[ Recent Advances in Chest Medicine ]

Identifying Patients With Sepsis on the

Hospital Wards
Poushali Bhattacharjee, MD; Dana P. Edelson, MD; and Matthew M. Churpek, MD, MPH, PhD

Sepsis contributes to up to half of all deaths in hospitalized patients, and early interventions,
such as appropriate antibiotics, have been shown to improve outcomes. Most research has
focused on early identification and treatment of patients with sepsis in the ED and the ICU;
however, many patients acquire sepsis on the general wards. The goal of this review is to
discuss recent advances in the detection of sepsis in patients on the hospital wards. We discuss
data highlighting the benefits and limitations of the systemic inflammatory response syndrome
(SIRS) criteria for screening patients with sepsis, such as its low specificity, as well as newly
described scoring systems, including the proposed role of the quick sepsis-related organ failure
assessment (qSOFA) score. Challenges specific to detecting sepsis on the wards are discussed,
and future directions that use big data approaches and automated alert systems are
highlighted. CHEST 2017; 151(4):898-907

KEY WORDS: clinical decision support systems; sepsis; SIRS; SOFA

The incidence of sepsis has been rising over such as appropriate antibiotic therapy,
the past decade, and it is one of the most improve outcomes in patients with sepsis,
common reasons for hospitalization, with an making early diagnosis critical.8,9 As such, the
estimated 1.6 million cases annually in the Surviving Sepsis Campaign (SSC) has made it
United States.1,2 This leads to approximately their mission to raise sepsis awareness and
$20 billion dollars in health-care spending in decrease sepsis-related mortality.10,11 Still,
the United States, which will likely continue recognition and treatment of sepsis remain
to increase as the population ages. Although a challenge given that more than one-half
most prior research has focused on patients of patients with severe sepsis are not
in the ICU or ED, up to 50% of patients with documented to have this diagnosis by their
sepsis are treated on the hospital wards.3,4 physicians.4 Therefore, work aimed at
improving the recognition of patients with
Longitudinal trends from observational data
sepsis is critical to improving their short- and
suggest that outcomes in patients with sepsis
long-term outcomes.
are improving. However, mortality remains
as high as 50% for those with septic shock.5-7 In this review, we discuss how the definition
It has been shown that early interventions, of sepsis has evolved over time and the

ABBREVIATIONS: EHR = electronic health record; ICD-9 = Interna- (4T32HS000078). M. M. C. is supported by a career development award
tional Classification of Diseases, Ninth Revision; NPV = negative from the National Heart, Lung, and Blood Institute (K08HL121080).
predictive value; PPV = positive predictive value; qSOFA = quick CORRESPONDENCE TO: Matthew M. Churpek, MD, MPH, PhD, Sec-
sepsis-related organ failure assessment; SIRS = systemic inflammatory tion of Pulmonary and Critical Care Medicine, University of Chicago
response syndrome; SSC = Surviving Sepsis Campaign Medical Center, 5841 South Maryland Ave, MC 6076, Chicago, IL
AFFILIATIONS: From the Department of Medicine, University of 60637; e-mail: [email protected]
Chicago, Chicago, IL. Copyright Ó 2016 American College of Chest Physicians. Published by
FUNDING AND SUPPORT: P. B. is supported by a postdoctoral training Elsevier Inc. All rights reserved.
grant from the Agency for Healthcare Research and Quality DOI: http://dx.doi.org/10.1016/j.chest.2016.06.020

898 Recent Advances in Chest Medicine [ 151#4 CHEST APRIL 2017 ]

potential benefits of active sepsis screening on the wards. of sepsis and septic shock.23 The new consensus
In addition, we review recent studies that use automated guidelines define sepsis as life-threatening organ
sepsis screening tools for patients on the general wards dysfunction, with suspected or documented infection
and their clinical utility. Finally, we end with gaps in and an acute two-point increase in the sepsis-related
knowledge and future directions in this important area organ failure assessment (SOFA) score. In addition,
of research. septic shock was redefined as the need for vasopressors
to maintain the mean arterial pressure at > 65 mm Hg
Defining Sepsis: An Evolution and a lactate level > 2 mmol/L after adequate fluid
resuscitation. It is important to note that the term
The concept of sepsis has been around since 700 BC
“severe sepsis” and the SIRS criteria are no longer part of
when Homer used the term “sepo” to describe human
the updated definition.
death as “rot” in his poems.12 Since then, the idea of
sepsis has undergone many iterations. Hippocrates
expanded on Homer’s idea in Corpus Hippocraticum, in Screening for Sepsis on the Wards
which he described it as a dangerous odiferous biological Because patients on the ward may become septic at
decay. It was not until 600 years later that Galen linked any point during hospitalization, screening is often
sepsis to blood infections and began treating patients performed longitudinally rather than at one point in
with sepsis using abscess drainage and bloodletting. time. This requires additional resource strain and
Even by the 1980s, sepsis literature reported significant burden on caregivers compared with a one-time
disparities in sepsis incidence and mortality due to the screening on admission. In addition, there are few data
fact that there was no universally accepted definition.13-17 to support optimal treatment strategies for patients with
Finally, in 1992, the American College of Chest sepsis identified on the wards, as discussed in further
Physicians and the Society of Critical Care Medicine detail by a recent review.24 Despite these considerations,
held a consensus conference that introduced the several studies suggest that actively screening patients on
systemic inflammatory response syndrome (SIRS) the wards is associated with improved process measures
and multiple organ dysfunction syndrome hoping that and patient outcomes, as discussed further on.
these definitions would improve clinicians’ aptitude for
early detection and treatment as well as standardize Manual Screening Tools
future sepsis-related clinical trials.18 Despite ongoing controversy surrounding the sepsis
definitions, prior studies suggest that sepsis screening
This definition was quickly met with criticism. Many
using the 2001 consensus definitions may decrease
felt that SIRS was too nonspecific and that it did not
mortality.25,26 For example, the SSC conducted a 2-year
adequately describe the underlying physiology of sepsis.19
retrospective study that evaluated quarterly bundle
In addition, there was concern that using this definition of
compliance and hospital mortality after initiation of
sepsis would lead to clinical trials with patient populations
sepsis screening programs in the ED, wards, and ICU.
that were too heterogeneous, thus precluding comparisons
Results from 252 hospitals showed a significant increase
of patients with sepsis and negating the original intent
in compliance rates with the 6-hour treatment bundle
behind SIRS.20 Furthermore, a survey of 1,100 critical care
(10.9% vs 31.3%) and 24-hour treatment bundle
physicians in 2000 revealed that 67% “cited no common
(18.4% vs 36.1%). This correlated temporally with a
definition of sepsis” and < 20% of respondents agreed on
decrease in unadjusted mortality rates from 37.0% to
any one definition.21 Together, these factors prompted the
30.8% at the end of the 2-year study.25 The most recent
2001 International Sepsis Definitions Conference to revisit
SSC study extended the follow-up period to 71/2 years.
the definitions proposed in 1991.22 However, apart from
This study reaffirmed the mortality benefit of high
expanding the list of signs and symptoms of sepsis, the
compliance with both treatment bundles by demonstrating
experts thought there was not enough new evidence to
a 25% relative risk reduction in mortality.27 As a result,
support a change to the original criteria.
the SSC advocates routine screening for sepsis with the
Because of these controversies and advances made in use of a paper-based screening tool to be used in all levels
understanding the pathophysiology, epidemiology, and of acute care.28,29 However, it is important to note that
treatment of sepsis, the Society of Critical Care Medicine before and after studies such as these and others that
and the European Society of Intensive Care Medicine are discussed further on run the risk of patient selection
convened a task force in 2015 to update the definitions bias, temporal bias, and investigator bias. Thus, future

journal.publications.chestnet.org 899
900 Recent Advances in Chest Medicine

TABLE 1 ] Characteristics of Studies Evaluating Automated Sepsis Screening Tools

Gold Standard for
Reference Study Design Test Sites Sepsis Definition of Alert Alert Frequency Primary Outcome Secondary Outcome
Thiel et al, Retrospective One ICD-9 codes for Decision trees with Continuous Diagnostic accuracy .
201048 cohort academic acute and without ABG
hospital infection þ data
acute organ
dysfunction þ
need for
vasopressors
Sawyer et al, Prospective One 2001 consensus Decision tree Continuous Rate of IV antibiotics/ Rate of ICU
201149 observational academic criteria without ABG fluids, respiratory transfer, rate of
pilot study hospital data support, ICU transfer
vasopressor within 12 h of
initiation, laboratory alert, in-hospital
results/imaging mortality, LOS,
within 12 h of alert LOS after alert
Buck, 201439 Prospective One ICD-9 code for $ 2 SIRS criteria þ Once per 24 h Diagnostic accuracy Any intervention
observational community sepsis, severe 1 sign of organ from MD or RN
pilot study hospital sepsis, or dysfunction in a
septic shock 24-h window
Palleschi et al, Retrospective 4 hospitals ICD-9 code for $ 2 SIRS  1 sign Continuous Lactate Timing of early
201433 before and within 1 sepsis, severe of organ determination, intervention in
after study medical sepsis, or dysfunction  blood culture ED vs acute care
center septic shock elevated lactate results, time to
level antibiotic treatment
McRee et al, Retrospective 1 community ICD-9 code for $ 2 SIRS þ manual . Stage of sepsis, length .
201442 before and hospital sepsis, severe nurse evaluation of stay, discharge
after study sepsis, or deeming patient location, mortality
septic shock “high risk”
[
151#4 CHEST APRIL 2017

Brandt et al, Prospective 1 academic ICD-9 code for Must have had Continuous Diagnostic accuracy Difference in time
201536 observational hospital sepsis, severe acute infection, of alert prior to
pilot study sepsis, or organ recognition by
septic shock dysfunction, or MD (determined
change in mental by chart review)
status prior to
screening for
SIRS; then must
have met $ 2
SIRS criteria

(Continued)
]
journal.publications.chestnet.org

TABLE 1 ] (Continued)
Gold Standard for
Reference Study Design Test Sites Sepsis Definition of Alert Alert Frequency Primary Outcome Secondary Outcome
Amland et al, Prospective 1 community ICD-9 code for $ 3 SIRS criteria Continuous Diagnostic accuracy .
201538 observational hospital septicemia, (including
study sepsis, severe elevated glucose
sepsis, or levels in patients
septic shock without
diabetes)  1
sign of organ
dysfunction þ
post-alarm
cross-check by
MD
Kurczewski Retrospective 1 academic ICD-9 code for $ 2 SIRS criteria . Time to any sepsis- Time to individual,
et al 201540 before and hospital sepsis, severe inclusive of related intervention sepsis-related
after study sepsis, or either abnormal (IV antibiotics/ intervention,
septic shock temperature or fluids, blood work) LOS, LOS in ICU,
WBC count mortality
Umscheid Retrospective Multicenter ICD-9 code for Risk score of $ 4 Continuous; Predictive ability for Rate of IV
et al,41 2015 before and sepsis points: 1 point stopped composite of ICU antibiotics/
after study for each SIRS once transfer, RRT data, fluids, blood
criterion, 1 point patient or death across all work, imaging
for SBP triggered 3 hospitals
< 100 mm Hg, alert
1 point for
lactate level
> 2.2 mmol/L
Amland and Retrospective Multicenter ICD-9 code for $ 3 SIRS Continuous Diagnostic accuracy Sepsis prevalence,
Hahn-Cover, cohort sepsis, severe (including incidence, and
201637 sepsis, or elevated glucose patient
septic shock levels in patients outcomes after
without the alert
diabetes)  1
sign of organ
dysfunction

ABG ¼ arterial blood gas; ICD-9 ¼ International Classification of Diseases, Ninth Revision; LOS ¼ length of stay; MD ¼ medical doctor; RN ¼ registered nurse; RRT ¼ rapid response team; SBP ¼ systolic blood
pressure; SIRS ¼ systemic inflammatory response syndrome.
901
randomized controlled trials are needed to confirm these was issued to a sepsis surveillance group consisting of an
findings. intensivist and critical care nurse, who performed a
chart review to determine if the primary team should
Automated Screening Tools
be notified. This automated method relied heavily on
Although manual sepsis screens are commonly used, adequate provider documentation (ie, the physician had
they have several disadvantages.30 First, they are to add an infection or acute organ dysfunction diagnosis
susceptible to transcription and calculation errors, which to the active problem list) and resulted in a positive
can lead to inaccurate screening results. In addition, predictive value (PPV) of 16.5% and sensitivity of
manual screens can only be performed intermittently, 100% for the diagnosis of severe sepsis based on expert
often once every nursing shift.31-33 This can lead to adjudication of all patients identified by the alert.
considerable delays in recognition and treatment.
Finally, manual screening typically requires a caregiver An initial study performed by Amland and Hahn-
to contact the physician to initiate a plan of care. Delays Cover37 triggered an alert in patients meeting either
in calling or failure to call may also impact patient SIRS criteria alone or with at least one sign of organ
outcomes. In contrast, automated screening tools have dysfunction. During a silent testing period, the alert
the potential to decrease diagnostic delays and increase resulted in a sensitivity of 83% and a PPV of 46% for the
screen accuracy. Several institutions have developed diagnosis of sepsis, using an International Classification
automated screening tools to expedite the diagnosis of of Diseases, Ninth Revision (ICD-9) code of septicemia,
sepsis and the delivery of subsequent sepsis bundles sepsis, severe sepsis, or septic shock. In a subsequent
(Table 1). It is important to note that all of these are study with the alert running live, a post-alarm cross-
based on before and after studies, with the exception of check had to be completed by the covering physician in
one randomized controlled trial that was conducted in the electronic health record (EHR).38 If the physician
patients in the ICU.34 checked a box labeled “suspected infection,” an
automated order set for blood cultures and lactate levels
SIRS-Based Screening Tools was populated. The addition of this second component
Many automated sepsis screening tools described in the to the screening tool increased the PPV to 94% and
literature are primarily based on SIRS criteria, with maintained similar sensitivity (81%).
additional specifications that are tailored to individual
The impact of SIRS-based automated screening tools
hospital systems. Modifying the SIRS criteria for
on improving sepsis-related interventions has also been
automated screening to improve specificity is an
studied (Tables 3, 4). For example, in a prospective pilot
important concept given that a recent study suggested
study, Buck39 noticed that 40% of the patients identified
that up to one-half of patients on the wards will meet
by the alert received escalated care in the form of
SIRS criteria at least once during their admission.35
repeated evaluation by a physician, additional
Several studies have investigated the diagnostic accuracy medications or intravenous fluids, laboratory tests,
of SIRS-based screening tools (Table 2). In a prospective respiratory support, or transfer to the ICU. Additionally,
pilot study, Brandt et al36 required the presence of in the pilot study mentioned earlier, Brandt et al36
infection, organ dysfunction, or altered mental status in showed that the alert resulted in a diagnosis of sepsis
the patient’s active problem list prior to allowing an approximately 27 minutes earlier when compared with
automated system to search for SIRS criteria. The alert the time of sepsis diagnosis based on chart review.

TABLE 2 ] Predictive Ability of Automated Sepsis Screening Tools

Diagnosis of Sepsis, Severe Sepsis, or Septic Shock
Reference Type of Alert Sensitivity (%) Specificity (%) PPV (%) NPV (%)
Thiel et al, 201048 (without ABG) Non-SIRS decision tree 17.1 96.0 20.5 95.9
36
Brandt et al, 2015 SIRS-based 100 62.0 16.5 100
Buck, 201439 SIRS-based . . 17 .
37
Amland and Hahn-Cover, 2016 SIRS-based 83 92 46 99
Amland et al, 201538 SIRS-based 81 87 94 63

NPV ¼ negative predictive value; PPV ¼ positive predictive value. See Table 1 legend for expansion of other abbreviation.

902 Recent Advances in Chest Medicine [ 151#4 CHEST APRIL 2017 ]

TABLE 3 ] Frequency of Initiation of Clinical Process Measures After Implementation of an Automated Sepsis
Screening Tool
Antibiotic Administer IV Draw for Lactate Blood for Cultures
Escalation (%) Fluids (%) Determination (%) Drawn (%)
Before After Before After Before After Before After
Reference Type of Alert Alert Alert Alert Alert Alert Alert Alert Alert
Umscheid et al, 201541 SIRS based 18 27 27 37 19 36 19 24
Sawyer et al, 201149 Non-SIRS decision tree 24 36 24 38 . . . .
33
Palleschi et al, 2014 SIRS based . . . . 50 89 72 75

See Table 1 legend for expansion of abbreviation.

In a before and after interventional study, Kurczewski provider, bedside nurse, and rapid response team
et al40 evaluated a tool that used at least two SIRS gathered at the patient’s bedside within 30 minutes. This
criteria to trigger the sepsis alert but modified it so group then had to assess and document the most likely
that one of them had to include an abnormal WBC condition that triggered the alert and whether clinical
count or temperature. Rather than diagnostic accuracy, management should be modified. The screening tool was
differences in time to sepsis-related therapies in patients then silenced for the remainder of that patient’s hospital
discharged with an ICD-9 code for sepsis, severe sepsis, stay. Using four points as the trigger threshold resulted
or septic shock were evaluated. Once a patient met the in a screen-positive rate of 6%, sensitivity of 17%,
criteria, providers were forced to address the alert in the specificity of 97%, PPV of 28%, and negative predictive
EHR before any other tasks could be performed, and value (NPV) of 95% for the composite outcome of
there were no limits to how frequently the alert could ICU transfer, rapid response activation, or death.
fire. In patients who triggered the alert, there was a After initiation of the alert on the wards, there was a
significant decrease in median time to any sepsis-related significant increase in sepsis-related interventions
intervention (0.6 hours vs 4.1 hours), blood culture within 3 hours of the alert, including ordering of
collection (1.1 hours vs 13.1 hours), and lactate antibiotics (10% vs 16%), determination of lactate levels
determination (2.4 hours vs 40.5 hours). In another (10% vs 23%), and blood product administration
study using a similar screening tool, with the added (5% vs 10%). Of note, review of the alert assessments
component of an interprofessional sepsis education revealed that one-half of the alerted clinicians did not
program, Palleschi et al33 also showed improvements think the patient was critically ill, > 30% believed the
in obtaining lactate levels (50% vs 89%; P < .001), and diagnosis was sepsis, and more than 90% knew of this
blood cultures (72% vs 75%). diagnosis prior to the alert.
Rather than using the traditional definition of sepsis Aside from the results of the most recent SSC study,
to develop an alert system, Umscheid et al41 performed the impact of automated sepsis screening on patient
a before and after study using a tool that translated outcomes, such as mortality, has been mixed (Table 5).
SIRS criteria into a risk score in which patients earned One study suggesting an improvement in patient
one point for each SIRS criterion as well as for a outcomes was performed by McRee et al,42 who used an
systolic blood pressure < 100 mm Hg or a lactate level alert triggered by two SIRS criteria with the additional
> 2.2 mmol/L. Once an alert was triggered, the covering component of a manual risk assessment by a nurse in

TABLE 4 ] Time to Initiation of Clinical Process Measures After Implementation of an Automated Sepsis Screening
Tool
Antibiotic Therapy Administer IV Lactate Levels Blood for Cultures
Escalation (h) Fluids (h) Determined (h) Drawn (h)
Before After Before After Before After Before After
Reference Type of Alert Alert Alert Alert Alert Alert Alert Alert Alert
Kurczewski et al, 201540 SIRS based 5.2 3.9 7.1 1.9 40.5 2.4 13.2 1.1
Palleschi et al, 201433 SIRS based 3.0 1.5 . . . . . .

See Table 1 legend for expansion of abbreviation.

journal.publications.chestnet.org 903
TABLE 5 ] Frequency of Patient Outcomes After Implementation of an Automated Sepsis Screening Tool
ICU Transfers (%) Mortality (%)
Reference Type of Alert Before Alert After Alert Before Alert After Alert
49
Sawyer et al, 2011 Non-SIRS decision tree 23 26 12 10
McRee et al, 201442 SIRS based . . 9.3 1.0
Kurczewski et al, 2015 40
SIRS based 47 27 . .
Umscheid et al, 201541 SIRS based 35 35 17 13

See Table 1 legend for expansion of abbreviation.

a before and after study. When compared with patients shock using a decision tree model that included vital
who were discharged with an ICD-9 code for sepsis, signs, such as systolic blood pressure, and laboratory
severe sepsis, or septic shock before alert implementation, tests, including blood urea nitrogen, albumin, and
those who had triggered the alert had a substantial bilirubin determinations. This resulted in a PPV of
decrease in mortality (9.35% vs 1.0%) and a higher 21.4% and a NPV of 96.1% for the diagnosis of septic
likelihood of being discharged home (25.3% vs 49.0%) but shock in one of their validation cohorts. They
no difference in length of stay. However, the two studies subsequently performed a prospective observational
noted previously by Kurczewski et al40 and Umscheid study using the developed model on the wards.49 Of
et al,41 which demonstrated an increase in sepsis-related patients who triggered the alert, 70.8% received at least
interventions, failed to show any differences in overall one sepsis-related intervention. In this group, there was
length of stay and mortality after alert implementation. a significant increase in escalation of antibiotic therapy
(36% vs 24%), IV fluid administration (38% vs 24%),
These results suggest that more data are needed from
and oxygen therapy (20% vs 8%). There was no
clinical trials to conclusively determine if automated
difference in the rate of transfers to the ICU and hospital
SIRS-based screening tools improve important clinical
mortality in this study.
outcomes such as mortality in patients on hospital
wards. A recently developed risk stratification tool that was
presented along with the new sepsis definitions is termed
Non-SIRS-Based Screening Tools the quick sepsis-related organ failure assessment
(qSOFA).50 This tool was developed using EHR data
Over the past several years, many groups have developed
from 12 hospitals within the University of Pittsburgh
risk-stratification tools for identifying high-risk patients
health system. In this study, suspicion of infection was
outside the ICU.43,44 The modified early warning score is
defined as antibiotic administration and culture orders
one example that is already in place in several hospitals
within a specific time window. Optimal cut points of
in the United States and around the world and has been
different vital signs were determined in univariate
shown to predict patient outcomes, including cardiac
analyses, and these variables were combined in a logistic
arrest, ICU transfers, and in-hospital mortality. Given
regression model. The final qSOFA score consisted of
the lack of specificity of the SIRS criteria, there has been
altered mental status, systolic blood pressure # 100 mm Hg,
significant interest in evaluating the performance of
and a respiratory rate of at least 22 breaths per minute.
early warning scores in predicting outcomes for patients
A score of 2 or higher had > 60% sensitivity for
with sepsis. To date, most work in this area has been
in-hospital mortality in the University of Pittsburgh
performed in the ED, with evidence that these scores
Medical Center validation cohort, which included
can accurately predict mortality.45,46 On the wards, one
patients in the ICU and patients not in the ICU. The
prospective study of patients with sepsis demonstrated
proposed use of qSOFA is at the bedside to identify
that the simple clinical score and rapid emergency
high-risk infected patients outside the ICU and to
medicine score, which are based on patient demographics
prompt clinicians to consider additional diagnostic tests
and vital signs, were accurate predictors of mortality
or escalation of therapy. However, it is not currently part
(area under the curve, 0.77 for both).47
of the recent consensus definition of sepsis. In addition,
Other groups have developed sepsis risk scores using the SSC still recommends screening with SIRS criteria
patient-level data from the EHR. For example, Thiel and using the qSOFA to screen for organ dysfunction in
et al48 developed an automated screening tool for septic those who meet the traditional definition of sepsis.51

904 Recent Advances in Chest Medicine [ 151#4 CHEST APRIL 2017 ]

Current Controversies and Gaps in Knowledge in a recent review, of the 122 studies used to develop
these recommendations, only one prospective clinical
Identifying Infected Patients
trial, which examined the duration of empirical
One of the biggest challenges facing sepsis research, antibiotics, included any patients on the wards.24 There
especially for studies that use retrospective EHR data, is were no patients on the wards in any trial supporting
to determine which patients are truly infected. Even aggressive fluid resuscitation, which not only is a
caregivers of the same patient may disagree on whether frequent intervention on the wards for septic patients
the patient might be infected. For example, in a single- but also is a commonly reported outcome for the efficacy
center prospective study by our group, we found that of sepsis screening tools. Future research in sepsis
bedside nurses and ordering providers agreed on the treatment for patients on the wards is clearly needed to
presence of infection only 17% of the time (k ¼ 0.12).52 optimize outcomes in this population.
Furthermore, progression to severe sepsis or shock
was significantly higher when both providers suspected
Future Directions
infection in a SIRS-positive patient (17.7%) and lowest
In addition to the gaps in knowledge noted earlier,
when neither suspected infection (1.5%), with single-
future work in the use of biomarkers and EHR data may
provider suspicion conferring intermediate risk (6.0%).
lead to important developments in the care of patients
Studies using EHR data to define infection are equally
with sepsis. For example, one of the most promising
challenging, as the accuracy of culture orders, antibiotic
biomarkers in use today is procalcitonin, as its values
prescriptions, and other interventions for defining
rise rapidly with invasive bacterial infection and
infection are unknown. The impact of different
decrease once that infection has cleared.53 However,
definitions of infection and their effect on variable
procalcitonin is also elevated in other conditions, such as
importance in screening models and algorithm accuracy
trauma. Thus, the current suggested use of procalcitonin
as well as their benefit to patients is unknown and an
is limited to guiding the cessation of antibiotic
important area of future research.
therapy.28,54,55 Another promising alternative includes
Optimal Screening Frequency an advanced polymerase chain reaction followed by
electrospray ionization and mass spectrometry, which
Ideally, automated screening tools perform continuous
can detect more than 800 pathogens in a single assay
background monitoring allowing for real-time sepsis
within 6 hours.56 As our understanding of sepsis
detection. Of the studies that allowed for continuous
pathophysiology continues to grow, discovery of novel
or repeated alerts, alert fatigue or high rates of false-
biomarkers that provide diagnostic information, as well
positive results were common.39,40 Some studies
as risk stratification, will allow providers to deliver
attempted to decrease alert fatigue by limiting the
optimal care to patients with sepsis. Finally, as EHR
number of times the screening tool could fire per day or
data that can be used for research becomes more
per hospitalization. This does not capture the dynamic
widespread, using advanced machine learning techniques
longitudinal nature of a patient’s clinical course on the
to develop highly accurate tools will become more
wards nor does it maximize the use of an automated
common.57,58
system. It also highlights the need for more specific
screening tools that have been validated on the wards
and that provide clinicians with meaningful information Limitations
that is actionable. None of the sepsis screening tools reviewed relied on
the new definition of sepsis, and it is unclear how this
Treatment of Screen-Positive Patients impacts patient care. Additionally, we did not perform a
As noted earlier, current data on automated sepsis standardized regimented literature search of all studies
screening on the wards suggest that these tools improve on sepsis detection because our article focused only on
delivery of the sepsis management bundle without studies performed in the past 5 years.
altering rates of ICU transfers and mortality. These
results raise the issue of the optimal treatment of Conclusions
patients on the wards who are flagged as high risk by the Because of the increased awareness of sepsis driven by
screening tools. The SSC’s recommendations for severe previous landmark studies and the SSC, there have been
sepsis and shock are based on evidence that primarily a multitude of efforts to improve sepsis detection in
includes patients in the ICU or ED. In fact, as outlined hospitalized patients. Many of the studies in this review

journal.publications.chestnet.org 905
offer promising screening tools for patient surveillance 15. Bihari DJ. Septicaemia—the clinical diagnosis. J Antimicrob
Chemother. 1990;25(suppl C):1-7.
and increased sepsis-specific interventions, but their
16. Sprung CL. Definitions of sepsis—have we reached a consensus? Crit
effect on patient outcomes is less clear. Future work is Care Med. 1991;19(7):849-851.
needed to determine the optimal way to identify patients 17. Matot I, Sprung CL. Definition of sepsis. Intensive Care Med. 2001;
with sepsis on the wards that will most likely benefit 27(suppl 1):S3-9.

from earlier and more aggressive interventions. 18. Bone RC, Balk RA, Cerra FB, et al. Definitions for sepsis and organ
failure and guidelines for the use of innovative therapies in sepsis.
The ACCP/SCCM Consensus Conference Committee. American
Acknowledgments College of Chest Physicians/Society of Critical Care Medicine. Chest.
1992;101(6):1644-1655.
Financial/nonfinancial disclosures: The authors have reported to
CHEST the following: M. M. C. and D. P. E. have a patent pending 19. Vincent JL. Dear SIRS, I’m sorry to say that I don’t like you. Crit
(ARCD P0535US.P2) for risk stratification algorithms for Care Med. 1997;25(2):372-374.
hospitalized patients. In addition, D. P. E. has received research 20. Marshall JC. SIRS and MODS: what is their relevance to the science
support and honoraria from Philips Healthcare (Andover, MA), and practice of intensive care? Shock. 2000;14(6):586-589.
research support from the American Heart Association (Dallas, TX) 21. Poeze M, Ramsay G, Gerlach H, Rubulotta F, Levy M. An
and Laerdal Medical (Stavanger, Norway), and an honorarium from international sepsis survey: a study of doctors’ knowledge and
Early Sense (Tel Aviv, Israel). She has an ownership interest in perception about sepsis. Crit Care. 2004;8(6):R409-413.
Quant HC (Chicago, IL), which develops products for risk
stratification of hospitalized patients. None declared (P.B.). 22. Levy MM, Fink MP, Marshall JC, et al. 2001 SCCM/ESICM/ACCP/
ATS/SIS International Sepsis Definitions Conference. Crit Care Med.
2003;31(4):1250-1256.
References 23. Singer M, Deutschman CS, Seymour CW, et al. The Third
1. Pfuntner A, Wier LM, Stocks C. Most frequent conditions in U.S. International Consensus Definitions for Sepsis and Septic Shock
hospitals, 2010. http://www.hcup-us.ahrq.gov/reports/statbriefs/ (Sepsis-3). JAMA. 2016;315(8):801-810.
sb148.pdf. Accessed February 15, 2016.
24. Odden AJ, Govindan S, Sheth J, Iwashyna TJ. A systematic
2. Lagu T, Rothberg MB, Shieh MS, Pekow PS, Steingrub JS, assessment of the surviving sepsis campaign’s evidence supporting
Lindenauer PK. Hospitalizations, costs, and outcomes of severe the care of patients with severe sepsis on the wards. Ann Am Thorac
sepsis in the United States 2003 to 2007. Crit Care Med. 2012;40(3): Soc. 2015;12(6):956-958.
754-761. 25. Levy MM, Dellinger RP, Townsend SR, et al. The Surviving Sepsis
3. Esteban A, Frutos-Vivar F, Ferguson ND, et al. Sepsis incidence and Campaign: results of an international guideline-based performance
outcome: contrasting the intensive care unit with the hospital ward. improvement program targeting severe sepsis. Crit Care Med. 2010;
Crit Care Med. 2007;35(5):1284-1289. 38(2):367-374.
4. Rohde JM, Odden AJ, Bonham C, et al. The epidemiology of acute 26. Moore LJ, Jones SL, Kreiner LA, et al. Validation of a screening
organ system dysfunction from severe sepsis outside of the intensive tool for the early identification of sepsis. J Trauma. 2009;66(6):
care unit. J Hosp Med. 2013;8(5):243-247. 1539-1546; discussion 1546-1547.
5. Liu V, Escobar GJ, Greene JD, et al. Hospital deaths in patients with 27. Levy MM, Rhodes A, Phillips GS, et al. Surviving Sepsis Campaign:
sepsis from 2 independent cohorts. JAMA. 2014;312(1):90-92. association between performance metrics and outcomes in a 7.5-year
study. Crit Care Med. 2015;43(1):3-12.
6. Dombrovskiy VY, Martin AA, Sunderram J, Paz HL. Rapid increase
in hospitalization and mortality rates for severe sepsis in the United 28. Dellinger RP, Levy MM, Rhodes A, et al. Surviving Sepsis Campaign:
States: a trend analysis from 1993 to 2003. Crit Care Med. 2007;35(5): international guidelines for management of severe sepsis and septic
1244-1250. shock, 2012. Intensive Care Med. 2013;39(2):165-228.
7. Angus DC, Linde-Zwirble WT, Lidicker J, Clermont G, Carcillo J, 29. Evaluation for severe sepsis screening tool. http://survivingsepsis.org/
Pinsky MR. Epidemiology of severe sepsis in the United States: SiteCollectionDocuments/ScreeningTool.pdf. Accessed May 24, 2016.
analysis of incidence, outcome, and associated costs of care. Crit 30. Surviving Sepsis Campaign: protocols and checklists. http://www.
Care Med. 2001;29(7):1303-1310. survivingsepsis.org/Resources/Pages/Protocols-and-Checklists.aspx.
8. Rivers E, Nguyen B, Havstad S, et al. Early goal-directed therapy in Accessed May 24, 2016.
the treatment of severe sepsis and septic shock. N Engl J Med. 31. Gyang E, Shieh L, Forsey L, Maggio P. A nurse-driven screening tool
2001;345(19):1368-1377. for the early identification of sepsis in an intermediate care unit
9. Kumar A, Roberts D, Wood KE, et al. Duration of hypotension setting. J Hosp Med. 2015;10(2):97-103.
before initiation of effective antimicrobial therapy is the critical 32. McKinley BA, Moore LJ, Sucher JF, et al. Computer protocol
determinant of survival in human septic shock. Crit Care Med. facilitates evidence-based care of sepsis in the surgical intensive care
2006;34(6):1589-1596. unit. J Trauma. 2011;70(5):1153-1166; discussion 1166-1167.
10. Dellinger RP, Levy MM, Carlet JM, et al. Surviving Sepsis Campaign: 33. Palleschi MT, Sirianni S, O’Connor N, Dunn D, Hasenau SM.
international guidelines for management of severe sepsis and septic An interprofessional process to improve early identification and
shock: 2008. Crit Care Med. 2008;36(1):296-327. treatment for sepsis. J Healthc Qual. 2014;36(4):23-31.
11. Action Demanded on World’s Oldest Killer. ’Barcelona Declaration’ 34. Hooper MH, Weavind L, Wheeler AP, et al. Randomized trial
Aims to Tackle 10th Most Common Cause of Death, 2002. http:// of automated, electronic monitoring to facilitate early detection
www.survivingsepsis.org/SiteCollectionDocuments/About-Barcelona- of sepsis in the intensive care unit*. Crit Care Med. 2012;40(7):
Declaration.pdf. Accessed May 24, 2016. 2096-2101.
12. Funk DJ, Parrillo JE, Kumar A. Sepsis and septic shock: a history. 35. Churpek MM, Zadravecz FJ, Winslow C, Howell MD, Edelson DP.
Crit Care Clin. 2009;25(1):83-101, viii. Incidence and prognostic value of the systemic inflammatory
13. Knaus WA, Sun X, Nystrom O, Wagner DP. Evaluation of definitions response syndrome and organ dysfunctions in ward patients.
for sepsis. Chest. 1992;101(6):1656-1662. Am J Respir Crit Care Med. 2015;192(8):958-964.
36. Brandt BN, Gartner AB, Moncure M, et al. Identifying severe sepsis
14. Bone RC, Fisher CJ, Clemmer TP, Slotman GJ, Metz CA,
via electronic surveillance. Am J Med Qual. 2015;30(6):559-565.
Balk RA. Sepsis syndrome: a valid clinical entity.
Methylprednisolone Severe Sepsis Study Group. Crit Care Med. 37. Amland RC, Hahn-Cover KE. Clinical decision support for early
1989;17(5):389-393. recognition of sepsis. Am J Med Qual. 2016;31:103-110.

906 Recent Advances in Chest Medicine [ 151#4 CHEST APRIL 2017 ]

38. Amland RC, Lyons JJ, Greene TL, Haley JM. A two-stage clinical 48. Thiel SW, Rosini JM, Shannon W, Doherty JA, Micek ST,
decision support system for early recognition and stratification of Kollef MH. Early prediction of septic shock in hospitalized patients.
patients with sepsis: an observational cohort study. JRSM Open. J Hosp Med. 2010;5(1):19-25.
2015;6(10):2054270415609004. 49. Sawyer AM, Deal EN, Labelle AJ, et al. Implementation of a real-
39. Buck KM. Developing an early sepsis alert program. J Nurs Care time computerized sepsis alert in nonintensive care unit patients.
Qual. 2014;29(2):124-132. Crit Care Med. 2011;39(3):469-473.
40. Kurczewski L, Sweet M, McKnight R, Halbritter K. Reduction in 50. Seymour CW, Liu VX, Iwashyna TJ, et al. Assessment of clinical
time to first action as a result of electronic alerts for early sepsis criteria for sepsis: for the Third International Consensus Definitions
recognition. Crit Care Nurs Q. 2015;38(2):182-187. for Sepsis and Septic Shock (Sepsis-3). JAMA. 2016;315(8):762-774.
51. Antonelli M, DeBacker D, Dorman T, Kleinpell R, Levy M, Rhodes
41. Umscheid CA, Betesh J, VanZandbergen C, et al. Development,
A. Surviving Sepsis Campaign Responds to Sepsis-3, 2016. http://
implementation, and impact of an automated early warning and
www.survivingsepsis.org/SiteCollectionDocuments/SSC-Statements-
response system for sepsis. J Hosp Med. 2015;10(1):26-31.
Sepsis-Definitions-3-2016.pdf. Accessed May 11, 2016.
42. McRee L, Thanavaro JL, Moore K, Goldsmith M, Pasvogel A. 52. Bhattacharjee P, Churpek MM, Howell MD, Edelson DP. Detecting
The impact of an electronic medical record surveillance program sepsis: are two opinions better than one? [abstract]. J Hosp Med.
on outcomes for patients with sepsis. Heart Lung. 2014;43(6): 2016;11(suppl 1).
546-549.
53. Wacker C, Prkno A, Brunkhorst FM, Schlattmann P. Procalcitonin
43. Churpek MM, Yuen TC, Park SY, Gibbons R, Edelson DP. Using as a diagnostic marker for sepsis: a systematic review and meta-
electronic health record data to develop and validate a prediction analysis. Lancet Infect Dis. 2013;13(5):426-435.
model for adverse outcomes in the wards*. Crit Care Med.
54. Becker KL, Snider R, Nylen ES. Procalcitonin assay in systemic
2014;42(4):841-848.
inflammation, infection, and sepsis: clinical utility and limitations.
44. Churpek MM, Yuen TC, Winslow C, et al. Multicenter development Crit Care Med. 2008;36(3):941-952.
and validation of a risk stratification tool for ward patients. Am J 55. Billeter A, Turina M, Seifert B, Mica L, Stocker R, Keel M. Early
Respir Crit Care Med. 2014;190(6):649-655. serum procalcitonin, interleukin-6, and 24-hour lactate clearance:
45. Corfield AR, Lees F, Zealley I, et al. Utility of a single early warning useful indicators of septic infections in severely traumatized patients.
score in patients with sepsis in the emergency department. Emerg World J Surg. 2009;33(3):558-566.
Med J. 2014;31(6):482-487. 56. Bacconi A, Richmond GS, Baroldi MA, et al. Improved sensitivity for
46. Keep JW, Messmer AS, Sladden R, et al. National early warning molecular detection of bacterial and Candida infections in blood.
score at emergency department triage may allow earlier J Clin Microbiol. 2014;52(9):3164-3174.
identification of patients with severe sepsis and septic shock: 57. Henry KE, Hager DN, Pronovost PJ, Saria S. A targeted real-time
a retrospective observational study. Emerg Med J. 2016;33(1): early warning score (TREWScore) for septic shock. Sci Transl Med.
37-41. 2015;7(299):299ra122.
47. Ghanem-Zoubi NO, Vardi M, Laor A, Weber G, Bitterman H. 58. Taylor RA, Pare JR, Venkatesh AK, et al. Prediction of in-hospital
Assessment of disease-severity scoring systems for patients with mortality in emergency department patients with sepsis: a local big
sepsis in general internal medicine departments. Crit Care. 2011; data-driven, machine learning approach. Acad Emerg Med.
15(2):R95. 2016;23(3):269-278.

journal.publications.chestnet.org 907