Criterios Clinicos de Sepsis Seymour

8/18/2019 Criterios Clinicos de Sepsis Seymour

1/13

Copyright 2016 American Medical Association. All rig hts reserved.

Assessment of Clinical Criteria for Sepsis

For the Third International Consensus Definitions

for Sepsis and Septic Shock (Sepsis-3)

Christopher W. Seymour, MD, MSc; Vincent X. Liu, MD, MSc; Theodore J.Iwashyna,MD, PhD; FrankM. Brunkhorst, MD;Thomas D.Rea,MD, MPH;

AndréScherag, PhD;Gordon Rubenfeld,MD, MSc;JeremyM. Kahn, MD,MSc; Manu Shankar-Hari, MD,MSc; Mervyn Singer, MD,FRCP;

CliffordS. Deutschman, MD, MS; Gabriel J.Escobar, MD;Derek C.Angus, MD, MPH

IMPORTANCE The Third InternationalConsensus Definitions Task Force defined sepsis

as “life-threatening organ dysfunction due to a dysregulated hostresponse to infection.”

Theperformanceof clinical criteria forthis sepsisdefinition is unknown.

OBJECTIVE To evaluate the validity of clinical criteria to identify patients with suspected

infectionwho are at risk of sepsis.

DESIGN, SETTINGS, AND POPULATION Among1.3 million electronichealth record encounters

from January 1, 2010, to December 31,2012, at 12 hospitalsin southwesternPennsylvania, we

identified those with suspected infectionin whom to compare criteria. Confirmatory analyses

were performed in 4 data sets of 706 399 out-of-hospital andhospital encounters at 165 US

andnon-US hospitalsrangingfrom January 1, 2008, until December 31,2013.

EXPOSURES Sequential[Sepsis-related] OrganFailure Assessment(SOFA)score, systemic

inflammatory response syndrome(SIRS) criteria,LogisticOrgan Dysfunction System (LODS)

score,and a newmodelderived usingmultivariablelogisticregressionin a split sample, thequick

Sequential[Sepsis-related] OrganFailure Assessment(qSOFA) score(range,0-3 points, with1

pointeach forsystolichypotension [100mm Hg], tachypnea[22/min],or altered mentation).

MAIN OUTCOMESAND MEASURES For constructvalidity, pairwise agreementwas assessed.

For predictive validity, the discrimination for outcomes (primary: in-hospital mortality;

secondary:in-hospital mortality or intensive care unit [ICU] length of stay3 days) more

common in sepsis thanuncomplicated infectionwas determined. Results were expressed as

thefold change in outcome over deciles of baseline risk of death andarea under thereceiver

operating characteristic curve (AUROC).

RESULTS In theprimary cohort, 148907 encounters hadsuspected infection(n = 74 453

derivation; n = 74454 validation), of whom 6347 (4%) died. Among ICUencounters in the

validation cohort (n = 7932 withsuspected infection, of whom1289 [16%] died),the predictive

validityfor in-hospital mortalitywas lowerfor SIRS (AUROC = 0.64; 95% CI, 0.62-0.66) and

qSOFA (AUROC = 0.66; 95% CI, 0.64-0.68) vs SOFA (AUROC = 0.74; 95%CI, 0.73-0.76;

P < .001 for both) or LODS(AUROC = 0.75; 95%CI, 0.73-0.76; P < .001 for both). Among

non-ICU encounters in thevalidationcohort (n = 66 522withsuspectedinfection, of whom

1886 [3%]died),qSOFA had predictive validity (AUROC = 0.81; 95%CI, 0.80-0.82) thatwas

greater thanSOFA (AUROC = 0.79; 95% CI, 0.78-0.80;P < .001) and SIRS (AUROC = 0.76; 95%

CI, 0.75-0.77; P < .001). Relativeto qSOFA scoreslowerthan 2, encounters with qSOFAscores of

2 or higherhad a 3- to 14-fold increase in hospitalmortality across baseline risk deciles. Findingswere similar in external data sets andfor thesecondary outcome.

CONCLUSIONS AND RELEVANCE Among ICU encounters with suspected infection, the

predictive validity for in-hospital mortality of SOFA was not significantly different than the

more complex LODS butwas statisticallygreater than SIRS andqSOFA, supporting itsuse in

clinical criteria for sepsis. Among encounters with suspectedinfection outside of theICU, the

predictive validity for in-hospital mortality of qSOFA was statistically greater thanSOFA and

SIRS, supporting itsuse as a promptto consider possible sepsis.

JAMA. 2016;315(8):762-774. doi:10.1001/jama.2016.0288

Editorial page 757

Author AudioInterview at

jama.com

Related articles pages775 and

801

Supplementalcontent at

jama.com

Author Affiliations: Authoraffiliations arelisted at theendof this

article.

Corresponding Author: Christopher

W.Seymour, MD,MSc, Departments

of Critical Care Medicineand

Emergency Medicine,Universityof

PittsburghSchool of Medicine,

Clinical Research,Investigation,and

Systems Modeling of AcuteIllness

(CRISMA) Center,3550 TerraceSt,

ScaifeHall, Ste639, Pittsburgh, PA

15261 ([email protected]).

Research

Original Investigation | CARING FOR THE CRITICALLY ILL PATIENT

762 (Reprinted) jama.com


wnloaded From: http://jama.jamanetwork.com/ by Jose Vergara on 02/29/2016

http://jama.jamanetwork.com/article.aspx?doi=10.1001/jama.2016.0288&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288http://jama.jamanetwork.com/article.aspx?doi=10.1001/jama.2016.0288&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288http://jama.jamanetwork.com/article.aspx?doi=10.1001/jama.2016.0290&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288http://jama.jamanetwork.com/article.aspx?doi=10.1001/jama.2016.0288&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288http://www./http://jama.jamanetwork.com/article.aspx?doi=10.1001/jama.2016.0287&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288http://jama.jamanetwork.com/article.aspx?doi=10.1001/jama.2016.0289&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288http://jama.jamanetwork.com/article.aspx?doi=10.1001/jama.2016.0288&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288http://www.jama.com/?utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288mailto:[email protected]:[email protected]:[email protected]://www.jama.com/?utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288http://www.jama.com/?utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288mailto:[email protected]://www.jama.com/?utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288http://jama.jamanetwork.com/article.aspx?doi=10.1001/jama.2016.0288&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288http://jama.jamanetwork.com/article.aspx?doi=10.1001/jama.2016.0289&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288http://jama.jamanetwork.com/article.aspx?doi=10.1001/jama.2016.0287&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288http://www./http://jama.jamanetwork.com/article.aspx?doi=10.1001/jama.2016.0288&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288http://jama.jamanetwork.com/article.aspx?doi=10.1001/jama.2016.0290&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288http://jama.jamanetwork.com/article.aspx?doi=10.1001/jama.2016.0288&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288


2/13


Although common and associated with high morbidity

and mortality,, sepsis and related terms remain diffi-

cult to define. Two international consensus confer-

encesin andusedexpertopinion togenerate thecur-

rentdefinitions.,However,advancesin theunderstanding of

thepathobiologyand appreciation that elements of thedefini-

tions may be outdated, inaccurate, or confusing prompted

the European Society of

Intensive Care Medicine

and the Society of Critical

Care Medicine to convene

a Third International Con-

sensus Task Force to re-

examine the definitions.

Like many syndromes,

there is no “gold stan-

dard” diagnostic test for

sepsis. Therefore, the task

force chose several meth-

odsto evaluatethe useful-

ness of candidate clinical criteria, including clarity, reliability

(consistency and availability), content validity (biologic ratio-

nale and face validity),constructvalidity (agreementbetween

similar measures), criterion validity (correlation with estab-

lished measures and outcomes), burden, and timeliness. Un-

like prior efforts, the task force used systematic literature re-

views and empirical data analyses to complement expert

deliberations.

Based on clarity and content validity and after literature

review and expert deliberation, the task force recommended

elimination of the terms sepsis syndrome, septicemia, and se-

vere sepsis and instead defined sepsis as “life-threatening or-

gan dysfunction due to a dysregulated host response to

infection.” Ofnote,the task forcedid notattempt to redefine

infection. Rather, it next sought to generate recommenda-tions for clinical criteria that could be used to identify sepsis

among patients with suspected or confirmed infection. The

purposeof thisstudywas to inform this stepby analyzingdata

from several large hospitaldatabases to explore the construct

validity andcriterionvalidityof existingand novel criteriaas-

sociated with sepsis.

Methods

This study was approved with waiver of informed consent by

the institutional review boardsof theUniversity of Pittsburgh,

Kaiser Permanente Northern California (KPNC), Veterans Ad-ministration (VA) Ann Arbor Health System, Washington State

Department of Health, King County Emergency Medical Ser-

vices (KCEMS), University of Washington, and Jena University

Hospital.

Study Design, Setting, and Population

A retrospective cohortstudywas performedamong adult en-

counters (age ≥ years) with suspected infection. The pri-

mary cohortwas all hospital encounters from to at

communityand academic hospitalsin theUPMChealth care

systemin southwesternPennsylvania. The cohortincluded all

medical and surgical encounters in the emergency depart-

ment, hospital ward, and intensive care unit (ICU). We cre-

ated a random split sample (/) fromthe UPMC cohort,the

derivation cohort for developing new criteria, and the valida-

tion cohort for assessment of new and existing criteria.

We also studied external data sets: () all inpatient

encounters at KPNC hospitalsfrom to ; ()all en-

counters in hospitals in the United States’ VA system

from to ; () all nontrauma, nonarrest emergency

medical services records from advanced life support agen-

ciesfrom - transported to hospitalswith commu-

nityinfectionin KingCounty, Washington(KCEMS);and()all

patients from - at German hospital enrolled with

hospital-acquired infection in the ALERTS prospective cohort

study.These cohorts were selected becausethey included pa-

tient encounters fromdifferent phases of acutecare (outof hos-

pital, emergency department, hospital ward) and countries

(United States and Germany) with different types of infection

(community and nosocomial). The UPMC, KPNC, and VA data

were obtained fromthe electronic healthrecords (EHRs)of the

respective health systems;KCEMSdata wereobtainedfrom the

administrative out-of-hospital record; and ALERTS data were

collected prospectively by research coordinators.

Defining a Cohort With Suspected Infection

For EHR data (UPMC, KPNC, and VA), the first episode of sus-

pected infection wasidentifiedas thecombination of antibiot-

ics(oral or parenteral) andbodyfluidcultures(blood, urine,ce-

rebrospinal fluid, etc). We required thecombination of culture

and antibiotic start time to occur within a specifictime epoch.

If theantibioticwas givenfirst,the culturesamplingmusthave

beenobtained within hours.If theculturesamplingwas first,

theantibioticmust have beenordered within hours. The“on-

set” of infection was defined as the time at which the first of these events occurred (eAppendix in the Supplement). For

non-EHR data in ALERTS, patients were included who met

US Centers for Disease Control and Prevention definitions or

clinical criteria for hospital-acquired infection more than

hoursafter admission asdocumentedbyprospectivescreening.

For non-EHR data in KCEMS, administrative claims identified

infection present on admission (Angus implementation of in-

fectionusing International Classification of Diseases, Ninth Re-

vision, Clinical Modification ( ICD--CM ) diagnosis codes).

Determining Clinical Criteriafor SepsisUsing ExistingMeasures

In UPMC derivation and validation data, indicators were gen-

erated for each component of the systemic inflammatory re-sponse syndrome (SIRS) criteria; the Sequential [Sepsis-

related] Organ Failure Assessment (SOFA) score; and the

Logistic Organ Dysfunction System (LODS) score, a weighted

organ dysfunction score (Table ). We used a modifiedversion

of the LODS score that did not contain urine output (because

ofpooraccuracy inrecording onhospital ward encounters), pro-

thrombin, or urea levels. The maximum SIRS criteria, SOFA

score, and modified LODS score were calculated for the time

window from hoursbeforeto hours after the onset ofin-

fection, aswell as on each calendar day. This window wasused

EHR electronic healthrecord

GCS GlasgowComa Scale

ICU intensivecare unit

LODS LogisticOrgan Dysfunction

System

qSOFA quickSequential

[Sepsis-related]Organ Function

Assessment

SIRS systemic inflammatory

response syndrome

SOFA Sequential [Sepsis-related]

OrganFunctionAssessment

Assessment of Clinical Criteria for Sepsis Original Investigation Research

jama.com (Reprinted) JAMA February23,2016 Volume 315, Number8 763



http://jama.jamanetwork.com/article.aspx?doi=10.1001/jama.2016.0288&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288http://jama.jamanetwork.com/article.aspx?doi=10.1001/jama.2016.0288&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288http://www.jama.com/?utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288http://www.jama.com/?utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288http://jama.jamanetwork.com/article.aspx?doi=10.1001/jama.2016.0288&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288


3/13


forcandidate criteriabecause organ dysfunction in sepsismay

occur prior to, near the moment of, or after infection is recog-

nized by clinicians or when a patient presents for care. More-

over, the clinical documentation, reporting of laboratory val-

ues in EHRs, and trajectory of organ dysfunction are

heterogeneous acrossencounters andhealth systems. In a post

hocanalysis requestedby thetask force,a change inSOFA score

was calculated of points or more from up to hours before

to up to hours after the onset of infection.

Deriving Novel Clinical Criteria for Sepsis

In the derivation cohort (UPMC),new, simple criteriawere de-

veloped accordingto theTransparent Reportingof a Multivari-able Prediction Model for Individual Prognosis or Diagnosis

(TRIPOD) recommendations.This entailed steps:() assess-

ingcandidatevariablequality andfrequencyof missingdataand

()developinga parsimoniousmodel andsimplepointscore.,,

Becauseof thesubjectivenature andcomplexityof variablesin

existing criteria, we soughta simple model that could easily be

used by a clinician at the bedside.

Based on the assumptionthat hospital mortality would be

farmorecommonin encounterswithinfectedpatients whohave

sepsis than in those who do not, all continuous variables were

dichotomized by defining their optimal cutoffsusing themini-

mum/ distanceon thearea under thereceiver operating char-

acteristic curve(AUROC)forin-hospitalmortality.

Cutoffswererounded to the nearest integer, and standard single-value im-

putation was used, withnormal value substitution if variables

were missing. The latter approach is standard in clinical risk

scores,, and mirrors how clinicians would use the score at

the bedside. Multiple logistic regression was used withrobust

standard errors and forward selection of candidate variables

using theBayesianinformation criterion to developthe “quick

SOFA” (qSOFA) model. The Bayesian information criterion is a

likelihood-based stepwise approach that retainsvariables that

improve the model’s overall ability to predict the outcome of

interest while incorporating a penalty for including too many

variables. Favoring simplicity over accuracy, a point score of

was assigned to each variable in the final model, irrespective

of the regression coefficients. Model calibration was assessed

by comparing clinically relevant differences in observed vs ex-

pected outcomes, as the Hosmer-Lemeshow test may be sig-

nificant due to large sample sizes.

Assessments of Candidate Clinical Criteria

Thetest:retestor interraterreliability ofindividualelementswas

not assessed, in part because most elements have known reli-

ability. However, the frequency of missing data was deter-

mined for each element because more common missing datafor individual elements will potentially affect the reliability of

integratedscoressuch asthe SOFA score. Constructvaliditywas

determined by examining the agreement between different

measures analogous to the multitrait-multimethod matrixap-

proach of Campbell and Fiske, using the Cronbach α to mea-

sureagreementor commonality.,Confidenceintervalswere

generated withthe bootstrap method ( replications).

Criterion validity was assessed using the predictive valid-

ity of the candidate criteriawith outcomes (primary outcome:

in-hospital mortality; secondary outcome: in-hospital mortal-

ityorintensivecareunit[ICU]lengthofstay≥days).Theseout-

comes are objective, easily measured across multiple hospi-

tals in US/non-US cohorts, and are more likely to be present inencounters with patients with sepsis than those with uncom-

plicatedinfection.To measurepredictivevalidity, a baselinerisk

model wascreated forin-hospital mortality based on preinfec-

tion criteria using multivariable logistic regression. The base-

linemodel included age(as a fractional polynomial),sex,race/

ethnicity ( black, white, or other), and the weighted Charlson

comorbidity score (as fractional polynomial) as a measure of

chronic comorbidities., Race/ethnicity was derived from

UPMC registration system data using fixed categories consis-

tent with the Centers for Medicare & Medicaid Services EHR

Table 1. Variables for CandidateSepsisCriteria Among Encounters WithSuspected Infection

SystemicInflammatoryResponse Syndrome(SIRS) Criteria(Range, 0-4 Criteria)

Sequential[Sepsis-related] Organ FailureAssessment (SOFA)(Range, 0-24 Points)

Logistic Organ DysfunctionSystem (LODS)(Range, 0-22 Points)a

Quick Sequential[Sepsis-related] Organ FailureAssessment (qSOFA)(Range, 0-3 Points)

Respiratory rate,breaths per minute

PaO2/FiO2 ratio PaO2/FiO2 ratio Respiratory r ate, b reathsper minute

White bloodcell

count, 109/L

Glasgow ComaScalescore Glasgow ComaScalescore Glasgow ComaScalescore

Bands,% Meanarterialpressure, mmHg Systolicbloodpressure, mm Hg Systolicbloodpressure,mmHg

Heartrate, beatsper minute

Administration of vasopressorswithtype/dose/rateof infusion

Heart rate, beatsper minute

Temperature, °C Serum creatinine, mg/dL,or urineoutput, mL/d

Serum creatinine,mg/dL

Arterial carbondioxide tension,mmHg

Bilirubin, m g/dL Bilirubin, m g/dL

Plateletcount, 109/L Platelet count, 109/L

White bloodcell count,109/L

Urine output, L/d

Serumurea, mmol/L

Prothrombin time,% of standard

Abbreviation:FiO2, fraction of

inspired oxygen.

a Measurement unitsfor LODS

variables per original descriptionby

LeGall etal.9

Research Original Investigation Assessment of Clinical Criteria for Sepsis

764 JAMA February23,2016 Volume315,Number 8 (Reprinted) jama.com



http://www.jama.com/?utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288http://www.jama.com/?utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288


4/13


meaningful use data set. Race/ethnicity was included in the

baseline modelbecauseof itsdescribedassociation withthe in-

cidence and outcomes of sepsis.

Encounterswerethen dividedinto decilesof baseline risk.

Within each decile, the rate of in-hospital mortality ± ICUlength of stay of days or longer was determined comparing

encounters with infection with or more SIRS, SOFA, LODS,

andqSOFA points vs encounterswith lessthan criteriaof the

same score (threshold of points was determined a priori).

Model discrimination was assessed with the AUROC for each

outcome using the continuous score(s) alone, then added to

the baseline risk model. Analyses were separately performed

in ICUencountersand non-ICU encounters at theonsetof in-

fection. New, simple criteria in external data sets were as-

sessed in both ICU and non-ICU encounters.

Because serum lactate is widely used as a screening tool

in sepsis,howitsmeasurement would improvepredictiveva-

lidity of newcriteriawas assessedin posthoc analyses. Evalu-ation included qSOFA modelsthatdid and didnot include se-

rum lactate at thresholds of ., ., and . mmol/L (, ,

and mg/dL)and as a continuousvariable.OnlyKPNC data

were used for these analyses because an ongoing quality im-

provement program promoting frequent serum lactate mea-

surement acrossthe healthsystem minimized confoundingby

indication.

Several sensitivity analyses were performed to assess ro-

bustness of the findings. These included a variety of restric-

tions to thecohort, more rigorous definitions of suspected or

presumedinfection,alternative ways to measureclinicalvari-

ables (such as altered mentation in theEHR), andmultiple im-

putation analyses for missing data. There are many possible

time windowsfor criteria aroundthe onset of infection. A va-

riety of windows differing from the primary analysis weretested,including () hours before to hours after;() hours

before to hours after;and () restricting to only the hours

after the onset of infection. Detailed descriptions are in the

Supplement.

All analyses were performed with STATA software, ver-

sion . (Stata Corp). All tests of significance used a -sided

P ≤ .. We considered AUROCs to be poor at . to ., ad-

equate at . to ., good at . to ., andexcellent at . or

higher.

Results

Cohorts and Encounter Characteristics

At hospitals in US and non-US data sets between

and (Table), encounterswere studied.In the

primarycohort of records(UPMCderivationand vali-

dation; Figure ), encounters had suspected infec-

tion, most often presenting outside of the ICU (n =

[%]). Asshownin Table, first infection wascommonlysus-

pected within hours of admission (%), most often pre-

senting in the emergency department (%) compared with

theward (%) or ICU(%), andmortality waslow (%).The

Table 2. Summary of Data Sets

Characteristics UPMCa KPNC VA ALERTS KCEMS

Years of cohort 2010-2012 2009-2013 2008-2010 2011-2012 2009-2010

No. of hospitals 12 20 130 1 14

Total No. of encounters 1 309 025 1 847 165 1 640 543 38 098 50 727

Data sourceand study design

Retrospective studyof EHRs

Retrospective study ofEHRs

Retrospective studyof EHRs

Prospective cohortstudy

Retrospective studyof administrative records

Setting Integrated healthsystem in southwesternPennsylvania

Integrated healthsystem in northernCalifornia

All hospitals in the USVA system

Single universityhospital, Jena,Germany

Out-of-hospital recordsfrom integratedemergency medicalservices system in KingCounty, Washington

Definition of suspectedinfection

Combination of bodyfluid culture andnonprophylacticantibiotic administrationin the EHRb



CDC criteriafor hospital-acquiredinfectionsc

ICD-9-CM codesfor infection, withpresent-on-admissionindicatorsd

No. with suspectedinfection (% of total)

148 907 (11) 321 380 (17) 377 325 (23) 1186 (3) 6508 (13)

Location at onset ofinfection, No. (%) infected

Intensive care unit 15 768 (11) 7031 (2) 73 264 (19) 300 (25) 0

Outside of intensivecare unit

133 139 (89) 314 349 (98) 304 061 (81) 886 (75) 6508 (100)

In-hospital mortality,No. (%) infectede

6347 (4) 16 092 (5) 22 593 (6) 210 (18) 700 (11)

Abbreviations: KCEMS, King County EmergencyMedical Services; KPNC,Kaiser

Permanente Northern California;EHR, electronic health record; ICD-9-CM,

International Classificationof Diseases, NinthRevision,ClinicalModification;

VA,VeteransAdministration.

a Referredto as theprimary cohort,further dividedintoderivation (n = 74453)

and validation (n = 74454) cohorts.

b SeetheeAppendix in theSupplementfor detailsabouttimewindows

specified between bodyfluid cultures and antibioticadministration.

c Patientswere enrolled in ALERTSif thein-hospitalstaywas longerthan

48 hoursand in-person prospectivescreeningrevealedhospital-acquired

infection criteria according to Centers for Disease Control and Prevention

(CDC)guidelines.7

d Required Angusimplementation ICD-9-CM code for infection accompanied

by present-on-admissionindicator, as previously validated.6

e AmongUPMC encounters,28 286 (19%)had in-hospitalmortality plus

intensivecareunit lengthof stay of 3 days or longer.





http://jama.jamanetwork.com/article.aspx?doi=10.1001/jama.2016.0288&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288http://www.jama.com/?utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288http://www.jama.com/?utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288http://jama.jamanetwork.com/article.aspx?doi=10.1001/jama.2016.0288&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288


5/13


mediantimefrom thestart ofthe encounteruntil theonset of

suspected infection(defined as cultureor antibiotics order) was

. hours (interquartile range, .-. hours). In KPNC hos-

pitals(eTable in the Supplement), firstsuspected infections

occurred outside theICU (%)with similar mortality(%)and

proportion identifiedwithin hours of admission(%). Se-

rum lactate was measured in % of suspected infection en-

counters in KPNC hospitals compared with less than % in

the other cohorts. In VA hospitals, encounters with sus-

pected infectionhadsimilarmortality(%)but weremore likely

to be firstidentified in the ICU(%).A minority of firstinfec-

tion episodes occurred following surgery, and positive blood

cultures were found in %to % of encounters. In the base-line risk model, using only demographics and comorbidities,

there was a -fold variation for in-hospital mortality across

deciles of baseline risk, ranging from .% to %(eFigure in

the Supplement).

Frequency of Missing Data Among Clinical

and Laboratory Variables

In theUPMC derivationcohort, SIRS criteriaand selectedlabo-

ratory tests in SOFA andLODS were variably measured in the

EHR near the onset of infection (eFigure in the Supple-

ment). Tachycardia, tachypnea, and hypotension, although

present in less than % of encounters, were the most com-

monclinical abnormalities. Encounters in theICU were morelikely to have SIRS and SOFA variables measured and values

were morelikely tobe abnormal. For encountersoutsideof the

ICU, laboratory data were less available, with total bilirubin,

ratioofPaO to fractionof inspiredoxygen,and platelet counts

absent in %, %, and %of encounters, respectively.

Performance of Existing Criteria in the ICU

in the UPMC Cohort

Among ICUencounters withsuspected infectionin theUPMC

validation cohort (n = [%]), most had or more LODS

points (%), SOFA points (%), or SIRS criteria (%) near

thetime of suspected infection, withmortalityratesof %for

allscoresat thisthreshold (Figure andeFigureinthe Supple-

ment). SOFA and LODShad greater statistical agreement with

each other (α = .; % CI, .-.) but lower with SIRS

(α = . [% CI,.-.] for SOFA; α = .[% CI,.-

.]forLODS)(Figure).EncountersintheICUwithormore

vslessthan SIRS criteriawerecomparedwithindecileof base-

line risk and observed a - to -fold increased rate of hospital

mortality compared with a - to -fold increase in mortality

comparing those with or more vs less than SOFA points

(Figure).Thefoldchangein the LODS scorewasevengreater

than that for SOFA.In the ICU, the predictive validity for hospital mortality

using SOFA (AUROC = .; % CI, .-.) and LODS

(AUROC = .; % CI, .-.; P = .) were not statisti-

cally different but were statistically greater than that of SIRS

(AUROC = .; % CI, .-.; P < . for either LODS

or SOFA vs SIRS) (Figure and eFigure and eTable in the

Supplement).Resultsfor a change in SOFA of points or more

weresignificantlygreatercompared with SIRS(AUROC = .;

% CI, .-.; P < . vs SIRS criteria). The SOFA score

was or more in %of decedents (% CI,%-%); among

survivors,the SOFA score wasless than in %(%CI, %-

%). These proportions were similar for a LODS threshold of

or (eTable in the Supplement). Among decedents, ormore SIRS criteriawerepresent in % (% CI,%-%). Re-

sults were consistent for the combined outcome (eFigures

and inthe Supplement).

Performance of Existing Criteria Outside the ICU

in the UPMCCohort

For encounters with suspected infection outside of the ICU

(n = [% of cohort]), (%) had no SIRS crite-

ria, (%) had no SOFA points, and (%) had

no LODSpoints(Figure ).Agreementfollowed a patternsimi-

Figure 1. Accrual of Encounters for Primary Cohort

1309025 Patient encounters at 12 UPMChospitals in 2010-2012

148907 With suspected infection in ED,ICU, ward, step-down unit, orPACU included in primary cohort

1160118 Excluded

1109402 No infection present

2117 Error in encounter start time

45628 Aged


6/13


7/13


Figure 2. Distribution of PatientEncounters Over SIRSCriteria and SOFA, LODS, andqSOFA Scores Among ICU Patientsand Non-ICU Patients

WithSuspected Infectionin theUPMC Validation Cohort (N = 74 454)

ICU encounters (n = 7932) Non-ICU encounters (n = 66 522)

50

40

30

20

10

0

E n c o u n t e r s ,

%

qSOFA Score

0 1 32

50

40

30

20

10

0


%

qSOFA Score

0 1 32

qSOFA scoreD

50

40

30

20

10

0

E n c o u n t e r

s ,

%

LODS Score

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14

50

40

30

20

10

0

E n c o u n t e r

s ,

%

LODS Score

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14

LODS scoreC

50

40

30

20

10

0

E n c o u n t e r s , %

SOFA Score

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

50

40

30

20

10

0

E n c o u n t e r s , %

SOFA Score

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

SOFA scoreB

50

40

30

20

10

0


%

SIRS Criteria

0 1 2 3 4

50

40

30

20

10

0


%

SIRS Criteria

0 1 2 3 4

SIRS criteriaA

ICU indicates intensivecare unit;LODS,Logistic OrganDysfunction System;

qSOFA, quickSequential [Sepsis-related]Organ Function Assessment; SIRS,

systemic inflammatory response syndrome;SOFA, Sequential[Sepsis-related]

OrganFunctionAssessment. Thex-axis is the scorerange, withLODS truncated

at 14points(of 22points) andSOFA truncatedat 16points(of 24points) for

illustration.







8/13


The discrimination of hospital mortality using SOFA

(AUROC = .;% CI,.-.),LODS (AUROC = .; %

CI, .-.), or change in SOFA (AUROC = .;%CI, .-

.) scores was significantly greater comparedwith SIRScri-

teria(AUROC = .; %CI, .-.; P < . forall) (Figure

andeFigureandeTableinthe Supplement). Sixty-eight per-cent (% CI, %-%) of decedents had or more SOFA

points and %(%CI, %-%)of survivorshad less than

SOFA points. In comparison, only % (% CI, %-%)

of decedents had or more SIRS criteria,whereas % of sur-

vivorshad less than SIRS criteria (% CI,%-%) (eTable

in the Supplement). Results were consistent for the com-

bined outcome (eFigures and in the Supplement).

Performance of New, Simple Criteria

The final qSOFA model included Glasgow Coma Scale (GCS)

scoreoforless,systolicbloodpressureofmmHgorless,

and respiratory rate of /min or more ( point each; score

range, -) (Table ). Most encounters with infection (%-%) hadless than qSOFA points,and mortality rangedfrom

%to % over the scorerange (eFigure in the Supplement).

Calibration plots showed similar observed vs expected pro-

portionof deathsacross qSOFA scores(eFigure in theSupple-

ment). The qSOFA agreed reasonably well with both SOFA

(α = .;% CI,.-.) andLODS (α = .;% CI,.-

.) and, unlike SOFA andLODS, also agreedmore with SIRS

(α = .; % CI, .-.) (Figure ). The % of encoun-

terswithinfection with or qSOFA points accountedfor %

of deaths, % of deaths or ICU stays of days or longer.

In the ICU, the predictive validity for hospital mortality of

qSOFA above baseline risk (AUROC = .; % CI, .-

.) was statistically greater than SIRS criteria ( P = .) but

significantly less than SOFA ( P < .) (Figure and eFigure

and eTable in the Supplement). Outside of the ICU, there

was a - to -fold increase in the rate of hospital mortalityacross the entire range of baseline risk comparing those with

or more vs less than qSOFA points (Figure ). The predic-

tive validity of qSOFA was good for in-hospital mortality

(AUROC = .; % CI, .-.), was not statistically dif-

ferent from LODS ( P = .) and was statistically greater than

SOFA or change in SOFA score ( P < . for both) (Figure ,

Figure, and eFigure and eTable inthe Supplement). Sev-

enty percent (% CI, %-%) of decedents had or more

qSOFA points and % (% CI, %-%) of survivors had

less than qSOFA points (eTable in the Supplement).

Results were consistent for the combined outcome (eFigures

and in the Supplement).

Among encounterswith or moreqSOFA points,% alsohad ormoreSOFA points (eFigure inthe Supplement). This

proportion was greater among decedents (%) and ICU en-

counters (%) and increased as the time window for evalu-

ationwas extendedto hours (%) and hours (%) af-

ter the onset of infection.

External Data Sets

The qSOFA was tested in external data sets comprising

patient encounters at hospitals in out-of-

hospital(n = ),non-ICU(n = ), andICU (n = )

Figure 3. AreaUnder theReceiver OperatingCharacteristicCurve and 95%Confidence Intervals forIn-Hospital Mortalityof CandidateCriteria

(SIRS, SOFA, LODS, and qSOFA) Among SuspectedInfection Encounters in theUPMC Validation Cohort (N = 74 454)

ICU encounters (n = 7932)A

SIRS 0.64

(0.62-0.66)

0.43

(0.41-0.46)

0.41

(0.38-0.43)

0.46

(0.43-0.48)

SOFA


9/13


settings (eTable in theSupplement). Among encounterswith

communityinfection(KCEMS) or hospital-acquiredinfection

(ALERTS), qSOFA had consistent predictive validity

(AUROC = . and ., respectively) (Table and eFigure

in the Supplement). Results were similar in the VA data set

(AUROC = .), in which no GCS data were available.

Serum LactateDuring model building in UPMC data, serum lactate did not

meet prespecifiedstatistical thresholdsfor inclusionin qSOFA.

In KPNC data,the post hocaddition of serum lactate levelsof

. mmol/L( mg/dL)or more toqSOFA (revised toa -point

score with added point forelevatedserumlactate level)sta-

tisticallychanged thepredictive validityof qSOFA (AUROCwith

lactate = .; % CI, .-. vs AUROC without lac-

tate = .; % CI, .-.; P < .) (eFigure A in the

Supplement).AsshownineTableintheSupplement,thiswas

consistent for higher thresholds of lactate (. mmol/L

[mg/dL], .mmol/L[ mg/dL])or using a continuousdis-

tribution ( P < .).However,the clinicalrelevance wassmall

as theratesof in-hospital mortalitycomparingencounters with

or more vs lessthan points across deciles of risk were nu-

merically similar whether or not serum lactate was included

in qSOFA (eFigure B in the Supplement).

Among encounters with qSOFA point but also a serum

lactate level of .mmol/L or more,in-hospitalmortality washigher than that for encounters with serum lactate levels of

lessthan. mmol/L acrossthe range of baseline risk. Therate

of in-hospitalmortality wasnumerically similar to that foren-

counters with qSOFA points using the model without se-

rumlactate(eFigureinthe Supplement). Becauseserumlac-

tate levels arewidely used for screening at many centers, the

distribution of qSOFA scores over strata of serumlactatelevel

was investigated. The qSOFA consistently identified higher-

risk encounters even at varying serum lactate levels (eFigure

in the Supplement).

Figure 4. FoldChangein Rateof In-Hospital Mortality(Log Scale) Comparing Encounters With≥2 vs


10/13


Time Windows for Measuring qSOFA Variables

When qSOFA variables were measured in the time window

from hours before/after or hours before/after the onset of

infection in KPNC data (eTable in the Supplement), results

were notsignificantly differentfrom theoriginalmodel( P

= .for hours and P = . for hours). When qSOFA variables

were restricted to only the -hour period after the onset of

infection,the predictive validityfor in-hospital mortalitywas

significantly greater (AUROC = .; % CI, .-.;

P < .) compared with the primary model.

Additionalsensitivityanalyses areshown ineTable in the

Supplement. The predictive validityof qSOFA wasnot signifi-

cantly differentwhen using moresimple measures,such as any

altered mentation (GCSscore


11/13


tion, lowsystolic blood pressure, andelevated respiratoryrate

hadstatisticallygreater predictivevalidity thanthe SOFA score

(Figure ).The predictive validityof qSOFA wasrobust toevalu-

ation under varied measurementconditions,in academic and

community hospitals, in international locations of care, for

community and hospital-acquired infections, and after mul-

tiple imputation for missing data. It was, however, statisti-

cally inferior compared with SOFA for encounters in the ICU

and has a statistically lower content validity as a measure of

multiorgan dysfunction. Thus, the task force recommended

use of a SOFA score of pointsor morein encounterswithin-

fection as criteria for sepsisand useof qSOFA in non-ICU set-

tings to consider the possibility of sepsis.

Criteria Outside of the ICU

For infected patients outside of the ICU, there is an increasing

focus on early recognitionof sepsis. Potentialcriteria fororgan

dysfunctionlike SOFA or LODSrequired clinical andlaboratory

variables that maybe missing anddifficultto obtain in a timely

manner.These characteristics mayincreasemeasurement bur-

denfor clinicians.In comparison, a simplemodel (qSOFA)uses

clinical variables, hasno laboratorytests,and hasa predictive

validity outside of the ICUthat is statistically greater than the

SOFA score ( P < .). TheqSOFA andSOFA scoresalso hadac-

ceptable agreement in the majority of encounters.

However, potentially controversial issues are worth not-

ing.First, qSOFA wasderived andtestedamongpatient encoun-

tersin whichinfection wasalreadysuspected. TheqSOFA is not

analertthatalone willdifferentiate patients withinfection from

those without infection. However, at least in many US and

European hospital settings, infection is usually suspected

promptly, as evidenced by rapid initiation of antibiotics.,

Second,mentalstatus is assessed variablyin different set-

tings, which may affect the performance of the qSOFA. Al-

though the qSOFA appeared robust in sensitivity analyses toalternative GCS cut points, further work is needed to clarify

itsclinical usefulness. In particular,the modelevaluated only

whether mental status was abnormal, not whether it had

changed from baseline, which is extremely difficult to opera-

tionalize and validate, both in the EHR and as part of routine

charting.An alternative to theGCS (eg, Laboratory andAcute

PhysiologyScore, version, in KPNCencounters)foundsimi-

lar results.

Third, serumlactate levels, which have been proposed as

a screening tool for sepsis or septic shock, were not retained

in the qSOFA during model construction. One reason may be

because serum lactatelevels were not measuredcommonly in

the UPMC data set. When serum lactate levels were added toqSOFA post hocin theKPNC healthsystem data set, in which

measurement of lactatelevels wascommon, thepredictive va-

lidity wasstatistically increasedbut withlittledifference inhow

encounters were classified. This analysis assessed only how

serumlactate levels at differentthresholdscontributed above

and beyondthe qSOFA model. However,among intermediate-

riskencounters(qSOFA score = ),the additionof a serum lac-

tate level of . mmol/L( mg/dL)or higheridentified those

with a risk profile similar to those with qSOFA points. Thus,

areas for further inquiry include whether serum lactate lev-

els could be used for patients with borderline qSOFA values

or as a substitute for individual qSOFA variables (particularly

mental status,given theinherentproblemsdiscussed above),

especially in health systems in which lactate levels are reli-

ably measured at low cost and in a timely manner.

Criteria in the ICU

Among ICU encounters, the diagnosis of sepsis may be chal-

lenging because of preexisting organ dysfunction, treatment

priorto admission,and concurrent organ support.In thisstudy,

as others have reported in a distinct geographic region and

health care system, traditional tools such as the SIRS crite-

ria have poor predictive validity among patients who are in-

fected. Yet in our study, SOFA and LODS scores had superior

predictive validity in the ICU and greater agreement, perhaps

because more variables were likely to be measured, abnor-

mal, and independent of ongoing interventions. These re-

sults areconsistent withpriorstudiesof SOFA andLODS in the

ICU., On average, only of infected decedents in the

ICU had a SOFA orLODSscoreof lessthan .TheqSOFA score

had statisticallyworse predictive validity in the ICU, likely re-

lated to theconfoundingeffectsof ongoingorgansupport(eg,

mechanical ventilation, vasopressors).

Advances Using EHRs

The data from these analyses provided the Third Interna-

tional Consensus Task Force with evidence aboutclinical cri-

teria for sepsis using EHRs from large health systems with

both academic and community hospitals. More than % of

US nonfederal, acute care hospitals(and all US federal hospi-

tals) now use advanced EHRs. Adoption of EHRs has in-

creased -fold since in the United States and will con-

tinue to increase. The EHR may present hospitals with an

opportunity to rapidly validate criteria for patients likely to

have sepsis, to testprompts or alerts among infected patientswithspecificEHR signaturessuggestive of sepsis,and to build

platforms for automated surveillance. In addition, criteria

such as in theqSOFA canbe measured quickly and easily and

assessed repeatedly over timein patientsat riskof sepsis,per-

haps even in developing countries without EHRs.

Limitations

Thisinvestigationhas severallimitations.First,we studiedonly

patients in whom infection was already suspected or docu-

mented. We did not addresshow to diagnose infectionamong

those in whomlife-threateningorgandysfunctionwas theini-

tialpresentation. Therefore, these dataalone do not mandate

thathospitalized patients with SOFA or qSOFA pointsbe evalu-ated for the presence of infection.

Second, we chose to develop simple criteria that clini-

cians could quickly use at the bedside, balancing timeliness

and content validity with greater criterion validity. We ac-

knowledge that predictive validity would be improved with

more complex models thatinclude interaction terms or serial

measurements over time.,, We tested how the change in

SOFA score over time would perform, andalthough similar to

the maximum SOFA score, the optimal time windows over

which change should be measured are not known.







12/13


Third, no organ dysfunction measurements evaluated in

this study distinguish between chronic and acute organ dys-

function, assess whether the organ dysfunction has an expla-

nation other than infection, or attribute dysfunction specifi-

cally to a dysregulated host response. For example, a patient

with dementiawithan abnormal GCS scoreat baselinewill al-

ways have qSOFA pointbut may notbe aslikely to have sep-

sisas a patientwith a normalbaseline sensorium. Assuch, we

illustrated the predictive validity of various criteria across a

fullrange of underlyingriskdeterminedfrom comorbidity and

demographics.

Fourth, we chose outcomes associated morecommonly

with sepsis than with uncomplicated infection. These out-

comeshavehigh contentvalidityand weregeneralizableacross

data sets, but there are certainly alternative choices.

Fifth, we compared predictive validity with tests of infer-

ence that may be sensitive to sample size. We found that sta-

tistically significant differencesin AUROC wereoften present,

yet these resulted in differences in classification with debat-

able clinical relevance. We reconciledthese databy reporting

the fold change in outcome comparing encounters of differ-

ent scores to provide more clinical context.

Sixth,the acute,life-threatening organdysfunctionin sep-

sis may also occur at different times in different patients

(before, during, or afterinfectionis recognized).Results were

unchanged overa variety of timewindows, includingboth long

(-hour)and short(-hour) windows around the onsetof in-

fection. Prospective validation in other cohorts, assessment

in low- to middle-income countries, repeated measurement,

and the contribution of individual qSOFA elements to predic-

tive validity are important future directions.

Conclusions

Among ICU encounters with suspected infection, the predic-

tive validityfor in-hospital mortalityof SOFA wasnot signifi-

cantly different than the more complex LODS but was statis-

tically greater than SIRS and qSOFA, supporting its use in

clinical criteria for sepsis. Among encounters with suspected

infection outside of the ICU, the predictive validity for in-

hospital mortality of qSOFA was statisticallygreaterthanSOFA

and SIRS, supporting its use as a prompt to consider possible

sepsis.

ARTICLE INFORMATION

Author Affiliations: Departmentof Critical Care

Medicine,Universityof PittsburghSchool of

Medicine, Pittsburgh, Pennsylvania (Seymour,

Kahn,Angus); Clinical Research,Investigation, and

Systems Modeling of AcuteIllness (CRISMA) Center,

Pittsburgh,Pennsylvania(Seymour, Kahn, Angus);

Divisionof Research,Kaiser Permanente, Oakland,

California (Liu);Department of Internal Medicine,

Universityof Michigan, Ann Arbor (Iwashyna,

Escobar); VeteransAffairs Center for Clinical

Management Research,Ann Arbor, Michigan

(Iwashyna,Escobar); Australia and New Zealand

IntensiveCare Research Centre, Departmentof Epidemiologyand PreventiveMedicine, Monash

University, Melbourne,Victoria, Australia

(Iwashyna,Escobar); Center for Clinical Studies,

JenaUniversityHospital,Jena, Germany

(Brunkhorst); Division of General Internal Medicine,

Universityof Washington, Seattle (Rea); Research

Group Clinical Epidemiology, IntegratedResearch

and Treatment Center,Center for SepsisControl

and Care,Jena UniversityHospital, Jena,Germany

(Scherag);Trauma,Emergency, and Critical Care

Program, Sunnybrook Health Sciences Centre;

InterdepartmentalDivision of Critical Care,

Universityof Toronto, Toronto,Ontario,Canada

(Rubenfeld); Critical Care Medicine,Guy’s and St

Thomas’ NHS FoundationTrust,London, England

(Shankar-Hari);BloomsburyInstituteof Intensive

Care Medicine,UniversityCollege London, London,England (Singer); FeinsteinInstitute for Medical

Research,Hofstra–North Shore–LongIsland Jewish

Schoolof Medicine, Steven and Alexandra Cohen

Children’s Medical Center,New HydePark,

New York (Deutschman).

Author Contributions: DrSeymour hadfull access

toall ofthedata inthestudyandtakes

responsibility forthe integrityof thedataand the

accuracy of thedataanalysis.

Study concept and design: Seymour,Iwashyna,

Rubenfeld,Kahn, Shankar-Hari, Deutschman,

Escobar,Angus.

Acquisition, analysis, or interpretation of data: Liu,

Iwashyna, Brunkhorst,Rea, Scherag, Kahn,Singer,

Escobar, Angus.

Drafting of themanuscript:Seymour, Singer,

Deutschman, Angus.

Critical revision of themanuscriptfor important

intellectual content: Liu, Iwashyna,Brunkhorst,Rea,

Scherag, Rubenfeld,Kahn, Shankar-Hari, Singer,

Deutschman, Escobar,Angus.

Statistical analysis: Seymour, Liu, Iwashyna,

Scherag.

Obtained funding: Escobar.

Administrative, technical, or material support:

Brunkhorst,Rea, Scherag, Deutschman, Escobar,Angus.

Study supervision: Deutschman, Escobar.

Conflict of Interest Disclosures: All authors have

completedand submittedtheICMJEFormfor

Disclosure of PotentialConflicts of Interest.Dr

Seymourreportsreceipt of personal fees from

Beckman Coulter.Dr Singer reports board

membershipswith InflaRx, Bayer, Biotest, and

Merck. Dr Deutschmanreports holding patents on

materials unrelatedto this work andreceiptof

personal fees fromtheCenters forDiseaseControl

and Prevention, theWorld Federationof Societies

of Intensive and Critical Care,the Pennsylvania

Assembly of Critical CareMedicine,the Society of

Critical Care Medicine,the Northern Ireland Society

of Critical Care Medicine,the International Sepsis

Forum, Stanford University, theAcute DialysisQuality Initiative,and the European Society of

Intensive Care Medicine.Dr Escobar reports receipt

of grantsfrom theNational Institutes of Health, the

Gordonand BettyMoore Foundation,Merck,

Sharpe& Dohme, and AstraZeneca-MedImmune.

No otherdisclosureswere reported.

Funding/Support: This work wassupported

in partby theNational Institutes of Health

(grants K23GM104022 and K23GM112018),

theDepartmentof VeteransAffairs (grantHSR&D

11-109),the Permanente Medical Group,

andthe Centerof SepsisControl andCare,funded

bythe GermanFederalMinistryof Educationand

Research(grant01 E01002/01E0 1502).

Roleof the Funder/Sponsor:Thefunding sources

hadno rolein thedesign andconduct of thestudy;

collection, management, analysis, and

interpretation of thedata; preparation,review, or

approval of themanuscript; anddecisionto submit

the manuscriptfor publication.

Disclaimer: Thisarticle does not necessarily

representthe view of theUS government or

Departmentof VeteransAffairs. Dr Angus,

Associate Editor, JAMA, had noroleinthe

evaluationof or decision to publishthisarticle.

Additional Contributions: We acknowledge the

European Society of IntensiveCare Medicine and

Society of Critical CareMedicine fortheirpartial

administrative supportof thiswork. We

acknowledgethe contributions of the 2016Third

International Consensus SepsisDefinitions Task

Forcemembers,who were notcoauthors, fortheir

reviewof themanuscript: John C.Marshall,MD,

Universityof Toronto, Toronto,Ontario,Canada;

DjilalliAnnane,MD, PhD, Critical Care Medicine,

Schoolof Medicine, Universityof Versailles,France;

Greg S.Martin, MD, Emory University Schoolof

Medicine,Atlanta, Georgia; Michael Bauer, MD,

Center for SepsisControl and Care,University

Hospital, Jena, Germany; Steven M. Opal,MD,

Infectious Disease Section, BrownUniversitySchool

of Medicine,Providence, RhodeIsland; RinaldoBellomo, MD,Australianand NewZealand Intensive

CareResearch Centre, Schoolof PublicHealth and

PreventiveMedicine, MonashUniversity,University

of Melbourne,and AustinHospital,Melbourne,

Victoria, Australia; GordonR. Bernard, MD,

VanderbiltInstitute for Clinical and Translational

Research,Vanderbilt University, Nashville,

Tennessee;Jean-Daniel Chiche, MD,PhD,

RéanimationMédicale-Hôpital Cochin, Descartes

University, CochinInstitute, Paris, France; CraigM.

Coopersmith,MD, Emory Critical Care Center,

Emory UniversitySchool of Medicine, Atlanta,







13/13

Georgia; Tomvan derPoll,MD,Academisch

Medisch Centrum, Amsterdam,the Netherlands;

Richard S. Hotchkiss,MD, WashingtonUniversity

Schoolof Medicine,St Louis,Missouri;Jean-Louis

Vincent, MD,PhD, UniversitéLibre de Bruxelles,

and Departmentof Intensive Care,Erasme

UniversityHospital, Brussels, Belgium;

andMitchellM. Levy, MD, Division of

Pulmonary and Critical Care Medicine,

BrownUniversitySchool of Medicine,Providence,

RhodeIsland. Thesecontributions wereprovided

without compensation.

REFERENCES

1. Liu V, EscobarGJ,Greene JD, etal. Hospital

deathsin patientswith sepsisfrom 2 independent

cohorts. JAMA. 2014;312(1):90-92.

2. Angus DC, Linde-Zwirble WT, Lidicker J,

Clermont G, CarcilloJ, PinskyMR. Epidemiology of

severesepsis in theUnitedStates:analysis of

incidence,outcome,and associatedcosts of care.

Crit Care Med . 2001;29(7):1303-1310.

3. Levy MM,FinkMP, Marshall JC,et al;

SCCM/ESICM/ACCP/ATS/SIS. 2001

SCCM/ESICM/ACCP/ATS/SIS International Sepsis

Definitions Conference. Crit Care Med . 2003;31(4):

1250-1256.

4. BoneRC, Balk RA,CerraFB, et al;ACCP/SCCM

Consensus ConferenceCommittee; American

Collegeof ChestPhysicians/Society of Critical Care

Medicine.Definitionsfor sepsisand organfailure

andguidelinesfor theuse of innovative therapiesin

sepsis. Chest . 1992;101(6):1644-1655.

5. Shankar-HariM, Phillips GS,LevyML, et al;

SepsisDefinitions Task Force. Developinga new

definition and assessing new clinical criteria for

septicshock: for theThird International Consensus

Definitionsfor Sepsis and Septic Shock(Sepsis-3).

JAMA. doi:10.1001/jama.2016.0289 .

6. Seymour CW, Rea TD, Kahn JM,WalkeyAJ, Yealy

DM,Angus DC.Severe sepsisin pre-hospital

emergencycare: analysis of incidence,care, and

outcome. Am J Respir Crit Care Med . 2012;186(12):

1264-1271.

7. HagelS, Ludewig K, FrosinskiJ, etal.

Effectiveness of a hospital-wide educational

programmefor infectioncontrol to reducethe rate

of health-care associated infections and related

sepsis(ALERTS)—methods and interim results

[in German]. DtschMedWochenschr . 2013;

138(34-35):1717-1722.

8. Vincent JL,Moreno R, Takala J,et al.The SOFA

(Sepsis-RelatedOrgan Failure Assessment) scoreto

describe organdysfunction/failure. IntensiveCare

Med . 1996;22(7):707-710.

9. LeGall JR,Klar J,Lemeshow S,et al;ICU Scoring

Group.The Logistic OrganDysfunction System:

a newwayto assessorgandysfunctionin the

intensivecare unit. JAMA. 1996;276(10):802-810.

10. CollinsGS, Reitsma JB,Altman DG,Moons KG.

Transparent Reporting of a MultivariablePrediction

Modelfor Individual Prognosis or Diagnosis

(TRIPOD): theTRIPOD statement. Ann InternMed .

2015;162(1):55-63.

11. Rangel-FraustoMS, Pittet D,Costigan M, Hwang

T, DavisCS, Wenzel RP. Thenaturalhistory of the

systemicinflammatoryresponse syndrome (SIRS):

a prospectivestudy. JAMA. 1995;273(2):117-123.

12. PerkinsNJ, Schisterman EF.The inconsistency

of “optimal”cutpointsobtained usingtwo criteria

basedon thereceiver operating characteristic

curve. Am J Epidemiol . 2006;163(7):670-675.

13. Seymour CW, Kahn JM,CookeCR, Watkins TR,

Heckbert SR,Rea TD. Predictionof critical illness

duringout-of-hospital emergencycare. JAMA.

2010;304(7):747-754.

14. KnausWA, ZimmermanJE, Wagner DP, Draper

EA, Lawrence DE. APACHE—AcutePhysiologyand

Chronic HealthEvaluation:a physiologically based

classification system. Crit Care Med . 1981;9(8):591-

597.

15. KramerAA, ZimmermanJE. Assessingthe

calibrationof mortality benchmarksin critical care:

theHosmer-Lemeshow testrevisited. Crit Care Med .

2007;35(9):2052-2056 .

16. Bland JM,Altman DG.Cronbach’sα. BMJ . 1997;

314(7080):572.

17. CampbellDT,Fiske DW. Convergent

and discriminantvalidationby the

multitrait-multimethod matrix. Psychol Bull .

1959;56(2):81-105.

18. Quan H,SundararajanV,Halfon P, etal. Coding

algorithms for defining comorbiditiesin ICD-9-CM

and ICD-10 administrative data. MedCare . 2005;43

(11):1130-1139.

19. Royston P, Sauerbrei W.Building multivariable

regression modelswith continuouscovariatesin

clinical epidemiology—withan emphasis on

fractional polynomials.Methods InfMed . 2005;44

(4):561-571.

20. Centers for Medicaid and Medicare Services.

Eligible Professional MeaningfulUse CoreMeasures:

Measure7 of13. 2014. https://www.cms.gov

/Regulations-and-Guidance/Legislation

/EHRIncentivePrograms/downloads/7_Record

_Demographics.pdf.Accessed January 1, 2016.

21. Barnato AE, Alexander SL, Linde-Zwirble WT,Angus DC.Racial variationin theincidence,care,

andoutcomesof severesepsis: analysis of

population, patient, and hospital characteristics.

Am J Respir Crit Care Med . 2008;177(3):279-284.

22. Dellinger RP, Levy MM,Rhodes A, etal;

Surviving SepsisCampaignGuidelinesCommittee

Including thePediatricSubgroup. Surviving Sepsis

campaign:international guidelinesfor management

ofseveresepsis andseptic shock:2012. Crit CareMed .

2013;41(2):580-637.

23. CecconiM, DeBacker D,Antonelli M,et al.

Consensus on circulatory shockand hemodynamic

monitoring: task force of theEuropeanSocietyof

IntensiveCare Medicine. IntensiveCare Med . 2014;

40(12):1795-1815.

24. LiuV,MorehouseJW, Soule J,Whippy A,Escobar GJ.Fluid volume, lactate values, and

mortality in sepsispatientswith intermediate

lactate values. Ann Am Thorac Soc . 2013;10(5):466-

473.

25. HanleyJA, McNeilBJ. The meaning anduse of

thearea undera receiver operating characteristic

(ROC)curve. Radiology . 1982;143(1):29-36.

26. FerrerR, Martin-Loeches I,Phillips G,et al.

Empiricantibiotictreatment reduces mortality in

severesepsis andseptic shockfrom thefirsthour:

results from a guideline-basedperformance

improvement program. Crit Care Med . 2014;42(8):

1749-1755.

27. Houck PM,Bratzler DW, NsaW,Ma A,

Bartlett JG. Timing of antibioticadministration and

outcomes for Medicare patients hospitalizedwith

community-acquired pneumonia. Arch InternMed .2004;164(6):637-644.

28. EscobarGJ,Gardner MN, GreeneJD,Draper D,

KipnisP. Risk-adjustinghospital mortality usinga

comprehensive electronicrecord in an integrated

health caredelivery system. MedCare . 2013;51(5):

446-453.

29. KaukonenKM, BaileyM, Pilcher D,Cooper DJ,

Bellomo R. Systemic inflammatoryresponse

syndrome criteria in defining severe sepsis. NEnglJ

Med . 2015;372(17):1629-1638.

30. VincentJL, deMendonçaA, CantraineF, etal.

Useof theSOFA score to assessthe incidenceof

organdysfunction/failure in intensivecare units:

results of a multicenter,prospectivestudy. CritCare

Med . 1998;26(11):1793-1800.

31. Ferreira FL,Bota DP, Bross A, Mélot C,Vincent JL.Serial evaluationof theSOFA score to

predictoutcomein critically ill patients. JAMA.

2001;286(14):1754-1758.

32. Charles D,Gabriel M,FurukawaMF. Adoption of

ElectronicHealthRecord SystemsamongUS

Non-Federal AcuteCare Hospitals: 2008-2013.

Washington, DC: Officeof the National Coordinator

for Health InformationTechnology;2014.ONC Data

Brief 16.

33. Rhee C, MurphyMV, Li L, PlattR, Klompas M;

Centers for Disease Controland Prevention

EpicentersProgram. Comparison of trends in sepsis

incidenceand codingusing administrative claimsvs

objective clinical data.Clin Infect Dis. 2015;60(1):

88-95.

34. ShapiroNI, Wolfe RE,MooreRB, SmithE,BurdickE, BatesDW.Mortalityin Emergency

DepartmentSepsis (MEDS) score:a prospectively

derived and validated clinical prediction rule.Crit

Care Med . 2003;31(3):670-675.

35. OpalSM. Concept of PIRO as a newconceptual

framework to understandsepsis. Pediatr CritCare

Med . 2005;6(3)(suppl):S55-S60.

36. Iwashyna TJ,Odden A, Rohde J,et al.

Identifyingpatientswith severe sepsisusing

administrative claims:patient-level validation of the

Angusimplementationof theInternational

Consensus Conferencedefinitionof severe sepsis.

MedCare . 2014;52(6):e39-e43.

37. Kellum JA,KongL, Fink MP, etal; GenIMS

Investigators. Understanding theinflammatory

cytokine response in pneumonia and sepsis: resultsof theGeneticand InflammatoryMarkers of Sepsis

(GenIMS) study. Arch InternMed . 2007;167(15):

1655-1663.


http://www.ncbi.nlm.nih.gov/pubmed/24838355http://www.ncbi.nlm.nih.gov/pubmed/24838355http://www.ncbi.nlm.nih.gov/pubmed/24838355http://www.ncbi.nlm.nih.gov/pubmed/11445675http://www.ncbi.nlm.nih.gov/pubmed/11445675http://www.ncbi.nlm.nih.gov/pubmed/11445675http://www.ncbi.nlm.nih.gov/pubmed/12682500http://www.ncbi.nlm.nih.gov/pubmed/12682500http://www.ncbi.nlm.nih.gov/pubmed/12682500http://www.ncbi.nlm.nih.gov/pubmed/12682500http://www.ncbi.nlm.nih.gov/pubmed/1303622http://www.ncbi.nlm.nih.gov/pubmed/1303622http://www.ncbi.nlm.nih.gov/pubmed/1303622http://jama.jamanetwork.com/article.aspx?doi=10.1001/jama.2016.0289&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288http://jama.jamanetwork.com/article.aspx?doi=10.1001/jama.2016.0289&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288http://jama.jamanetwork.com/article.aspx?doi=10.1001/jama.2016.0289&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288http://www.ncbi.nlm.nih.gov/pubmed/23087028http://www.ncbi.nlm.nih.gov/pubmed/23087028http://www.ncbi.nlm.nih.gov/pubmed/23087028http://www.ncbi.nlm.nih.gov/pubmed/23087028http://www.ncbi.nlm.nih.gov/pubmed/23934590http://www.ncbi.nlm.nih.gov/pubmed/23934590http://www.ncbi.nlm.nih.gov/pubmed/23934590http://www.ncbi.nlm.nih.gov/pubmed/23934590http://www.ncbi.nlm.nih.gov/pubmed/8844239http://www.ncbi.nlm.nih.gov/pubmed/8844239http://www.ncbi.nlm.nih.gov/pubmed/8844239http://www.ncbi.nlm.nih.gov/pubmed/8844239http://www.ncbi.nlm.nih.gov/pubmed/8769590http://www.ncbi.nlm.nih.gov/pubmed/8769590http://www.ncbi.nlm.nih.gov/pubmed/8769590http://www.ncbi.nlm.nih.gov/pubmed/25560714http://www.ncbi.nlm.nih.gov/pubmed/25560714http://www.ncbi.nlm.nih.gov/pubmed/25560714http://www.ncbi.nlm.nih.gov/pubmed/25560714http://www.ncbi.nlm.nih.gov/pubmed/7799491http://www.ncbi.nlm.nih.gov/pubmed/7799491http://www.ncbi.nlm.nih.gov/pubmed/7799491http://www.ncbi.nlm.nih.gov/pubmed/16410346http://www.ncbi.nlm.nih.gov/pubmed/16410346http://www.ncbi.nlm.nih.gov/pubmed/16410346http://www.ncbi.nlm.nih.gov/pubmed/20716737http://www.ncbi.nlm.nih.gov/pubmed/20716737http://www.ncbi.nlm.nih.gov/pubmed/20716737http://www.ncbi.nlm.nih.gov/pubmed/20716737http://www.ncbi.nlm.nih.gov/pubmed/7261642http://www.ncbi.nlm.nih.gov/pubmed/7261642http://www.ncbi.nlm.nih.gov/pubmed/7261642http://www.ncbi.nlm.nih.gov/pubmed/7261642http://www.ncbi.nlm.nih.gov/pubmed/17568333http://www.ncbi.nlm.nih.gov/pubmed/17568333http://www.ncbi.nlm.nih.gov/pubmed/17568333http://www.ncbi.nlm.nih.gov/pubmed/17568333http://www.ncbi.nlm.nih.gov/pubmed/9055718http://www.ncbi.nlm.nih.gov/pubmed/9055718http://www.ncbi.nlm.nih.gov/pubmed/9055718http://www.ncbi.nlm.nih.gov/pubmed/9055718http://www.ncbi.nlm.nih.gov/pubmed/13634291http://www.ncbi.nlm.nih.gov/pubmed/13634291http://www.ncbi.nlm.nih.gov/pubmed/13634291http://www.ncbi.nlm.nih.gov/pubmed/13634291http://www.ncbi.nlm.nih.gov/pubmed/16224307http://www.ncbi.nlm.nih.gov/pubmed/16224307http://www.ncbi.nlm.nih.gov/pubmed/16224307http://www.ncbi.nlm.nih.gov/pubmed/16224307http://www.ncbi.nlm.nih.gov/pubmed/16342923http://www.ncbi.nlm.nih.gov/pubmed/16342923http://www.ncbi.nlm.nih.gov/pubmed/16342923http://www.ncbi.nlm.nih.gov/pubmed/16342923https://www.cms.gov/Regulations-and-Guidance/Legislation/EHRIncentivePrograms/downloads/7_Record_Demographics.pdfhttps://www.cms.gov/Regulations-and-Guidance/Legislation/EHRIncentivePrograms/downloads/7_Record_Demographics.pdfhttps://www.cms.gov/Regulations-and-Guidance/Legislation/EHRIncentivePrograms/downloads/7_Record_Demographics.pdfhttps://www.cms.gov/Regulations-and-Guidance/Legislation/EHRIncentivePrograms/downloads/7_Record_Demographics.pdfhttp://www.ncbi.nlm.nih.gov/pubmed/17975201http://www.ncbi.nlm.nih.gov/pubmed/17975201http://www.ncbi.nlm.nih.gov/pubmed/17975201http://www.ncbi.nlm.nih.gov/pubmed/23353941http://www.ncbi.nlm.nih.gov/pubmed/23353941http://www.ncbi.nlm.nih.gov/pubmed/23353941http://www.ncbi.nlm.nih.gov/pubmed/23353941http://www.ncbi.nlm.nih.gov/pubmed/25392034http://www.ncbi.nlm.nih.gov/pubmed/25392034http://www.ncbi.nlm.nih.gov/pubmed/25392034http://www.ncbi.nlm.nih.gov/pubmed/25392034http://www.ncbi.nlm.nih.gov/pubmed/24004068http://www.ncbi.nlm.nih.gov/pubmed/24004068http://www.ncbi.nlm.nih.gov/pubmed/24004068http://www.ncbi.nlm.nih.gov/pubmed/24004068http://www.ncbi.nlm.nih.gov/pubmed/7063747http://www.ncbi.nlm.nih.gov/pubmed/7063747http://www.ncbi.nlm.nih.gov/pubmed/7063747http://www.ncbi.nlm.nih.gov/pubmed/24717459http://www.ncbi.nlm.nih.gov/pubmed/24717459http://www.ncbi.nlm.nih.gov/pubmed/24717459http://www.ncbi.nlm.nih.gov/pubmed/24717459http://www.ncbi.nlm.nih.gov/pubmed/15037492http://www.ncbi.nlm.nih.gov/pubmed/15037492http://www.ncbi.nlm.nih.gov/pubmed/15037492http://www.ncbi.nlm.nih.gov/pubmed/15037492http://www.ncbi.nlm.nih.gov/pubmed/23579354http://www.ncbi.nlm.nih.gov/pubmed/23579354http://www.ncbi.nlm.nih.gov/pubmed/23579354http://www.ncbi.nlm.nih.gov/pubmed/23579354http://www.ncbi.nlm.nih.gov/pubmed/25776936http://www.ncbi.nlm.nih.gov/pubmed/25776936http://www.ncbi.nlm.nih.gov/pubmed/25776936http://www.ncbi.nlm.nih.gov/pubmed/25776936http://www.ncbi.nlm.nih.gov/pubmed/9824069http://www.ncbi.nlm.nih.gov/pubmed/9824069http://www.ncbi.nlm.nih.gov/pubmed/9824069http://www.ncbi.nlm.nih.gov/pubmed/9824069http://www.ncbi.nlm.nih.gov/pubmed/11594901http://www.ncbi.nlm.nih.gov/pubmed/11594901http://www.ncbi.nlm.nih.gov/pubmed/11594901http://www.ncbi.nlm.nih.gov/pubmed/11594901http://www.ncbi.nlm.nih.gov/pubmed/25258352http://www.ncbi.nlm.nih.gov/pubmed/25258352http://www.ncbi.nlm.nih.gov/pubmed/25258352http://www.ncbi.nlm.nih.gov/pubmed/25258352http://www.ncbi.nlm.nih.gov/pubmed/12626967http://www.ncbi.nlm.nih.gov/pubmed/12626967http://www.ncbi.nlm.nih.gov/pubmed/12626967http://www.ncbi.nlm.nih.gov/pubmed/12626967http://www.ncbi.nlm.nih.gov/pubmed/15857560http://www.ncbi.nlm.nih.gov/pubmed/15857560http://www.ncbi.nlm.nih.gov/pubmed/15857560http://www.ncbi.nlm.nih.gov/pubmed/15857560http://www.ncbi.nlm.nih.gov/pubmed/23001437http://www.ncbi.nlm.nih.gov/pubmed/23001437http://www.ncbi.nlm.nih.gov/pubmed/23001437http://www.ncbi.nlm.nih.gov/pubmed/17698689http://www.ncbi.nlm.nih.gov/pubmed/17698689http://www.ncbi.nlm.nih.gov/pubmed/17698689http://www.ncbi.nlm.nih.gov/pubmed/17698689http://www.jama.com/?utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288http://www.jama.com/?utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288http://www.ncbi.nlm.nih.gov/pubmed/17698689http://www.ncbi.nlm.nih.gov/pubmed/17698689http://www.ncbi.nlm.nih.gov/pubmed/23001437http://www.ncbi.nlm.nih.gov/pubmed/15857560http://www.ncbi.nlm.nih.gov/pubmed/15857560http://www.ncbi.nlm.nih.gov/pubmed/12626967http://www.ncbi.nlm.nih.gov/pubmed/12626967http://www.ncbi.nlm.nih.gov/pubmed/25258352http://www.ncbi.nlm.nih.gov/pubmed/25258352http://www.ncbi.nlm.nih.gov/pubmed/11594901http://www.ncbi.nlm.nih.gov/pubmed/11594901http://www.ncbi.nlm.nih.gov/pubmed/9824069http://www.ncbi.nlm.nih.gov/pubmed/9824069http://www.ncbi.nlm.nih.gov/pubmed/25776936http://www.ncbi.nlm.nih.gov/pubmed/25776936http://www.ncbi.nlm.nih.gov/pubmed/23579354http://www.ncbi.nlm.nih.gov/pubmed/23579354http://www.ncbi.nlm.nih.gov/pubmed/15037492http://www.ncbi.nlm.nih.gov/pubmed/15037492http://www.ncbi.nlm.nih.gov/pubmed/24717459http://www.ncbi.nlm.nih.gov/pubmed/24717459http://www.ncbi.nlm.nih.gov/pubmed/7063747http://www.ncbi.nlm.nih.gov/pubmed/24004068http://www.ncbi.nlm.nih.gov/pubmed/24004068http://www.ncbi.nlm.nih.gov/pubmed/25392034http://www.ncbi.nlm.nih.gov/pubmed/25392034http://www.ncbi.nlm.nih.gov/pubmed/23353941http://www.ncbi.nlm.nih.gov/pubmed/23353941http://www.ncbi.nlm.nih.gov/pubmed/17975201https://www.cms.gov/Regulations-and-Guidance/Legislation/EHRIncentivePrograms/downloads/7_Record_Demographics.pdfhttps://www.cms.gov/Regulations-and-Guidance/Legislation/EHRIncentivePrograms/downloads/7_Record_Demographics.pdfhttps://www.cms.gov/Regulations-and-Guidance/Legislation/EHRIncentivePrograms/downloads/7_Record_Demographics.pdfhttps://www.cms.gov/Regulations-and-Guidance/Legislation/EHRIncentivePrograms/downloads/7_Record_Demographics.pdfhttp://www.ncbi.nlm.nih.gov/pubmed/16342923http://www.ncbi.nlm.nih.gov/pubmed/16342923http://www.ncbi.nlm.nih.gov/pubmed/16224307http://www.ncbi.nlm.nih.gov/pubmed/16224307http://www.ncbi.nlm.nih.gov/pubmed/13634291http://www.ncbi.nlm.nih.gov/pubmed/13634291http://www.ncbi.nlm.nih.gov/pubmed/9055718http://www.ncbi.nlm.nih.gov/pubmed/9055718http://www.ncbi.nlm.nih.gov/pubmed/17568333http://www.ncbi.nlm.nih.gov/pubmed/17568333http://www.ncbi.nlm.nih.gov/pubmed/7261642http://www.ncbi.nlm.nih.gov/pubmed/7261642http://www.ncbi.nlm.nih.gov/pubmed/20716737http://www.ncbi.nlm.nih.gov/pubmed/20716737http://www.ncbi.nlm.nih.gov/pubmed/16410346http://www.ncbi.nlm.nih.gov/pubmed/7799491http://www.ncbi.nlm.nih.gov/pubmed/25560714http://www.ncbi.nlm.nih.gov/pubmed/25560714http://www.ncbi.nlm.nih.gov/pubmed/8769590http://www.ncbi.nlm.nih.gov/pubmed/8844239http://www.ncbi.nlm.nih.gov/pubmed/8844239http://www.ncbi.nlm.nih.gov/pubmed/23934590http://www.ncbi.nlm.nih.gov/pubmed/23934590http://www.ncbi.nlm.nih.gov/pubmed/23087028http://www.ncbi.nlm.nih.gov/pubmed/23087028http://jama.jamanetwork.com/article.aspx?doi=10.1001/jama.2016.0289&utm_campaign=articlePDF%26utm_medium=articlePDFlink%26utm_source=articlePDF%26utm_content=jama.2016.0288http://www.ncbi.nlm.nih.gov/pubmed/1303622http://www.ncbi.nlm.nih.gov/pubmed/12682500http://www.ncbi.nlm.nih.gov/pubmed/12682500http://www.ncbi.nlm.nih.gov/pubmed/11445675http://www.ncbi.nlm.nih.gov/pubmed/24838355

Date post:	07-Jul-2018
Category:	Documents
Upload:	carlos-poveda
View:	216 times
Download:	0 times

Criterios Clinicos de Sepsis Seymour

Documents