for investigators. First and foremost, the new criteria for both sepsis and septic shock require further validation in additional large datasets. Four of the five datasets used to derived and validate qSOFA and to validate SOFA ≥2 were compiled using data from health systems in the United States. Data from regions outside the United States with highly developed health systems should be more thoroughly examined. It will also be especially important to validate the new criteria in under-resourced regions because none of the sepsis-3 clinical criteria have been examined in these areas. The criteria also need to be tested or derived independently in children. Further, the clinical criteria need to be validated in animal models. There is a long-standing debate on how accurately small and large animal models of sepsis replicate the human condition (29). Determining the value of the elements in qSOFA, SOFA, and the variables outlined in sepsis-2 in animal models would also be of significant interest to researchers who use these models to develop new insights into human sepsis. Looking forward, the development and validation of biomarkers that could be used to direct therapy in an individualized fashion is a critical frontier in transitioning sepsis care from “one size fits all” to precision medicine.
The clinical criteria for septic shock also require validation. These variables were selected because they have face validity for both cardiovascular and cellular/metabolic abnormalities and sufficient predictive validity to identify patients with a significantly higher mortality rate. The presence of hypotension and/or an ongoing need for vasopressors to maintain blood pressure and serum lactate concentration indeed identified a patient cohort with poor survival. However, the choice of these variables reflects expert opinion only. Therefore, researchers might want to test the predictive validity of additional parameters. Just what constitutes “hypotension” should also be investigated, especially because this parameter might vary between patients based upon basal blood pressure and other, currently unmeasured, factors. In addition, while lactate is easy to measure and provides a crude assessment of global tissue perfusion, it is certainly possible that there are more specific markers for cellular and metabolic abnormalities. Finally, as with the clinical criteria for sepsis, it is pragmatically important to determine if the new clinical criteria for septic shock can serve as entry criteria into clinical trials.
Both the new definitions and clinical criteria of sepsis and septic shock open up a number of important avenues for investigation (Table 1). Undoubtedly, others will emerge. We believe that it is particularly incumbent upon basic and translational investigators to provide additional information regarding the “undefined” components of the definitions—for example, “organ dysfunction” or “a dysregulated host response” and to identify biomarkers that can improve upon the predictive validity of the clinical criteria detailed in sepsis-3. Ideally, the insights gleaned from the research community will move the field forward in a way that ultimately benefits patients and will prove invaluable in helping shape sepsis-4.
The authors are indebted to the other members of the SCCM/ESICM Task Force for Redefinition of Sepsis and Septic Shock: Mervyn Singer, MD, Christopher W. Seymour, MD, MSc, Manu Shankar-Hari, MSc, MD, FFICM, Djillali Annane, MD, PhD, Michael Bauer, MD, Rinaldo Bellomo, MD, Gordon R. Bernard, MD, Jean-Daniel Chiche, MD, PhD, Richard S. Hotchkiss, MD, Mitchell M. Levy, MD, John C. Marshall, MD, Greg S. Martin, MD, MSc, Steven M. Opal, MD, Gordon D. Rubenfeld, MD, MS, Tom van der Poll, MD, PhD, Jena-Louis Vincent, MD, PhD, and Derek C. Angus, MD, MPH.
1. Singer M, Deutschman CS, Seymour CW, Shankar-Hari M, Annane D, Bauer M, Bellomo R, Bernard GR, Chiche J-D, Coopersmith CM, et al. The third international consensus definitions for sepsis
and septic shock
2016; 315 8:801–810.
2. Seymour C, Liu VX, Iwashyna TJ, Brunkhorst FM, Rea TD, Scherag A, Rubenfeld G, Kahn JM, Shankar-Hari M, Singer M, et al. Assessment of clinical criteria
: for the third international consensus definitions for sepsis
and septic shock
2016; 315 8:762–774.
3. Shankar-Hari M, Phillips GS, Levy ML, Seymour CW, Liu VX, Deutschman CS, Angus DC, Rubenfeld GD, Singer M. Developing a new definition
and assessing new clinical criteria
for septic shock
: for the third international consensus definitions for sepsis
and septic shock
2016; 315 315:775–787.
4. Bone RC, Balk RA, Cerra FB, Dellinger RP, Fein AM, Knaus WA, Schein RMH, Sibbald WJ. American College of Chest Physicians/Society of Critical Care Medicine Consensus Conference: definitions for sepsis
and organ failure and guidelines for the use of innovative therapies in sepsis
. Crit Care Med
5. Levy MM, Fink MP, Marshall JC, Abraham E, Angus D, Cook D. SCCM/ESICM/ACCP/ATS/SIS International Sepsis
Definitions Conference. Crit Care Med
6. Bone RC. The pathogenesis of sepsis
. Ann Intern Med
7. Dinarello CA. The proinflammatory cytokines interleukin-1 and tumor necrosis factor and treatment of the septic shock
syndrome. J Infect Dis
8. Damas P, Ledoux D, Nys M, Vrindts Y, De Groote D, Franchimont P, Lamy M. Cytokine serum levels during severe sepsis
in humans: IL-6 as a marker of severity. Ann Surg
9. Dofferhoff AS, Bom VJJ, de Vries-Hospers HG, van Ingen J, van der Meer J, Hazenberg BPC, Mulder POM, Weits J. Patterns of cytokines, plasma endotoxin, plasminogen activator inhibitor and acute phase proteins in severe sepsis
. Crit Care Med
10. Fisher CJ, Opal SM, Dhainaut JF, Stephens S, Zimmerman JL, Nightengale P, Harris SJ, Schein RMH, Panacek EA, Vincent JL, et al. Influence of an anti-tumor necrosis factor monoclonal antibody on cytokine levels in patients with sepsis
. Crit Care Med
11. Pinsky MR, Vincent JL, Deviere J, Alegre M, Kahn RJ, Dupont E. Serum cytokine levels in human septic shock
: relation to multiple system organ failure and mortality. Chest
12. Beal AL, Cerra FB. Multiple organ failure syndrome in the 1990 s: systemic inflammatory response and organ dysfunction. JAMA
13. Hotchkiss RS, Monneret G, Payen D. Sepsis
-induced immunosuppression: from cellular dysfunction to immunotherapy. Nat Rev Immunol
2013; 13 12:862–874.
14. van der Poll T, Marchant A, Buurman WA, Berman L, Keogh CV, Lazarus DD, Nguyen L, Goldman M, Moldawer LL, Lowry SF. Endogenous IL-10 protects mice from death during septic peritonitis. J Immunol
15. Hotchkiss RS, Swanson PE, Freeman BD, Tinsley KW, Cobb JP, Matuschak GM. Apoptotic cell death in patients with sepsis
, shock, and multiple organ dysfunction. Crit Care Med
16. Boomer JS, To K, Chang KC, Takasu O, Osborne DF, Walton AH, Bricker TL, Jarman SD 2nd, Kreisel D, Krupnick AS, et al. Immunosuppression in patients who die of sepsis
and multiple organ failure. JAMA
2011; 306 23:2594–2605.
17. Hotchkiss RS, Sherwood ER. Immunology. Getting sepsis
therapy right. Science
18. Vanhorebeek I, Van den Berghe G. The neuroendocrine response to critical illness is a dynamic process. Crit Care Clin
19. Deutschman CS, Raj NR, McGuire EO, Kelz MB. Orexinergic activity modulates altered vital signs and pituitary hormone secretion in experimental sepsis
. Crit Care Med
20. Pavlov VA, Tracey KJ. The vagus nerve and the inflammatory reflex—linking immunity and metabolism. Nat Rev Endocrinol
2012; 8 12:743–754.
21. Vincent JL, de Mendonça A, Cantraine F, Moreno R, Takala J, Suter PM. Use of the SOFA score to assess the incidence of organ dysfunction/failure in intensive care units: results of a multicenter, prospective study. Working group on “sepsis
-related problems” of the European Society of Intensive Care Medicine. Crit Care Med
22. Le Gall JR, Klar J, Lameshow S, Saulnier F, Alberti C, Artigas A, Teres D. The logistic organ dusfynction system. A new way to assess organ dysfunction in the intensive care unit. ICU Scoring Group. JAMA
1996; 276 10:802–810.
23. Dellinger RP, Levy MM, Rhodes A, Annane D, Gerlach H, Opal SM, Sevransky JE, Sprung CL, Douglas IS, Jaeschke R, et al. Surviving Sepsis
Campaign: international guidelines for management of severe sepsis
and septic shock
. Crit Care Med
2013; 41 2:580–637.
24. Klingensmith NJ, Coopersmith CM. The gut as the motor of multiple organ dysfunction in critical illness. Crit Care Clin
2016; 2 2:203–212.
25. Mittal R, Coopersmith CM. Redefining the gut as the motor of critical illness. Trends Molec Med
26. Shanmugam MK, Bhatia M. The role of pro-inflammatory molecules and pharmacolical agents in acute pancreatitis and sepsis
. Inflamm Allergy Drug Targets
2010; 9 1:20–31.
27. Pierce A, Pitet JF. Inflammatory response to trauma; implications for coagulation and resuscitation. Current Opin Anaestheiol
2014; 27 2:246–252.
28. Giannoudis PV. Current concepts of the inflammatory response after major trauma: an update. Injury
2003; 34 6:397–404.
29. Efron PA, Mohr AM, Moore FA, Moldawer LL. The future of murine sepsis
and trauma research
models. J Leukoc Biol
2015; 98 6:945–952.