Georeferencing of deaths from sepsis in the city of São Paulo

(1)

w w w . e l s e v i e r . c o m / l o c a t e / b j i d

The

Brazilian

Journal

of

INFECTIOUS

DISEASES

Original

article

Georeferencing

of

deaths

from

sepsis

in

the

city

of

São

Paulo

Décio

Diament

∗

_,

_Fernando

_Colombari,

_Adriana

_Serra

_Cypriano,

_Luis

_Fernando

_Lisboa,

Bento

Fortunato

Cardoso

dos

Santos,

Miguel

Cendoroglo

Neto,

Ary

Serpa

Neto,

Eliezer

Silva

DepartmentofCriticalCareMedicineandAtlasdaSaúdedeSãoPaulo,HospitalIsraelitaAlbertEinstein,SãoPaulo,SP,Brazil

a

r

t

i

c

l

e

i

n

f

o

Articlehistory:

Received22August2015 Accepted7November2015 Availableonline2February2016

Keywords: Sepsis Mortality Infection Georeferencing

a

b

s

t

r

a

c

t

Objective:Theaimofthepresentstudywastoobtaininformationaboutdeathsduetosepsis inSãoPaulofrom2004to2009andtheirrelationshipwithgeographicaldistribution.

Methods:Causesofdeath,bothmainandsecondary,weredefinedaccordingtothecodes oftheInternationalClassificationofDiseaseversion10(ICD-10)containedinthedatabase. Sepsis,septicshock,multipleorganfailure,pneumonia,urinarytractinfection, peritoni-tisandotherintraabdominalinfections,skinandsofttissueinfections(includingsurgical woundinfection)andmeningitiswereconsideredasimmediatecauseofdeathorasthe conditionleadingtotheimmediatecauseofdeathrelatedorassociatedtosepsis.

Results:Intheanalyzedperiod,therewasa15.3%increaseintheabsolutenumberofdeaths fromsepsisinSãoPaulo.Themeannumberofdeathsduringthisperiodwas28,472±1566. Mostdeathsduetosepsisandsepsis-relateddiseasesoverthestudiedperiodoccurredin ahospitalorhealthcarefacility,showingthatmostofthepatientsreceivedmedicalcare duringtheeventthatledtodeath.Weobservedasignificantconcentrationofdeathsinthe mostpopulousregions,tendingmoretowardthecenterofthecity.

Conclusions: Georeferencingdatafromdeathcertificatesorothersourcescanbea power-fultooltouncoverregionalepidemiologicaldifferencesbetweenpopulations.Ourstudy revealedanevendistributionofsepsisallovertheinhabitedareasofSãoPaulo.

Introduction

Sepsisisasyndromeresultingfromsystemicmanifestations of infection associated with high incidence and mortality rate. Also, care of patients with sepsis costs as much as

∗ _{Corresponding}_author_at_:_Av._Albert_Einstein,₇₀₀_São_Paulo,_Brazil.

E-mailaddress:decio.diament@einstein.br(D.Diament).

$50,000perpatient,resultinginaneconomicimpactfor soci-etyofnearly$17billionannuallyintheUnitedStatesalone.1,2

Demographicandsocioeconomicvariables,pre-existing dis-eases, and accesstohealthcareare factors thataffect the incidenceand mortalityofpatients withseveresepsis and septicshock.2–6_For_instance,_an_American_{retrospective}_study

http://dx.doi.org/10.1016/j.bjid.2015.11.010

(2)

showedahigherincidenceofsepsisamongblacks,men,and peoplefromabjectpovertyandurbanization.Thesamegroups alsohadhigherratesofmortality.3

Disparitiesalsooccurasthesourceofinfection.Menhave morerespiratoryinfections,whilewomenhavemore geni-tourinaryinfections.Regardlessoftheinfectionsiteinmen, infectionswithGram-positivebacteriaaremoreprevalent.5

ThesameistrueforblackAmericans,whohavemore comor-bidities and longer hospitalization. The reasons for these disparitiesarenotwellunderstood.InBrazil,studiesonthe epidemiologyofsepsisarelimitedtointensivecareunitsfrom somepublicandprivatehospitals.Thosestudieshaveshowed thatthemortalityrateishigherinBrazilcomparedtoother countries.2,7

Georeferencing, which relates information to the geo-graphic location, is an important and innovative tool in epidemiological studies8,9 _that _has _been _used _in _various

scenarios,10–12_aiming_to_clarify_aspects_of_geographic

distri-bution.Theapplicationofthemethodologyofgeoreferencing assists in studies of population mobility correlating with socioeconomicandgeographicsegmentationofagivenarea. Infact,theBrazilianpopulationhasimportantsocioeconomic disparitiesinaccesstohealthcare,whichcouldimpactsepsis outcomes.Hence,georeferencingmaybeusedfor understand-ingtherelationshipbetweensocioeconomicattributesandthe outcomesofsepticpatients.

Theaimofthe presentstudy wastoobtain information aboutdeathsduetosepsisinSãoPaulofrom2004to2009and theirrelationshipwithgeographicaldistribution.

Materials

and

methods

Deathsdatabasewasextractedfrom“Programade Aprimora-mentodasInformac¸õesdeMortalidade”(PRO-AIM;“Mortality InformationImprovementProgram”),coordinatedbytheSão PauloCountyHealthAuthorityandDataProcessingCompany (PRODAM). Death certificate information was collected for residentswhodiedwithinthecountyboundaries.Addresses were standardized and cross-match coded with a street databaseusingfuzzytechniquesaimingtocleanstrange char-acters or errors. Death certificates from 2004to 2009were analyzedforage,sex,primaryandsecondarycauseofdeath, patients’address,andplaceofdeath.

Causesofdeath,bothmainandsecondary,weredefined accordingtothecodes oftheInternationalClassificationof Diseaseversion10(ICD-10)containedinthedatabase.Sepsis, septicshock,multipleorganfailure,pneumonia,urinarytract infection, peritonitis and other intra-abdominal infections, skinandsofttissueinfections(includingsurgicalwound infec-tion),andmeningitiswereconsideredasimmediatecauseof deathorastheconditionleadingtotheimmediatecauseof deathrelatedorassociatedtosepsis.

Whendeathoccurred inhospitalsorhealthinstitutions, theseestablishmentswereidentifiedbytheNationalHealth Establishments Registry (Cadastro Nacional de Estabeleci-mentos de Saúde (CNES)) or Health System Ambulatory Information(SistemadeInformac¸õesAmbulatoriaisdoSUS (SIASUS)) databases.Correspondenceofhealthinstitutions’ addressesbetweenthetwodatabaseswasmadetoobtaincode

uniformity inorder toavoid duplicateormissing informa-tion.Thespatialdistributionofdeathswasperformedusing patients’residenceaddressandplaceofdeath.

Humandevelopmentindex(HDI;UnitedNations Develop-mentProgram)wasusedtogradethesocioeconomicstatus ofSãoPaulo Countydistricts.Differencesbetweendistricts werecomparedfornumberofdeaths,theircauses,andtheir relation to the HDI and mortality coefficients. Data from theBrazilian2000Census(InstitutoBrasileirodeGeografiae Estatística(IBGE))wasusedtogeneratemortalitycoefficients foreverydistrictinSãoPauloCounty,accordingtopopulation estimatesforthatyear.

ComparisonoftwoproportionswasdoneusingZ statis-ticswithnormaldistributionand95%confidenceintervalsor Chi-squaretestswhenappropriate.Spearmanlinear regres-sionwasusedforcorrelationbetweentheHDIandmortality rates.MapsweregeneratedusingtheMAPINFOProfessional 9.0,usingdatafromtheyear2009asanexample.All statisti-calanalyseswereperformedwithExcel2010(Microsoft,USA) andSPSSv.13.0(IBMCorporation,NewYork,USA).Forall anal-yses, two-sided p-values<0.05 were considered significant. ThestudywasapprovedbytheEthicsCommitteeinResearch oftheSecretariaMunicipaldeSaúde(CEP-SMS).Allpatient recordsandinformationwereanonymizedandde-identified priortoanalysis.

Results

Totaldeathsduetosepsis

Intheanalyzedperiod,therewasa15.3%increaseinthe abso-lutenumberofdeathsfromsepsisinSãoPaulo,risingfrom 27,135in2004to31,286in2009.Themeannumberofdeath duringthisperiodwas28,472±1566.Foranestimatedmean populationof10,965,000inhabitantsduringthisperiod,the mortalitycoefficientwas259.55/100,000inhabitants.

Distributionofdeathsaccordingtogender,age,andage group

ThenumberofdeathsfromsepsisinSãoPauloincreasedwith ageofthepatients,being8905(5.2%)upto18years,51,166 (29.9%)from19to64years,and110,080(64.4%)for65years orolder.Ageinformationwasnotavailablefor680patients (0.4%). The distribution ofdeathsaccording to genderwas nearlyequalforbothsexes:51%inmalesand49%infemales. Thisdistributionremainedstableduringallperiodsfrom2004 to2009.However,whenweanalyzeddeathsbysex,stratifying thembyagegroups,thereweremoredeathsamongmalesin theagegroupupto18years(53.9%vs.46.1%)and19–64years (61%vs. 39%).Intheagegroupabove65 years,the propor-tionreversedto46.3%formalesand53.7%forfemales.These differenceswerestatisticallysignificant(Table1).

Distributionofdeathsaccordingtoregionsandthe humandevelopmentindex

Wechoosetheyear2009tocalculatethemortalityrateper 100,000 inhabitants in different regions of São Paulo. The

(3)

Table1–Distributionofdeathsaccordingtoagegroups from2004to2009.

Agegroup Female Male Total

Upto18years 4111 4809 8920

19to64years 19,963 31,203 51,166

65yearsorolder 59,416 51,139 110,555

Notavailable 33 155 188

Total 83,523 87,306 170,829

mortality rate was higher in regions where the HDI was higher.Onepossibleexplanationforthisfactisthatinregions withahigherHDIlifeexpectancywashigher.Consequently, therewasagreaterlikelihoodofpatientsdevelopingsepsisor sepsis-relatedinfections,especiallypneumonia,sincethese conditionsweremorefrequentintheelderly.Themortality rateforpneumoniavariedfrom71per100,000inhabitantsin thecounty’ssouthernregion,whichwasthemostpopulated andhadthelowestHDI(0.794),to135per100,000inhabitants inthecounty’scentralregion,whichwaslesspopulatedand hadthehighestHDI(0.906)(r=0.87;p<0.05).

Weanalyzedsepsisdeathsintwopopulationgroups:onein districtswithHDIlessthan0.76andtheotherindistrictswith anHDIhigherthan0.96.DistrictswithalowerHDIhada popu-lationof1,139,942inhabitants,having39%ofthepopulationin thebelow-18-yearsagegroup,58%between19and64years, and3%65yearsorolder.Higher-HDIdistricts,with146,127 inhabitants,hadadifferentagegroupdistribution:17%upto 18years,67%between19and64years,and15%65yearsor older.IndistrictswithalowerHDI,percentageofdeathswere 10%intheagegroupupto18years,44%fortheagegroup between19and64years,and47%forthegroup65yearsor older.Mortalitycoefficientswere43.23,131.59,and2809.23per 100,000,respectively.FordistrictswithahigherHDI, percent-ageofdeathswere1%fortheagegroupupto18years,12%

between19and64years,and86%for65yearsorolder. Mor-talitycoefficientswere27.41,74.52,and2231.06per100,000, respectively.

WhencomparedtodistrictswithahighHDI,low-HDI dis-tricts had ahigher percentageofdeathsintheage groups up to18years (1%vs.10%;difference=0.9;95%confidence interval[CI]:0.07–0.12;Z=6.99;p<0.0001)andbetween19and 64years(12%vs.44%;diff.=0.32;95%CI=0.28–0.36;Z=14.07;

p<0.0001).However, inhigh-HDIdistricts deathrateswere higherfortheagegroupabove65years(47%vs.86%;diff.= -0.39;95% CI=−0.44to −0.34;Z=16.66;p<0.0001).Mortality coefficientwasnotstatisticallydifferentintheagegroupup to18yearsbetweenlow-andhigh-HDIdistricts,butitwas sig-nificantlyhigherintheagegroupbetween19and64yearsin thelow-HDIdistricts(p<0.001).Fortheagegroupof65years orolder,themortalitycoefficientwashigherinthehigh-HDI districts(p<0.001).Nevertheless,themortalitycoefficientfor the wholepopulationwashigherforthe high-HDIdistricts (Table2).

Distributionofdeathsaccordingtothelocaleofoccurrence Mostdeathsduetosepsisoverthestudiedperiodoccurred in a hospital or health carefacility, showing that mostof thepatientsreceivedmedicalcareduringtheeventthatled to death. However, about 6% of cases died before reach-ing a medical facility or receiving proper medical care, most of them at a patient’s home. Most deaths occurred inhospitalsandhealthcarefacilities;52.4% wereinpublic hospitals (112institutions, 33%), 46.4% inprivate hospitals (178 institutions, 53%), and 1.2% in non-identified places (48 institutions, 14%). Theaverageload ofdeathswas 79% higher in public institutions, which had a mean of 737 deaths per institution compared to 411 in private institu-tions.

Populational density (Inhab/Km2)

A

B

C

Human development index - HDI 2000 Above 11.900

9.000 a 11.900 7.000 a 9.000

0.930 a 0.961 Sepsis related deathsin 2009 - all ages. (1 dot=1 death)

0.881 a 0.930 0.827 a 0.881 0.701 a 0.827 Up to 7.000

Fig.1–(A)EstimatedSãoPaulocitypopulationdensityin2009;(B)estimatedhumandevelopmentindexbycitydistrictsin 2009;(C)totaldistributionofdeathsduetosepsisandillnessespotentiallyrelatedtosepsisinSãoPauloin2009.

(4)

Table2–Distributionofpopulationandsepsisdeathsbyagegroupsinlow-andhigh-HDIdistrictsinthecityofSão Paulointheyearof2009.

Agegroup LowerHDI(<0.76) HigherHDI(>0.96) Difference

(95%CI)

pvalue

Population % Sepsis death

% Mortalitya _Population _% _Sepsis death % Mortalitya ≤18years 441,800 38.8 191 9.5 43.23 25,538 17.5 7 1.2 27.41 15.82(37.03to−5.39) n.s. 18–65years 664,930 58.3 875 43.8 131.59 97,954 67.0 73 12.5 74.52 57.07(76.25to37.89) <0.005 ≥65years 33,212 2.9 933 46.7 2809.23 22,635 15.5 505 86.3 2231.06 −578.2(−840.1to−316.2) <0.005 Total 1,139,942 100 1999 100 175.36 146,127 100 585 100 400.34 <0.005

HDI,humandevelopmentindex;CI,confidenceinterval;n.s.,non-significant. Chi-square:289,91;p<0.001(fornumberofdeaths).

a _Mortality_coefficient_is_per_100,000_inhabitants.

Distributionofdeathsaccordingtoagegroupandaddress Usingageoreferencingtechnique,deathsduetosepsisand ill-nessespotentiallyrelatedtosepsisthatoccurredinSãoPaulo in2009wereplottedonamap(Fig.1).Weobservedasignificant concentrationofdeathsinthemostpopulousregions, ten-dingmoretowardthecenterofthecity.However,insparsely populatedextremestherewerefewcasesofdeaths.

Distributionofdeathsaccordingtopublicorprivate hospitals

The distribution of deaths by private or public hospitals showedatendencyforthe concentrationofcasesofdeath byregionscoveredbythehospital,asseeninFig.2AandB. However,asseveralhospitalsservingtheprivatenetworkwere locatedinthedowntownarea,therewasagreatmovementof patientsinsomeregions,afactlessevidenttopublic hospi-tals,wherecasestendedtofocusinthecoverageareaofthese services.

Whenwedistributedthecasesofdeathbyplaceof occur-rence(privateorpublichospital)andbyagegroup,therewas agreaterpercentageofdeathsintheagegroupof65yearsor

olderinprivatehospitals,followingthegeneraldistributionof deaths.However,inpublichospitals,despitethesimilar dis-tribution,thereweremoredeathsintheagegroupbetween 19and64years.

Distributionofdeathsaccordingtocause

Pneumonia,sepsis,andmultipleorganfailurewerethemain causesofdeathreportedondeathcertificatesfortheperiod studied(Table3).However,whenwestudiedtheproportion ofeachcauseindifferentagegroups,wenotedthatthe pro-portionvaried.Intheagegroupupto18years,mostdeaths wereduetoinfectionoftheurinarytract,whileattheageof65 yearsorolder,pneumonia,sepsis,andmultipleorganfailure werethemostcommoncauses.

Welookedatsecondarycausesofdeathtoexplain differ-encesindeathratesforhigher-andlower-HDIdistricts.For theagegroupupto18years,therewasasignificantdifference insecondarycausesofdeathonlyforpulmonaryand neuro-logicaldiseases,beinghigherinlower-HDIdistricts(p<0.001). Inlower-HDIdistricts,fortheagegroupbetween19 and64 years,theproportionofdeathsduetosecondarycauseswas significantlyhigherforpulmonary,neoplastic,cardiovascular,

Sepsis related deaths in 2009

A

B

Selected private hospitals Sepsis related deaths in 2009_{Selected public hospitals}

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

Fig.2–(A)Georeferencingorigin–destinyofdeathsoccurringinthesevenprivatehospitalswiththehighestmortality;(B) Georeferencingorigin–destinyofdeathsoccurringinthesevenpublichospitalswiththehighestmortality.

(5)

Table3–Maincausesofsepsisbyagegroupfrom2004to2009.

Sepsis-causes Agegroups Total

≤18years 18–65years ≥65years

Pneumonia 2532 16,872 46,266 65,670

Sepsis(nosource) 4878 16,378 34,488 55,744

Multipleorganfailure 950 14,633 26,117 41,700

Peritonitis 38 677 800 1515

Meningitis 364 1140 587 2091

Otherabdominalinfections 38 1273 1540 2851

Skinandsofttissueinfections 7 187 280 474

Urinarytractinfections 98 6 2 106

Total 8905 51,166 110,080 170,151

digestive,infectious,endocrine,neurologic,andgenitourinary diseases (p<0.001). For the agegroup of 65 years or older, wefoundahigherdeathratefromsecondarycausesin dis-tricts with higher HDI (p<0.001) ofpulmonary, neoplastic, cardiovascular, digestive, infectious, endocrine, neurologic, genitourinary,andboneandjointdiseases.

Discussion

Asfarasweknow,nostudy hasdealtwithgeoreferencing ofsepsistodate.However, georeferencingis awell-known andwidelyusedtechniqueinmanyconditionsinthemedical literature.Duringthesix-yearstudyperiod,theabsolute num-berofsepsisorsepsis-relatedinfection deathshada15.6% increase.Thiscan beattributedtothe populationincrease associatedandwithbetterdiagnosisoftheseconditions,as duringthestudiedperiodeducationalcampaigneffortswere maderegardingtheSurvivingSepsisCampaign(SSC).13 _The

number ofdeathsincreasedwithage asexpectedbut was distributedalmostequallybetweengenders.However,forthe youngeragegroupsdeathsweremorenumerousforthemale gender,whichisperhapsareflexofhormonalprotectionin thefemalegender.Fortheoldergroup,withalargeramount ofwomenbeinginmenopause,theproportionoffemaleswas higher.14

WeobservedapositivecorrelationbetweenHDIandthe mortalitycoefficient.HDIiscomposedofmanyvariables,one ofthembeinglifeexpectancyatbirth.Thus,HDIvarieswith populationage,andinaregionwithlongerlifeexpectancyit willbehigherbecauseagreaterproportionofthepopulation willbeolder.Asthemortalitycoefficientishigherinolder populationgroups,correlationbetweenthe twovariables is positive.IndistrictswithgreaterHDI,thepopulationisolder andthemortalitycoefficientishigher.

Sepsismortalityishigherinlow-incomecountries, proba-blybecauseoflackofaccesstohealthcarefacilities.15_Sepsis

acquisitionriskishigherinpoorer neighborhoods.16 _Sepsis

deathratewashigherindistrictswithlowerHDI,butintheage groupof19to64yearsitwasdisproportionatelyhigher.Thus, sepsisreacheshigherdeathratesinyoungerandpoorer popu-lationsin São Paulo.However, the mortalitycoefficient for thewholepopulationwashigherinhigh-HDIdistricts,again reflectinganolderpopulationwithgreaternumberofdeaths. Themajorityofdeathsoccurredinhealthcarefacilities,but

the burdenwas higherforthe publichealthsystem, which attendsthemajorityoftheBrazilianpopulation.

Ourstudyshowedanevendistributionofsepsisamongall inhabiteddistrictsoftheSãoPaulomunicipality,evenwhen splitbyagegroups.Accesstohealthcarefacilitieswasalmost universalincaseofseverediseasesandtendedtoconcentrate aroundthenearesthospital,indifferentlyofpublicorprivate institutions.Greatdisplacementswerenottherule,sincethe cityisrelativelywellservedbymanyhospitals.Still,thismay notoccurinotherlocalitiesinthecountry.Thedemandfor high-complexityhealthservicestendstoconcentrateinlarge urbancenters,generatinggreatdisplacementsofpatientsall overthecountry.17

When consideringthe causeofdeathdeclaredondeath certificates,pneumoniarankedfirst,followedbysepsisand multipleorganfailure.IntheBrazilianSepsis Epidemiologi-calStudy(BASES),2_respiratory _infections_accounted_for_the

majorityofhospitaladmissionsasacauseofsepsisor sep-ticshock.ThesamewasobservedintheSepsisBrazilStudy,7

wherepulmonaryinfectionswereresponsiblefor69%of sep-sis or septic shock cases. However, when we look at the relative frequencyofeach cause byage group,pneumonia and sepsis were relatively morefrequent in the older age groups, and in younger patients they were an important cause,despitetheirlowabsolutenumbers.AEuropean pedi-atric community-acquired septic shock study showed that pneumococcaldiseasecomprised10.8%ofallinfectionsand 9.8%ofurinarytractinfections,butthemajorityofthecases in this sample had meningococcal disease (23.6%).18 _This

studysuggestedthaturinarytractinfectionshadalmostequal importanceofpneumoniaasacauseofsepticshockin Euro-peanchildren.

The secondary cause analysis revealed some striking differencesbetweenpopulationslivingindistrictswithhigh orlowHDIinthethreeagegroups.Fortheagegroupupto 18yearsold,moreindividualswhodiedinthelow-HDIgroup had a secondary cause inthe pulmonaryand neurological ICD-10 categories. Similarly, in the age group between 19 and 64 years in the low-HDI group, a significantly higher proportion ofindividuals had secondary causes forseveral ICD-10 categories (pulmonary, neoplastic, cardiovascular, digestive, infectious, endocrine, neurologic, and genitouri-nary diseases). However, when we looked atthe above-65 group,theseproportionswerehigherforthehigh-HDIgroup in the very same ICD-10 categories. This demonstrated

(6)

that the patients in the 19-to-64-years group had more comorbidities,suggestingthatthispopulationhadtheworst health conditions, probably related to poverty. This could partiallyexplainthe higherdeathrateobservedinthisage group.

Inconclusion,georeferencingdatafromdeathcertificates orothersourcescanbeapowerfultooltouncoverregional epidemiologicaldifferencesbetweenpopulations.Ourstudy revealed an even distribution of sepsis and sepsis-related infectionsallovertheinhabitedareasofthecityofSãoPaulo. However,differences betweenlow- and high-income areas andagegroupsweredemonstratedbelowthisveilof equal-case geographic distribution. This kind ofstudy is able to showrelevantepidemiologicalinformationthatcanbeused toimplementpublichealthmeasuresbygovernmentaland privatesectorsofthehealthindustry.

Conflicts

of

interest

Theauthorsdeclarenoconflictsofinterest.

r

e

f

e

r

e

n

c

e

s

1. AngusDC,Linde-ZwirbleWT,LidickerJ,etal.Epidemiologyof severesepsisintheUnitedStates:analysisofincidence, outcome,andassociatedcostsofcare.CritCareMed. 2001;29:1303–10.

2. SilvaE,PedroMA,SogayarAC,etal.Braziliansepsis epidemiologicalstudy(BASESstudy).CritCare. 2004;8:R251–60.

3. BarnatoAE,AlexanderSL,Linde-ZwirbleWT,AngusDC. Racialvariationintheincidence,care,andoutcomesof severesepsis:analysisofpopulation,patient,andhospital characteristics.AmJRespirCritCareMed.2008;177:279–84. 4. DombrovskiyVY,MartinAA,SunderramJ,PazHL.Occurrence

andoutcomesofsepsis:influenceofrace.CritCareMed. 2007;35:763–8.

5.EsperAM,MossM,LewisCA,etal.Theroleofinfectionand comorbidity:factorsthatinfluencedisparitiesinsepsis.Crit CareMed.2006;34:2576–82.

6.MartinGS,ManninoDM,EatonS,MossM.Theepidemiology ofsepsisintheUnitedStatesfrom1979through2000.NEngl JMed.2003;348:1546–54.

7.SalesJuniorJAL,DavidCM,HatumR,etal.An

epidemiologicalstudyofsepsisinintensivecareunits– sepsisBrazilstudy.RevBrasTerIntens.2006;18:9–17. 8.HillLL.Georeferencing:thegeographicassociationsof

information.Cambridge,MA:MassachusettsInstituteof Technology;2006.

9.WieczorekWF,DelmericoAM.Geographicinformation systems.ComputStat.2009;1:167–86.

10.FangL,YanL,LiangS,etal.Spatialanalysisofhemorrhagic feverwithrenalsyndromeinChina.BMCInfectDis.2006;6:77. 11.PeledR,ReuveniH,PliskinJS,etal.Defininglocalitiesof

inadequatetreatmentforchildhoodasthma:aGISapproach. IntJHealthGeogr.2006;5.

12.ReinhardtM,EliasJ,AlbertJ,etal.EpiScanGIS:anonline geographicsurveillancesystemformeningococcaldisease. IntJHealthGeogr.2008;7:33.

13.TelesJM,SilvaE,WestphalG,FilhoRC,MachadoFR.Surviving sepsiscampaigninBrazil.Shock.2008;30:47–52.

14.RajuR,ChaudryIH.Sexsteroids/receptorantagonist:their useasadjunctsaftertrauma-hemorrhageforimproving immune/cardiovascularresponsesandfordecreasing mortalityfromsubsequentsepsis.AnesthAnalg. 2008;107:159–66.

15.GavidiaR,FuentesSL,VasquezR,etal.Lowsocioeconomic statusisassociatedwithprolongedtimestoassessmentand treatment,sepsis,andinfectiousdeathinpediatricfeverinEl Salvador.PLoSONE.2012;7:e43639.

16.MenduML,ZagerS,GibbonsFK,ChristopherKB.Relationship betweenneighborhoodpovertyrateandbloodstream infectioninthecriticallyill.CritCareMed.2012;40:1427–36. 17.OliveiraEXG,CarvalhoMS,TravassosC.Theterritorialbasis oftheBrazilianNationalHealthSystem:mappinghospital networks.CadSaudePublica.2004;20:386–402.

18.VandeVoordeP,EmersonB,GomezB,etal.Paediatric community-acquiredsepticshock:resultsfromtheREPEM networkstudy.EurJPediatr.2013;172:667–74.