Sepsis Screening: Current Evidence and Available Tools

Abstract

Background:

Early recognition of sepsis is challenging, especially in the surgical patient. Because of the non-specific nature of the initial signs and symptoms, delays in recognition are all too common. To improve the early identification of sepsis, screening tools have been developed, and several papers have described their results. This article reviews the available sepsis screening tools.

Methods:

A PubMed search was performed using the search terms “sepsis” and “shock,” “electronic alert,” “clinical decision support,” and “early warning systems.” The papers found were reviewed to determine their relevance to the topic of sepsis screening, and outcome data were extracted from appropriate papers.

Results:

Multiple sepsis screening tools were identified with differing performance characteristics. These tools are reviewed individually along with a summary of their sensitivity, specificity, and positive and negative predictive values.

Conclusions:

Clearly, sepsis screening has the potential to improve patient outcomes by aiding clinicians in the early recognition of the condition, enabling early implementation of evidence-based therapies. However, significant challenges remain, including identifying an optimal screening tool. Continued research is needed into the development and integration of automated screening tools that will be effective in a variety of clinical settings.

Sepsis continues to be a significant cause of morbidity and death worldwide. It is estimated that, globally, sepsis accounts for 8 million deaths annually [1]. A June 2016 statistical brief from the Healthcare Cost and Utilization Project reported that hospital expenditures for sepsis have risen dramatically, to $24 billion per year, making sepsis the most expensive health condition to treat. Although sepsis mortality rates have decreased in the past decade, the rate remains high, ranging from 11% to 40%, depending on the severity of the infection [2 –4].

One of the major contributors to the decrease in sepsis-related death has been the early implementation of evidence-based therapies for initial resuscitation and management. Early, targeted therapies such as fluid resuscitation and broad-spectrum antibiotics have had a tremendous impact on patient outcomes [5 –8]. However, implementation of these useful therapies is dependent on the early recognition of sepsis by healthcare providers.

Unfortunately, early recognition of sepsis is challenging, especially in the surgical patient. Because of the non-specific nature of the signs and symptoms of early sepsis, delays in recognition are all too common. In an effort to improve the early identification of sepsis, several screening tools have been developed. With the recent focus on early identification of sepsis, there has been an emergence of literature describing various tools for sepsis screening. This article reviews the available screening tools.

Methods

A PubMed search was performed using the search terms “sepsis” and “shock,” “electronic alert,” “clinical decision support,” and “early warning systems.” Articles were reviewed to determine their relevance to the topic of sepsis screening. Case reports were excluded.

Results

Moore et al. described their initial experience with the implementation and validation of a sepsis screening tool in a surgical intensive care unit (SICU)[9]. The tool is based on the four variables that comprise the systemic inflammatory response syndrome (SIRS). This tool tabulates scores for routinely collected SIRS criteria—heart rate, ventilatory rate, minimum and maximum temperatures, and white blood cell (WBC) count—that are derived from a five-point ordinal scale (0–4) on the basis of increasing deviation from normal values (Fig. 1). The total score ranges from 0 to 17. A score of ≥4 is considered the threshold for a positive early sepsis screen; this score triggers an evaluation by a “second responder,” a clinician who attempts to identify the source of infection. If the patient is found to be positive in both steps, the ICU physician determines whether to implement the sepsis resuscitation bundle, which is done if the clinician suspects that the physiologic derangement is secondary to infection. The SIRS component of the screening tool is applied to all patients admitted to the SICU, where patients are screened twice daily by the bedside ICU nursing staff. In the initial trial, all patients admitted to the SICU over a period of 5 months were screened. A total of 4,991 screens were performed, and the tool yielded a sensitivity of 96.5%, a specificity of 96.7%, a positive predictive value (PPV) of 80.2%, and a negative predictive value (NPV) of 99.5%. The prevalence of sepsis in the cohort was 12.2%. Subsequent to the implementation of this two-step sepsis screening tool and sepsis management protocol, the mortality rate from severe sepsis and septic shock decreased from 35.1% to 24.2%. Following this initial experience, use of the screening tool was expanded to the general surgery floor. On this floor with 9,332 screens, the sensitivity was 99.9%, the specificity 91.3%, the PPV 16.3%, and the NPV 99.9%. Lastly, the sepsis screening protocol was implemented in the trauma population at a different academic hospital with similar results. A total of 5,485 screens in trauma patients yielded a sensitivity of 97.7%, a specificity of 91.8%, a PPV of 52.2%, and an NPV of 99.8%.

FIG. 1.

Sepsis screening tool for use in a surgical intensive care unit.

Because of this very promising initial experience with the tool, the decision was made to automate it to create data for the electronic medical record (EMR). Jones et al. performed this task. The automated tool was studied in several hospital units covering 56,190 screens in 9,718 patients. Between 2006 and 2014, the duration of this study, there were 15,353 sepsis-associated inpatient stays. After implementation of the automated sepsis screening tool, the combined inpatient sepsis-associated death rate decreased from 29.7% to 21.1%, with a $6.5 million annual decrease in inpatient costs [10].

In another follow-up study, Wawrose et al. evaluated the performance characteristics of the St. John's Sepsis Agent (SJSA) developed by Cerner (Kansas City, MO) compared with the Sepsis Screening Score (SSS) developed by Moore et al. as described above. The comparison study occurred in the surgical intermediate care unit (SIMU) at Memorial Hermann Hospital, a tertiary referral hospital in Houston, Texas. The SJSA is continuously running in the background of the unit and evaluates temperature, heart rate, ventilatory rate, serum glucose concentration, and WBC count for a primary screen and serum lactate concentration, systolic blood pressure or mean arterial pressure, serum creatinine, and bilirubin for a secondary screen. If the primary and secondary screens are positive, heparin concentration, SIRS, and signs of end-stage renal disease are determined. A head-to-head comparison of the SSS and SJSA was performed with all patients being screened prospectively. More than 1,700 sepsis screens were completed on 348 patients. The incidence of sepsis was 13.5%. The SJSA had a sensitivity of 44.7%, specificity of 84.7%, PPV of 31.3%, and NPV of 90.7%, whereas the SSS had a sensitivity of 74.5%, specificity of 86.4%, PPV of 46.1%, and NPV of 95.6% [11]. The differences in sensitivity (p < 0.001), PPV (p < 0.001), and NPV (p = 0.011) were statistically significant. The SJSA was implemented by constant surveillance of patients' EMRs, whereas the SSS was performed twice daily by a nurse. Regardless of the continuous, automated screening process, the SJSA still detected fewer truly septic patients than the SSS.

Although the performance characteristics of the SJSA were inferior to those of the non-automated SSS, the appeal of an automated screening tool remains. An automated Early Warning and Response System (EWRS) for sepsis was developed at the University of Pennsylvania Health System by Umscheid et al. This screen utilized the SIRS criteria in addition to signs of organ dysfunction to tabulate a score in adult patients admitted to acute inpatient units. Thresholds were created based on derivation cohorts, and when these measures were surpassed, a rapid response coordinator was notified, and an immediate bedside patient evaluation was performed. The automated EWRS yielded a sensitivity of 16%, specificity if 97%, PPV of 26%, and NPV of 94% [12]. The results of this study showed an increase in early sepsis care and in ICU transfers, as well as a trend toward a decrease in sepsis-related deaths.

The Modified Early Warning Score (MEWS) was developed in the 2000s to detect medical patients at risk of catastrophic deterioration [13]. The MEWS tool evaluates five physiological features: Systolic blood pressure, pulse rate, ventilatory rate, temperature, and the level of consciousness based on the Alert, Voice, Pain, Unresponsive (AVPU) scale. Subbe et al. evaluated the ability of MEWS to identify medical patients at risk and to examine the feasibility of using MEWS as a screening tool to trigger early assessment and admission to a high-dependency unit or ICU. Those investigators found that a score of ≥5 was associated with a higher risk of death (odds ratio [OR] 5.4; 95% confidence interval [CI] 2.8–10.7)[14]. However, the authors did not report whether identification of critically ill patients, selection of a higher level of care, and consecutive changes in management had an effect on outcomes.

Despite the fact that MEWS was never intended as a sepsis screening tool, it has been used for that purpose, with mixed results. In 2015, a systematic review of 18 articles was performed by Roney et al. to evaluate the use of MEWS for sepsis identification and impact on the mortality rate [14]. They concluded that although the MEWS is recommended for assessment of inpatients at risk for clinical deterioration, there is a lack of randomized controlled trials or clinical outcomes data regarding its utility as a screening tool for sepsis. Therefore, the value of the use of MEWS for identification of sepsis is unclear. Multi-center trials are needed for a more reliable assessment.

Although much of the early focus on sepsis screening occurred in the in-patient setting, there is an increasing body of literature on the use of sepsis screening in the emergency department (ED). Today, two thirds of sepsis cases are evaluated initially in the ED. Therefore, developing a robust sepsis screening program for the ED could have a tremendous impact on patient outcomes. Ideally, this screening would occur at triage, but screening in this setting is difficult because of the limited patient information. Another challenge to sepsis screening in the ED is the higher nursing workload and pace of the work flow. These unique challenges make electronic sepsis alert systems particularly appealing for the ED, allowing continuous sepsis screening as additional clinical data become available.

Al Solamy et al. described their results with an electronic alert tool screening for severe sepsis and septic shock in the ED. In their tool, a positive screen consists of either two SIRS criteria and one organ dysfunction or two organ dysfunction criteria. Positive screens trigger an alert on the nursing task list. Subsequently, the nurse notifies the physician via a paging system. Review of this tool a yielded a sensitivity of 93%, specificity of 98%, PPV of 21%, and NPV of 99% [16]. The electronic alert preceded ICU referral by a median of 40.2 h. However, the impact of this tool on patient outcomes was not studied. Furthermore, the use of organ dysfunction criteria in the identification of sepsis may be flawed, as by the time these events occur, the patient has progressed beyond sepsis.

In an attempt to develop a sepsis screening tool that could be implemented at ED triage using only vital signs and non-invasive monitoring, Goerlich et al. created a modified screening tool using the near-infrared spectroscopy-derived tissue hemoglobin saturation (StO₂). The StO₂ monitor measures oxygen saturation by placement of a point-of-care device on the thenar eminence for a few seconds. The StO₂ readings correlate with mixed venous oxygen blood saturation (SvO₂) and central venous oxygen saturation (ScvO₂) values, with lower values indicating the presence of tissue hypoperfusion [16]. The Modified Sepsis Screening Tool includes heart rate, ventilatory rate, and temperature. The StO₂ values were incorporated into this tool to generate a cumulative score, based on % tissue oxygen saturation cutoffs of <75 and >90. A total of 500 patients were screened in the ED, with an 8.4% incidence of sepsis. This approach yielded a sensitivity of 86%, specificity of 79%, PPV of 27%, and NPV of 99% [17].

Discussion

Although there is an increasing body of evidence related to sepsis screening, there is is a lack of randomized trials related to such screening. The majority of the published work has been performed in surgical patients, and the applicability to other populations remains unknown. What is clear is that when an effective sepsis screening tool is used, there is an increase in sepsis recognition, allowing implementation of early evidence-based care. Sepsis screening thus is the initial critical step in the evaluation and management of a patient with sepsis. In order to be effective, sepsis screening must be paired with early, aggressive implementation of evidence-based care, including broad-spectrum antibiotic administration, fluid resuscitation for hypotension, and assessment of tissue hypoperfusion with serum lactate measurements. One of the current barriers to early implementation of evidence-based sepsis therapies is failure of clinicians to recognize the condition. A recent study by Assuncao demonstrated that physicians were able to recognize sepsis only 27% of the time [18]. This failure to recognize sepsis might be overcome by implementing a sepsis screening program.

The optimal sepsis screening tool should have the following characteristics. It should present minimal risk to the patient. It must be affordable, be easy to implement in a variety of hospital settings, and utilize readily available data. The ideal tool will recognize sepsis early, ideally prior to the onset of organ dysfunction. Finally, as a screening tool, it should have a high NPV.

The majority of the screening tools discussed here incorporate some element of the SIRS criteria (see Table 1). These criteria have been notoriously non-specific for identifying sepsis, but by weighting the individual components of the SIRS criteria based on the degree of physiologic derangement, it is possible to use the SIRS criteria for sepsis screening. However, SIRS criteria-based screening may not work in all populations. In the ED, sepsis screening ideally should occur at the time of initial triage. However, WBC counts typically are not known at the time of ED triage and therefore cannot be used for screening. Another potential limitation of some of the screening tools is the inclusion of evidence of organ dysfunction. Waiting for a patient to develop clinical or laboratory evidence of organ dysfunction leaves him or her with sepsis that may progress in severity, even to septic shock, before the alert triggers.

Table 1.

Available Screening Tools

Series (year)	Criteria	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)
Moore et al. (2009) [9]	Modified SIRS criteria (weighted based on degree of physiologic derangement)	96.5	96.7	80.2	99.5
Wawrose et al. (2016) [11]	Temperature, HR, VR, glucose, WBC, SBP, lactate, MAP, creatinine, bilirubin, heparin, SIRS, ESRD	44.7	84.7	31.3	90.7
Umscheid et al. (2015) [12]	SIRS criteria, signs of organ dysfunction	16	97	26	94
Subbe et al. (2001) [13]	SBP, HR, VR, temperature, consciousness based on the Alert, Voice, Pain, Unresponsive (AVPU) scale	NR	NR	NR	NR
Al Solamy et al. (2014) [15]	2 SIRS criteria plus 1 organ dysfunction OR 2 organ dysfunctions	93	98	21	99
Goerlich et al. (2014) [14]	HR, VR, temperature, StO2	86	79	27	99

HR = heart rate; MAP = mean arterial pressure; NR = not reported; PPV = positive predictive value; NPV = negative predictive value; SBP = systolic blood pressure; SIRS = systemic inflammatory response syndrome; VR = ventilatory rate; WBC = white blood cell count.

The 2016 release of the new Sepsis-3 definitions has redefined sepsis as “life-threatening organ dysfunction caused by a dysregulated host response to infection” [19]. With these new definitions, there is an increasing focus on the presence of organ dysfunction as a key discriminator of sepsis vs. infection without physiologic derangement. Assessment of the presence of SIRS criteria has been eliminated from the definitions. The updated sepsis definitions include a more focused assessment of organ dysfunction utilizing the Sequential (Sepsis-related) Organ Failure Assessment (SOFA) score as well as an abbreviated version of the SOFA score called the qSOFA. In addition, the new sepsis definitions have eliminated the term “severe sepsis.”

The qSOFA score has three components: (1) Systolic blood pressure <100 mm Hg; (2) ventilatory rate >22/min; and (3) presence of altered mental status. A person with two or more qSOFA criteria is considered to have a positive score. Sepsis is defined as life-threatening organ dysfunction caused by a dysregulated host response to infection. Organ dysfunction can be identified as an acute change in total SOFA score ≥2 points consequent to the infection or a positive qSOFA score. Septic shock is defined as sepsis plus either a vasopressor requirement or a serum lactate concentration >2 mmol/L.

The SOFA scores were designed originally to predict death in critically ill patients with severe organ dysfunction [20]. The SOFA score and qSOFA score based on the new Sepsis-3 definitions were never intended to be utilized as screening tools for sepsis. There is concern that clinicians will begin utilizing qSOFA as a screening tool for sepsis. In fact, the only component of the new Sepsis-3 definition of sepsis that even selects patients with sepsis is the requirement for the clinician to “suspect infection.” A recent study by Haydar et al. compared the use of SIRS criteria-based screening with qSOFA-based screening in 200 patients with sepsis treated in the ED [21]. In this cohort, 94.5% met the SIRS criteria, whereas in the ED, only 58.3% met the qSOFA criteria. The mean time from arrival to SIRS documentation was 47.1 min (95% CI 36.5–57.8) compared with 84.0 min (95% CI 62.2–105.8) for qSOFA. The median ED “door” to positive SIRS criteria was 12 min and 29 min for qSOFA. Based on these results, the authors concluded that qSOFA performed poorly as a sepsis screening tool.

Conclusion

Clearly, sepsis screening has the potential to improve patient outcomes by aiding clinicians in the early recognition of sepsis, enabling prompt implementation of evidence-based therapies. However, there still are significant challenges to overcome, including identifying an optimal screening tool. Continued research is needed into the development and integration of automated screening tools that will be effective in a variety of clinical settings.

Footnotes

Author Disclosure Statement

No competing financial interests exist for any of the authors.

References

Dugani

, Laxminarayan

, Kissoon

. The quadruple burden of sepsis. CMAJ Can Med Assoc J, 2017; 189:E1128–E1129.

Dellinger

, Levy

, Rhodes

, et al. Surviving Sepsis Campaign: International guidelines for management of severe sepsis and septic shock: 2012. Crit Care Med, 2013; 41:580–637.

Mayr

, Yende

, Angus

. Epidemiology of severe sepsis. Virulence, 2014; 5:4–11.

Kempker

, Martin

. The changing epidemiology and definitions of sepsis. Clin Chest Med, 2016; 37:165–179.

ProCESS

Investigators

, Yealy

, Kellum

, Huang

, et al. A randomized trial of protocol-based care for early septic shock. N Engl J Med, 2014; 370:1683–1693.

Nguyen

, Jaehne

, Jayaprakash

, et al. Early goal-directed therapy in severe sepsis and septic shock: Insights and comparisons to ProCESS, ProMISe, and ARISE. Crit Care 2016;20:160. Available at: https://www-ncbi-nlm-nih-gov.ezproxyhost.library.tmc.edu/pmc/articles/PMC4929762/

Kumar

, Roberts

, Wood

, et al. Duration of hypotension before initiation of effective antimicrobial therapy is the critical determinant of survival in human septic shock. Crit Care Med, 2006; 34:1589–1596.

Ferrer

, Martin-Loeches

, Phillips

, et al. Empiric antibiotic treatment reduces mortality in severe sepsis and septic shock from the first hour: Results from a guideline-based performance improvement program. Crit Care Med, 2014; 42:1749–1755.

Moore

, Jones

, Kreiner

, et al. Validation of a screening tool for the early identification of sepsis. J Trauma, 2009; 66:1539–1546.

10.

Jones

, Ashton

, Kiehne

, et al. Reductions in sepsis mortality and costs after design and implementation of a nurse-based early recognition and response program. Jt Comm J Qual Patient Saf, 2015; 41:483–491.

11.

Wawrose

, Baraniuk

, Standiford

, et al. Comparison of sepsis screening tools' ability to detect sepsis accurately. Surg Infect, 2016; 17:525–529.

12.

Umscheid

, Betesh

, VanZandbergen

, et al. Development, implementation, and impact of an automated early warning and response system for sepsis. J Hosp Med, 2015; 10:26–31.

13.

Subbe

, Kruger

, Rutherford

, Gemmel

. Validation of a modified Early Warning Score in medical admissions. QJM Mon J Assoc Physicians, 2001; 94:521–526.

14.

Roney

, Whitley

, Maples

, et al. Modified early warning scoring (MEWS): Evaluating the evidence for tool inclusion of sepsis screening criteria and impact on mortality and failure to rescue. J Clin Nurs, 2015; 24:3343–3354.

15.

Al Solamy

, Al Salamah

, Al Thagafi

, et al. Diagnostic accuracy of a screening electronic alert tool for severe sepsis and septic shock in the emergency department. BMC Med Inform Decis Mak, 2014; 14:105.

16.

Hernandez

, Regueira

, Bruhn

, et al. Relationship of systemic, hepatosplanchnic, and microcirculatory perfusion parameters with 6-hour lactate clearance in hyperdynamic septic shock patients: An acute, clinical–physiological, pilot study. Ann Intensive Care, 2012; 2:44.

17.

Goerlich

, Wade

, McCarthy

, et al. Validation of sepsis screening tool using StO₂ in emergency department patients. J Surg Res, 2014; 190:270–275.

18.

Assunção

, Akamine

, Cardoso

, et al. Survey on physicians' knowledge of sepsis: Do they recognize it promptly?. J Crit Care, 2010; 25:545–552.

19.

Singer

, Deutschman

, Seymour

, et al. The Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). JAMA, 2016; 315:801–810.

20.

Vincent

, de Mendonça

, Cantraine

, et al. 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, 1998; 26:1793–1800.

21.

Haydar

, Spanier

, Weems

, et al. Comparison of qSOFA score and SIRS criteria as screening mechanisms for emergency department sepsis. Am J Emerg Med, 2017; 35:1730–1733.