Professional Documents
Culture Documents
Sepsis 3
Sepsis 3
Sepsis 3
MAIN OUTCOMES AND MEASURES Evidence for and agreement on septic shock definitions
and criteria.
RESULTS The systematic review identified 44 studies reporting septic shock outcomes (total of
166 479 patients) from a total of 92 sepsis epidemiology studies reporting different cutoffs
and combinations for blood pressure (BP), fluid resuscitation, vasopressors, serum lactate level,
and base deficit to identify septic shock. The septic shock–associated crude mortality was 46.5%
(95% CI, 42.7%-50.3%), with significant between-study statistical heterogeneity (I2 = 99.5%;
τ2 = 182.5; P < .001). The Delphi process identified hypotension, serum lactate level,
and vasopressor therapy as variables to test using cohort studies. Based on these 3 variables
alone or in combination, 6 patient groups were generated. Examination of the SSC database
demonstrated that the patient group requiring vasopressors to maintain mean BP 65 mm Hg
or greater and having a serum lactate level greater than 2 mmol/L (18 mg/dL) after fluid
resuscitation had a significantly higher mortality (42.3% [95% CI, 41.2%-43.3%]) in risk-adjusted
comparisons with the other 5 groups derived using either serum lactate level greater than
2 mmol/L alone or combinations of hypotension, vasopressors, and serum lactate level 2 mmol/L Author Affiliations: Author
or lower. These findings were validated in the UPMC and KPNC data sets. affiliations are listed at the end of this
article.
CONCLUSIONS AND RELEVANCE Based on a consensus process using results from a systematic Group Information: Members of the
Sepsis Definitions Task Force are
review, surveys, and cohort studies, septic shock is defined as a subset of sepsis in which listed at the end of this article.
underlying circulatory, cellular, and metabolic abnormalities are associated with a greater risk of
Corresponding Author: Manu
mortality than sepsis alone. Adult patients with septic shock can be identified using the clinical Shankar-Hari, MD, MSc, Department
criteria of hypotension requiring vasopressor therapy to maintain mean BP 65 mm Hg or greater of Critical Care Medicine, Guy’s and St
and having a serum lactate level greater than 2 mmol/L after adequate fluid resuscitation. Thomas’ NHS Foundation Trust,
London SE1 7EH, United Kingdom
JAMA. 2016;315(8):775-787. doi:10.1001/jama.2016.0289 (manu.shankar-hari@kcl.ac.uk).
(Reprinted) 775
C
onsensus definitions, generated in 19911 and revisited in and epidemiology and limits of human studies; adults 19 years
2001,2 describe septic shock as a state of cardiovascu- or older; English-language publications; and publication dates
lar dysfunction associated with infection and unex- between January 1, 1992 (1991 definitions1), and December 25,
plained by other causes. The increasing availability of large elec- 2015. For full-text review, only noninterventional studies re-
tronic health record (EHR) data sets, registries, national case mix porting sepsis epidemiology and all-cause mortality were in-
programs, trial data sets, and claims databases using Interna- cluded. Randomized clinical trials were excluded, because the
tional Classification of Diseases codes have since generated mul- additional inclusion and exclusion criteria might confound the
tiple observational studies reporting septic shock epidemiol- effect of criteria on mortality (the study objective).8 To avoid
ogy. However, variable interpretation and application of the variability in outcomes related to specific pathogens, specific
consensus definitions1,2 have contributed to variable esti- patient groups, and interventional before-and-after studies,
mates of both incidence and outcomes.3-8 It is unclear to what studies reporting these populations were also excluded. Data
extent these variations represent true differences or an artifact were extracted on cohort recruitment period, cohort charac-
attributable to inconsistent use of definitions.8,9 Furthermore, teristics, setting, criteria used to identify septic shock, and acute
emerging insights into sepsis pathophysiology10-13 warrant a re- mortality. Detailed methods, including search strategy, are pre-
view of the current septic shock definition and the criteria used sented in eMethods 1 and eTable 1 in the Supplement.
to identify it clinically.
Against this background, the Society of Critical Care Medi- Delphi Study
cine (SCCM) and the European Society of Intensive Care Med To generate consensus on the septic shock definition and cri-
(ESICM) convened an international task force to review defini- teria, 3 face-to-face meetings, 3-round sequential pretested
tions of sepsis and septic shock in January 2014. To support the questionnaires, and email discussions among the task force par-
task force deliberations on redefining septic shock, a series of ticipants were conducted. One task force member did not par-
activities was performed: a systematic review and meta- ticipate in these surveys because of lack of content expertise,
analysis of criteria used in observational studies reporting sep- and 1 did not respond to the first 2 surveys. Questionnaires were
sis epidemiology in adults; a Delphi study to achieve consen- developed, refined, and administered consisting of single- and
sus; cohort studies using the Surviving Sepsis Campaign (SSC) multiple-answer questions, free-text comments, and a 5-point
registry; and subsequent testing of the applicability of the new Likert agreement scale. For consensus discussions and not-
criteria in patients with suspected infection from 2 large EHR- ing agreement, the 5-point Likert agreement scales were
derived data sets. The aims of this study were to develop an up- grouped at the tails of the scale choices (ie, “strongly dis-
dated septic shock definition and to derive clinical criteria for agree” grouped with “disagree”; “strongly agree” grouped with
identifying patients with septic shock meeting this updated defi- “agree”). All outputs from the systematic review, surveys, and
nition. Specifically, this updated definition and these criteria are the results of cohort studies were made available to partici-
intended to provide a standard classification to facilitate clini- pants throughout the Delphi study.
cal care, future clinical research, and reporting. In the first round (August 2014), using 26 questions in 4
domains, agreement and opinions were explored on (1) com-
ponents of the new septic shock definition; (2) variables and
their cutoffs identified by the systematic review; (3) defini-
Methods tions of, and criteria for, hypotension, persistent hypoten-
In this article, “definition” refers to a description of septic shock sion, adequacy of resuscitation, and resuscitation end points;
and “clinical criteria” to variables used to identify adult pa- and (4) septic shock severity scoring. In the second round
tients with septic shock. (November 2014), 4 questions were used to generate state-
ments for key terms (persistent hypotension, adequacy of re-
Task Force suscitation, and septic shock) and to reach agreement on test
The SCCM and ESICM each nominated cochairs of the task force variables and outcomes for subsequent analysis of predictive
and provided unrestricted funding support toward the work con- validity. The objectives of the third round (January 2015) were
ducted. The 2 cochairs then selected 17 other task force partici- to establish a consensus definition of septic shock and re-
pants based on their scientific expertise in sepsis epidemiol- lated clinical criteria. In the third survey, the task force mem-
ogy, clinical trials, and basic or translational research. Task force bers were given 4 choices for the septic shock updated crite-
participants are listed at the end of the article. The task force ria ([1] serum lactate level alone; [2] hypotension alone;
retained complete autonomy for all decisions. ESICM and SCCM [3] vasopressor-dependent hypotension or serum lactate
had no role in study design, conduct, or analysis but were con- level; [4] vasopressor-dependent hypotension and serum lac-
sulted for peer review and endorsement of the manuscript.14 tate level) and were asked to provide their first and second
choices. The cumulative first or second choices were used to
Systematic Review and Meta-analysis agree on the reported septic shock criteria.
The aims of the systematic review were to assess the differ- Questionnaire items were accepted if agreement ex-
ent criteria used to identify adult patients with septic shock ceeded 65%. Choices for which agreement was less than 65%
and whether these criteria were associated with differences in were rediscussed to achieve consensus or were eliminated, as
reported outcomes. MEDLINE was searched using search appropriate to achieve the project aims. The survey question-
terms, MeSH headings, and combinations of sepsis, septic shock, naires are presented in eMethods 2 in the Supplement.
776 JAMA February 23, 2016 Volume 315, Number 8 (Reprinted) jama.com
Cohort Studies
Figure 1. Study Identification and Selection Process Used
The institutional review boards of Cooper University Hospital in the Systematic Review
(Camden, New Jersey),15 University of Piitsburgh Medical Cen-
ter (UPMC; a network of hospitals in western Pennsylvania), 1017 Records identified and screened
and Kaiser Permanente Northern California (KPNC)16 pro- 982 MEDLINE
35 Other sources a
vided ethics approvals for research using the SSC and EHR data
sets, respectively.
The SSC registry includes data collected from 218 hospi- 915 Excluded
894 Did not meet screening
tals in 18 countries on 28 150 patients with suspected infec- criteria
tion who, despite adequate fluid resuscitation as judged by 21 Duplicate
the collecting sites, still had 2 or more systemic inflamma-
tory response syndrome criteria and 1 or more organ dysfunc-
102 Met full-text review criteria
tion criteria (eMethods 3 in the Supplement). The SSC data-
base setup, inclusion, and reporting items are described in
detail elsewhere.6,17 To select clinical criteria for the new sep- 36 Excluded b
16 Specific population
tic shock definition, an analysis data set was created that in- 10 Included all age groups
cluded all patients with a serum lactate level measurement or 10 Interventional study
jama.com (Reprinted) JAMA February 23, 2016 Volume 315, Number 8 777
Table 1. Summary of Septic Shock Definitions and Criteria Reported in the Studies Identified by the Systematic Reviewa
2 generalized estimating equation population-averaged logistic tential criterion variables on hospital mortality adjusted for other
regression models with exchangeable correlation structure, covariates. These models also included an a priori adjustment
where hospital site was the panel variable. variable for covariates including region (United States and
The first model used the potential septic shock groups 1 to Europe), location where sepsis was suspected (emergency de-
6 derived from these variables (eTable 5 in the Supplement), with partment, ward, or critical care unit), antibiotic administration,
group 1 as the referent group and adjusted for other covariates steroid use, organ dysfunction (pulmonary, renal, hepatic, and
to assess true mortality difference between these groups. The sec- acutely altered mental state), infection source (pneumonia, uri-
ond model assessed the independent association of these 3 po- nary tract infection, abdominal, meningitis and other), hyper-
778 JAMA February 23, 2016 Volume 315, Number 8 (Reprinted) jama.com
Patients With
Septic Shock Septic Shock, Mortality, %
Source Deaths, No. No. (95% CI)
Consensus Definition
Degoricija et al,46 2006 90 125 72.0 (64.1-79.9)
Angkasekwinai et al,38 2007 41 78 52.6 (41.5-63.6)
Nesseler et al,27 2013 30 93 32.3 (22.8-41.8)
Sakr et al,25 2013 85 145 58.6 (50.6-66.6)
23
Goncalves-Pereira et al, 2014 418 856 48.8 (45.5-52.2)
Leligdowicz et al,5 2014 4146 7974 52.0 (50.9 -53.1)
Ortiz et al,19 2014 144 319 45.1 (39.7-50.6)
Hypotension
Laupland et al,47 2004 81 159 50.9 (43.2-58.7)
Gaspraovic et al,45 2006 44 129 34.1 (25.9-42.3)
Shapiro et al,44 2006 15 53 28.3 (16.2-40.4)
Povoa et al,35 2009 202 458 44.1 (39.6-48.7)
7
Klein Klowenberg et al, 2012 52 98 53.1 (43.2-62.9)
Kaukonen et al,22 2014 14 609 51 079 28.6 (28.2-29.0)
Hypotension + Perfusion Abnormalities and/or Vasopressor Therapy
Rangel-Frausto et al,56 1995 51 110 46.4 (37.0-55.7)
Salvo et al,55 1995 27 33 81.8 (68.7-95.0) Forty-four studies report septic
Alberti et al,52 2002 752 1180 63.8 (60.7-67.0) shock–associated mortality5-7,19-59
Hypotension + Vasopressor Therapy and were included in the quantitative
Rodriguez et al,31 2001 129 283 45.6 (39.8-51.4) synthesis using random-effects
Silva et al,48 2004 106 203 52.2 (45.3-59.1) meta-analysis. The Surviving
Laupland et al,49 2005 28 57 49.1 (36.5-61.8) Sepsis Campaign (SSC) database
Vincent et al,43 2006 250 462 54.1 (49.6-58.7)
analyses with similar data are
Karlsson et al,40 2007 90 363 24.8 (20.4-29.2)
reported in 2 studies6,29; therefore,
Sakr et al,39 2007 250 462 54.1 (49.6-58.7)
only one of these was used in the
Kauss et al,34 2010 185 255 72.5 (67.1-78.0)
meta-analysis reported.6 Levy et al
Levy et al,6 2010 915 2494 36.7 (34.8-38.6)
32
Phua et al, 2011 441 939 47.0 (44.3-49.7) report 3 septic shock subsets,6
Ogura et al,20 2014 117 282 41.5 (35.7-47.2) Klein Klowenberg et al report 2
GiViTI database, 2015a 15 935 26 295 60.6 (60.0-61.2) (restrictive and liberal),7 Zahar et al
Hypotension + Vasopressor Therapy + Serum Lactate Level >2 mmol/L report 3 (community-acquired,
Group 1b 3602 8520 42.3 (41.2-43.3) ICU-acquired, and nosocomial
Hypotension + Perfusion Abnormalities + Vasopressor Therapy infection–associated septic shock),30
Lundberg et al,54 1998 19 41 46.3 (31.1-61.6) and Phua et al report 2 groups,32
Levy et al,6 2010 3428 7436 46.1 (45.0-47.2) which were treated as separate
26
Quenot et al, 2013 728 1495 48.7 (46.2-51.2) data points in the meta-analysis.
Hypotension ± Vasopressor Therapy or Metabolic Abnormalities Studies under “consensus definition”
Peake et al,36 2009 75 324 23.1 (18.6-27.7) cite the Sepsis Consensus
Hypotension or Vasopressor Therapy Definitions.1,2 The categorization
Dahmash et al,59 1993 14 36 38.9 (23.0-54.8) used to assess heterogeneity does
McLauchlan et al,58 1995 73 101 72.3 (63.5-81.0) not fully account for septic shock
Pittet et al,57 1995 7 12 58.3 (30.4-86.2) details in individual studies.
Schoenberg et al,53 1998 32 80 40.0 (29.3-50.7)
Engel et al,42 2007 119 190 62.6 (55.8-69.5) SI conversion factor: To convert
Esteban et al,41 2007 27 59 45.8 (33.1-58.5) serum lactate values to mg/dL, divide
Khwannimit and Bhuayanontachai,37 2009 164 303 54.1 (48.5-59.7) by 0.111.
Moore et al,33 2011 22 61 36.1 (24.0-48.1) a
Data obtained from GiViTI database
Zahar et al,30 2011 (community) 215 530 40.6 (36.3-44.8) provided by Bertolini et al
Zahar et al,30 2011 (ICU) 123 232 53.0 (47.1-59.0) (published 20158).
Zahar et al,30 2011 (nosocomial) 233 580 40.2 (36.1-44.2) b
7 The mortality data of Group 1
Klein Klowenberg et al, 2012 29 47 61.7 (47.8-75.6)
Park et al,28 2012 228 740 30.8 (27.5-34.1) patients (new septic shock
Hypotension or Serum Lactate Any Value or Vasopressor Therapy population) and the overall
Liu et al,21 2014 827 2536 32.6 (30.8-34.4) potential septic shock patient
SSC database,16 2016b 6556 18 840 34.8 (34.1-35.5) populations (n = 18 840) described
International Classification of Diseases Codes in the manuscript from the current
Annane et al,51 2003 13 269 26 172 50.7 (50.1-51.3) study using the Surviving SSC
Flaatten,50 2004 457 1562 29.3 (27.1-31.6) database are also included in the
Whittaker et al,24 2013 117 321 36.4 (31.2-41.7) meta-analysis. Septic shock–specific
Serum Lactate Level >4 mmol/L data were obtained from Australian
Levy et al,6 2010 242 811 29.8 (26.7-33.0) & New Zealand Intensive Care
Phua et al,32 2011 219 466 47.0 (42.0-52.0) Society Adult Patient Database
Overall (I2 = 99.5%; P = .000) 46.5 (42.7-50.3) (ANZICS), from a previously
0 20 40 60 80 100 published report.22 This results in
Mortality, % (95% CI) 52 data points for random-effects
meta-analysis.
thermia (>38.3°C), hypothermia (<36°C), chills with rigor, tachyp- (plasma glucose level >120 mg/dL [6.7 mmol/L), platelet count
nea (>20/min), leukopenia (<4000 cells/μL), hyperglycemia <100 ×103/μL, and coagulopathy.
jama.com (Reprinted) JAMA February 23, 2016 Volume 315, Number 8 779
These models were used to estimate acute hospital mor- between resuscitation status and end points to shock diagno-
tality odds ratios (ORs) and adjusted ORs for mortality per-unit sis were seldom reported. The studies differed in the descrip-
increase in the serum lactate level using continuous natural tion of resuscitation, persistent hypotension, and in their
log–transformed serum lactate level. The operating character- vasopressor definitions when using the cardiovascular
istics (sensitivity/specificity over hospital mortality curves; Sequential [Sepsis-related] Organ Failure Assessment (SOFA)
positive and negative predictive values) of different serum score categories.113 Diverse infection and organ dysfunction
lactate cutpoints (2, 3, and 4 mmol/L) were also tested using codes were also used in the International Classification of
the logistic regression model. Multiple imputations (n = 20) Diseases–based derivations.63,70,79,90 Variables highlighted in
were used to assess the statistical effect of missing serum lac- Table 1 and in eTable 2 in the Supplement informed the Del-
tate values. phi survey questions.
P < .05 (2-sided) was considered statistically significant. All The random-effects meta-analysis showed significant
analyses were performed using Stata version 13.1 (StataCorp). heterogeneity in septic shock mortality (mean mortality,
46.5% [95% CI, 42.7%-50.3%], with a near 4-fold variation
from 23.0% to 81.8%; I2 = 99.5%; τ2 = 182.5; and P < .001)
(Figure 2). Statistically significant heterogeneity was also
Results observed in random-effects meta-analysis by clinical criteria
Systematic Review and Meta-analysis reported for septic shock case definition in studies (Table 2).
The systematic review identified 44 studies (166 479 The meta-regression models described could not explain this
patients) reporting septic shock mortality5-7,19-59 from a total heterogeneity (eTable 3A and eTable 3B in the Supplement).
of 92 studies reporting sepsis cohorts between 1987 and
20155-7,19-107 (Figure 1; eTable 2 in the Supplement). Different Delphi Study
shock criteria were used for systolic blood pressure In the first round, informed by the systematic review, 15 task
(<90 mm Hg; <100 mm Hg; decrease >40 mm Hg; or decrease force members (88%) voted to include persistent hypoten-
>50% of baseline value if hypertensive), mean arterial sion, vasopressor therapy, and hyperlactatemia in the
pressure (<70; <65; <60 mm Hg), serum lactate level (>4, updated criteria. There was no agreement on the lower cutoff
>2.5, >2, >1 mmol/L) and base deficit (−5 mmol/L) (Table 1; for serum lactate level in this round. Eleven members (65%)
eTable 2 in the Supplement). Temporal relationships voted that including fluid resuscitation would improve the
780 JAMA February 23, 2016 Volume 315, Number 8 (Reprinted) jama.com
Table 3. Distribution of Septic Shock Cohorts and Crude Mortality From Surviving Sepsis Campaign Database (n = 18 840 patients)
Lactate Mortality,
Category, No. (% of total) Acute Hospital Mortality, χ2 Test Adjusted OR
a
Cohorts mmol/Lb [n = 18 840] No. (%) [95% CI] for Trend (95% CI)c P Valuec
Group 1 (hypotensive after fluids
and vasopressor therapy and serum lactate
levels >2 mmol/L) >2 to ≤3 2453 (13.0) 818 (33.3) [31.5-35.3] <.001 1 [Reference]
>3 to ≤4 1716 (9.1) 621 (36.2) [33.9-38.5]
>4 4351 (23.1) 2163 (49.7) [48.2-51.2]
All 8520 (45.2) 3602 (42.3) [41.2-43.3]
Group 2 (hypotensive after fluids ≤2 3985 (21.2) 1198 (30.1) [28.6-31.5] NAd 0.57 (0.52-0.62) <.001
and vasopressor therapy and serum lactate
levels ≤2 mmol/L)
Group 3 (hypotensive after fluids
and no vasopressors and serum lactate
levels >2 mmol/L) >2 to ≤3 69 (0.4) 15 (21.7) [12.7-33.3] .04 0.65 (0.47-0.90) .009
>3 to ≤4 57 (0.3) 14 (24.6) [14.1-37.8]
>4 97 (0.5) 35 (36.1) [26.6-46.5]
All 223 (1.2) 64 (28.7) [22.9-35.1]
Group 4 (serum lactate levels >2 mmol/L
and no hypotension after fluids
and no vasopressors) >2 to ≤3 860 (4.6) 179 (20.8) [18.1-23.7] <.001 0.71 (0.62-0.82) <.001
>3 to ≤4 550 (2.9) 105 (19.1) [15.9-22.6]
>4 1856 (9.9) 555 (29.9) [27.8-32.0]
All 3266 (17.3) 839 (25.7) [24.2-27.2]
Group 5 (serum lactate levels between
2-4 mmol/L and no hypotension before fluids
and no vasopressors) >2 to ≤3 1624 (8.6) 489 (30.1) [27.9-32.4] NAd 0.77 (0.66-0.90) .001
>3 to ≤4 1072 (5.7) 313 (29.2) [26.5-32.0]
>4 790e
All 2696 (14.3) 802 (29.7) [28.0-31.5]
Group 6 (hypotensive after fluids and no ≤2 150 (0.8) 28 (18.7) [12.8-25.8] NAd 0.32 (0.20-0.51) <.001
vasopressors and serum lactate ≤2 mmol/L)
c
Abbreviations: NA, not available; OR, odds ratio. Refers to the adjusted OR generated using generalized estimating equation
SI conversion factor: To convert serum lactate values to mg/dL, divide by 0.111. regression model (eTable7 in the Supplement).
d
a
Mean arterial pressure less than 65 mm Hg was used to define hypotension. χ2 test for trend could only be performed if there were 3 or more serum
“After fluids” was defined using the field “crystalloids” coded as a binary term lactate categories.
e
within the Surviving Sepsis Campaign database. Excluded from full case analysis.
b 2
Using χ tests, trends in mortality across serum lactate categories within
groups (>2 to ⱕ3 mmol/L; >3 to ⱕ4 mmol/L and >4 mmol/L) were assessed.
criteria. The task force determined that neither a severity (77%) agreed on acute hospital mortality as the outcome vari-
grading for septic shock nor criteria for either adequacy of able. The test variables could be present either alone or in com-
fluid resuscitation or persistent hypotension should be pro- binations, thus identifying 6 potential groups of patients with
posed because of the nonstandardized use of hemodynamic septic shock (Table 3; eTable 5 in the Supplement).
monitoring, resuscitation protocols, and vasopressor dosing Prior to the final round of the Delphi process, all analyses
in clinical practice. (Other results are reported in eTable 4 in from the SSC data set and the EHR data sets were provided.
the Supplement.) These findings generated the new definition—“septic shock is
In Delphi round 2, the task force was provided with a pre- defined as a subset of sepsis in which underlying circulatory,
liminary descriptive analysis from the SSC database. With cellular, and metabolic abnormalities are associated with a
agreement on the description of the septic shock illness con- greater risk of mortality than sepsis alone”—and the clinical
cept, 3 test variables (hypotension after fluid resuscitation, va- criteria described below.
sopressor therapy, and serum lactate level) were agreed on for
predictive validity analyses. The “after fluids” field in the SSC Cohort Studies
database was used as a proxy for resuscitation. The need for SSC Database
vasopressors was agreed as a proxy for persistent hypoten- Patients with serum lactate levels greater than 4 mmol/L who
sion by 95% of the task force. Twelve members (71%) voted that did not receive fluids as recommended by the SSC guidelines111
a minimum vasopressor dose should not be proposed in view (n = 790 [2.8%]) were excluded. Patients without any serum
of the variability in blood pressure targets and resuscitation lactate values measured were excluded initially for full case
protocols identified by the systematic review, and because of analysis (n = 4419 [15.7%]) but were reassessed in the miss-
variable sedation use. Vasopressor therapy was therefore ing data analysis. Of the 22 941 remaining patients, 4101 coded
treated as a binary variable within the analysis. To derive an as having severe sepsis were excluded from this analysis, gen-
optimal cutoff for serum lactate level, 13 task force members erating the analysis set of 18 840 patients who were either hy-
jama.com (Reprinted) JAMA February 23, 2016 Volume 315, Number 8 781
Hypotension was defined as mean arterial pressure less than 65 mm Hg. as “cryptic shock.” Missing serum lactate level measurements (n = 4419 [15.7%])
Vasopressor therapy to maintain mean arterial pressure of 65 mm Hg or higher and patients with serum lactate levels greater than 4 mmol/L (36 mg/dL) who
is treated as a binary variable. Serum lactate level greater than 2 mmol/L (18 did not receive fluids as per SSC guidelines (n = 790 [2.8%]) were excluded
mg/dL) is considered abnormal. The “after fluids” field in the Surviving Sepsis from full case analysis. Of the 22 941 patients, 4101 who were coded as having
Campaign (SSC) database was considered equivalent to adequate fluid severe sepsis were excluded. Thus, the remaining 18 840 patients were
resuscitation. “Before fluids” refers to patients who did not receive fluid categorized within septic shock groups 1 to 6.
a
resuscitation. Serum lactate level greater than 2 mmol/L after fluid resuscitation Patients with screening serum lactate levels coded as greater than 2 mmol/L
but without hypotension or need for vasopressor therapy (group 4) is defined (n=3342) were included in the missing-data analysis.
potensive after fluids or required vasopressors or had a se- Derivation of Serum Lactate Cutoff Value and Missing Data Analysis
rum lactate level measurement (Figure 3 and Table 3). In the generalized estimating equation model (shown in eTable
Hypotension was reported in 83.1%, serum lactate level greater 8 in the Supplement), serum lactate level was associated with
than 2 mmol/L in 78.1%, and receipt of vasopressors in 66.4%. mortality, and the adjusted OR for hospital mortality in-
Overall, crude hospital mortality was 34.7%. Cohort charac- creased linearly with increasing serum lactate level. An in-
teristics by setting are shown in eTable 6 in the Supplement. crease in serum lactate level from 2 to 10 mmol/L increased
the adjusted OR for hospital mortality from 1.4 (95% CI, 1.35-
Predictive Validity of Potential Septic Shock Groups 1.45) to 3.03 (95% CI, 2.68-3.45) (referent lactate = 1; Figure 4).
Of the 6 groups of potential patients with septic shock (Table 3), A serum lactate level greater than 2 mmol/L was chosen as the
the most prevalent was group 1 (hypotension + vasopressor preferred cutoff value for the new septic shock criteria, the ra-
therapy + serum lactate level >2 mmol/L) (n = 8520); followed tionale being the trade-off between highest sensitivity (82.5%
by groups 2 (n = 3985) and 4 (n = 3266). Crude hospital mor- when using the n = 18 840 subset, and 74.9% when using pa-
tality rates in these 3 groups were 42.3%, 30.1%, and 25.7%, re- tients in groups 1 and 2 combined [n = 12 475]), and the deci-
spectively. Statistically significant increasing trends in crude sion from the Delphi process to identify the lowest serum lac-
mortality were observed over increasing serum lactate level cat- tate level independently associated with a greater risk of death
egories within groups (χ2 test of trend: P < .001 for groups 1 and (OR of 1.4 at a lactate value of 2 mmol/L) (Table 4; eTable 9,
4, P = .04 for group 3). The adjusted OR for hospital mortality eFigure 1, and eFigure 2 in the Supplement).
using group 1 for reference was significantly lower in all other Predicated on this understanding of the SSC database struc-
groups (P < .01 for groups 2 to 6), suggesting that group 1 rep- ture and the regression analyses completed (eTable 6, eTable
resents a distinct subpopulation with a significantly greater risk 7, and eTable 8 in the Supplement), we assumed that data were
of death (eTable 7 in the Supplement). By a majority (cumula- missing at random; ie, any difference between observed val-
tive first choice, 72.2%; second choice, 55.6%) (eTable 4 in the ues and missing values did not depend on unobserved data.
Supplement), the task force agreed that group 1 was most con- Complete case analysis was therefore performed, followed by
sistent with the proposed septic shock definition, thus gener- multiple imputation analysis to support the missing-at-
ating the new septic shock criteria. random assumption.114 The ORs for mortality per unit in-
782 JAMA February 23, 2016 Volume 315, Number 8 (Reprinted) jama.com
crease in serum lactate level using complete case analysis MAP of 65 mm Hg or greater and having a serum lactate level
(n = 18 840) and imputed analyses (n = 22 182) were similar greater than 2 mmol/L persisting after fluid resuscitation.
(1.09 [95% CI, 1.08-1.10]; P < .001 vs 1.09 [95% CI, 1.08-1.09]; The proposed definition and criteria of septic shock differ
P < .001, respectively). The imputed and complete case analy- from prior definitions1,2,111 in 2 respects: (1) the need for both a
sis probabilities of hospital mortality were also similar (36.4% serum lactate level and vasopressor-dependent hypotension
and 35.5%, respectively). (ie, cardiovascular SOFA score ≥2) instead of either alone
and (2) a lower serum lactate level cutoff of 2 mmol/L vs
EHR Data Sets
The UPMC and KPNC EHRs included 148 907 and 321 380 adult
patients with suspected infection, respectively (eTable 10 in Figure 4. Serum Lactate Level Analysis
the Supplement). Forty-six percent (n = 5984) of UPMC pa-
5.0
tients and 39% (n = 54 135) of KPNC patients with 1 or more
Table 4. Characteristics of Serum Lactate Level Cutoff Values for Complete Case Analysis and Imputation Analysis Using Surviving Sepsis
Campaign Database
jama.com (Reprinted) JAMA February 23, 2016 Volume 315, Number 8 783
Table 5. Crude Mortality in Septic Shock Groups From UPMC and KPNC Data sets
4 mmol/L as currently used in the SSC definitions. In the new horts) were iterative and concurrent with the consensus
septic shock definition, an increase in serum lactate level is po- process, a significant step forward from previous definitions.
sitioned as a proxy for a cellular metabolic abnormality, and as This study also has several limitations. First, the systematic
a variable independently associated with acute mortality (pre- review did not formally assess study quality and was restricted
dictive validity), which is consistent with the published to MEDLINE publications, adult populations, and observational
literature.115-118 An elevated serum lactate level is not specific studies reporting epidemiology. Second, only the Delphi-derived
for cellular dysfunction in sepsis118,119 but has face validity given variables were tested in multiple data sets to generate the pro-
the lack of a superior yet readily available alternative. This posed septic shock criteria. Other variables, including tissue per-
present study identifies a lower serum lactate level cutoff as fusion markers (eg, base deficit, oliguria, acute alteration in men-
an independent prognostic variable when compared with a re- tation), blood pressure characteristics (eg, diastolic pressure),
cent analysis of the entire SSC database. This disparity is ex- resuscitation end points (eg, central venous saturation, lactate
plained by using a data set of 18 840 patients in the analysis in clearance), and numerous biomarkers reported in the literature,17
this study rather than the total 28 150-patient SSC data set used could potentially improve on the proposed septic shock criteria
by Casserly et al.17 From this subpopulation 6 groups were iden- but were not included. However, operationalizing the definition
tified and analyzed as risk strata within the generalized esti- of septic shock with 3 commonly measured variables should in-
mating equation model and performance-tested for various se- crease both generalizability and clinical utility. Third, the lack of
rum lactate level cutoffs. The group with a significantly greater agoldstandarddiagnosticcriteriaforsepticshock8 precludescom-
risk of death was then selected. In contrast, Casserly et al17 re- parative assessment of these proposed criteria. Fourth, all data
ported the independent relationship of hypotension and se- sets had missing data that could potentially introduce a form of
rum lactate levels with mortality in severe sepsis. selection bias.120 In the primary data set (SSC database) this is-
The 6 potential septic shock patient groups analyzed in this sue was addressed by demonstrating that full case analysis is an
study also provide an explanation for the heterogeneity in sep- appropriate method (see “Derivation of Serum Lactate Cutoff
tic shock mortality highlighted by the meta-analysis. Depend- Value and Missing Data Analysis”). Fifth, serum lactate measure-
ing on the group selected, septic shock mortality ranged from ments are not universally available, especially outside of a criti-
12.8% to 51.2% within the SSC data set and from 7.0% to 64.0% cal care setting or in resource-limited environments. Although
in the EHR data sets. The KPNC EHR data set corroborated the feasibility is a quality indicator for a definition,8 identification of
consistent trends of higher mortality associated with a higher a critically ill patient would generally trigger obtaining a serum
serum lactate level, even in a population with a wider range lactate measurement, both to stratify risk and to monitor the re-
of illness severity captured by more prevalent measurement sponse to treatment.17 Sixth, although the proposed new defini-
of serum lactate levels. tion and clinical criteria for sepsis are arbitrary, these do have pre-
The key strengths of the present study are in the method- dictive validity for mortality, alongside face and content validity.8
ology used to arrive at the new definition and clinical criteria This study represents one step in an ongoing iterative pro-
for septic shock, a clinical syndrome with a range of signs, cess and provides a resourceful structure and a predictive va-
symptoms, and biochemical abnormalities that are not pathog- lidity standard for future investigations in this area. Prospec-
nomonic. Furthermore, the supporting studies (systematic re- tive validation of the clinical criteria may improve on the
view, Delphi process, and analyses of the SSC and EHR co- variables and cutoffs proposed herein, and identification and
784 JAMA February 23, 2016 Volume 315, Number 8 (Reprinted) jama.com
validation of novel markers of organ dysfunction and shock defined as a subset of sepsis in which underlying circula-
may replace lactate level.8 tory, cellular, and metabolic abnormalities are associated
with a greater risk of mortality than sepsis alone. Adult
patients with septic shock can be identified using the clini-
cal criteria of hypotension requiring use of vasopressors to
Conclusions maintain mean blood pressure of 65 mm Hg or greater and
Based on a consensus process using results from a system- having a serum lactate level greater than 2 mmol/L persist-
atic review, surveys, and cohort studies, septic shock is ing after adequate fluid resuscitation.
ARTICLE INFORMATION reported receiving K grant support from the 5. Leligdowicz A, Dodek PM, Norena M, et al.
Author Affiliations: Division of Asthma, Allergy, National Institutes of Health (K23GM112018). Dr Association between source of infection and
and Lung Biology, King’s College London, London, Deutschman reported holding patents on materials hospital mortality in patients who have septic
United Kingdom (Shankar-Hari); Department of not related to this work and receiving travel/ shock. Am J Respir Crit Care Med. 2014;189(10):
Critical Care Medicine, Guy’s and St Thomas’ NHS accommodations and related expenses for 1204-1213.
Foundation Trust, London SE17EH, United Kingdom participation in meetings paid by the Centers for 6. Levy MM, Dellinger RP, Townsend SR, et al.
(Shankar-Hari); The Ohio State University College of Disease Control and Prevention, World Federation The Surviving Sepsis Campaign: results of an
Medicine, Department of Biomedical Informatics, of Societies of Intensive and Critical Care, international guideline-based performance
Center for Biostatistics, Columbus (Phillips); Pennsylvania Assembly of Critical Care Medicine/PA improvement program targeting severe sepsis.
Rhode Island Hospital, Brown University School of Chapter, Society of Critical Care Medicine (SCCM)/ Intensive Care Med. 2010;36(2):222-231.
Medicine, Providence, Rhode Island (Levy); Clinical Penn State–Hershey Medical Center, Society of
Critical Care Medicine, Northern Ireland Society of 7. Klein Klouwenberg PM, Ong DS, Bonten MJ,
Research, Investigation, and Systems Modeling of Cremer OL. Classification of sepsis, severe sepsis
Acute Illness Center, Department of Critical Care Critical Care Medicine, International Sepsis Forum,
Department of Anesthesiology, Stanford University, and septic shock: the impact of minor variations in
and Emergency Medicine, University of Pittsburgh, data capture and definition of SIRS criteria.
Pittsburgh, Pennsylvania (Seymour, Angus); Acute Dialysis Quality Initiative, and European
Society of Intensive Care Medicine (ESICM). Dr Intensive Care Med. 2012;38(5):811-819.
Division of Research, Kaiser Permanente, Oakland,
California (Liu); Department of Pediatrics, Singer reported serving on advisory boards for 8. Shankar-Hari M, Bertolini G, Brunkhorst FM,
Hofstra-North Shore-Long Island Jewish-Hofstra Bayer and Biotest and that he is a recipient of a UK et al. Judging quality of current septic shock
School of Medicine, Steven and Alexandra Cohen NIHR Senior Investigator Fellowship (2009-2017). definitions and criteria. Crit Care. 2015;19(1):445.
Children’s Medical Center, New Hyde Park, No other disclosures were reported. 9. Iwashyna TJ, Govindan S. Did they just prove
New York (Deutschman); Department of Molecular Members of the Sepsis Definitions Task Force that a diagnosis of “septic shock” is meaningless?
Medicine, Hofstra-North Shore-Long Island and Delphi Study: Derek Angus, Djilalli Annane, Am J Respir Crit Care Med. 2014;189(10):1156-1157.
Jewish-Hofstra School of Medicine, Steven and Michael Bauer, Rinaldo Bellomo, Gordon Bernard, 10. Singer M. The role of mitochondrial dysfunction
Alexandra Cohen Children’s Medical Center, Jean-Daniel Chiche, Craig Coopersmith, in sepsis-induced multi-organ failure. Virulence.
New Hyde Park, New York (Deutschman); Feinstein Cliff Deutschman (cochair), Richard Hotchkiss, 2014;5(1):66-72.
Institute for Medical Research, Manhasset, Mitchell Levy, John Marshall, Greg Martin,
New York (Deutschman); Associate Editor, JAMA Steve Opal, Gordon Rubenfeld, Christopher 11. Hotchkiss RS, Monneret G, Payen D.
(Angus); Interdepartmental Division of Critical Care Seymour (co-opted), Manu Shankar-Hari Sepsis-induced immunosuppression. Nat Rev
Medicine, University of Toronto, Toronto, Ontario, (co-opted), Mervyn Singer (cochair), Immunol. 2013;13(12):862-874.
Canada (Rubenfeld); Sunnybrook Health Sciences Tom van der Poll, Jean-Louis Vincent. 12. Takasu O, Gaut JP, Watanabe E, et al.
Centre, Toronto, Ontario, Canada (Rubenfeld); Funding/Support: The Sepsis Definitions Task Mechanisms of cardiac and renal dysfunction in
Bloomsbury Institute of Intensive Care Medicine, Force received unrestricted funding support from patients dying of sepsis. Am J Respir Crit Care Med.
University College London, London, the European Society of Intensive Care Medicine 2013;187(5):509-517.
United Kingdom (Singer). and the Society of Critical Care Medicine 13. Rudiger A, Dyson A, Felsmann K, et al.
Author Contributions: Dr Shankar-Hari had full (sponsors). Early functional and transcriptomic changes in the
access to all of the data in the study and takes Role of the Funders/Sponsors: The funders/ myocardium predict outcome in a long-term rat
responsibility for the integrity of the data and the sponsors had no role in the design and conduct of model of sepsis. Clin Sci (Lond). 2013;124(6):391-401.
accuracy of the data analysis. the study; collection, management, analysis, and 14. Singer M, Deutschman C, Seymour CW, et al.
Study concept and design: Shankar-Hari, Levy, interpretation of the data; preparation, review, or The Third International Consensus Definitions for
Deutschman, Angus, Rubenfeld, Singer. approval of the manuscript; and decision to submit Sepsis and Septic Shock (Sepsis-3). JAMA.
Acquisition, analysis, or interpretation of data: the manuscript for publication. doi:10.1001/jama.2016.0287.
Shankar-Hari, Phillips, Levy, Seymour, Liu, Singer.
Drafting of the manuscript: Shankar-Hari, Phillips, Disclaimer: Dr Angus, JAMA Associate Editor, had 15. US Department of Health and Human Services
Levy, Deutschman, Angus, Singer. no role in the evaluation of or decision to publish (DHHS). Frequently Asked Questions About Human
Critical revision of the manuscript for important this article. Research. DHHS website. http://www.hhs.gov
intellectual content: All authors. /ohrp/policy/faq/index.html. May 10, 2010.
Statistical analysis: Shankar-Hari, Phillips, REFERENCES Accessed Feburary 1, 2016.
Seymour, Liu. 1. Bone RC, Balk RA, Cerra FB, et al. Definitions for 16. Seymour CW, Liu V, Iwashyna TJ et al.
Obtained funding: Levy. sepsis and organ failure and guidelines for the use Assessment of clinical criteria for sepsis: for the
Administrative, technical, or material support: of innovative therapies in sepsis. Chest. 1992;101 Third International Consensus Definitions for Sepsis
Shankar-Hari, Levy, Deutschman, Angus. (6):1644-1655. and Septic Shock (Sepsis-3). JAMA. doi:10.1001
Study supervision: Seymour, Deutschman, Singer. 2. Levy MM, Fink MP, Marshall JC, et al. /jama.2016.0288.
Conflict of Interest Disclosures: All authors have 2001 SCCM/ESICM/ACCP/ATS/SIS International 17. Casserly B, Phillips GS, Schorr C, et al.
completed and submitted the ICMJE Form for Sepsis Definitions Conference. Crit Care Med. 2003; Lactate measurements in sepsis-induced tissue
Disclosure of Potential Conflicts of Interest. Drs 31(4):1250-1256. hypoperfusion. Crit Care Med. 2015;43(3):567-573.
Shankar-Hari and Singer reported receiving support 3. Rhee C, Gohil S, Klompas M. Regulatory 18. World Health Organisation (WHO). Health
from the UK NIHR Biomedical Research Centre mandates for sepsis care—reasons for caution. Statistics and Information Systems: Definition of
schemes. Dr Levy reported receiving a grant from N Engl J Med. 2014;370(18):1673-1676. Region Groupings. WHO website. http://www.who
ImmuneExpress. Dr Seymour reported receiving .int/healthinfo/global_burden_disease/definition
personal fees from Beckman Coulter. Dr Liu 4. Iwashyna TJ, Angus DC. Declining case fatality
rates for severe sepsis. JAMA. 2014;311(13):1295-1297. _regions/en/. July 5, 2015.
jama.com (Reprinted) JAMA February 23, 2016 Volume 315, Number 8 785
19. Ortíz G, Dueñas C, Rodríguez F, et al. 36. Peake SL, Bailey M, Bellomo R, et al. 54. Lundberg JS, Perl TM, Wiblin T, et al.
Epidemiology of sepsis in Colombian intensive care Australasian resuscitation of sepsis evaluation Septic shock: an analysis of outcomes for patients
units. Biomedica. 2014;34(1):40-47. (ARISE). Resuscitation. 2009;80(7):811-818. with onset on hospital wards versus intensive care
20. Ogura H, Gando S, Saitoh D, et al. 37. Khwannimit B, Bhurayanontachai R. units. Crit Care Med. 1998;26(6):1020-1024.
Epidemiology of severe sepsis in Japanese The epidemiology of, and risk factors for, mortality 55. Salvo I, de Cian W, Musicco M, et al. The Italian
intensive care units. J Infect Chemother. 2014;20 from severe sepsis and septic shock in a SEPSIS study: preliminary results on the incidence
(3):157-162. tertiary-care university hospital setting. Epidemiol and evolution of SIRS, sepsis, severe sepsis and
21. Liu V, Escobar GJ, Greene JD, et al. Hospital Infect. 2009;137(9):1333-1341. septic shock. Intensive Care Med. 1995;21(suppl 2):
deaths in patients with sepsis from 2 independent 38. Angkasekwinai N, Rattanaumpawan P, S244-S249.
cohorts. JAMA. 2014;312(1):90-92. Thamlikitkul V. Epidemiology of sepsis in Siriraj 56. Rangel-Frausto MS, Pittet D, Costigan M,
22. Kaukonen KM, Bailey M, Suzuki S, Pilcher D, Hospital 2007. J Med Assoc Thai. 2009;92(suppl 2): Hwang T, Davis CS, Wenzel RP. The natural history
Bellomo R. Mortality related to severe sepsis and S68-S78. of the systemic inflammatory response syndrome
septic shock among critically ill patients in Australia 39. Sakr Y, Vincent JL, Schuerholz T, et al. (SIRS). JAMA. 1995;273(2):117-123.
and New Zealand, 2000-2012. JAMA. 2014;311 Early- versus late-onset shock in European 57. Pittet D, Rangel-Frausto S, Li N, et al.
(13):1308-1316. intensive care units. Shock. 2007;28(6):636-643. Systemic inflammatory response syndrome, sepsis,
23. Gonçalves-Pereira J, Pereira JM, Ribeiro O, et al. 40. Karlsson S, Varpula M, Ruokonen E, et al. severe sepsis and septic shock. Intensive Care Med.
Impact of infection on admission and of the process Incidence, treatment, and outcome of severe sepsis 1995;21(4):302-309.
of care on mortality of patients admitted to the in ICU-treated adults in Finland: the Finnsepsis 58. McLauchlan GJ, Anderson ID, Grant IS,
intensive care unit: the INFAUCI study. Clin study. Intensive Care Med. 2007;33(3):435-443. Fearon KC. Outcome of patients with abdominal
Microbiol Infect. 2014;20(12):1308-1315. 41. Esteban A, Frutos-Vivar F, Ferguson ND, et al. sepsis treated in an intensive care unit. Br J Surg.
24. Whittaker SA, Mikkelsen ME, Gaieski DF, Koshy Sepsis incidence and outcome: contrasting the 1995;82(4):524-529.
S, Kean C, Fuchs BD. Severe sepsis cohorts derived intensive care unit with the hospital ward. Crit Care 59. Dahmash NS, Chowdhury NH, Fayed DF.
from claims-based strategies appear to be biased Med. 2007;35(5):1284-1289. Septic shock in critically ill patients. J Infect. 1993;
toward a more severely ill patient population. Crit 42. Engel C, Brunkhorst FM, Bone HG, et al. 26(2):159-170.
Care Med. 2013;41(4):945-953. Epidemiology of sepsis in Germany. Intensive Care 60. Whittaker SA, Fuchs BD, Gaieski DF, et al.
25. Sakr Y, Elia C, Mascia L, et al. Epidemiology and Med. 2007;33(4):606-618. Epidemiology and outcomes in patients with severe
outcome of sepsis syndromes in Italian ICUs. 43. Vincent JL, Sakr Y, Sprung CL, et al. Sepsis in sepsis admitted to the hospital wards. J Crit Care.
Minerva Anestesiol. 2013;79(9):993-1002. European intensive care units: results of the SOAP 2015;30(1):78-84.
26. Quenot JP, Binquet C, Kara F, et al. study. Crit Care Med. 2006;34(2):344-353. 61. Cross G, Bilgrami I, Eastwood G, et al.
The epidemiology of septic shock in French 44. Shapiro N, Howell MD, Bates DW, Angus DC, The epidemiology of sepsis during rapid response
intensive care units. Crit Care. 2013;17(2):R65. Ngo L, Talmor D. The association of sepsis team reviews in a teaching hospital. Anaesth
27. Nesseler N, Defontaine A, Launey Y, Morcet J, syndrome and organ dysfunction with mortality in Intensive Care. 2015;43(2):193-198.
Mallédant Y, Seguin P. Long-term mortality and emergency department patients with suspected 62. Vincent JL, Marshall JC, Namendys-Silva SA,
quality of life after septic shock. Intensive Care Med. infection. Ann Emerg Med. 2006;48(5):583-590. et al; ICON Investigators. Assessment of the
2013;39(5):881-888. 45. Gasparović V, Gornik I, Ivanović D. worldwide burden of critical illness. Lancet Respir
28. Park DW, Chun BC, Kim JM, et al. Sepsis syndrome in Croatian intensive care units: Med. 2014;2(5):380-386.
Epidemiological and clinical characteristics of piloting a national comparative clinical database. 63. Stevenson EK, Rubenstein AR, Radin GT,
community-acquired severe sepsis and septic Croat Med J. 2006;47(3):404-409. Wiener RS, Walkey AJ. Two decades of mortality
shock. J Korean Med Sci. 2012;27(11):1308-1314. 46. Degoricija V, Sharma M, Legac A, et al. trends among patients with severe sepsis. Crit Care
29. Levy MM, Artigas A, Phillips GS, et al. Survival analysis of 314 episodes of sepsis in Med. 2014;42(3):625-631.
Outcomes of the Surviving Sepsis Campaign in medical intensive care unit in university hospital. 64. Koupetori M, Retsas T, Antonakos N, et al;
intensive care units in the USA and Europe. Lancet Croat Med J. 2006;47(3):385-397. Hellenic Sepsis Study Group. Bloodstream
Infect Dis. 2012;12(12):919-924. 47. Laupland KB, Zygun DA, Doig CJ, et al. infections and sepsis in Greece. BMC Infect Dis.
30. Zahar JR, Timsit JF, Garrouste-Orgeas M, et al. One-year mortality of bloodstream 2014;14:272.
Outcomes in severe sepsis and patients with septic infection-associated sepsis and septic shock among 65. Bouza C, López-Cuadrado T, Saz-Parkinson Z,
shock [published correction appears in Crit Care patients presenting to a regional critical care Amate-Blanco JM. Epidemiology and recent trends
Med. 2011;39(10):2392]. Crit Care Med. 2011;39(8): system. Intensive Care Med. 2005;31(2):213-219. of severe sepsis in Spain. BMC Infect Dis. 2014;14:
1886-1895. 48. Silva E, Pedro MdeA, Sogayar AC, et al. 3863.
31. Rodríguez F, Barrera L, De La Rosa G, et al. Brazilian Sepsis Epidemiological Study (BASES 66. Storgaard M, Hallas J, Gahrn-Hansen B, et al.
The epidemiology of sepsis in Colombia: study). Crit Care. 2004;8(4):R251-R260. Short- and long-term mortality in patients with
a prospective multicenter cohort study in ten 49. Laupland KB, Davies HD, Church DL, et al. community-acquired severe sepsis and septic
university hospitals. Crit Care Med. 2011;39(7):1675- Bloodstream infection-associated sepsis and septic shock. Scand J Infect Dis. 2013;45(8):577-583.
1682. shock in critically ill adults. Infection. 2004;32(2): 67. Stiermaier T, Herkner H, Tobudic S, et al.
32. Phua J, Koh Y, Du B, et al. Management of 59-64. Incidence and long-term outcome of sepsis on
severe sepsis in patients admitted to Asian 50. Flaatten H. Epidemiology of sepsis in Norway general wards and in an ICU at the General Hospital
intensive care units. BMJ. 2011;342:d3245. in 1999. Crit Care. 2004;8(4):R180-R184. of Vienna: an observational cohort study. Wien Klin
33. Moore LJ, McKinley BA, Turner KL, et al. Wochenschr. 2013;125(11-12):302-308.
51. Annane D, Aegerter P, Jars-Guincestre MC,
The epidemiology of sepsis in general surgery Guidet B; CUB-Réa Network. Current epidemiology 68. Rohde JM, Odden AJ, Bonham C, et al.
patients. J Trauma. 2011;70(3):672-680. of septic shock: the CUB-Réa Network. Am J Respir The epidemiology of acute organ system
34. Kauss IA, Grion CM, Cardoso LT, et al. Crit Care Med. 2003;168(2):165-172. dysfunction from severe sepsis outside of the
The epidemiology of sepsis in a Brazilian teaching intensive care unit. J Hosp Med. 2013;8(5):243-247.
52. Alberti C, Brun-Buisson C, Burchardi H, et al.
hospital. Braz J Infect Dis. 2010;14(3):264-270. Epidemiology of sepsis and infection in ICU patients 69. Gray A, Ward K, Lees F, et al. The epidemiology
35. Póvoa PR, Carneiro AH, Ribeiro OS, Pereira AC; from an international multicentre cohort study. of adults with severe sepsis and septic shock in
Portuguese Community-Acquired Sepsis Study Intensive Care Med. 2002;28(2):108-121. Scottish emergency departments. Emerg Med J.
Group. Influence of vasopressor agent in septic 2013;30(5):397-401.
53. Schoenberg MH, Weiss M, Radermacher P.
shock mortality. Crit Care Med. 2009;37(2):410-416. Outcome of patients with sepsis and septic shock 70. Gaieski DF, Edwards JM, Kallan MJ, Carr BG.
after ICU treatment. Langenbecks Arch Surg. 1998; Benchmarking the incidence and mortality of
383(1):44-48.
786 JAMA February 23, 2016 Volume 315, Number 8 (Reprinted) jama.com
severe sepsis in the United States. Crit Care Med. Intensivos. Incidence, organ dysfunction and severe sepsis in the United States: analysis of
2013;41(5):1167-1174. mortality in severe sepsis: a Spanish multicentre incidence, outcome, and associated costs of care.
71. Czupryna P, Garkowski A, Moniuszko A, et al. study. Crit Care. 2008;12(6):R158. Crit Care Med. 2001;29(7):1303-1310.
Patients with sepsis in infectious diseases 88. Beovic B, Hladnik Z, Pozenel P, Siuka D; 104. Wichmann MW, Inthorn D, Andress HJ,
department in years 1997-2010—epidemiology and Slovenian Severe Sepsis Study Group. Schildberg FW. Incidence and mortality of severe
clinical features. Przegl Epidemiol. 2013;67(3):429- Epidemiology of severe sepsis in Slovenian sepsis in surgical intensive care patients. Intensive
434. intensive care units for adults. J Chemother. 2008; Care Med. 2000;26(2):167-172.
72. Seymour CW, Rea TD, Kahn JM, Walkey AJ, 20(1):134-136. 105. Sands KE, Bates DW, Lanken PN, et al.
Yealy DM, Angus DC. Severe sepsis in pre-hospital 89. Andreu Ballester JC, Ballester F, González Epidemiology of sepsis syndrome in 8 academic
emergency care. Am J Respir Crit Care Med. 2012; Sánchez A, et al. Epidemiology of sepsis in the medical centers. JAMA. 1997;278(3):234-240.
186(12):1264-1271. Valencian Community (Spain), 1995-2004. Infect 106. Sasse KC, Nauenberg E, Long A, Anton B,
73. Sancho S, Artero A, Zaragoza R, et al. Impact of Control Hosp Epidemiol. 2008;29(7):630-634. Tucker HJ, Hu TW. Long-term survival after
nosocomial polymicrobial bloodstream infections 90. Wang HE, Shapiro NI, Angus DC, Yealy DM. intensive care unit admission with sepsis. Crit Care
on the outcome in critically ill patients. Eur J Clin National estimates of severe sepsis in United States Med. 1995;23(6):1040-1047.
Microbiol Infect Dis. 2012;31(8):1791-1796. emergency departments. Crit Care Med. 2007;35 107. Brun-Buisson C, Doyon F, Carlet J, et al;
74. Lagu T, Rothberg MB, Shieh MS, et al. (8):1928-1936. French ICU Group for Severe Sepsis. Incidence, risk
Hospitalizations, costs, and outcomes of severe 91. Cheng B, Xie G, Yao S, et al. Epidemiology of factors, and outcome of severe sepsis and septic
sepsis in the United States 2003 to 2007. Crit Care severe sepsis in critically ill surgical patients in ten shock in adults. JAMA. 1995;274(12):968-974.
Med. 2012;40(3):754-761. university hospitals in China. Crit Care Med. 2007; 108. Mayr FB, Yende S, Angus DC. Epidemiology of
75. Bastani A, Galens S, Rocchini A, et al. 35(11):2538-2546. severe sepsis. Virulence. 2014;5(1):4-11.
ED identification of patients with severe 92. Tanriover MD, Guven GS, Sen D, Unal S, 109. Moss M, Martin GS. A global perspective on
sepsis/septic shock decreases mortality in a Uzun O. Epidemiology and outcome of sepsis in a the epidemiology of sepsis. Intensive Care Med.
community hospital. Am J Emerg Med. 2012;30(8): tertiary-care hospital in a developing country. 2004;30(4):527-529.
1561-1566. Epidemiol Infect. 2006;134(2):315-322.
110. Linde-Zwirble WT, Angus DC. Severe sepsis
76. Vesteinsdottir E, Karason S, Sigurdsson SE, 93. Strehlow MC, Emond SD, Shapiro NI, Pelletier epidemiology: sampling, selection, and society. Crit
et al. Severe sepsis and septic shock: a prospective AJ, Camargo CA Jr. National study of emergency Care. 2004;8(4):222-226.
population-based study in Icelandic intensive care department visits for sepsis, 1992 to 2001. Ann
units. Acta Anaesthesiol Scand. 2011;55(6):722-731. Emerg Med. 2006;48(3):326-331. 111. Dellinger RP, Levy MM, Rhodes A, et al.
Surviving Sepsis Campaign: international guidelines
77. Rosado V, Pérez L, Guerra H, et al. Outcomes 94. Harrison DA, Welch CA, Eddleston JM. for management of severe sepsis and septic shock:
associated with conventional management of The epidemiology of severe sepsis in England, 2012. Crit Care Med. 2013;41(2):580-637.
severe sepsis at Damas Hospital. Bol Asoc Med P R. Wales and Northern Ireland, 1996 to 2004. Crit Care.
2011;103(2):35-38. 2006;10(2):R42. 112. Bernard GR, Vincent JL, Laterre PF, et al;
Recombinant human protein C Worldwide
78. Davis JS, Cheng AC, McMillan M, Humphrey 95. Záhorec R, Firment J, Straková J, et al. Evaluation in Severe Sepsis (PROWESS) Study
AB, Stephens DP, Anstey NM. Sepsis in the tropical Epidemiology of severe sepsis in intensive care Group. Efficacy and safety of recombinant human
Top End of Australia’s Northern Territory. Med J Aust. units in the Slovak Republic. Infection. 2005;33(3): activated protein C for severe sepsis. N Engl J Med.
2011;194(10):519-524. 122-128. 2001;344(10):699-709.
79. Wilhelms SB, Huss FR, Granath G, Sjöberg F. 96. Sundararajan V, Macisaac CM, Presneill JJ, 113. Vincent JL, Moreno R, Takala J, et al; Working
Assessment of incidence of severe sepsis in Sweden Cade JF, Visvanathan K. Epidemiology of sepsis in Group on Sepsis-Related Problems of the European
using different ways of abstracting International Victoria, Australia. Crit Care Med. 2005;33(1):71-80. Society of Intensive Care Medicine. The SOFA
Classification of Diseases codes. Crit Care Med. 97. Adrie C, Alberti C, Chaix-Couturier C, et al. (Sepsis-related Organ Failure Assessment) score to
2010;38(6):1442-1449. Epidemiology and economic evaluation of severe describe organ dysfunction/failure. Intensive Care
80. Shen HN, Lu CL, Yang HH. Epidemiologic trend sepsis in France. J Crit Care. 2005;20(1):46-58. Med. 1996;22(7):707-710.
of severe sepsis in Taiwan from 1997 through 2006. 98. van Gestel A, Bakker J, Veraart CP, 114. Li P, Stuart EA, Allison DB. Multiple imputation:
Chest. 2010;138(2):298-304. van Hout BA. Prevalence and incidence of severe a flexible tool for handling missing data. JAMA.
81. Husak L, Marcuzzi A, Herring J, et al. National sepsis in Dutch intensive care units. Crit Care. 2015;314(18):1966-1967.
analysis of sepsis hospitalizations and factors 2004;8(4):R153-R162. 115. Mikkelsen ME, Miltiades AN, Gaieski DF, et al.
contributing to sepsis in-hospital mortality in 99. Finfer S, Bellomo R, Lipman J, French C, Dobb Serum lactate is associated with mortality in severe
Canada. Healthc Q. 2010;13(Spec No):35-41. G, Myburgh J. Adult-population incidence of severe sepsis independent of organ failure and shock. Crit
82. Bateman BT, Schmidt U, Berman MF, sepsis in Australian and New Zealand intensive care Care Med. 2009;37(5):1670-1677.
Bittner EA. Temporal trends in the epidemiology of units. Intensive Care Med. 2004;30(4):589-596. 116. Vincent JL, Ince C, Bakker J. Clinical review:
severe postoperative sepsis after elective surgery. 100. Brun-Buisson C, Meshaka P, Pinton P, Vallet B; Circulatory shock—an update: a tribute to Professor
Anesthesiology. 2010;112(4):917-925. EPISEPSIS Study Group. EPISEPSIS: a reappraisal of Max Harry Weil. Crit Care. 2012;16(6):239.
83. Vogel TR, Dombrovskiy VY, Lowry SF. Trends in the epidemiology and outcome of severe sepsis in 117. Nichol AD, Egi M, Pettila V, et al. Relative
postoperative sepsis: are we improving outcomes? French intensive care units. Intensive Care Med. hyperlactatemia and hospital mortality in critically
Surg Infect (Larchmt). 2009;10(1):71-78. 2004;30(4):580-588. ill patients. Crit Care. 2010;14(1):R25.
84. Beale R, Reinhart K, Brunkhorst FM, et al. 101. Padkin A, Goldfrad C, Brady AR, Young D, 118. Levy B, Gibot S, Franck P, Cravoisy A,
Promoting Global Research Excellence in Severe Black N, Rowan K. Epidemiology of severe sepsis Bollaert PE. Relation between muscle Na+K+
Sepsis (PROGRESS): lessons from an international occurring in the first 24 hrs in intensive care units in ATPase activity and raised lactate concentrations in
sepsis registry. Infection. 2009;37(3):222-232. England, Wales, and Northern Ireland. Crit Care Med. septic shock. Lancet. 2005;365(9462):871-875.
85. Baharoon S, Al-Jahdali H, Al Hashmi J, Memish 2003;31(9):2332-2338.
119. Garcia-Alvarez M, Marik P, Bellomo R.
ZA, Ahmed QA. Severe sepsis and septic shock at 102. Martin GS, Mannino DM, Eaton S, Moss M. Sepsis-associated hyperlactatemia. Crit Care. 2014;
the Hajj. Travel Med Infect Dis. 2009;7(4):247-252. The epidemiology of sepsis in the United States 18(5):503.
86. Rezende E, Silva JM Jr, Isola AM, et al. from 1979 through 2000. N Engl J Med. 2003;348
(16):1546-1554. 120. Grimes DA, Schulz KF. Bias and causal
Epidemiology of severe sepsis in the emergency associations in observational research. Lancet.
department and difficulties in the initial assistance. 103. Angus DC, Linde-Zwirble WT, Lidicker J, 2002;359(9302):248-252.
Clinics (Sao Paulo). 2008;63(4):457-464. Clermont G, Carcillo J, Pinsky MR. Epidemiology of
87. Blanco J, Muriel-Bombín A, Sagredo V, et al;
Grupo de Estudios y Análisis en Cuidados
jama.com (Reprinted) JAMA February 23, 2016 Volume 315, Number 8 787
Shankar-Hari M, Phillips GS, Levy ML, et al; Sepsis Definitions Task Force. Developing
a new definition and assessing new clinical criteria for septic shock: For the Third
International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). JAMA.
doi:10.1001/jama.2016.0289.
This supplementary material was provided by the authors to give readers additional
information about their work.
Study outcome(s)
The primary outcome was to explore the relationship between criteria and reported
mortality. The secondary outcomes were the different criteria used in septic shock case
definitions reported in the literature.
Search strategy
The search involved the Medline database, between 01/01/1992 and 12/25/2015. The first
author of the paper (MSH) conducted the literature search, with input from co-authors
and members of the Task Force. The search strategy including keywords, exploded fields
and limits are presented in eTable-1. Eligible studies were identified with title and
abstract screening. Eligibility criteria for full text review were reporting of sepsis
epidemiology. All the reference lists of studies eligible for full text review were hand
searched for additional files. In addition, the references lists of systematic review5,
editorials, and review article files identified by this search were also checked and
included for full text assessment.
Statistics
For meta-analysis and meta-regression, only studies reporting septic shock mortality were
included. A random effects mata-analysis of septic shock mortality, by definitional
groups was completed. In addition, we included septic shock specific data from ANZICS
[provided by Bellomo R et al]; GiViTI [provided by Bertolini G et al]; Cub-REA
[provided by Aegerter P, Guidet B, Annane D for the Cub-REA network] databases.4 In
the meta-analysis, data from the current paper is also included - SSC refers to n=18,840
patients included in the study and SSC-group 1 refers to the group of patients meeting the
new septic shock criteria. In one study, the sepsis definition could not be ascertained16. In
studies where septic shock patients were reported using multiple criteria, this data is
presented with the criteria in parenthesis if separate outcomes could be ascertained. When
two authors had the same surname or when multiple papers from same authors are
presented, the publication year in parenthesis is reported. Details are provided in eTable2
study summaries.
From the total population of sepsis patients reported, the number of septic shock patients
and the corresponding number of deaths were calculated from the studies. This data was
then used to generate the proportion and standard error of proportions. Random effects
meta-analysis was done by grouping definitions into either ‘Consensus’ definitions6,7 or
‘other’ definitions [ICD based derivations1-3,17,18; APACHE codes19, based on previous
clinical trials20,21] and by septic shock criteria as described above. Two meta-regression
models were generated to address statistical heterogeneity in the septic shock mortality,
which involved recoding of variables as described. Statistical heterogeneity was assessed
using I2 and Tau2 statistics. Meta analysis and meta regression were done using ‘Stata
13.1 (StataCorp, College Station, Texas).
In the original 1992 version, septic shock was explicitly considered as a subset of severe
sepsis and defined as “sepsis-induced hypotension, persisting despite adequate fluid
resuscitation, along with the presence of hypoperfusion abnormalities or organ
dysfunction” 6. Sepsis-induced hypotension was defined “by the presence of a systolic BP
<90 mm Hg or its reduction by ≥40 mmHg from the baseline, in the absence of other
causes of hypotension (e.g. cardiogenic shock).”Septic shock was re-defined by the 2001
International Sepsis Definitions Conference as “a state of acute circulatory failure
characterized by persistent arterial hypotension unexplained by other causes.” For
adults, hypotension was defined as “a systolic arterial pressure (SAP) <90 mm Hg, a
mean arterial pressure (MAP) <60, or a reduction in SAP of >40 mm Hg from baseline,
despite adequate volume resuscitation, in the absence of other causes for hypotension”.
No definition was provided to characterize ‘adequate’ fluid resuscitation. The 2012
Surviving Sepsis Campaign (SSC) 22 guidelines modified this definition so that septic
shock became “sepsis-induced hypotension persisting despite adequate fluid
resuscitation.” The SSC defined sepsis-induced hypotension as either a SAP <90 mmHg,
or MAP <70 mmHg, or a SBP decrease >40 mmHg or <2 standard deviations below
normal for age in the absence of other causes of hypotension. The SSC also separately
defined “sepsis-induced tissue hypoperfusion as infection-induced hypotension, elevated
lactate, or oliguria”.
Observational studies have used different criteria to identify sepsis and septic shock
(highlighted in Table 1 main manuscript) that, in sum, can be classified as the presence of
infection (presumed or confirmed) in conjunction with varying combinations of:
(i) hypotension (SAP <90 mmHg OR MAP <60 or <70 mmHg OR fall in SAP
pressure >40 mmHg from baseline) …
AND/OR
(ii) … that persists despite ‘adequate fluid resuscitation’ (either unspecified OR after
challenges of 20 ml/kg OR 1000 ml)
AND/OR
(iii)abnormal biochemical variables (e.g. lactate >2 or >4 mmol/L or base deficit >5
mmol/l)
AND/OR
(iv) use of inotropes and/or vasopressors [not necessarily above a pre-specified dose]
AND/OR
(v) new onset organ dysfunction (defined variably using various scoring systems such
as APACHE II, APACHE III, or the cardiovascular component of the SOFA
score).
Please refer to main manuscript for results from random effects meta-analyses. eTables
3a and 3b for meta-regression results and eFigures1 and eFigures2 for additional results.
Introduction:
The 2001 International Sepsis Definitions Conference defined septic shock as “a state of
acute circulatory failure characterized by persistent arterial hypotension unexplained by
other causes.” Within this definition [in adults], hypotension was defined as “a systolic
arterial pressure (SAP) <90 mm Hg, a mean arterial pressure (MAP) <60, or a
reduction in SAP of >40 mm Hg from baseline, despite adequate volume resuscitation, in
the absence of other causes for hypotension”. The MAP definitions in this document
differ from those used by both the Surviving Sepsis Campaign (<70mmHg) and the
SOFA score (also MAP <70mmHg).
Secondly, the utility and cut-off values for lactate may need to be clarified within a new
septic shock definition. There is an obvious disconnect between the high ED department
lactates yet relatively low mortalities recorded in ARISE and ProCESS, and the much
lower lactates and much higher mortalities seen in recent ICU patient studies (e.g. the
Scandinavian TRISS and Italian ALBIOS studies).
Thirdly, other parameters used to qualify the definition of hypotension are currently
either not considered, stated, or are somewhat variable, namely (a) adequacy of fluid
resuscitation and (b) duration of hypotension.
Fourth, the use of vasopressors, ± an accompanying (variably defined) low blood
pressure definition, are also used by some to define shock.
Aim
To assess current consensus on the following domains:
Domain 1: Components of the septic shock definition
Domain 2: Hypotension, persisting hypotension, adequacy of resuscitation
Domain 3: Use of vasopressors
Domain 4: Use of lactate
Domain 5: Severity of septic shock
Your feedback will be used to inform the variables required for analyses and to support
the new definition.
B. PERSISTING HYPOTENSION
C. ADEQUACY OF RESUSCITATION
Q15: If you ticked ‘yes’ to ‘Ongoing need for vasopressors after predefined fluid
bolus’ in Q13 above, please state what amount of fluid you would give (ml, ml/kg)
Q16. In your opinion, should the need for vasopressor therapy be used as a variable
to define septic shock?
ANSWER: Likert scale agreement 5 point: Agree
Q17. If you agree/strongly agree to Q16, should a minimum dose of vasopressor be
stated?
ANSWER: Likert scale agreement 5 point: Disagree
Q18: If you ticked ‘agree/strongly agree’ to ‘minimum dose’ in Q17 above, please
state what dose level you would use
If you would to us to consider other aspects of vasopressor use, please highlight in the
free text section.
Q21. In your opinion, is there a need for a severity grading of septic shock?
ANSWER: Likert scale agreement 5 point: Nether agree nor disagree
Q22. If you agree/strongly agree to Q21, what variables would you like to see within
the definition of shock severity (please tick all you feel should be included)?
Vasopressor dose
Lactate level
Other [Please specify]
Q23. If you agree/strongly agree to Q21, how many points should be there on the
severity scale (assuming 0 = no septic shock)
1
2
3
4
Q24: If you answered ‘yes’ to vasopressor dose in Q22, how should dose of
vasopressor be defined?
- Spot dose of vasopressor
- Average dose of vasopressor over a defined period
o if ‘yes’ please specify how long (hours)
- Peak dose of vasopressor over a defined period
- Other [Please specify]
Q25: If you answered ‘yes’ to lactate in Q22, how should the lactate level be used as
a variable in defining the severity of septic shock?
- Spot level of lactate (when BP ± vasopressor dose recorded)
- Average level of lactate over a defined period
o if ‘yes’ please specify how long (hours)
- Peak level of lactate over a defined period
- Other way of looking at lactate [Please specify]
Q26. Should a composite score be used, taking into account both the BP target and
the vasopressor dose being given? (thus, if aiming for a MAP of 75 mmHg in a
particular patient, much more vasopressor would be needed compared to a target
MAP of 60 mmHg)
ANSWER: Likert scale agreement 5 point: Nether agree nor disagree
If you would to us to consider other variables in defining septic shock, please highlight in
the free text section below.
SURVEY 2
Conclusions from the first Septic Shock definition survey:
• 88% of taskforce (TF) members wanted persistent hypotension, requirement for
vasopressors and raised lactate to be included in the updated definition.
Q2. “The TF members could not agree on endpoint[s] that define adequate
resuscitation in patients with septic shock. As resuscitation is an iterative process this
should be separated from the concept of persistent hypotension.”
Would you agree with this statement regarding adequacy of resuscitation?
ANSWER: Likert scale agreement 5 point: Agree
Please make any comments in the free text box below
Q4. Approximately half the TF voted for including lactate in the definition, wanted
a cut-off value to be determined by mining ‘big data’ to determine
sensitivity/specificity of acute hospital mortality? Are you happy with this proposal?
ANSWER: Likert scale agreement 3 point: Agree
Please comment in the free text box below.
Data collection
Data were entered into the SSC database locally at individual hospitals into pre-
established, non-modifiable fields. Data stripped of private health information were
submitted every 30 days to the secure master server at the Society of Critical Care
Medicine (Mount Prospect, Ill.).
Whittaker 1853 2005 and 2009 USA Single ICU and Bone6 and Levy7 SSC based 5.2% in non-
SA24 (2015) center non-ICU ICU and
22.5% in
ICU patients
with severe
sepsis
Cross G25 159 1st April to Australia Single Hospital Bone6 Not described 25.0% sepsis
(2015) with 30th June center cases
sepsis 2011
Vincent JL26 10,069 2012 International Multicent ICU Study specific CVS SOFA score 35·3%
(2014) [84 er [730] >2 (33·5–37·1)
countries] for sepsis
Stevenson Variabl 1993 to 2009 USA Multicent Hospital ICD-9 Severe sepsis ICD-9 septic shock Trends in
EK27 (2014) e er [NIS including case identification code 785.52 change
sample data] ICU Angus algorithm reported.
sizes patients and Martin
algorithm
Ortiz G28 826 2007 – 2008 Colombia Multi ICU CDC definitions for CVS SOFA score 45.1%
(2014) center infection and Bone6 >2
(n=10)
Ogura H29 14,417 06/2010 to Japan Multicent ICU Bone6 and Levy7 CVS SOFA score 41.5%
(2014) 05/2011 er [15] >2
Liu V30 55,008 2012 USA Multicent Hospital ICD-9 vs. Angus ICD-9 septic shock 23.3%
(2014) to [Northern er including Algorithm code 785.52
80,678 California] [NIS ICU +
[KPNC] data] patients Hypotension and/or
& Lactate>4mmol/L
280,663
to
717,718
[NIS]
Leligdowicz 7,974 1989 to 2008 Canada, the Multicent ICU Bone6 No description 52.0%
A31 (2014) United er [N=29]
States, and
Saudi Arabia
Koupetori 754 2006-2013 Greece Multicent ICU and Levy7 No description 30.5%
M32 (2014) er (n=46) non-ICU severe sepsis
Goncalves- 1652 May 2009 to Portugal Multi ICU Levy7 No description 48.8%
Pereira J33 with December center (Bone Definitions
(2014) sepsis 2010 (n=14) cited)
Kaukonen 101,064 2000 to 2012 Australia Multicent ICU Bone6 CVS failure: lowest 40.3% in
KM19 (2014) and New er MAP <65 mmHg or 2000
Zealand [N=171] lowest systolic 22.0% in
pressure <90 mmHg 2012
APACHE III codes
for septic shock
Bouza C34 240,939 2006 to 2011 Spain Multicent Hospital ICD-9 codes CVS: 785.5 with all 43.0% for
(2014) er using sub codes severe sepsis
National Shock without
Minimum trauma: 785.1,
Basic Data 785.52, 785.9
Set Hypotension: 458
with sub codes
(458.0, 458.8 458.9)
Nonspecific low
blood pressure
reading: 796.3
Storgaard 212 with Since 1994 Denmark National Hospital CDC definitions for CVS: Systolic blood 33.0%
M35 (2013) severe registries admission infection and Bone6 pressure <90mmHg (25.%-41%)
sepsis/sh s or fall >40mmHg for severe
ock sepsis
Whittaker 1,735 2005 to 2009 University of Single ICU ICD-9 vs. Angus ICD-9 septic shock 36.4%
SA36 (2013) Pennsylvania center admission Algorithm (785.52)
s from Shock was defined
Emergenc as systolic blood
y pressure less than 90
Departme mm Hg after fluid
nt resuscitation (1500
mL) or use of
vasoactive agents.
Stiermaier 139 with 2007 General Single Hospital ICD-10 and Bone6 in No description 48.8 [severe
T37 (2013) sepsis Hospital of center admission ICU admissions sepsis]
Vienna s
Sakr Y38 446 with 2006 Piedmont Multi- ICU Bone6 No description 58.6 (50.5-
(2013) sepsis region, Italy center 66.3)
[N=24]
Rohde JM39 3,146 2009 to 2010 University of Single Hospital ICD-9 Levy et al cited for Not
(2013) with Michigan center admission organ dysfunction extractable
severe s definitions
sepsis
Quenot JP20 1,495 2009 to 2011 North-East Multi ICU Only Septic shock Infection requiring 48.7%
(2013) with France center population included initiation of
septic [n=14] as described vasopressors despite
shock adequate vascular
filling, with at least
one of the following
hypoperfusion
criteria: (1)
metabolic acidosis
(base excess ≥5
mEq/L, alkaline
reserve <18 mEq/L
or lactate ≥2.5
mmol/L); (2)
oliguria/ renal
insufficiency (<0.5
mL/kg/h for 3 h or
elevation >50% of
baseline creatinine);
or (3) hepatic
dysfunction (AST or
ALT >500 IU/L or
bilirubin >34
mol/L).
Nesseler N40 96 2008 to 2010 France Single ICU Bone6 Bone et al cited for 32.0%
(2013) center septic shock
Gray A41 626 2009 Scotland Multi Emergenc SSC No description 28.3%
(2013) center y [severe
[N=20] Departme sepsis +
nt septic shock]
Gaieski DF17 Variable 2004 to 2009 USA Multicent Hospital ICD-9 vs. Angus vs. ICD-9 septic shock 14.7% to
(2013) er admission Martin vs. Wang vs. (785.52) 29.9% for
(NIS s Dombrovskiy severe sepsis
data) Algorithms
Czupryna P42 107 1997 to 2010 Poland Single Hospital Levy7 SBP<90; MAP<60 30.9% for
(2013) center admission or Fall in SBP>40 severe sepsis
s
Seymour 13,249 2000 to 2009 USA (Pre- Multi Emergenc ICD-9 [Angus No description 19.6% for
CW43 (2012) severe hospital center y Medical Algorithm] severe sepsis
sepsis Kings Services
County) encounters
Sancho S44 371 10/1998 to Valencia, Single Hospital Bone6 + Nosocomial No description 59.3% ICU
(2012) 02/2010 Spain center admission bacteremia mortality
s severe
sepsis/shock
Klein 1,072 01/2009 to Netherlands Single ICU Multiple algorithms Refractory 45%
Klouwenberg 10/2010 center tested hypotension 62%
PMC45 Refractory
(2012) hypotension with
other organ system
failure
Park DW46 1,192 2005 to 2009 Korea Multi ICU Bone6 Either SBP ≤ 90 30.8%
(2012) center mmHg or MAP ≤ 70
[n=12] mmHg for at least 1
hour after adequate
fluid resuscitation or
use of vasopressors
to maintain systolic
blood pressure > 90
mmHg or mean
arterial pressure >
70 mmHg.
Levy MM47 25375 2005-2010 International Multi ICU; non- Levy7 Surviving sepsis As
(2012) center ICU; campaign previously
Emergenc reported
y
departmen
t
Lagu T48 4,799,56 2003 to 2007 USA Multi Hospital ICD-9-CM CVS: 785.5 with all Not
(2012) 5 sepsis center sub codes extractable
cases (NIS Shock without
database) trauma: 785.1,
785.52, 785.9
Hypotension: 458
with sub codes
(458.0, 458.8 458.9)
Nonspecific low
blood pressure
reading: 796.3
Bastani A49 267 2007 to 2010 USA Single ICU SSC No description but 33.3% for
(2012) center Early Goal Directed severe
therapy referenced sepsis/shock
Zahar JR50 3,588 1996 to 2009 France Multi ICU Bone6 Sepsis-induced 46.0%
(2011) center hypotension
[N=12] persisting despite
adequate fluid
resuscitation
together with organ
dysfunction OR
Patients receiving
inotropic or
vasoactive agents
who had organ
dysfunction but who
were no longer
hypotensive
Vesteinsdotti 115 2008 to 2009 Iceland Multi ICU Bone6 Sepsis-induced 29.6%
r E51 (2011) severe center hypotension severe
sepsis/ [N=20] persisting despite sepsis/septic
shock adequate fluid shock
resuscitation
together with organ
dysfunction;
Patients receiving
inotropic or
vasoactive agents to
maintain MAP>65
or SBP>90mmHg
Rosado V52 30 2007 to 2009 Peurto Rico Single ICU ICD-9 Not described 66.0%
(2011) center severe
sepsis/shock
Rodriguez 2,681 2007 to 2008 Columbia Multi Hospital CDC definitions Sepsis-induced 45.6%
F53 (2011) center cohort hypotension
[N=10] persisting despite
adequate fluid
resuscitation + need
for vasoactive
agents to maintain
MAP>65 or
SBP>90mmHg
Phua J54 1285 2009 Asian Multi ICU Levy7 SSC Hypotension or on H+V=77.1%
(2011) countries center vasopressors [H/V] LCT=38.3%
[N=150] Hyperlactatemia
>4mmol/L [LCT]
Moore LJ55 231 2007-2009 Houston, Single ICU Bone6 SIRS + infection + 36.0%
(2011) Texas, USA center [Surgical] acute cardiac
dysfunction that is
defined by the
following (must
meet both criteria):
(1) intravenous fluid
challenge >20
mL/kg/ideal body
weight of isotonic
crystalloid infusion
or central venous
pressure >8 mm Hg
or PCWP >12 mm
Hg and (2)
requirement of
vasopressors to
increase MAP >65
mm Hg.
Davis JS56 1191 2007 to 2008 Northern Single Hospital PROWESS Not described 17.1%
(2011) territory of center admission Definition severe
Australia s sepsis/septic
shock
Wilhems 37,990 01/1987 to Sweden Multi Hospital ICD-9 and ICD-10 Angus algorithm, 22.1 to
SB18 (2010) 12/2005 center admission codes mirroring Bone Martin algorithm 29.2% for
Swedish s includes Definition and Flatten H sepsis
Hospital ICU description of ICD depending
Discharge based data on algorithm
Register extraction
Septic shock is not
described
Shen HN57 5,258 1997 to 2006 Taiwan Multi Hospital ICD-9 + Angus No description 30.8%
(2010) center admission Algorithm severe
National s sepsis/shock
Health
Insurance
Program
Levy MM23 15,022 2005 to 2008 International Multi ICU; Bone6 Lactate>4 [LCT] LCT=29.9%
(2010) center Emergenc SSC Vasopressors [V] V=36.7%
[N=165] y Lactate>4 + LCT/V=46.1
Departme vasopressors %
nt; Ward [LCT/V] Overall
38.4%
Kauss IA58 1,179 2004 to 2005 Brazil Single ICU Bone6 Sepsis + arterial 72.7% [68.1
(2010) center hypotension needing – 76.9%]
vasopressors,
despite initial
volume
resuscitation.
Husak L59 30,587 04/2004 to Canada Multi Hospital ICD-10 No description 45.2%
(2010) 03/2009 [excluding center and severe sepsis
Quebec] (discharg Emergenc
e abstract y
database) Departme
nt
Bateman 2,039,77 1997 to 2006 USA Multicent Post ICD-9 ICD-9 septic shock 44.4%
BT60 (2010) 6 er operative (785.52) in1997 to
(NIS sepsis 34.0% in
data) Post 2006 for
operative severe sepsis
sepsis
Vogel TR61 N=1,276 1990 to 2006 New Jersey, Multi Post ICD-9 and surgical CVS failure 45.7%
(2009) ,451 USA center operative diagnosis related (785.50, 785.51, severe sepsis
(HCUP sepsis group 785.59, 458.0,
data) 458.8, 458.9, 796.3,
427.5)
Septic shock NOT
explicitly stated
Povoa PR62 897 2004 to 2005 Portugal Multi ICU Bone6 State of acute 44.0% [ICU
(2009) center circulatory failure mortality]
[n=17] characterized by
persistent arterial
hypotension (SBP
<90mmHg,
MAP<60mmHg,or a
reduction in SBP of
>40 mm Hg from
baseline) despite
adequate fluid
resuscitation
Peake SL63 324 2006 to 2007 Australia Multi Emergenc Levy7 EITHER: Blood BP=23.0%
(2009) and New center y pressure (BP) criteria LCT=26.9%
Zealand [n=32] Departme (one or more of the Overall:
nt following): (i) SBP 23.1%
<90 mmHg OR >40
mmHg fall below
premorbid SBP OR
MAP<65 mmHg, after
≥500 ml intravenous
fluid challenge over
30–60min; (ii)
requirement for
vasoactive infusion for
≥30 min to maintain
BP OR: Metabolic
acidosis criteria (one
or more of the
following): arterial or
venous: (i) blood
lactate >4.0 mmol/L
[LCT]; (ii) anion gap
>20 mEq/L; (iii)
serum bicarbonate
<16.0 mmol/L.
Khwannimit 390 2004 to 2006 Thailand Single ICU Bone6 SBP<90 mmHg or 54.1%
B64 (2009) center MAP<65 mmHg]
for at least 1 h
despite adequate
fluid resuscitation
(CVP>8 mmHg or
PAOP >12 mmHg)
or use of a
vasopressor
(dopamine >5
mg/kg per min or
norepinephrine,
epinephrine any
dose) for >1 h in an
attempt to maintain
SBP >90 mmHg or
MAP>65mmHg
Beale R65 12,570 2002 to 2005 International Multi ICU Bone6 CVS dysfunction is 49.6%
(2009) adults (13 center defined as severe sepsis
countries) [n=276) hypotension in the
absence of causes
other than sepsis.
Baharoon S66 165 2004 Makkah, Multi ICU Levy7 No description 54.7% severe
(2009) Saudi Arabia center sepsis/septic
(n=2) shock
Angkasekwi 3541 2007 Thailand Single Hospital Bone6 No data available 52.6%
nai N16 center admission
(2009) s
Rezende E67 5,332 2004 Brazil Single Emergenc Bone6 No description 64.0%
(2008) center y CVS dysfunction severe sepsis
Departme defined as
nt hypotension or need
for vasoactive drugs
Blanco J68 4,317 2002 Spain Multi ICU PROWESS Criteria Shock: SBP ≤90 54.3%
(2008) center mmHg or MAP ≤70 severe sepsis
(N=14) mmHg, during at least + shock
1 hour despite
adequate resuscitation
with fluids or adequate
intravascular volume;
or use of vasopressors
(dopamine ≥5
µg/Kg/minute;
noradrenalin or
adrenalin at any dose;
dobutamine was not
taken into account).
Unexplained
metabolic acidosis (pH
<7.30 or base excess
≤-5 mmol/l) associated
with an arterial lactate
concentration ≥ 2
mmol/l with no other
apparent cause.
Beovic B69 701 2004 Slovenia Multi ICU Bone6 No description 45.1%
(2008) center severe sepsis
[n=28]
Andreu 33,767 1995 to 2004 Valencia, Multi Hospital ICD-9 CVS (785.5 with all 42.5%
BallesterJC70 Spain center admission sub codes, 427.5, severe sepsis
(2008) [n=26] s 458.0, 458.8, 458.9, and shock
and 796.3)
Septic shock is not
described
Wang HE3 331.5 2001 to 2004 USA Multi National ICD-9 and Levy CVS (785.5 with Not
(2007) million center Hospital sub codes, 458.0, extractable
ED visits Ambulator 458.8, 458.9)
y Medical
Care
Survey
Sakr Y71 3,147 2002 Europe (24 Multi ICU Bone6 Referred to Bone 54.1%
(2007) countries) center Criteria
(N=198)
Karlsson S72 4,500 11/2004 to Finland Multi ICU Bone6 Not described 24.7%
(2007) 02/2005 center [refractive
(N=24) shock]
Esteban A73 15,852 03/2003 to Madrid, Multi Ward and Bone6 Sepsis and at least 45.7%
(2007) 06/2003 Spain center ICU one of the following
[N=3] criteria: SBP <90
mm Hg despite
adequate fluid
administration, need
for inotropes or
vasopressors
(excluding
dopamine >5
g/kg/min), or SBP
decrease of >40 mm
Hg from usual
baseline level.
Engel C74 3,877 1 day in 2003 Germany Multi ICU Bone6 Sepsis or severe 62.4% [55.2
(2007) center sepsis with – 69.2]
[n=454] persistent
hypotension despite
adequate fluid
resuscitation or the
necessity of
vasopressor
administration to
maintain a MAP>=
70 mmHg
Cheng B75 3,6665 12/2004 to China Multi Surgical Bone6 and Levy7 No description 48.7%
(2007) 11/2005 center ICUs severe sepsis
[n=10]
Vincent JL76 3,147 2 weeks in Europe (24 Multi ICU Bone6 No description 54.1% ICU
(2006) 05/2002 countries) center mortality
(N=198)
Tanriover 69 01/2002 to Turkey Single Hospital Diagnostic code of No description Not
MD77 (2006) 06/2003 center charts sepsis/septicaemia/ba extractable
cteremia
Strehlow MC 2.8 1992 to 2001 USA Multi National ICD-9 codes Martin ICD-9 codes for Not
78
(2006) million center Hospital Algorithm organ dysfunction extractable
sepsis Ambulator & Bone6 Definitions
y Medical
Care
Survey
Shapiro N79 3,102 02/2000 to Boston, USA Single Emergenc Bone Suspected infection 27.8%
(2006) 02/2001 center y and hypotension
Departme (SBP<90) which
nt persisted after a
crystalloid fluid
challenge of 20-30
cc/kg.
Harrison 92,672 12/1995 to United Multi ICU PROWESS criteria No description. CVS 48.3 to 44.7
DA80 (2006) 01/2005 Kingdom center dysfunction is coded severe
ICNARC as severe sepsis sepsis/shock
database
Sundararajan 33,741 07/1999 to Victoria, Multi Hospital ICD-10 ICD-10 codes based 29.5%
V84 (2005) 06/2003 Australia center admission Bone6 on Martin severe sepsis
Populatio s Algorithm
n 785.59
database
Laupland 251 07/1999 to Calgary, Multi ICU SIRS + bacteremia BSI-associated 51.0% [28
KB85 (2005) 03/2002 Canada center sepsis patients that day
Populatio required a mortality]
n databse vasopressor infusion
at any dose
Adrie C86 1698 04/1997 to France Multi ICU Bone6 CVS failure = need 39% severe
(2005) 12/2000 center for vasopressors sepsis/shock
[n=6] and/or inotropic
drugs, and/or SBP
of <90 mm Hg, and/
or drop in SBP >40
mm Hg from
baseline
Van Gestel 151 One day Netherlands Multi ICU Bone6 Septic shock = Not
A87 (2004) [11/12/2001] center Sepsis + CVS organ extractable
[N=47] failure + metabolic
dysfunction during
the first 24 hours of
ICU admission.
CVS failure: SBP ≤
90 mmHg or MAP≤
70 mmHg for 1
hour, despite
adequate fluid
resuscitation, or the
need to administer
vasopressors in
order to maintain
SBP≥ 90 mmHg or
MAP≥ 70 mmHg.
Metabolic
dysfunction:
metabolic acidosis
(pH ≤ 7.30 or base
deficit ≥ 5.0 mmol/l)
in association with a
plasma lactate level
> 3.0 mmol/l
Silva E88 1383 05/2001 to Brazil Multi ICU Bone6 Septic shock was 52.2%
(2004) 01/2002 center defined as severe
[n=5] sepsis and
vasoactive drug
requirement (SOFA
3–4).
Laupland 1,981 05/1999 to Calgary, Multi ICU SIRS + bacteremia BSI-associated 49.0%
KB89 (2004) 04/2000 Canada center septic shock was the
Populatio subset of BSI-
n associated sepsis
database patients that
presented with
hypotension
[MAP<60mmHg].
Flaatten H90 6665 1999 Norway Multi Hospital ICD-10 A41.9 – septic 29.3%
(2004) center data shock
Norwegia I50.9 – Unspecified
n patient heart failure
registry
Finfer S91 5878 05/1999 to Australia/ Multi ICU Bone6 PROWESS SBP<90 mmHg or a 32.4%
(2004) 07/1999 New center MAP<70 mmHg for severe
Zealand [N=21] at least 1 h despite sepsis/ shock
adequate fluid
resuscitation,
adequate
intravascular
volume status,
and/or need for
vasopressors to
maintain systolic
blood pressure >90
mmHg or MAP >70
mmHg for >1 h
Brun- 3738 2 weeks in France Multi ICU Bone6 No description 35.0%
Buisson 92 11/2001 center severe
(2004) [n=206] sepsis/shock
Padkin A93 56,673 1995 to 2000 England/Wal Multi ICU PROWESS trial No description 47.3%
(2003) es and center severe
Northern [n=91] sepsis/shock
Ireland
Martin GS2 10,319,4 1979 to 2000 USA Multi Hospital ICD-9 ICD-9 CVS codes 70.0%
(2003) 18 sepsis center admission 458.0 - severe
Hospital s Hypotension, sepsis/shock
discharge postural; 785.5
data Shock; 785.51
cardiogenic shock
and 785.59 septic;
796.3 Transient
hypotension
Annane D94 100,554 1993 to 2000 France Multi ICU Bone6 Not described 60.1%
(2003) center
[n=22]
Alberti C95 14,364 05/1997 to International Multi ICU Bone6 Sepsis-induced 47.2% to
(2002) 05/1998 – Europe, center hypotension 63.8%
Canada and (n=28) persisting despite
Israel adequate fluid
resuscitation along
with the presence of
hypoperfusion
abnormalities or
organ dysfunction
(patients receiving
inotropic or
vasopressor agents
may no longer be
hypotensive by the
time they manifest
hypoperfusion
abnormalities or
organ dysfunction
but would be
considered to have
septic shock).
Angus DC1 192,980 1995 USA Multi Hospital ICD-9 CVS organ Sepsis
(2001) center discharge Angus derivation dysfunction 28.6%
[n=847] records Shock without trauma
ICD codes (785.5 and
458)
Wichmann 48,136 02/1991 to Germany Single Surgical Bone6 Need for 65.7%
MW96 06/1998 center ICU catecholamine severe
(2000) therapy despite sepsis/shock
adequate fluid
support
Schoenberg 664 1997 Germany Single Surgical Bone6 Severe sepsis 40.0%
MH97 (1998) center ICU concomitant with
arterial hypotension
with a SBP<90
mmHg and/or the
need for vasopressor
treatment with
noradrenaline and/or
adrenaline (infusion
rate >0.1 g/kg/min)
and/or dopamine
(infusion rate >5
g/kg/min) despite
adequate volume
resuscitation
Lundberg 41 08/1992 to USA Single ICU Bacteremia + Bone6 Septic shock = a) at 46.0%
JS98 (1998) 04/1993 center least two criteria
SIRS; b) positive
blood
culture(s); c)
systematic
manifestation of
peripheral
hypoperfusion,
defined as lactic
acidosis, oliguria, or
acute alteration of
mental status; and d)
hypotension, defined
as systolic blood
pressure (BP) <90 mm
Hg unresponsive to a
500-mL fluid bolus or
requiring use of
vasopressors to keep
the systolic BP of >90
mm Hg. If the patient
had been hypertensive,
a decrease of 40 mm
Hg systolic BP was
considered the
criterion for
hypotension.
Sands KE99 1342 01/1993 to USA Multi Hospital Bone6 Presence of sepsis 34.0% 28
(1997) 04/1994 center admission syndrome plus a day
[n=8] s >40mmHg drop in mortality
SBP unresponsive to sepsis
a fluid challenge, syndrome
occurring within 24
hours before or
6hours after the
onset of sepsis
syndrome.
Sasse KC100 153 01/1987 to USA Single ICU Bone6 Not described 51.0%
(1995) 03/1991 center severe sepsis
Salvo I101 1101 04/1993 to Italy Multi ICU Bone6 Sepsis with 81.8%
(1995) 03/1994 center hypotension, despite
[n=99] 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 on inotropic
or vasopressor agents
may not be
hypotensive at the
time that per- fusion
abnormalities are
measured
Hypotension: A
systolic BP of <90
mmHg or a reduction
of >40mmHg from
baseline in the absence
of other causes for
Hypotension.
Rangel- 3708 08/1992 to USA Multi ICU Bone6 Septic shock was 46.0%
Frausto 04/1993 center defined as a patient
MS102 (1995) [n=3] with hypotension
not responsive to a
500-mL intravenous
fluid challenge plus
manifestations of
peripheral
hypoperfusion.
Patients without
hypotension but
receiving more than
5mcg/kg/minute of
dopamine or any
other vasopressor
Pittet D103 170 04/1992 USA Single Surgical Bone6 Shock was defined 58%
(1995) center ICU as SBP<90 mmHg
OR a fall in SBP
equal to or greater
than 50 percent for
hypertensive
patients; or blood
pressure requiring a
continuous infusion
of vasopressors.
McLauchlan 125 01/1990 to Edinburgh, Single ICU Study specific Not described 72.0%
GJ21 (1995) 06/1993 Scotland center
Brun- 11,828 01/1993 to France Multi ICU Bone6 Shock was defined 59.0%
Buisson104 02/1993 center as hypotension severe
(1995) [n=170] persisting at least 1 sepsis/shock
hour despite
administration of
fluids, associated
with signs of organ
dysfunction or
hypoperfusion.
Hypotension =
SBP<90 or drop
>40mmHg from
baseline
Dahmash NS 45 02/1989 to Saudi Arabia Single ICU Study specific Not described 40.0%
105
(1993) 05/1991 center
The study specific checklist included: cohort recruitment period, cohort size, cohort location [geographic; Emergency department [ED]; Ward;
Intensive Care unit [ICU]], sepsis and septic shock definitions and acute mortality as reported in the document. Angus algorithm1 = ICD-9-CM
code for infection + Organ Dysfunction code; Martin algorithm2 = ICD-9-CM codes for septicaemia, bacteraemia, or fungaemia + Organ
Dysfunction code. Wang3 and Dombrovskiy algorithms use ICD-9 infection along with ICD-9 severe sepsis codes; PROWESS criteria are the trial
based criteria, which uses 3 systemic inflammatory response syndrome and one or more organ dysfunction to define severe sepsis. USA – United
The table shows the summary of the random effects meta-regression model described in
eMethods1, where per-capita ICU beds data was available to be included as a covariate.
The number of observations in this model was 36, due to either lack of per capita bed
data or international cohorts. The estimate of between study variance (tau2) was 0.02 and
the proportion of residual variation due to heterogeneity (I2_res) was 98.54%. The
proportion of between-study variance explained by covariates included in this model was
2.11%. The joint test for all covariates in the model was not significant (p=0.39)
The three variables proposed from the Delphi process could be present as single
variables, simultaneously, or in pairs i.e. hypotension AND/OR need for vasopressor
therapy AND/OR abnormal lactate. Abnormal lactate refers to concentrations greater
than 2mmol/L. Hypotension refers to mean arterial pressure less than 65mmHg in SSC
database and systolic blood pressure less than 100mmHg in HER datasets. The field
‘crystalloids’ coded as a binary variable within the SSC database defined the ‘after fluids’
term, which refers to resuscitation. We defined cryptic shock as persistently elevated
lactate ‘after fluids’. Hypotension and Vasopressor therapy are related variables.
Hypotension was assumed when vasopressor therapy was administered. Therefore, the
groups were collapsed down to six.
Group 1: Hypotensive after fluids requiring vasopressor therapy with lactate >2
mmol/L
Group 2: Hypotensive after fluids requiring vasopressor therapy with lactate <2
mmol/L
Group 3: Hypotensive after fluids with lactate >2 mmol/L
Group 4: Raised lactate after fluids = ‘cryptic’ shock
Group 5: Isolated lactate >2 mmol/L without hypotension or need for vasopressor,
thus doesn’t mandate resuscitation coding within the SSC database and therefore does
not meet ‘cryptic shock’ definition
Group 6: Hypotension after fluids and lactate<2mmol/L and no use of vasopressor
Black lines are based on the complete case analysis septic shock population (N = 18,840)
while red lines are based on the imputed missing serum lactate population (N = 22,182).
Probability of hospital mortality expressed as a proportion (range 0 – 1) and is based on a
logistic regression model where the dependent variable is hospital mortality and the
independent variable is continuous serum lactate (mmol/L) (eTable8). The lines
produced by the logistic regression model show the relationship between predicted
hospital mortality and serum lactate (mmol/L). The serumn lactate level is truncated at
6mmol/L since approximately 85% of the observations are ≤ 6 mmol/L