Detecting Sepsis in an Emergency Department: SIRS vs. qSOFA

Abstract

Sepsis is a condition that can progress to serious illness and even death. The diagnosis of sepsis is difficult because no unique biomarker exists. With this, health care providers must rely on clinical diagnostic criteria to guide diagnosis. Systemic Inflammatory Response Syndrome (SIRS) criteria have been used for diagnosis since 1992. The more recent attempt to replace SIRS with the quick Sequential Organ Failure Assessment (qSOFA) for assessment of potentially septic patients is troublesome. The qSOFA was designed as a prognostic and not diagnostic tool. Using established processes of evidence-based medicine, it is shown herein that qSOFA fails to meet the definition of a diagnostic assessment tool. Thus, the SIRS assessment should remain the gold standard tool for detecting patients at risk of “sepsis.”

Introduction

Sepsis can lead to an overwhelming bodily response to infection, resulting in tissue damage, organ failure, and death.¹ Approximately 87% of sepsis cases originate outside of the hospital,¹ making the Emergency Department (ED) a key location for its diagnosis.

Centers for Disease Control data show that each year, about 1.7 million Americans develop sepsis and between 250,000² and 270,000³ of these, die. Approximately 35% of in-hospital deaths are due to sepsis.⁴ Sepsis is also the leading cause of unplanned readmissions to the hospital, with 19 percent of patients hospitalized with sepsis requiring re-hospitalization within 30 days after discharge.⁵

Early, accurate detection and treatment of sepsis is a high-stakes matter. Mortality due to sepsis increases by as much as 8% for every hour of treatment delay. Experts believe that up to 80% of sepsis deaths could be prevented with more rapid diagnosis and treatment.⁶

This manuscript will examine recently-emerged concepts regarding the timely diagnosis and treatment of sepsis using an Evidence-Based Medicine (EBM)-informed perspective. It is the purpose of this paper to inform readers why ED physicians should NOT abandon the use of SIRS criteria for the detection of septic patients in an ED.

Toward Defining and Reporting of Sepsis and Sepsis Outcomes

The concept of “sepsis” was first clearly defined in 1992 by Bone et al.⁷, reporting on the proceedings of an August 1991 consensus conference involving representatives of the American College of Chest Physicians (ACCP) and the Society of Critical Care Medicine (SCCM). Its purpose was to derive agreement upon “…a set of definitions that could be applied to patients with sepsis and its sequelae.”

Sepsis was newly defined as the presumed presence of infection plus at least two of the four possible “Systemic Inflammatory Response Syndrome” or “SIRS” Criteria (Table 1). The panel agreed that clinical signs and symptoms consistent with severe sepsis and septic shock could also be observed in other critically-ill patients with pro-inflammatory conditions such as trauma, severe burns and pancreatitis.⁷ Bone et al. graphically displayed this complexity of differentiating various clinical states in their original manuscript (Figure 1).⁷ However, these diagnostic criteria for sepsis, while lacking specificity, are highly sensitive.

Table 1.

Definitions of Sepsis and Systemic Inflammatory Response Syndrome (SIRS) as Initially Framed by Bone et al.⁷

“Bacteremia” = the presence of viable bacteria in the blood. “Sepsis” = Presumed presence of infection plus at least two positive Systemic Inflammatory Response Syndrome (“SIRS”) criteria “SIRS Criteria” include: ○ A temperature greater than 38°C or less than 36°C ○ An elevated heart rate greater than 90 beats per minute ○ Tachypnea, manifested by a respiratory rate greater than 20 breaths per minute or hyperventilation, as indicated by a PaC02 of less than 32 mm Hg ○ A White Blood Cell (WBC) count greater than 12,OOO/cu mm, a WBC count less than 4,OOO/cu mm; or the presence of more than 10 percent immature neutrophils in the qperipheral smear. “Severe sepsis” = sepsis associated with organ dysfunction, hypoperfusion abnormality, or sepsis-induced hypotension. ○ Hypoperfusion abnormalities include lactic acidosis, oliguria, and acute alteration of mental status. ○ Sepsis-induced hypotension is defined by the presence of microorganisms or the invasion of normally sterile host tissue by those organisms. “Septic shock” = sepsis-induced with hypotension despite adequate fluid resuscitation along with the presence of perfusion abnormalities that may include, but are not limited to, lactic acidosis, oliguria, or an acute alteration in mental status. ○ Patients who are receiving inotropic or vasopressor agents may not have persistent hypotension at the time that perfusion abnormalities are measured.

Figure 1.

The Intersection of Variable Disease States⁷

In 2003⁸ and again in 2012,⁹ new consensus conference updates of understandings of sepsis were reported. In both, the ACCP and the SCCM were involved, augmented by European expertise. SIRS criteria were retained when defining “sepsis” in these updates.

Why is Sepsis Such a Clinical Challenge?

Sepsis is a disease with a biomarker (lactate) suggestive of host physiologic distress. However, that biomarker is not unique to the processes of sepsis. For example, a patient with abdominal pain due to mesenteric ischemia is likely to manifest a lactic acidosis, despite not yet having bacteremia.

Sepsis has myriad presentations, depending in part upon the organ system(s) principally impacted. Again, Bone et al. specifically recognized this in the original definition of “sepsis” (Figure 1).⁷ Thus, due to variance in initial presentations, “sepsis” can be difficult to recognize early in its course, before its sequelae become clinically overt.

Liability Concerns, Explained

Retrospective reviews of cases can lead clinicians, laypersons, and juries to fault clinicians who did not act earlier and more definitively, for patients who progress from minor symptoms to sepsis, and then ultimately to death by septic shock. For example, Levy reports a case of a 12-year-old boy, who presented to an ED for treatment of vomiting and a cut on his arm, sustained while playing basketball. He was discharged from an ED with a “…diagnosis of an upset stomach and dehydration.” He died three days later from septic shock.¹⁰ In Minneapolis, a woman presented to the ED with infectious symptoms four days after giving birth. She was treated and released by an advance practice nurse-practitioner, but died two days later of sepsis-related complications.¹¹ Malpractice was alleged and a $20 million award resulted.

Despite these examples, experienced clinicians know that in many cases, patients who eventually succumb to septic shock do not manifest important clinical clues predictive of death, especially if they initially present to an ED early in their clinical course. As previously noted, sepsis lacks a specific biomarker and a specific singularity of symptoms, obscuring early diagnosis.

The Most Recent Update

Recently, a third revision of the Surviving Sepsis Campaign guidelines was convened. A new proposed definition of “sepsis,” which does NOT include the use of SIRS criteria, was promulgated.¹² Proceedings of this “Third International Consensus Conference on Sepsis” were reported in 2016.

The stated goal of this conference was to update the definition of “sepsis” in light of new scientific information and insights regarding the topic. To update a definition implies an intent to provide an updated means to diagnose sepsis. Directly following from this conference, the qSOFA score (Table 3) was advocated as a tool for clinicians who diagnose and treat sepsis.¹²

Table 3.

The “quick Sequential (Sepsis-related) Organ Failure Assessment (“qSOFA”) Score

Elements of the qSOFA Score ○ Glasgow Coma Score of 13 or less ○ Systolic Blood Pressure of 100 mm Hg or less ○ Respiratory rate of 22/min or more Score 1 point for each positive finding

Seymour et al. reported that the established Intensive Care Unit (ICU) scoring tool called “SOFA” (Sequential Organ Failure Assessment) was superior to the newly-defined “qSOFA” tool, as well as to SIRS, when predicting in-hospital mortality in the ICU.¹² However, they also reported that qSOFA was most predictive of death among patients who had not yet been admitted to the ICU.

They wrote, “Among non-ICU encounters in the validation cohort [n=66,522 with suspected infection, of whom 1,886 (3%) died], qSOFA had a predictive validity [Area under the Receiver Operating Characteristic Curve (AUROC) = 0.81, 95% Confidence Interval (CI) = 0.80–0.82]. This AUROC was greater than for SOFA (AUROC = 0.79, 95% CI 0.78–0.80, p< .001) and greater than for SIRS (AUROC 0.76, 95% CI 0.75–0.77, P<.001).” The “take-away” is that qSOFA proved to be the best predictive tool of mortality when used outside of an ICU.¹²

Seymour et al. also reported that for all patients, “Relative to qSOFA scores lower than 2, encounters with qSOFA scores of 2 or higher had a 3- to 14-fold increase in hospital mortality across baseline risk deciles.”¹² This adds that qSOFA is a useful tool to predict near-term risk of mortality, both in and out of the ICU.

This ability to prognosticate mortality led influential “thought leaders” to quickly advocate for abandoning SIRS criteria in favor of qSOFA as a diagnostic instrument to facilitate early detection of sepsis. For example, the commonly-utilized website MDCalc.com substituted qSOFA for SIRS within one month of the publication of “Sepsis-3.”

How were “Sepsis-3” Patients Studied?

The qSOFA score was derived and validated via a retrospective, cohort study. The authors created a random 50/50 split of a cohort of 148,907 emergency department, hospital ward, or ICU patients from a cohort assembled from 14 University of Pittsburgh Medical Center Hospitals. Half of the patients were randomized into the “derivation” cohort. Of these, 7,836 had data associated with an ICU stay, and 66,617 were studied in an ED or hospital ward. The other half comprised the “validation” cohort. Of these, 7,932 provided data while receiving care in an ICU, and 66,522 were in the ED or a hospital ward.¹²

Was This a Properly-Executed Trial?

The design of the trial, as a study to derive a clinical prognosis rule, was above reproach. Any study designed to derive a clinical decision rule should have a derivation cohort from whom a rule is developed, followed by a validation cohort among whom the new rule is tested. The “Sepsis-3” manuscript met this test.

However, of most importance is that the authors did not derive a diagnostic tool. They derived a prognostic instrument, because the primary outcome being evaluated was mortality. To predict mortality is to be prognostic, not diagnostic. This is a crucial difference.

An Evidence-Based Medicine Perspective

Because the qSOFA clinical decision tool is a prognostic instrument, ED physicians who might use qSOFA alone for diagnostic purposes are demonstrating flawed clinical reasoning. To employ qSOFA as a diagnostic tool in an ED violates two basic principles of Evidence-Based Medicine (EBM).

First, Straus et al., in an authoritative EBM text, state that for a manuscript regarding prognosis to have validity, a defined and representative sample of correctly diagnosed patients must have been assembled at a common point in the course of their disease.¹³ To satisfy this requirement, all patients must have been correctly diagnosed at the inception of the cohort. In other words, within the Sepsis-3 cohort that contributed to the qSOFA prognostic tool, all patients must have had a known and correct diagnosis upon entry into the study.

If the “Sepsis-3” investigators had derived a valid diagnostic tool, the final diagnosis of sepsis for the patients entered into the cohort may have been suspected, but could not have been known at the time of study entry. It follows that application of logic to the basic rules of EBM require that a patient cohort cannot simultaneously have had both a known and an unknown diagnosis. Therefore, simple logic leads to the inescapable conclusion that the qSOFA tool cannot simultaneously exist as both a prognostic and a diagnostic tool.

This explains unequivocally why qSOFA should not be used as a tool for diagnosis. To utilize the prognostic qSOFA score as reported out from the Consensus Conference for diagnostic purposes would violate basic principles of logic and would represent “circular reasoning”.

A second reason also exists to show why qSOFA has not been validated for diagnostic use in any ED. For manuscripts regarding diagnosis, the diagnostic test must have been evaluated in an appropriate spectrum of patients, a group of patients similar to those in whom this tool would be used in practice. The qSOFA score was not derived exclusively from study of a set of ED patients. Hospitalized ICU and non-ICU patients also contributed to the Sepsis-3 cohort. Because of the dissimilarity between the Sepsis-3 cohort and an undifferentiated cohort of ED patients, qSOFA fails on a second count as a potential tool for emergency physicians to use toward diagnosis of sepsis.

An understanding of these crucial principles of EBM would have precluded anyone from advocating for use of qSOFA as a diagnostic tool. Thus, unnecessary confusion could have been avoided. The “Sepsis-3” group began with the stated mission of deriving a new diagnostic tool that would incorporate new understandings of septic processes. However, the group fulfilled a different mission. They produced a prognostic tool that can signal that a patient is at high risk of clinical deterioration, not a diagnostic tool.

Today, MDCalc.com presents a broader view, offering links both to qSOFA scoring and a restored linkage to a “SIRS” calculator.^{14, 15}

Which Doctors Value Which Score, and Why

Tasks and priorities of emergency physicians and critical care specialists differ. An understanding of how their tasks differ helps explain why emergency physicians will naturally find SIRS more useful, while the qSOFA score may be favored by critical care physicians.

Emergency physicians attempt to make a correct diagnosis while not overlooking potentially lethal diagnoses during a very brief portion of the patient’s entire illness. Their worst error would be to prematurely discharge a patient who will soon deteriorate. Whether treating adults or children, emergency physicians conceptually “sort” patients into “bins”; the ICU versus non-ICU with telemetry versus “floor” versus “discharge to home”. This is based upon the apparent diagnosis and severity of patient illness.

The critical care physicians’ tasks differ. These physicians must adjudicate appropriateness of accepting a patient into an ICU, a scarce hospital resource. Critical care physicians then provide longitudinal management of critically-ill patients whom they accept for care.

Emergency physicians will naturally place greatest value on a sensitive tool that provides an early signal to diagnose and initiate treatment of possible sepsis. Emergency physicians initially need a diagnostic tool with few false negative results. Later, a tool with high ability to detect those who might deteriorate if sent home prematurely becomes of value to them, as they consider whom to discharge. Once a patient has been treated and released, there is no “rapid response team” to re-assess the patient and upgrade them to an ICU.

It may surprise some that emergency physicians, as understood by using SIRS criteria toward defining sepsis, have long treated and released some patients who are “septic”. If this is surprising, consider the febrile tachycardic patient with acute pyelonephritis or community acquired pneumonia who appears sufficiently “stable” to be discharged home. Some patients with these diagnoses can be treated with one dose of intravenous antibiotics and then released from an ED with prescriptions. These patients’ qSOFA scores could serve as a useful adjunct to the emergency physician who is considering treating and releasing such a septic patient. A qSOFA score of 2 or 3 might warn the ED physician that discharge is inappropriate.

We are currently studying this hypothesis, among a retrospective cohort of 1067 SIRS positive patients who were treated and released from EDs with a diagnosis of acute pyelonephritis, pneumonia or influenza. All 1067 were documented to still be alive both 7 and 30 days after study entry. Only 9 had a qSOFA score of 2 or 3, and none of these returned to the ED within 7 days of release. Our manuscript will have more details, and is currently under review by an emergency medicine journal.

Critical care specialists will naturally place greatest value on a clinical diagnostic tool that can discern who should be admitted to the scarce resource, the ICU. Thus, the critical care doctor places greatest value on a test that can predict risk of deterioration or death with high specificity.

Unifying these points, it becomes clear and logical why critical care physicians will have greater enthusiasm for qSOFA’s prognostic capabilities and specificity. Emergency physicians, who evaluate patients for only a brief portion of the disease course, should continue to favor SIRS Criteria. SIRS, due to its high sensitivity, is useful toward early diagnosis of possible sepsis in the patients’ clinical course. It is not known whether designers of electronic health record systems considered the points we cite, but we believe they have correctly retained SIRS criteria to enable sending automated alerts that a patient may have sepsis.

Unifying the Data

The saga of the diagnosis of sepsis and the implementation of a sepsis care “bundle”, designed to optimize care of the septic patient, is a story that has evolved. For example, both Drotrecogin (Xigris™) and a special central venous catheter used to be part of the sepsis “bundle”. Drotrecogin was rapidly abandoned after the “PROWESS-SHOCK” study showed it to be an expensive, yet ineffective agent.¹⁶ The use of a central venous catheter to monitor mixed central venous oxygen saturation was abandoned after negative findings from the PrioMISe¹⁷, ARISE¹⁸, and ProCESS¹⁹ trials.

However, ED physicians should not cast aside the use of SIRS Criteria for the diagnosis of sepsis. SIRS remains a useful diagnostic tool, unique for its sensitivity. The qSOFA tool advocated by the Sepsis-3 consensus conference team¹² may be a useful adjunct to guide caution regarding potentially treating and releasing septic patients from an emergency department a patient. However, because it is a valid prognostic instrument, qSOFA cannot be employed as a diagnostic tool.

Conclusion

Sepsis remains a potentially deadly clinical syndrome with no unique and unequivocal biomarker. Sepsis may not yet be clinically apparent and can easily be overlooked early in its course.

SIRS remains as the proper and valuable tool to help detect patients at risk of “sepsis” because of its high sensitivity.

The qSOFA tool has prognostic value. However, no peer-review literature exists validating qSOFA as a diagnostic tool for sepsis.

A qSOFA score of 2 or 3 may provide a signal that to discharge a particular septic patient from the ED may be contraindicated, but this application remains to be validated.

Electronic Health Record systems that retain use of SIRS to trigger an alert for possible sepsis are reasonable and proper, because SIRS (and not qSOFA) has high sensitivity for detecting sepsis.