Elucidating the Spectrum of Disease Severity Encompassed by Sepsis

The term sepsis commonly evokes a critically ill patient with multiorgan failure who requires immediate and aggressive lifesaving care. In the Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3), sepsis is defined conceptually as “life-threatening organ dysfunction caused by a dysregulated host response to infection.”^1(p801) The Sepsis-3 task force recommended operationalizing this definition as suspected infection and a concurrent increase in the Sequential Organ Failure Assessment (SOFA) score by 2 or more points. A growing body of work is making it increasingly clear, however, that these clinical criteria identify many patients with mild infections that respond quickly to treatment in addition to individuals with life-threatening critical illness.²

Peltan and colleagues³ reported the prevalence, characteristics, and outcomes of adult patients with sepsis who were discharged alive from the emergency departments (EDs) of 4 hospitals in Utah. The investigators used detailed electronic health record data to identify 12 333 ED patients with suspected infection (defined as the collection of a clinical culture and administration of intravenous antibiotics) and organ dysfunction (defined as an increase in SOFA score by 2 or more points higher than baseline), in accordance with Sepsis-3.¹ Peltan and colleagues³ found that 16.1% of patients who met the sepsis criteria were discharged alive from the ED.

Similar observations have previously been reported. A national study of ED visits between 2009 and 2011 that defined sepsis using diagnosis codes and/or clinical criteria reported that approximately 20% of patients with sepsis did not require inpatient admission.⁴ Another study reported that, among 67 733 patients with sepsis diagnosis who were admitted to 110 US hospitals, 1 in 10 were discharged alive within 3 days.⁵

The analysis by Peltan and colleagues³ expands on these studies by providing the clinical characteristics and 30-day outcomes of patients with sepsis who were discharged from the ED. These patients tended to be younger and less severely ill compared with those who were admitted to the hospital, with fewer perturbations in their vital signs, milder organ dysfunction, and higher rates of arrival by private vehicle over ambulance. Two-thirds had urinary infections, whereas those who were admitted to the hospital more commonly had pulmonary or intra-abdominal infections. The 30-day mortality rate in patients who were discharged was only 0.9% compared with 8.3% in those who were hospitalized. Although both crude and risk-adjusted 30-day mortality rates were lower in patients with sepsis who were discharged from the ED vs admitted to the hospital, Peltan and colleagues³ recognized the massive potential for residual confounding and were careful not to overstate these findings as support for routine outpatient sepsis management. Rather, the authors concluded that most ED clinicians synthesize the entire constellation of objective and subjective clinical data to appropriately triage patients who are at low risk for poor outcomes to outpatient follow-up care.

These observations can be interpreted in 1 of 2 ways. They can be seen as evidence that we should broaden our mental models of sepsis to include mild transient illnesses in addition to severe infections that lead to critical illness. Alternatively, they can prompt us to reconsider how best to operationalize the Sepsis-3 conceptual definition.

The first interpretation has major implications for epidemiologic surveillance given that most previous studies have described sepsis incidence, characteristics, and outcomes using either hospital discharge diagnosis codes or electronic clinical criteria that are generated exclusively during inpatient care. For example, the clinical surveillance definition used in a 2017 Centers for Disease Control and Prevention–sponsored study that generated annual estimates of sepsis incidence and mortality in the US (1.7 million adult sepsis cases, and 270 000 associated deaths) required new organ dysfunction combined with blood culture orders and at least 4 days of antibiotic treatment in the hospital (with <4 days allowed if patients died, were discharged to hospice, or were transferred to another hospital).⁶ Clearly, if many patients with sepsis are discharged from the ED or hospital after a short stay, these numbers substantially underestimate the true burden of sepsis in the US.

Before updating the epidemiologic estimates, however, it is worth taking a close look at the sepsis definition applied by Peltan et al³ to ascertain whether there were any limitations that may have inflated their estimates of patients with sepsis discharged from the ED. To the authors’ great credit, they meticulously applied Sepsis-3 clinical criteria to their cohort by conducting medical record reviews to verify ED clinicians’ suspicion of infection, manually resolving missing data, and imputing baseline SOFA scores using pre-encounter data when available. To our knowledge, this approach is among the most rigorous to date among studies that have electronically applied Sepsis-3 criteria.

Nonetheless, there are inherent limitations to any attempt to operationalize Sepsis-3 criteria. First, assuming that patients without pre-encounter data have baseline SOFA scores of 0 will inevitably lead to misclassifying some chronic organ dysfunction as new and, therefore, referring to some simple infections as sepsis. Second, the proxies for suspected infection (clinical cultures and intravenous antibiotics) do not necessarily equate to true infection; in one study, approximately one-third of ED patients who were treated with intravenous antibiotics did not have bacterial infections.⁷ Third, there is no way to ascertain electronically whether infection (if truly present) is associated with organ dysfunction (if it is new); even if the organ dysfunction is attributable to infection, it is difficult to identify whether the infection is associated with a dysregulated host response, direct invasion of the affected organ by the infecting pathogen (eg, pneumonia and hypoxemia), or a concurrent condition (eg, mucous plugging leading to atelectasis). Consider, for example, a young woman who presents with pyelonephritis that is complicated by mild hypotension and acute kidney injury that reverse quickly after intravenous fluids, the prototypical patient with sepsis who is discharged from the ED. Does this patient have organ dysfunction from a dysregulated immune response, or does she have volume depletion from several days of vomiting and poor oral intake?

These issues beg the question of why we have the term sepsis at all instead of simply calling all such events infections. In our opinion, the most important reason to call an infection sepsis is to emphasize the severity of the condition and trigger immediate aggressive care. We do not know the extent to which this reasoning applies to the 16.1% of patients with sepsis who were discharged from the ED in the study by Peltan and colleagues.³ Did these patients do well because of immediate aggressive care or because they had mild presentations to begin with? If the answer is the latter, then 1 way to better calibrate clinical criteria to match the conceptual definition of sepsis would be to require more severe organ failure or evidence of multiple organ failures. In practice, this change could mean increasing the threshold value for sepsis to a SOFA score of more than 2 or requiring some degree of sustained organ dysfunction despite initial fluids or other measures.

Ultimately, the investigation by Peltan and colleagues³ is an important contribution to the growing literature that elucidates the breadth of illness severities encompassed by current sepsis clinical criteria. It shows that the current operationalization of Sepsis-3 criteria includes mild presentations that respond quickly to ED care and can safely be discharged from the ED. Findings from this study should prompt further research and deliberation on how best to define sepsis to better distinguish patients who require immediate aggressive care from patients who can safely tolerate a more tempered approach.

Published: February 10, 2022. doi:10.1001/jamanetworkopen.2021.47888

Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2022 Rhee C et al. JAMA Network Open.

Corresponding Author: Chanu Rhee, MD, MPH, Department of Population Medicine, Harvard Medical School and Harvard Pilgrim Health Care Institute, 401 Park Dr, Ste 401, Boston, MA 02215 (crhee@bwh.harvard.edu).

Conflict of Interest Disclosures: Dr Rhee reported receiving personal fees from UpToDate, grants from the Centers for Disease Control and Prevention (CDC), grants from Agency for Healthcare Research and Quality (AHRQ), and personal fees from Pfizer outside the submitted work. Dr Klompas reported receiving grants from the CDC, AHRQ, and Massachusetts Department of Public Health and personal fees from UpToDate outside the submitted work.