Reliability of N-terminal pro-brain natriuretic peptide

NT-proBNP测定的可靠性 大数据分析

Primarycare

ReliabilityofN-terminalpro-brainnatriureticpeptideassayindiagnosisofheartfailure:cohortstudyinrepresentativeandhighriskcommunitypopulations

FDRHobbs,RCDavis,AKRoalfe,RHare,MKDavies,JEKenkre

Abstract

diagnosisisneededtoguideappropriatetreatmentObjectiveToinvestigatetheperformanceofanovelstrategies.DespitetheneedforaccuratediagnosisassayforN-terminalpro-brainnatriureticpeptidemanydoctors,especiallyinprimarycare,relyondiag-(NT-proBNP)indiagnosingheartfailureinvariousnosingheartfailureonclinicalgroundsalone.5

randomlyselectedgeneralandhighriskcommunityClinicaldiagnosisofheartfailureisnotpossibleonpopulations.

thebasisofEuropeanandAmericanguidelines,67asDesignCommunitycohortstudy(substudyoftheobjectiveevidenceofcardiacdysfunctionisneeded,echocardiographicheartofEnglandscreeningstudy).alongwiththepresenceofappropriatesymptoms,SettingFourrandomlyselectedgeneralpracticesinbeforeadiagnosisisconfirmed.Improvingthereliabil-theWestMidlandsofEngland.

ityofdiagnosisisessential,asdeterminingtheParticipants591randomlysampledpatientsovertheaetiologyandstageofheartfailureleadstodifferentageof45,stratifiedforageandsocioeconomicstatusmanagementchoices,suchasangiotensinconvertingandfallingintofourcohorts(generalpopulation,enzymeinhibitorsand blockersinmostpatientswithpatientswithanexistingclinicallabelofheartfailure,leftventricularsystolicdysfunction,89spironolactonepatientsprescribeddiuretics,andpatientsdeemedatinpatientswithsevereheartfailure,10orsurgeryifhighriskofheartfailure).

appreciablevalvediseaseexists.ThesetreatmentsMainoutcomemeasureSensitivity,specificity,improvesymptoms,qualityoflife,andprognosisofpositiveandnegativepredictivevalues,likelihooddiseaseandreducehealthcareutilisationandcosts.

ratios,andareaunderreceiveroperating

TheneedforeffectivediagnosticandtreatmentcharacteristiccurveforNT-proBNPassayinthestrategiesinheartfailureisimmense:theconditiondiagnosisofheartfailure.

occursinatleast2.3%oftheadultpopulationagedResultsForNT-proBNPinthediagnosisofheartover45,risingto4%inover75yearolds.2Symptomaticfailureinthegeneralpopulation(populationscreen),heartfailurehasamajorimpactonpatientsandalevelof>36pmol/lhadasensitivityof100%,ahealthcaresystems:itsprognosisisworsethanthoseofspecificityof70%,apositivepredictivevalueof7%,abreastcancerorprostatecancer,anditstreatmentcostsnegativepredictivevalueof100%,andanareaunderaresecondonlytothoseforstroke,mainlyowingtothereceiveroperatingcharacteristiccurveof0.92highadmissionrates.11Furthermore,diagnosisneeds(95%confidenceinterval0.82to1.0).Similarnegativetobemadeearly,whentheremaybenosymptoms,aspredictivevalueswerefoundforpatientsfromthetreatmentcandelayorreverseprogressionofdisease.12threeotherpopulationsscreened.

DiagnosticmethodsmaythereforeneedtoencompassConclusionsThisNT-proBNPassayseemstohavescreeningstrategies,1314aswellasidentificationofvalueinthediagnosisofheartfailureinthe

patientswithsymptoms.

community.HighnegativepredictivevaluesindicateTheinvestigationmostcommonlyusedtoconfirmthattheassay’schiefusewouldbetoruleoutheartthediagnosisofheartfailureisechocardiography,failureinpatientswithsuspectedheartfailurewithwhichoffersadetailed,semiobjectiveassessmentofnormalconcentrationsofNT-proBNP.Positiveresultsventricularandvalvefunction.However,accesstomayidentifypatientswhoneedcardiacimaging.

echocardiographyislimitedinmanyhealthcaresystems,5especiallyforprimarycarephysicians,whoIntroduction

seemostpotentialcases.Notsurprisingly,thevalidityofaclinicaldiagnosisofheartfailureinprimarycareisHeartfailureisanincreasinglycommondisorder,1poor,withhighratesofmisdiagnosiswhenpatientsarewithaprevalenceofaround2%indevelopedassessedagainstobjectivecriteria,rangingfrom25%tocountries.2Itischaracterisedbyverypoorprognosis50%accuracy.1516

andqualityoflifeforpatients,34andisresponsibleforDoalternativemethodsexistfordiagnosingheartveryhighhealthcarecosts.Asoutcomesinheartfailurefailureorenablingappropriatetriageofpatientsforarelinkedtothestageofdisease,earlyandaccurate

echocardiography?Oneoptioniselectrocardiography,

BMJVOLUME324

22JUNE2002

DepartmentofPrimaryCareandGeneralPractice,PrimaryCareClinicalSciencesBuilding,UniversityofBirmingham,BirminghamB152TTFDRHobbsheadofdepartmentRCDavis

clinicalresearchfellowAKRoalfestatisticianRHare

researchassociateJEKenkre

seniorresearchfellowDepartmentofCardiology,SellyOakHospital,BirminghamB292PEMKDaviesconsultantcardiologist

Correspondenceto:FDRHobbsf.d.r.hobbs@bham.ac.uk

2002;324:1

page1of5

NT-proBNP测定的可靠性 大数据分析

Primarycare

page2of5

asanormalrecordingwill,inmostcases,excludeleftventriculardysfunction.17However,changesontheelectrocardiogrammaybesubtle,andprimarycarephysicians18(andhospitaldoctors19)areunreliableinassessingsuchchanges.Interpretationofelectrocar-diogramsmay,therefore,stillneedreferraltoacardiologistforaspecialistopinion.20Chestradio-graphyisanotheroption,butnodataareavailableonitsreliabilityindiagnosingheartfailure.15

Onepotentialdiagnosticaidistheassessmentofpatientsbyassayofbrainnatriureticpeptide.Brainnatriureticpeptideisabiologicallyactivepeptideof32aminoacids,withvasodilatorandnatriureticproper-ties,whichiscleavedfromthe108aminoacidpro-brainnatriureticpeptidereleasedfromthecardiacventriclesinresponsetostretchingofthechamber.Thesecondremnantaftercleavage,N-terminalpro-brainnatriureticpeptide(NT-proBNP),isa76aminoacidpeptidewithnoknownbiologicalfunction,whichcirculatesathigherconcentrationsthanbrainnatriureticpeptideandmayrepresentcardiacstatusoverlongerperiods.Studiesofbrainnatriureticpeptidetodatehavebeensmall,anddataindifferentsubgroupsofpatientsareconflicting.2122Noreliabledataexistontheperformanceofassaysforbrainnatriureticpeptideassaysinthediagnosisofheartfail-ureinthegeneralpopulation.NT-proBNPhasbeenevenlessinvestigated.16ThisstudyprovidesoriginaldataontheutilityofanovelNT-proBNPassayinthediagnosisofheartfailure.

Methods

Thisisaprospectivesubstudyoftheechocardio-graphicheartofEnglandscreening(ECHOES)studyoftheprevalenceofleftventriculardysfunctionandheartfailure.2Themainstudywascarriedoutin16randomlyselectedprimarycarepracticepopulationsinEnglandafterstratificationforageandsocioeco-nomicstatus.Patientswererandomlyselectedfromeachoffourpopulationcohorts,identifiedineachpracticefromcomputerisedpracticeregisters:ran-domlysampledpatientsfromthoseaged45yearsandolder(generalpopulationscreen);patientswithaclini-caldiagnosticlabelofheartfailure;patientsprescribeddiureticdrugs;andpatientsathighriskofheartfailure(historyofmyocardialinfarction,angina,hypertension,ordiabetes).

Weconductedthissubstudyinthelastfourpractices,containingthefinal607consecutivelyscreenedpatientsrandomlyselectedfromthefourtargetgroups(307inthegeneralpopulation,103withanexistingclinicaldiagnosisofheartfailure,87takingdiuretics,and134athighriskofheartfailure,withsomesampledfrommorethanonecohort).Thesepracticeswererepresentativeofthesocioeconomicspreadofpatientsachievedbytheinitialstratification.Allpatientsgaveinformedconsent,andthestudyhadfullethicalapproval.

Wescreenedpatientsbyhistory,NewYorkHeartAssociationfunctionalclass,clinicalexamination,qual-ityoflife(SF-36healthstatusquestionnaire),spirometry,resting12leadelectrocardiography,andechocardiography,includingDopplerstudies.Wediag-nosedheartfailureonthebasisoftheagreedgoldstandardoftheEuropeanandAmericanguidelinecri-teria(box).Threeexperiencedcardiovascularcliniciansconductedblindedadjudicationofclinicalrecordsinequivocalcases.Aetiologiesforheartfailureincludedleftventricularsystolicdysfunction,atrialfibrillation,andsignificantvalvedisease.Wemadenoattempttodefinediastolicheartfailureinthisstudy.

Aresearchfellowtookbloodforthepeptideassayfromtherightantecubitalfossaofconsentingpatientsafterfiveminutes’supinerest,into5mlK+EDTAtubes.Bloodwaskeptatroomtemperatureforupto24hoursbeforetransporttoalocallaboratoryforcentrifugationandfreezingofplasmato 20°C.Onceaweekthefrozensampleswerecollectedforcentralstorageat 70°C.AssaysofNT-proBNP(RocheDiag-nostics,Germany)weresubsequentlyperformedatacentralindependentlaboratory,blindedtotheresultsofthescreeningassessments.TheNT-proBNPimmunoassay—anenzymelinkedimmunosorbantassay(ELISA)—requiredonlythreeoperatordepend-entstepsandtotalincubationofundertwohours.Analysis

Weassessedthediagnosticperformanceoftheassaybyusingreceiveroperatingcharacteristiccurves,formedbyplottingsensitivityontheyaxisand1 specificityonthexaxisforallpossiblecut-offvaluesofeachdiag-nostictest.Theoveralldiscriminatoryabilityofeachtestisshownbytheareaunderthecurve.

Inadditiontotheareaunderthecurve,weidentifiedthecut-offvaluefromthegeneralpopulationthatmaximisedsensitivitywithoutmuchlossofspecifi-city.Suchananalysismaximisesthenegativepredictivevalueofatest,whichisanappropriateaimforatestperformedinprimarycareonsymptomaticpatients.Wealsocalculatedlikelihoodratiosasameasureoftheextenttowhichthepre-testoddsarealteredbythetestresults(values>1increasetheodds;values<1decreasetheodds).Wethenappliedthecut-offvaluestotheothergroups.Wepresenttheresultingpositiveandnegativepredictivevaluesandlikelihoodratios,withconfidenceintervals.

Weusedmultiplelogisticregressionanalysistodeterminewhetherthepeptideassaygaveimproveddiagnosisofheartfailureovertheknownclinicalpredictorsofsex,electrocardiogramabnormality,andhistoryofmyocardialinfarctionordiabetes.Concen-trationsofNT-proBNPwerenotnormallydistributedandwerelogtransformedbeforeanalysis.Wecomparedtheresultantmodelsbyusingloglikelihood 2tests.WeusedSPSSandMicrosoftExcelforstatisti-calanalyses.

BMJVOLUME324

22JUNE2002

NT-proBNP测定的可靠性 大数据分析

Primarycare

Characteristic

Ageinyears(mean(SD))Sex:No(%)male

Heightinm(mean(SD))(n=590)Weightinkg(mean(SD))(n=587)Bodymassindex(mean(SD))(n=587)Ethnicity:No(%)whiteEversmoked(No(%))

Historyofmyocardialinfarction(No(%))Historyofangina(No(%))Historyofhypertension(No(%))Historyofdiabetes(No(%))Heartrate(mean(SD))(n=583)

SystolicbloodpressureinmmHg(mean(SD))DiastolicbloodpressureinmmHg(mean(SD))(n=590)FEV1inlitres(mean(SD))(n=583)FVCinlitres(mean(SD))(n=583)

65.8(10.7)316

(53.5)1.7(0.1)75.4(15.3)27.2(4.7)5733518712723268

(97.0)(59.4)(14.7)(21.5)(39.3)(11.5)

Table1Characteristicsofpatients(n=591)

70.9(13.4)153.5(22.6)84.4(11.5)2.21(0.81)2.63(0.93)

1 – Specificity

FEV1=forcedexpiratoryvolumein1second;FVC=forcedvitalcapacity.

Results

Fivehundredandninetyone(97%)of607eligiblepatientsconsentedtosupplyabloodsample.Table1liststhebasicdemographicsofthepopulationscreened,andtable2givestheoverallperformancecharacteristicsoftheassay.

Theassaydetecteddefiniteheartfailureinatotalof52peoplefromthefourdiagnosticgroups—sevenfromtherandomgeneralpopulation,35fromthepatientswithanexistingdiagnosisofheartfailure,14frompatientsondiuretictreatment,andnineinthehighriskgroup(13patientswereinmorethanonegroup).Themostfrequentcausesofheartfailurewereleftventricularsystolicdysfunction,atrialfibrillation,andsignificantvalvedisease,orcombinationsofthethree.

Inthegeneralpopulationsample,anNT-proBNPconcentrationof>36pmol/lhadasensitivityof100%,aspecificityof70%,apositivepredictivevalueof7%,andanegativepredictivevalueof100%.Thelikeli-hoodratioofapositivetestresultwas3.37,andthelikelihoodratioofanegativeresultwas0(table2).Theareaunderthereceiveroperatingcharacteristiccurvewas0.92(95%confidenceinterval0.82to1.00)(fig1).

Similarperformancecharacteristicsfordiagnosisofheartfailurewerefoundwhenweusedthesamecut-offvaluesintheotherthreescreenedpopulationcohorts(figs2,3,and4).Negativepredictivevaluesrangedfrom97%to100%,andlikelihoodratiosofanegativeresultrangedfrom0to0.18(table2).

Fig1ReceiveroperatingcharacteristiccurvesforNT-proBNPinthediagnosisofdefiniteheartfailureinthegeneralpopulation

1 – Specificity

Fig2ReceiveroperatingcharacteristiccurvesforNT-proBNPinthediagnosisofdefiniteheartfailureinthepopulationwithanexistingclinical(notvalidated)diagnosisofheartfailuremadeinprimarycare

Logisticregressionmodelspredictingdefiniteheartfailure,withexplanatoryvariablesofsex,historyofmyocardialinfarctionordiabetes,andQwavesor

Table2Areaunderreceiveroperatingcharacteristiccurve,sensitivity,specificity,predictivevalues,andlikelihoodratiosforNT-proBNP,withacut-offvalueof36pmol/l,inthediagnosisofdefiniteheartfailure.Valuesinparenthesesare95%confidenceintervals

Positivepredictivevalue

(%)7(3to14)39(28to49)

Negativepredictive

value(%)100(99to100)100(78to100)

Likelihoodratioofpositiveresult3.371.21

(2.83to4.01)(1.09to1.36)

Likelihoodratioofnegativeresult00

(0to0.89)(0to0.90)

Cohortsampled

Generalpopulationagedover45(n=307)

Patientswithexisting

diagnosisofheartfailure(n=103)

Patientstakingdiuretics(n=87)

Patientsathighriskforheartfailure(n=133)

Areaundercurve0.920.8

(0.82to1)(0.72to0.88)

Sensitivity(%)100(65to100)100(92to100)

Specificity(%)70(65to75)18(10to29)

0.870.84

(0.76to0.99)(0.76to0.93)

93(66to100)100(72to100)

40(28to52)44(35to54)

23(13to36)12(5to21)

97(83to100)100(96to100)

1.541.8

(1.22to1.95)(1.54to2.1)

0.180

(0.03to1.21)(0to1.16)

page3of5

NT-proBNP测定的可靠性 大数据分析

Primarycare

page4of5

1 – Specificity

Fig3ReceiveroperatingcharacteristiccurvesforNT-proBNPinthediagnosisofdefiniteheartfailureinthepopulationtakingprescribeddiuretics

bundlebranchblockpatternonelectrocardiogram,weresignificantlyimprovedwiththeadditionoflogNT-proBNP( 2=62.0,1df,P<0.001).

Discussion

Althoughseveralstudiesonassaysforbrainnatriureticpeptideinselectpatientgroupshavebeenpublished,thesearethefirstdataontheperformancecharacteris-ticsofanassayforNT-proBNPinalargegeneralisableseriesofrandomlyselectedadultswithvalidateddiag-nosesofheartfailureandwithacomparatornormativepopulationrandomlyselectedfromthesamepopulationsasthecases.Theassaywasalsousedunderthesameconditionsandconstraintsasoccurinroutineclinicalpractice.

Weavoidedspectrumbiasinthisanalysisbypresentingtheperformancecharacteristicsinthefour

1 – Specificity

Fig4ReceiveroperatingcharacteristiccurvesforNT-proBNPinthediagnosisofdefiniteheartfailureinthehighriskpopulation

screenedpopulationsseparately.Thereducednum-berslimittheprecisionoftheestimates,butthenega-tivepredictivevaluesandlikelihoodratiosofanegativeresultareverysimilaracrossthefourpopulations.Othercommonconfoundersofdiagnosticstudies—namely,verificationbias,treatmentparadox,andrefer-encestandarderror—wereminimised.Inthecaseoftreatmentparadox,anyinfluenceofdrugtreatmentwouldhavebeentoreducetheconcentrationsofnatriureticpeptideandthusleadtounderestimationofthetruetestperformance.

WehaveshownNT-proBNPassaystobeofvalueindiagnosingheartfailurewithinarepresentativecommunitybasedgeneraladultpopulationandalsowithinatriskpopulations.TheareaunderthereceiveroperatingcharacteristiccurvesforNT-proBNPinthediagnosisofdefiniteheartfailureinthegeneralpopu-lation(0.92)contrastswiththeareaunderthecurveforcervicalcytologyof0.723andthatforbreastmammog-raphyof0.85.24

Importantly,wehaveshownthatthenatriureticpeptideassayretainsitsutilitydespitesamplesbeinghandledastheywouldbeintheeverydaypracticeofprimarycare—thatis,collectionbypractisingnursesandstorageatroomtemperatureforsomehoursbeforetransfertolaboratorysettingsorfreezingatdomesticfreezertemperatures.Previousstudieshavemostlyusedspecialisedin-hospitalhandlingofsamples,includingimmediatechilledcentrifugationandrapidfreezingofsamplesto 70°C.

AnotherimportantfindingisthatNT-proBNPlevelsareelevatedinallmajorcausesofheartfailure—leftventricularsystolicdysfunction,atrialfibrillation,andvalvedisease.However,weprovidenodataontheutilityoftheassayindiagnosingdiastolicheartfailure.

Theveryhighnegativepredictivevalues,andcorrespondinglowlikelihoodratiosofanegativeresult,oftheassaymakeheartfailureveryunlikelywithconcentrationsbelowthecut-offvalue,suggestingthatthemostappropriateuseoftheassayinroutineprac-ticewouldbeasaruleoutorexclusiontest.Thelowlikelihoodratiosofanegativeresultarecomparabletothoseforcolposcopy,whichrangefrom0.002to0.38.25ElevatedconcentrationsofNT-proBNPshouldthere-foretriggerfurthercardiacinvestigation,includingechocardiography.Ifleftventricularsystolicdysfunc-tionisexcluded,elevatedconcentrationsofNT-proBNPmayindicateotherproblems,suchasdiastolicdysfunction,26leftventricularhypertrophy,unstableangina,orpulmonaryhypertension.27

Themainlimitationofthestudyrelatestoitbeingbasedonamajorepidemiologicalstudy.Thefindingsarereliableforassayperformanceinageneralpopula-tionscreenofpeopleagedover45andforscreeninginthethreeothertargetpopulationsoftheECHOESstudy,althoughtheindividualpopulationnumbersbecomesmall.However,generalpopulationscreeningisnotlikelytobearealisticproposal,despitethepreva-lenceandprognosisofheartfailure.Thesedatasuggestthat,inclinicalpractice,theassaywouldhavethreepracticaluses:screeningpatientswithexistingclinicallabelsofheartfailure(70ofthe103patientssocategorisedinthisstudyhadheartfailureruledoutonNT-proBNPtesting);triagingpatientspresentingwithsymptomssuggestiveofheartfailure(shortnessofbreath,lethargy)forechocardiography;andscreening

BMJVOLUME324

22JUNE2002

NT-proBNP测定的可靠性 大数据分析

patientsathighriskofheartfailure.Wesuspecttheassaywouldperformwellinthesesettings,butthefirstindicationwasnotformallytestedinthisstudy,andthethirdindicationwastestedinonly134patients.Thesedataaresufficientlypositivetojustifyformaltrialsoftheassaysinreallifestrategiesforscreeningsymptomaticpatientsandthoseathighrisk.

Contributors:FDRHandMKDwerejointprincipalinvestiga-torsandgrantholdersonthemainstudy,withFDRHasgrantholderofthissubstudy.FDRHledthedesignandexecutionofthestudyandanalysis.RCDconductedmostoftheclinicalassessmentsandreportingandmadeamajorcontributiontoanalysis.JEKandRHcontributedtostudydesignandanalysisandoverallstudymanagement.AKRdidthestatisticalanalysis.Allauthorscontributedtodraftsofthepaper.FDRHistheguarantor.

Funding:TheECHOESstudywasfundedbytheNHSR&Dcardiovasculardiseaseandstrokeprogramme.Thecostsofcol-lectingandhandlingthesamplesweresupportedbyanNHSRegionalR&Dgrant,andtheassaysandtechnicalsupportweredonatedbyJTrawinskiandJBaumannofRocheDiagnosticSolutions.

Competinginterests:FDRHisamemberoftheEuropeanSocietyofCardiology(ESC)WorkingPartyonHeartFailure,chairoftheBritishPrimaryCareCardiovascularSociety,andTreasureroftheBritishSocietyforHeartFailure.MKDischair-manoftheBritishSocietyforHeartFailure.FDRHandMKDhavereceivedtravelsponsorshipandhonorariumsfromseveralbiotechnologyandpharmaceuticalcompanieswithcardiovas-cularproductsforplenarytalksandattendanceatmajorcardi-ologyscientificcongressesandconferences.

1

BonneauxL,BarendregtJJ,MeeterK,BonselGJ,vanderMaasPJ.Estimatingclinicalmorbidityduetoischemicheartdiseaseand

congestiveheartfailure:thefutureriseofheartfailure.AmJPublicHealth1994;84:20-8.

2

DaviesMK,HobbsFDR,DavisRC,KenkreJE,RoalfeAK,HareR,etal.Prevalenceofleftventricularsystolicdysfunctionandheartfailureinthegeneralpopulation:mainfindingsfromtheECHOES(echocardio-graphicheartofEnglandscreening)ncet2001;358:439-45.

3

HoKKL,AndersonKM,KannelWB,GrossmanW,LevyD.SurvivalafteronsetofcongestiveheartfailureinFraminghamheartstudysubjects.Cir-culation1993;88:107-15.

4

DavisR,HobbsFDR,KenkreJE,RoalfeAK,McLeodS,HareR,etal.Qualityoflifeinheartfailure,asmeasuredbySF-36healthstatusques-tionnaire.EurHeartJ1998;19:S639.

5

HobbsFDR,JonesMI,AllanTF,WilsonS,TobiasR.Europeansurveyofprimarycarephysicianperceptionsonheartfailurediagnosisandman-agement(Euro-HF).EurHeartJ2000;21:1877-87.

6TheTaskForceonHeartFailureoftheEuropeanSocietyofCardiology.Guidelinesforthediagnosisofheartfailure.EurHeartJ1995;16:741-51.7

AmericanCollegeofCardiology,AmericanHeartAssociation.Guidelinesfortheevaluationandmanagementofheartfailure.ReportoftheAmericanCollegeofCardiology/AmericanHeartAssociationTaskForceonPracticeGuidelines(CommitteeonEvaluationandManage-mentofHeartFailure).Circulation1995;92:2764-84.

8

GargR,YusufSfortheCollaborativeGrouponACEInhibitorTrials.Overviewofrandomisedtrialsofangiotensin-convertingenzymeinhibi-torsonmortalityandmorbidityinpatientswithheartfailure.JAMA1995;273:1450-56.

ncet1999;353:9-13.

10

PittB,ZannadF,RemmWJ,CodyR,CastaigneA,PerezA,etalfortheRandomisedAldactoneEvaluationStudyInvestigators.Theeffectofspironolactoneonmorbidityandmortalityinpatientswithsevereheartfailure.NEnglJMed1999;341:709-17.

11GillumRF.Heartandstrokefacts.AmHeartJ1993;126:1042-7.

12

PfefferMA,BraunwaldE,MoyLA,BastaL,BrownEJ,CuddyTEfortheSAVEInvestigators.Effectofcaptoprilonmortalityandmorbidityinpatientswithleftventriculardysfunctionaftermyocardialinfarction:resultsofthesurvivalandventricularenlargementtrial.NEnglJMed1992;327:669-77.

13

HobbsFDR,DavisRC,McLeodS,MarshallT,KenkreJ,LancashireR,etal.Prevalenceofheartfailurein“highrisk”groups.JAmCollCardiol1998;31(S5):85C.

14

HobbsFDR.Primarycarephysicians:championsoforanimpedimenttooptimalcareofthepatientwithheartfailure?EurJHeartFail1999;1:11-5.

15

AgencyforHealthCarePolicyandResearch.Heartfailure:evaluationandcareofpatientswithleft-ventricularsystolicdysfunction.Rockville,MD:AgencyforHealthCarePolicyandResearch,1994.

16

HuntPJ,RichardsAM,NichollsMG,YandleTG,DoughtyRN,EspinerEA.Immunoreactiveamino-terminalprobrainnatriureticpeptide(NT-proBNP):anewmarkerofcardiacimpairment.ClinEndocrinol1997;47:287-96.

17

DavieAP,FrancisCM,LoveMP,CaruanaL,StarkeyIR,ShawTRD,etal.Valueofanelectrocardiograminidentifyingheartfailureduetoleftven-tricularsystolicdysfunction.BMJ1996;312:222.

18SurDK,KayeL,MikusM,GoadJ,MorenaA.Accuracyofelectrocardio-gramreadingbyfamilypracticeresidents.FamMed2000;32:315-9.

19

WestonMJ,BettJHN,OverR.Consensusopinionandobserveraccuracyinelectrocardiographywithreferencetocoronaryarteriographicinformation.AustNZJMed1976;6:429-32.

20HobbsFDR.Diagnosingheartfailureinprimarycare:isitpossible?BMJ2000;321:188-9.

21

McClureSJ,CaruanaL,DavieAP,GoldthorpS,McMurrayJJ.Cohortstudyofplasmanatriureticpeptidesforidentifyingleftventricularsysto-licdysfunctioninprimarycare.BMJ1998;317:516-9.

22

LandrayMJ,LehmanR,ArnoldI.Measuringbrainnatriureticpeptideinsuspectedleftventricularsystolicdysfunctioningeneralpractice:cross-sectionalstudy.BMJ2000;320:985-6.

23FaheyMT,IrwigL,MacaskillP.Meta-analysisofPaptestaccuracy.AmJEpidemiol1995;141:680-9.

24SwetsJA,GettyDJ,PickettRM,D’OrsiCJ,SeltzerSE,McNeillBJ.Enhanc-ingandevaluatingdiagnosticaccuracy.MedDecisMaking1991;11:9-18.25

MitchellMF,SchottenfeldD,Tortolero-LunaG,CantorSB,Richards-KortumR.Colposcopyforthediagnosisofsquamousintraepitheliallesions:ameta-analysis.ObstetGynecol1998;91:626-31.

26

FruhwaldF,FahrleitnerA,WatzingerN,FruhwaldS,DobnigH,SchumacherM,etal.Natriureticpeptidesinpatientswithdiastolicdysfunctionduetodilatedcardiomyopathy.EurHeartJ1999;20:1415-23.27

NagayaN,NishikimiT,OkanoY,UematsuM,SatohT,KyotaniS,etal.Plasmabrainnatriureticpeptidelevelsincreaseinproportiontotheextentofrightventriculardysfunctioninpulmonaryhypertension.JAmCollCardiol1998;31:202-8.

(Accepted4April2002)

Primarycare

page5of5

本文来源：https://www.bwwdw.com/article/qhoe.html