Air Pollution and Nonmalignant Respiratory Mortality in 16 Cohorts within the ESCAPE Project

内容需要下载文档才能查看 内容需要下载文档才能查看

AirPollutionandNonmalignantRespiratoryMortalityin16CohortswithintheESCAPEProject

KonstantinaDimakopoulou1,EvangeliaSamoli1,RobBeelen2,MassimoStafoggia3,ZoranaJovanovicAndersen4,5,BarbaraHoffmann6,PaulFischer7,MarkNieuwenhuijsen8,9,PaoloVineis10,WeiXun10,GerardHoek2,OleRaaschou-Nielsen4,AnnaOudin11,BertilForsberg11,LarsModig11,PekkaJousilahti12,TimoLanki13,AnuTurunen13,BenteOftedal14,PerNafstad14,15,PerE.Schwarze14,JohannaPenell16,LauraFratiglioni17,

NiklasAndersson16,18,NancyPedersen19,MichalKorek16,UlfDeFaire16,KirstenThorupEriksen4,AnneTjønneland4,ThomasBecker20,MengWang2,BasBueno-de-Mesquita7,10,21,Ming-YiTsai22,23,24,MarloesEeftens2,22,23,6PetraH.Peeters10,25,KeesMeliefste2,AlessandroMarcon26,UrsulaKramer¨,ThomasA.J.Kuhlbusch27,

MohammadVossoughi6,TimothyKey28,KeesdeHoogh10,ReginaHampel29,AnnettePeters29,JoachimHeinrich30,GudrunWeinmayr6,31,HansConcin32,GabrieleNagel31,32,AlexIneichen22,23,Ben´edicte´Jacquemin8,33,34,

MorganeStempfelet35,AliceVilier34,36,37,FulvioRicceri38,CarlottaSacerdote39,XanthiPedeli1,MichalisKatsoulis40,AntoniaTrichopoulou40,BertBrunekreef2,25,andKleaKatsouyanni1,41

DepartmentofHygiene,EpidemiologyandMedicalStatistics,MedicalSchool,UniversityofAthens,Athens,Greece;2InstituteforRiskAssessmentSciences,UtrechtUniversity,Utrecht,TheNetherlands;3DepartmentofEpidemiology,LazioRegionalHealthService,Italy;4DanishCancerSocietyResearchCenter,Copenhagen,Denmark;5CenterforEpidemiologyandScreening,DepartmentofPublicHealth,UniversityofCopenhagen,Copenhagen,Denmark;6IUF–LeibnizResearchInstituteforEnvironmentalMedicine,Dusseldorf,¨Germany;7NationalInstituteforPublicHealthandtheEnvironment,Bilthoven,TheNetherlands;8CentreforResearchinEnvironmentalEpidemiology(CREAL),Barcelona,Spain;9ConsortiumforBiomedicalResearchinEpidemiologyandPublicHealth(CIBERenEpidemiolog´?aySaludPublica-CIBERESP),´Madrid,Spain;10MRC-PHECentreforEnvironmentandHealth,DepartmentofEpidemiologyandBiostatistics,SchoolofPublicHealth,ImperialCollege,London,UnitedKingdom;11DivisionofOccupationalandEnvironmentalMedicine,DepartmentofPublicHealthandClinicalMedicine,Umea?University,Umea,?1213Sweden;NationalInstituteforHealthandWelfare,DepartmentofChronicDiseasePrevention,Helsinki,Finland;DepartmentofEnvironmentalHealth,NationalInstituteforHealthandWelfare,Kuopio,Finland;14DivisionofEnvironmentalMedicine,NorwegianInstituteofPublicHealth,Oslo,Norway;15InstituteofHealthandSociety,UniversityofOslo,Oslo,Norway;16InstituteofEnvironmentalMedicine,17AgingResearchCenter,DepartmentofNeurobiology,CareSciencesandSociety,and19DepartmentofMedicalEpidemiologyandBiostatistics,KarolinskaInstitutet,Stockholm,Sweden;18CentreforOccupationalandEnvironmentalMedicine,StockholmCountyCouncil,Stockholm,Sweden;20DepartmentofEnvironmentalScience,AarhusUniversity,Roskilde,Denmark;21DepartmentofGastroenterologyandHepatology,UniversityMedicalCentre,Utrecht,TheNetherlands;22SwissTropicalandPublicHealthInstitute,Basel,Switzerland;23UniversityofBasel,Basel,Switzerland;24DepartmentofEnvironmentalandOccupationalHealthSciences,UniversityofWashington,Seattle,Washington;25DepartmentofEpidemiology,JuliusCenterforHealthSciencesandPrimaryCare,UniversityMedicalCenterUtrecht,Utrecht,TheNetherlands;26UnitofEpidemiologyandMedicalStatistics,DepartmentofPublicHealthandCommunityMedicine,UniversityofVerona,Verona,Italy;27InstituteofEnergyandEnvironmentalTechnologye.V.,AirQuality&SustainableNanotechnology,Duisburg,Germany;28CancerEpidemiologyUnit,Nuf?eldDepartmentofClinicalMedicine,UniversityofOxford,Oxford,UnitedKingdom;29InstituteofEpidemiologyIIand30InstituteofEpidemiologyI,HelmholtzCenterMunich,GermanResearchCenterofEnvironmentalHealth,Neuherberg,Germany;31InstituteofEpidemiologyandMedicalBiometry,UlmUniversity,Ulm,Germany;32AgencyforPreventiveandSocialMedicine,Bregenz,Austria;33RespiratoryandEnvironmentalEpidemiologyTeamand36Nutrition,HormonesandWomen’sHealthTeam,Inserm,CentreforResearchinEpidemiologyandPopulationHealth(CESP),Villejuif,France;34UniversityParisSud,Villejuif,France;35InVS,FrenchInstituteforPublicHealthSurveillance,SaintMaurice,France;37IGR,Villejuif,France;38HumanGeneticsFoundation,Turin,Italy;39UnitofCancerEpidemiology,AOCitta’dellaSaluteedellaScienza-UniversityofTurinandCenterforCancerPrevention,Turin,Italy;40HellenicHealthFoundation,Athens,Greece;and41EnvironmentalResearchGroupandDepartmentofPrimaryCare&PublicHealthSciences,King’sCollegeLondon,London,UnitedKingdom1

(ReceivedinoriginalformOctober7,2013;acceptedin?nalformFebruary5,2014)

Thisworkwassupportedbythefollowinginstitutions.TheFinnishpartofthestudywasfundedbytheAcademyofFinland(projectnumber129317).ForHUBRO,thedatacollectionwasconductedaspartoftheOsloHealthStudy2000-2001and?nancedbytheNorwegianInstituteofPublicHealth,theUniversityofOsloandtheMunicipalityofOslo.FinancialsupportforthecombinedworkwiththeStockholmstudieswasreceivedfromtheSwedishEnvironmentalProtectionAgency,theSwedishHeart-LungFoundationandtheSwedishCouncilforWorkingLifeandSocialResearch.TheSwedishMinistryforHigherEducation?nanciallysupportstheSwedishTwinRegister.SALTwassupportedbytheSwedishCouncilforWorkingLifeandSocialResearchandbyNationalInstitutesofHealthgrantAG-08724.TwinGenewassupportedbytheSwedishResearchCouncil(M-2005-1112),GenomEUtwin(EU/QLRT-2001-01254;QLG2-CT-2002-01254),NationalInstitutesofHealthgrantDKU01-066134,TheSwedishFoundationforStrategicResearch,andHeartandLungFoundationgrant20070481.FinancialsupportandmortalitydataforEPIC-MORGENandEPIC-PROSPECTwerereceivedbytheDutchMinistryofPublicHealth,WelfareandSports(V.W.S.),NetherlandsCancerRegistry(N.K.R.),LKResearchFunds,DutchPreventionFunds,DutchZON(ZorgOnderzoekNederland),WorldCancerResearchFund(WCRF),andStatisticsNetherlands(TheNetherlands).Thebaselinestudyandthemortalityfollow-upofSALIAwerefundedbytheMinistryoftheEnvironmentofNorth-Rhine-Westfalia(Germany).TheKORAresearchplatformandtheMONICAAugsburgstudieswereinitiatedand?nancedbytheHelmholtzZentrumMunchen,¨GermanResearchCenterforEnvironmentalHealth,whichisfundedbytheGermanFederalMinistryofEducationandResearchandbytheStateofBavaria.TheVHM&PPissupportedbytheStateofVorarlberg,Austria.AmJRespirCritCareMedVol189,Iss6,pp684–696,Mar15,2014

Copyright©2014bytheAmericanThoracic

Society

684AmericanJournalofRespiratoryandCriticalCareMedicineVolume189Number6|March152014

内容需要下载文档才能查看

ORIGINALARTICLE

Prospectivecohortstudies,especiallyfromtheUnitedStatesandafewsinglecountrystudiesinEurope,haveprovidedsuf?cientevidencethatchronicexposureto

particulatematter(PM)andtraf?c-relatedairpollutionisassociatedwithall-causeandcardiovascularmortality(1–8).However,theevidenceforlong-termexposureeffectsonrespiratorymortalityislimitedandinconsistent,possiblyrelatedtodifferencesinevaluatedpollutants,exposure

assessmentmethods,andstatisticalanalysismethods.Thereismoreevidencethatthereareshort-termeffectsofPMwithanaerodynamicdiameter,2.5mm(PM2.5)and,10mm(PM10)onrespiratorymortalityandhospitaladmissions,

especiallyafterrelativelylongerexposure;thiseffectismorepronouncedforPMfromtraf?c,andlargereffectestimatesareobservedduringwarmperiods(9–11).Short-termeffectsofnitrogendioxide(NO2)onrespiratorymortalityhavealsobeenreported(12).

Arecentreviewonexposuretoairpollutionandlong-termcardiorespiratorymortalityeffects(13)identi?ed14cohortstudies,ofwhichfourwerefromthreeEuropeancountries,investigatingthelong-termeffectsofPMmetrics(e.g.,PM10,PM2.5,andblacksmoke)and/ornitrogen

AuthorContributions:K.D.contributedtothedesign,exposureassessment,statisticalscript,dataanalyses,anddraftedthemanuscript.E.S.andC.S.contributedtothestatisticalscript.R.B.contributedtothedesign,exposureassessment,statisticalscript,anddataanalyses.M.StafoggiaandG.W.contributedtothestatisticalscriptanddataanalyses.Z.J.A.contributedtothedesign,statisticalscript,anddataanalyses.B.H.contributedtothestatisticalscriptandprovidedlocalcohortdata.O.R.-N.andB.B.contributedtothedesign.P.J.,N.A.,A.Tjønneland,T.A.J.K.,M.V.,B.J.,M.Stempfelet,andA.V.contributedtodataanalyses.P.F.,M.N.,L.M.,M.Korek,K.T.E.,M.E.,K.M.,M.W.,K.d.H.,A.I.,andM.-Y.T.contributedtoexposureassessment.P.V.contributedtothedesignand

providedlocalcohortdata.W.X.contributedtothedesignanddataanalyses.A.O.,A.Turunen,B.O.,J.P.,F.R.,M.Katsouliscontributedtothedataanalyses.B.F.,P.N.,U.d.F.,N.P.,L.F.,P.H.P.,B.B.-d.-M.,U.K.,J.H.,T.K.,A.P.,H.C.,andA.Trichopoulouprovidedlocalcohortdata.T.L.contributedtoexposureassessmentandprovidedlocalcohortdata.P.E.S.andT.B.providedlocalcohortdata.R.H.andX.P.contributedtoexposureassessmentanddataanalyses.A.M.contributedtoexposureassessment.G.N.contributedtodataanalysesandprovidedlocalcohortdata.G.H.contributedtothedesignandstatisticalscript.K.K.contributedtothedesignanddraftedthemanuscript.Allauthorscontributedtocriticalreadingofandcommentstothemanuscriptandinterpretationofdata,andapprovedthe?naldraft.

CorrespondenceandrequestsforreprintsshouldbeaddressedtoKleaKatsouyanni,Ph.D.,DepartmentofHygiene,EpidemiologyandMedicalStatistics,UniversityofAthensMedicalSchool,75MikrasAsiasStreet,11527Athens,Greece.E-mail:kkatsouy@med.uoa.gr

Thisarticlehasanonlinedatasupplement,whichisaccessiblefromtheissue’http://wendang.chazidian.com

OriginallyPublishedinPressasDOI:10.1164/rccm.201310-1777OConFebruary12,2014Internetaddress:http://wendang.chazidian.com

Dimakopoulou,Samoli,Beelen,etal.:ExposuretoAirPollutionandMortality685

ORIGINALARTICLE

oxides(NOx)/NO2onrespiratorymortality.Thisreviewreportsthatrespiratory

mortalityeffectswereinvestigatedlessintheUnitedStatescomparedwithEuropeancohorts,thePMeffectstendedtobeweakerthanthoseofNOx,andgenerallytheevidencewasheterogeneous,oftenwithnonstatisticallysigni?canteffects.Aneedforfurtherinvestigationofrespiratoryeffectswasunderlined.

FewcohortstudiesinEuropewereinitiatedwiththeobjectivetostudytheeffectsofairpollutionexposure.Therefore,thecorrespondingresultswerebasedonvariablemethodologiesforexposureassessment,resultinginlimitedcomparability.

WithintheframeworkoftheEuropeanStudyofCohortsforAirPollutionEffects(ESCAPE)project,acommonairpollutionexposureassessmentandepidemiologicalanalysisprotocolwasusedfor16cohorts(from11Europeancountries)includingnonmalignantrespiratorymortalitycases,providingscopetoinvestigatethe

associationoflong-termexposuretoairpollutionandrespiratorymortality.

Someoftheresultsofthisstudyhavebeenpreviouslyreportedintheformofanabstract(14).

questionnairedata.Speci?cally,thecohorts

?,SNACK-K,wereinSweden(EPIC-Umea

SALT,andSixty),Finland(FINRISK),Norway(HUBRO),Denmark(DCH),TheNetherlands(EPIC-MORGENandEPIC-PROSPECT),Germany(SALIAandKORA),theUnitedKingdom(EPIC-Oxford),Austria(VHMandPP),France(E3N),Italy(EPIC-Turin),andGreece(EPIC-Athens)(Figure1).SixofthesecohortswerepartoftheEPICstudy.Themeanfollow-upperiodrangedfrom6.3to18.6years(Table1).

Nonmalignantrespiratorymortalitywasde?nedonthebasisoftheunderlyingcauseofdeathrecordedondeathcerti?cates.NonmalignantrespiratorymortalityincludedICD-9codes460to519orICD-10codesJ00toJ99.Inmostcohorts,follow-upwasbasedonlinkagetonationalmortalityregistries;forEPIC-Athens,activefollow-upwasimplemented(16).

AirPollutionExposureAssessment

Airpollutionconcentrationswereassignedtobaselineresidentialaddressesofstudyparticipantsafterextensiveand

standardizedpollutantmeasurement

campaignsandtheuseofstandardizedLandUseRegression(LUR)modelingproceduresinallstudyareas,whicharedescribedindetailelsewhere(17,18).Inbrief,airpollutionmonitoringcampaignswere

performedbetweenOctober2008andMay2011inallstudyareastoobtainannualaverageconcentrationsofnitrogenoxides(NO2andNOx),PM2.5absorbance(determinedasthere?ectanceofPM2.5?lters),andPM2.5andPM10(19,20).Themeasurementcampaignsinvolved20sitesforPMand40sitesforNOxusingthesamemeasurementtechniqueandcoveredallseasons.AllinstrumentswerepreparedandanalyzedbyonecentrallaboratoryattheInstituteforRiskAssessmentSciences,Utrecht,usingtheESCAPE

protocols.

Methods

TheepidemiologicalanalysisprotocoloftheESCAPEprojectincludedaprioriatwo-stageanalysis.First,eachcohortwasanalyzedseparatelyusingseveralmodelsunderacommonanalysisprotocolwithwell-de?nedexposures,outcomes,andconfoundersandwithananalyticalapproach.Aworkshopforalllocalanalystswasorganizedtoexplainthestatisticalmethods(15).Itwasnotpossibletoformacommondatabaseforpooledanalysisduetocon?dentialityissues.Subsequently,cohort-speci?ceffectestimateswerecombinedbyrandomeffectsmetaanalysis.

StudyPopulations,Mortality

Follow-up,andOutcomeDe?nition

Cohortswereincludedinthestudyifthenumberofrespiratorymortalitycases

exceededsevenduringthefollow-upperiodandifdataforthemostimportantpotentialconfounderswereavailable.Individual-levelconfounderswereavailablefrom686

Figure1.Locationofallcohorts.Circlesmarkthecohortareasinwhichbothparticulatematter(PM)andnitrogenoxidesweremeasured.SquaresindicatecohortareaswhereonlyNO2andnitrogenoxides(NOx)weremeasured.

AmericanJournalofRespiratoryandCriticalCareMedicineVolume189Number6|March152014

ORIGINALARTICLE

Table1:DescriptionoftheCohortStudies

TotalTimeatRiskinPerson-Years(AverageFollow-up

Time[yr])

281,711(12.7)109,396(10.6)175,062(9.6)17,311(6.3)47,887(8.7)42,168(11.2)469,571(13.2)217,722(13.2)

CohortName*EPIC-Umea,?Sweden

FINRISK,FinlandHUBRO,Oslo,NorwaySNAC-K,Stockholm,Sweden

SALT,Stockholm,Sweden

Sixty,Stockholm,Sweden

DCH,Copenhagen,Denmark

EPIC-MORGEN,TheNetherlandsEPIC-PROSPECT,TheNetherlandsSALIA,Germany

Subjects(n)22,13610,33018,2362,7685,5113,75835,45816,446

RespiratoryMortalityCases(n)

33279043431826536

BaselineRecruitment

Period

1992–19961992,1997,2002,2007

2000–20012001–20041998–20021997–19991993–19971993–1997

StudyAreaDescriptionCityofUmea?and

surroundingruralareasGreaterHelsinkiAreaandTurkucityanditsruralsurroundingsCityofOsloCityofStockholmStockholmCountStockholmCounty

CityofCopenhagenandsurroundingareasCitiesofAmsterdam,MaastrichtandDoetinchemand

surroundingruralareasCityofUtrechtand

surroundingruralareasAreasinthecitiesofDortmund,Duisburg,

Essen,GelsenkirchenandHernesituatedintheRuhrAreaandadjacenttownsBorkenandDulmenUrbanandruralareasinabufferof400kmaroundLondon-OxfordareaCityofAugsburgandtwoadjacentruralcountiesStateofVorarlberg,

excludinghighmountainareas(.600m)andareaswithin300mofstateborder

CitiesofParis,Grenoble,Lyon,andMarseilleandsurroundingruralareasCityofTurin

16municipalitiesofthegreaterAthensarea

15,6704,352

7932

1993–19971985–1987,1990–1994

202,809(12.9)81,093(18.6)

EPIC-Oxford,UKKORA,GermanyVHM&PP,Voralberg,Austria

38,9418,402103,097

15152619

1993–20011995–1995,1999–2001

1985–2005

491,542(12.6)97,168(11.6)1,769,491(17.2)

E3N,France?EPIC-Turin,ItalyEPIC-Athens,Greece

10,993(8,356)7,2634,192

29(22)812

1993–19961993–19981994–1999

151,423(13.8)(115,269[13.8])

93,515(12.9)46,852(11.2)

*Theorderofcohortsisbasedonanorth-to-southgradientinEurope.?

Numbersforanalysesforwhichparticulatematterdatawereavailable.

AdetaileddescriptionofsamplingandanalysisintheESCAPEprojectcanbefoundinEeftensandcolleagues(18)andCyrysandcolleagues(19).PMcoarsewascalculatedasPM10minusPM2.5.PM

measurementswererestrictedto15ofthe16studyareas.OnlyNOxandNO2weremeasuredinUmea(EPIC-UMEA),Swedenforbudgetaryreasons.

Study-areaspeci?cLURmodelsweredevelopedtoexplainthewithin-areaspatialvariationofmeasuredannualaverageairpollutionconcentrationsusingtraf?candlandusepredictorvariablesinaGeographicInformationSystem(17,20).A

standardizedapproachdescribedintheESCAPEexposuremanual(21)wasusedtodevelopLURmodelsinallareas.Brie?y,foreachsiteanannualaverageconcentrationwascalculated.The

coordinatesofthemonitoringsiteswereusedtoderivethevaluesforthepredictorvariablesthroughEurope-wideandlocalGeographicInformationSystemdatabases.LURmodelsweredevelopedusinglinearregression,withtheannualaverage

concentrationsasthedependentvariableandanextensivelistofgeographical

attributesaspossiblepredictorsrelatedtotraf?cintensity,roadlength,populationdensityandproximitytonaturalorurbangreenareas,proximitytoaportorindustrialareas,andaltitude;theseparametersmaximizetheadjustedpercentageexplainedvariance(R2)inasupervised,forward,stepwiseprocedure.

687

Dimakopoulou,Samoli,Beelen,etal.:ExposuretoAirPollutionandMortality

ORIGINALARTICLE

Thesetofpredictorschosendependedonthestudyareaandthespeci?cpollutant.

TheresultsoftheLURmodelswereusedtoestimateambientairpollutantconcentrationsattheparticipants’baselineaddresses.Anevaluationofthe?tofLURmodelsdevelopedwithintheESCAPEprojecthasbeendescribedelsewhere(22).Inadditiontoairpollutionconcentrations,traf?cintensityonthenearestroad(vehicles/d)andtotaltraf?cloadonallmajorroadswithina100-mbufferaroundeachsubject’sresidence(vehicles3m/d)wereusedasindicatorsoftraf?cattheresidence.Airpollutionmeasurementswereperformedfrom2008to2011,butbaselineaddresseswereinearliertimeperiods.Therefore,thepredictedconcentrationswereextrapolatedbackattherecruitmenttime(1985–2007withmoststudiesinthemid-90s)usingtheabsolutedifferenceandtheratiomethodbetweenthetwoperiods(21).Inthepresentanalysis,onlydataforNO2andNOxwereavailablefromroutinebackgroundmonitoringnetworksite(s)toperformback-extrapolationofthepredictedconcentrations.Moreover,theprocedurewasapplicabletoonlysomecohorts:10cohortswithback-extrapolatedconcentrationsforNO2andeightcohortsforNOx.

StatisticalAnalyses

Cohort-speci?canalyses.Coxproportional

employed,unemployed,stayathome,retired),fruitintake(continuousing/d),vegetablesintake(continuousing/y),

maritalstatus(assingle,married/livingwithapartner,divorced/separated,widowed),educationallevel(de?nedas“low”[primaryeducation],“medium”[secondary

education],and“high”[tertiaryeducation])andbodymassindex(BMI)(BMI[kg/m2]andBMI2wereusedtotakeintoaccountnonlineareffects).Model3included,inadditiontomodel2,area-levelsocioeconomicstatusvariables(e.g.,meanincomeoreducationallevelonaneighborhoodormunicipalityscale).Model3waschosenasthemainconfoundermodel.

Assensitivityanalyses,wetested

whetherback-extrapolationofthemodeledconcentrationstothebaselineyearhad

any

hazardsmodelswereusedforthecohort-speci?canalyses.Agewasusedasthe

underlyingtimescalebecauseofevidenceofbetteradjustmentforpotentialconfounding(23).Censoringoccurredatthetimeofdeath,emigration,lossoffollow-up,orendoffollow-up,whichevercame?rst.Airpollutionexposurewasanalyzedasalinearvariableinthreeapriorispeci?ed

confoundermodels(seebelow),whichhadanincreasinglevelofadjustmentforindividualandarea-levelconfounders.

Model1includedgender,calendartime(year(s)ofenrollment),andage(timeaxis).Model2included,inadditiontothe

variablesinModel1,smokingstatus(never/former/current),smokingintensity(aslife-timeaveragetobaccosmokingintensity,ing/d),smokingduration(thenetnumberofsmokingyears),environmentaltobaccosmokeexposure(variousde?nitionsindifferentcohorts),occupationalexposure(awhite/bluecollarclassi?cation),employmentstatus(employed,self-

Figure2.DescriptionofmodeledoutdoorconcentrationsofNO2(a)andparticulatematterwithanaerodynamicdiameter,2.5mm(PM2.5)(mg/m3)(b)atparticipantaddressesineachcohort.ParticulatematterdatanotavailableforEPIC-Umea.?

688

AmericanJournalofRespiratoryandCriticalCareMedicineVolume189Number6|March152014