消化道出血

Acutegastrointestinal(GI)hemorrhageisacommonclinicalproblemwithdiversemanifestations.SuchbleedingmayrangefromtrivialtomassiveandcanoriginatefromvirtuallyanyregionoftheGItract,includingthepancreas,liver,andbiliarytree.Althoughnodemographicgroupisspared,theannualincidenceofabout170casesper100,000adultsincreasessteadilywithadvancingage,andthediseaseisslightlymorecommoninmenthanwomen.[1]Furthermore,GIhemorrhageaccountsfor1%to2%ofacuteadmissions,resultinginmorethan300,000annualhospitalizationsintheUnitedStates.[2]Itisalsoacommoncomplicationinpatientshospitalizedforotherillness,especiallysurgicalpatients.AlthoughthetotaleconomicburdenofGIhemorrhagehasnotbeenformallyassessed,annualestimatessuggestthatdiverticularbleedingalonecostsinexcessof1.3billiondollars.[3]Managementofthesepatientsisfrequentlymultidisciplinary,involvingemergencymedicine,gastroenterology,intensivecare,surgery,andinterventionalradiology.Theimportanceofearlysurgicalconsultationinthecareofthesepatientscannotbeoveremphasized.[4]Inadditiontoaidingintheresuscitationoftheunstablepatient,insomesettingsthesurgicalendoscopistestablishesthediagnosisandinitiatestherapy.Evenwhenthegastroenterologistassumesthisrole,theearlycollaborationofthesurgeonpermitstheestablishmentofgoalsandlimitsforinitialnonoperativetherapy.Ultimately5%to10%ofpatientshospitalizedforbleedingrequireoperativeintervention,andpromptsurgicalconsultationpermitsmoretimeforpreoperativepreparationandevaluationaswellaspatientandfamilyeducationshouldurgentsurgerybecomenecessary.[1]MostpatientswithanacuteGIhemorrhagestopbleedingspontaneously.Thisallowstimeforamoreelectiveevaluation.However,inalmost15%ofcases,majorbleedingpersists,requiringemergentresuscitation,evaluation,andtreatment.[5]Improvementsinthemanagementofthesepatients,primarilybyearlyendoscopyanddirectedtherapy,hassignificantlyreducedthelengthofhospitalization.Despitethis,themortalityrateremainsgreaterthan5%andissignificantlyhigherinthosepatientsinitiallyhospitalizedforotherreasons.Thisdiscrepancybetweentherapeuticadvancesandoutcomesisprobablyrelatedtotheagingofthepopulationwithanincreaseinitscomorbidities.Currently,thepatientrequiringoperativeinterventionisbotholderandsickerthaninthepast.HemorrhagecanoriginatefromanyregionoftheGItractandistypicallyclassifiedbasedonthelocationrelativetotheligamentofTreitz.UpperGIhemorrhage(proximaltotheligamentofTreitz)accountsformorethan80%ofacutebleeding.[1]Pepticulcerdisease(PUD)andvaricealhemorrhagearethemostcommonetiologies.MostlowerGIbleedingoriginatesfromthecolon,withdiverticulaandangiodysplasiasaccountingforthemajorityofcases.Inlessthan5%ofpatients,thesmallintestineisresponsible.[1]Obscurebleedingisdefinedashemorrhagethatpersistsorrecursafternegativeendoscopy.Occultbleedingisnotapparenttothepatientuntilpresentationwithsymptomsrelatedtotheanemia.Determinationofthesiteofbleedingisimportantfordirectingdiagnosticinterventionswithminimaldelay.However,attemptstolocalizethesourceneverprecedeappropriateresuscitativemeasures.APPROACHTOTHEPATIENTWITHACUTEGASTROINTESTINALHEMORRHAGEInpatientswithGIbleeding,severalfundamentalprinciplesofinitialevaluationandmanagementmustbefollowed.Awell-definedandlogicalapproachtothepatientwithGIhemorrhageisoutlinedinFigure46-1.Onpresentation,arapidinitialassessmentpermitsadeterminationoftheurgencyofthesituation.Resuscitationisinitiatedwithstabilizationofthepatient'shemodynamicstatusandtheestablishmentofameansformonitoringongoingbloodloss.Acarefulhistoryandphysicalexaminationprovidescluestotheetiologyandsourceofthebleedingandidentifiesanycomplicatingconditionsormedications.Specificinvestigationthenproceedstorefinethediagnosis.Therapeuticmeasuresaretheninitiated,andbleedingiscontrolledandrecurrenthemorrhageprevented. Figure46-1 GeneralapproachtothepatientwithacuteGIhemorrhage. InitialAssessmentAdequacyofthepatient'sairwayandbreathingtakesfirstpriority.Afterthisisassured,thepatient'shemodynamicstatusbecomesthedominantconcernandformsthebasisforfurthermanagement.ThepresentationofGIbleedingisvariable,rangingfromhemoccult-positivestoolonrectalexamtoexsanguinatinghemorrhage.Initialevaluationfocusesonrapidassessmentofthemagnitudeofboththepreexistingdeficitsandongoinghemorrhage.Continuousreassessmentofthepatient'scirculatorystatusdeterminestheaggressivenessofsubsequentevaluationandintervention.Thehistoryofthebleeding,bothitsmagnitudeandfrequency,alsoprovidessomeguidance.Theseverityofthehemorrhagecanbegenerallydeterminedbasedonsimpleclinicalparameters.Obtundation,agitation,andhypotension(systolicbloodpressure<90mmHginthesupineposition),associatedwithcool,clammyextremities,areconsistentwithhemorrhagicshockandsuggestalossofmorethan40%ofthepatient'sbloodvolume.Arestingheartrateofmorethan100beats/minutewithadecreasedpulsepressureimpliesa20%to40%volumeloss.Inpatientswithoutshock,posturalchangesareelicitedbyallowingthepatienttositupwiththelegsdanglingfor5minutes.Afallinbloodpressureofmorethan10mmHgoranelevationofthepulseofmorethan20beats/minuteagainreflectsatleasta20%bloodloss.Patientswithlesserdegreesofbleedingmayhavenodetectablealterationsintheirhemodynamicstatus.Thehematocritisnotausefulparameterforassessingthedegreeofhemorrhageintheacutesettingbecausetheproportionofredbloodcells(RBCs)andplasmainitiallylostisconstant.Thehematocritdoesnotfalluntilplasmaisredistributedintotheintravascularspaceandresuscitationwithcrystalloidsolutionisbegun.Likewise,theabsenceoftachycardiamaybemisleading;somepatientswithseverebloodlossmayactuallyhavebradycardiasecondarytovagalslowingoftheheart.Hemodynamicsignsarelessreliableintheelderlyandpatientstakingβ-blockermedications.RiskStratificationNotallpatientswithGIbleedingrequirehospitaladmissionoremergentevaluation.Forexample,thepatientwithasmallamountofrectalbleedingthathasceasedcangenerallybeevaluatedonanoutpatientbasis.Clearly,inmanypatients,thedecisionmakingislessstraightforward.Othersrequireadmissionandobservationbutmaybefurtherevaluatedwithendoscopyonamoreelectivebasis.Severalprognosticfactors,showninBox46-1,havebeenassociatedwithadverseoutcomes,includingtheneedforemergentoperationanddeath.[6]ThesefactorsareconsideredduringtheinitialassessmentandresuscitationofpatientswithGIhemorrhage.Forinstance,patientsolderthan60yearsofagehavehighermortalityratesthantheiryoungercounterpartsandareevaluatedmorecautiously.Thisincreasedmorbiditymaybeareflectionofconcomitantdisease.Thedeleteriouseffectsofcardiac,renal,pulmonary,andhepaticcomorbidityallareconsideredwhenevaluatingpatientswithGIbleeding.Forexample,onestudyestimatedthatbleedingpatientswithsignificantrenaldiseasehaveamortalityrateofnearly30%;thisincreasesto65%inthepresenceofacuterenalfailure.[7]Otherfactors,includingthemagnitudeoftheinitialhemorrhage,persistenceorrecurrenceofthebleeding,andtheonsetofbleedingduringhospitalizationforanotherillness,alsocontributetoincreasedmorbidityandmortality.Box46-1  RiskFactorsforMorbidityandMortalityinAcuteGastrointestinalHemorrhage    Age>60yr    Comorbiddisease    Renalfailure    Liverdisease    Respiratoryinsufficiency    Cardiacdisease    Magnitudeofthehemorrhage    Systolicbloodpressure<100mmHgonpresentation    Transfusionrequirement    Persistentorrecurrenthemorrhage    Onsetofhemorrhageduringhospitalization    NeedforsurgeryConsiderablerecentefforthasbeendevotedtothedevelopmentofriskscoringtoolstofacilitatepatienttriage.Thesescoringsystemshavebeenusedtopredicttheriskforrebleedingandmortality,toevaluatetheneedforintensivecareunit(ICU)admission,andtodeterminetheneedforurgentendoscopy.Forexample,theBLEEDclassificationschemausesfivecriteria[8]:ongoingbleeding;asystolicbloodpressureoflessthan100mmHg;aprothrombintimeofgreaterthan1.2timescontrol;alteredmentalstatus;andanunstablecomorbiddiseaseprocessthatwouldrequireICUadmission.Ifanyoneofthesecriteriaispresent,themodelpredictsanaboutthree-foldincreaseintheriskforeitherrecurrenthemorrhage,theneedforsurgicalintervention,ordeath.Suchscoringsystemshavebeenalmostexclusivelyusedinresearchstudiesandaresignificantlymoreaccuratewhenendoscopicfindingsareincluded.Untiltheseschemahavebeenprospectivelyvalidatedforeverydayclinicalpractice,theyareonlyappliedinthecontextofclinicaljudgment.ResuscitationThemoreseverethebleeding,themoreaggressivetheresuscitation.Infact,thesingleleadingcauseofmorbidityandmortalityinthesepatientsismultiorganfailurerelatedtoinadequateinitialorsubsequentresuscitation.Intubationandventilationareinitiatedearlyifthereisanyquestionofrespiratorycompromise.Inpatientswithevidenceofhemodynamicinstabilityandthoseinwhomongoingbleedingissuspected,twolarge-boreintravenouslinesareplaced,preferablyintheantecubitalfossae.Unstablepatientsreceivea2Lbolusofcrystalloidsolution,usuallylactatedRinger's,whichmostcloselyapproximatestheelectrolytecompositionofwholeblood.Theresponsetothefluidresuscitationisnoted.Bloodissentimmediatelyfortypeandcrossmatch,hematocrit,plateletcount,coagulationprofile,routinechemistries,andliverfunctiontests.AFoleycatheteralsoisinsertedforassessmentofend-organperfusion.Inelderlypatientsandthosewithsignificantcardiac,pulmonary,orrenaldisease,placementofacentralvenousorpulmonaryarterycatheterisconsideredforclosermonitoring.Theoxygen-carryingcapacityofthebloodcanbemaximizedbyadministeringsupplementaloxygen.Frequently,thesepatientsbenefitfromearlyadmissiontoandmanagementintheICU.Thedecisiontotransfuseblooddependsontheresponsetothefluidchallenge,theageofthepatient,whetherconcomitantcardiopulmonarydiseaseispresent,andwhetherthebleedingcontinues.Theinitialeffectsofcrystalloidinfusionandthepatient'songoinghemodynamicparametersaretheprimarycriteria.Onceagain,thisprocessrequiresclinicaljudgment.Forexample,ayoung,healthypatientwithanestimatedbloodlossof25%whorespondstothefluidchallengewithanormalizationofhemodynamicsmaynotneedanybloodproducts,whereasanelderlypatientwithasignificantcardiachistoryandthesamebloodlossprobablyrequirestransfusion.Althoughthehematocritmayrequire12to24hourstofullyequilibrate,itiscommonlyemployedasoneindexoftheneedforbloodreplacement.Ingeneral,thehematocritismaintainedabove30%inelderlypatientsandabove20%inyoung,otherwisehealthypatients.Likewise,thepropensityofthesuspectedlesiontocontinuebleedingortorebleedmustplayaroleinthisdecision.Forexample,esophagealvaricesareverylikelytocontinuetobleed,andtransfusionmightbeconsideredearlierthaninapatientwithaMallory-Weisstear,whichhasalowrebleedingrate.Ingeneral,packedRBCsarethepreferredformoftransfusion,althoughwholeblood,preferablywarmed,maybeemployedincircumstancesofmassivebloodloss.HistoryandPhysicalExaminationAftertheseverityofthebleedingisassessedandresuscitationinitiated,attentionisdirectedtothehistoryandphysicalexamination.Thehistoryhelpstomakeapreliminaryassessmentofthesiteandcauseofbleedingandofsignificantmedicalconditionsthatmaydetermineoralterthecourseofmanagement.Obviously,thecharacteristicsofthebleedingprovideimportantclues.Thetimeofonset,volume,andfrequencyareimportantinestimatingbloodloss.Hematemesis,melena,andhematocheziaarethemostcommonmanifestationsofacutehemorrhage.HematemesisisthevomitingofbloodandisusuallycausedbybleedingfromtheupperGItract,althoughrarelybleedingfromthenoseorpharynxcanberesponsible.Itmaybebrightredorolderandthereforetakeontheappearanceofcoffeegrounds.Melena,thepassageofblack,tarry,andfoul-smellingstool,generallysuggestsbleedingfromtheupperGItract.Althoughthemelanoticappearancetypicallyresultsfromboththegastricacid,whichconvertshemoglobintohematin,andtheactionsofdigestiveenzymesandluminalbacteriainthesmallintestine,bloodlossfromthedistalsmallbowelorrightcolonmayhavethisappearance,particularlyiftransitisslow.Melenamustnotbeconfusedwiththegreenishcharacterofthestoolinpatientstakingironsupplements.Thesecanbedistinguishedbyperformingaguaiactest,whichtestsnegativeinthoseonironsupplementation.Hematocheziareferstobrightredbloodfromtherectumthatmayormaynotbemixedwithstool.Althoughthistypicallyreflectsadistalcolonicsource,ifthemagnitudeissignificant,evenupperGIhemorrhagemayproducehematochezia.Themedicalhistorymayprovidecluestothediagnosis.Chronicbloodlossmayleadtonon-GIend-organsymptomssuchassyncope,angina,andevenmyocardialinfarction.AntecedentvomitingmaysuggestaMallory-Weisstear,whereasweightlossraisesthespecterofmalignancy.Evendemographicdatamayproveuseful:elderlypatientsbleedfromlesionssuchasangiodysplasias,diverticula,ischemiccolitis,andcancer,whereasyoungerpatientsbleedfrompepticulcers,varices,andMeckel'sdiverticulum.AhistoryofGIdisease,bleeding,oroperationimmediatelybeginstofocusthedifferentialdiagnosis.Antecedentepigastricdistressmaypointtoapepticulcer,whereaspreviousaorticsurgerysuggeststhepossibilityofanaortoentericfistula.Ahistoryofliverdiseasepromptsaconsiderationofvaricealbleeding.Medicationusemayalsoberevealing.Ahistoryofingestionofsalicylates,nonsteroidalanti-inflammatorydrugs(NSAIDs),orselectiveserotonin-reuptakeinhibitors(SSRIs)iscommon,particularlyinelderlypatients.[9]ThesemedicationsareassociatedwithGImucosalerosionsseentypicallyintheupperGItract,butoccasionallytheycanoccurinthesmallbowelandcolon.GIbleedinginthesettingofanticoagulationtherapy,eitherwarfarinorlow-molecular-weightheparin,isstillmostcommonlytheresultofGIpathologyandisnotascribedtotheanticoagulationalone.[10]Physicalexaminationmayalsoberevealing.Bleedingfromtheoropharynxandnosecanoccasionallysimulatesymptomsofamoredistalsourceandmustalwaysbeexamined.Abdominalexaminationisonlyoccasionallyhelpful,butitisimportanttoexcludemasses,splenomegaly,andadenopathy.Epigastrictendernessissuggestive,butnotdiagnostic,ofgastritisorpepticulceration.Thestigmataofliverdisease,includingjaundice,ascites,palmarerythema,andcaputmedusae,maysuggestbleedingrelatedtovarices,althoughthesepatientscommonlybleedfromothersourcesaswell.OccasionallythephysicalexammayrevealcluestomoreobscurediagnosessuchasthetelangiectasiasofOsler-Weber-RendudiseaseorthepigmentedlesionsoftheoralmucosainPeutz-Jegherssyndrome.Arectalexamandanoscopyareperformedtoexcludealow-lyingrectalcancerorbleedingfromhemorrhoids.LocalizationSubsequentmanagementofthepatientwithacuteGIhemorrhagedependsonlocalizationofthesiteofthebleeding.AnalgorithmforthediagnosisofacuteGIhemorrhageisshowninFigure46-2. Figure46-2 AlgorithmforthediagnosisofacuteGIhemorrhage.EGD,esophagogastroduodenoscopy;RBC,redbloodcell. AlthoughmelenaisusuallytheresultofbleedingfromtheupperGItract,itcanbetheresultofbleedingfromthesmallbowelorcolon.Likewise,hematocheziaissometimestheconsequenceofbriskupperGIbleeding.Thefirststepindistinguishingthesepossibilitiesistheinsertionofanasogastric(NG)tubeandexaminationoftheaspirate.AlthoughhematemesisisusuallydiagnosticofanupperGIbleed,thetubeisstillusefultoassesstherateofongoingbleedingandtoremovebloodfromthestomachtopermitendoscopy.Iftheaspirateispositive,thiseffectivelylocalizesthelesion.Thepresenceofredbloodorcoffee-groundappearancesuggestsanuppersource.Testingforoccultbloodisrarelynecessary.Thereturnofbilefromagastricaspiratesuggeststhattheduodenumhasbeensampled.AlthoughabiliousnonbloodygastricaspirategenerallyexcludestheupperGItract,thesefindingscanoccasionallybemisleading.Onestudyfoundthatonly6of10yellow-greenNGaspiratestestedpositiveforbile.[11]Likewise,almost20%ofpatientswithaclearaspiratearestillbleedingfromanupperGIsource.[2]Inpatientswithmelenaorevenhematocheziafromanupperlesion,theNGaspiratemaybenegativeinthepresenceofsignificantduodenalbleedingandacompetentpyloruspreventingduodenogastricreflux.Theseconsiderationssuggestthat,althoughthefindingsoftheNGaspiratecanbehelpful,virtuallyallpatientswithsignificantbleedingneedtoundergoupperendoscopy.Esophagogastroduodenoscopy(EGD)underthesecircumstancesishighlyaccurateinbothidentifyinganupperGIlesionand,ifnegative,directingattentiontoalowerGIsource.Tomaximizeefficacy,EGDisperformedwithin24hourseveninstablepatients.[12]EarlyEGDwithdirectedtherapyhasbeenshowntoreduceresourceutilizationandtransfusionrequirementsandtoshortenhospitalstay.Accuracymaybelimitedbyeitheractivebleedingorabnormalanatomyasaresultofprevioussurgery.Occasionallypatientsmayhavemultiplelesions,andifthereisnoactivebleeding,specificityisreduced.Forexample,patientswithvaricesfrequentlyhavemucosalerosionsthatmayhavebeentheoriginalsource.CliniciansneedtobeawarethatEGDintheurgentoremergentsettingisassociatedwithasignificantincreaseintheincidenceofcomplications,includingaspiration,respiratorydepression,andGIperforation,whencomparedwithelectiveprocedures.Airwayprotectioniscriticalandmayrequireendotrachealintubationifithasnotbeenperformedpreviously.Volumeresuscitationmustnotbeinterruptedbytheexamination.AsshowninFigure46-2,subsequentevaluationdependsontheresultsoftheEGDandthemagnitudeofthebleeding.Angiographyorevensurgerymayprovenecessaryformassivehemorrhage,fromeithertheupperorlowerGItract.ForsloworintermittentbleedingfromthelowerGItract,colonoscopyisnowtheinitialdiagnosticmaneuverofchoice.Whenthisisnondiagnostic,thetaggedRBCscanisusuallyemployed.Forobscurebleeding,usuallyfromthesmallbowel,capsuleendoscopyisbecomingtheappropriatestudy.Thesediagnosticproceduresarediscussedingreaterdetaillater.TherapyDependingonthesourceofthebleeding,avarietyoftherapeuticoptionsareavailable.Theseincludepharmacologic,endoscopic,angiographic,andsurgicalmodalities.Pharmacologic,endoscopic,andsurgicaltherapiesare,forthemostpart,sitespecificandarediscussedfurtherinappropriatesectionslater.Angiographictechniquesaresomewhatmoregenericandincludeselectiveangiographywitheitherinfusionofavasoconstrictor,typicallyvasopressin,orembolization.Embolicagentsincludetemporarymaterialssuchasgelatinsponge(e.g.,Gelfoam)andautologousclotorpermanentdevicessuchascoils.Therearefewdatacomparingtheefficacyofthesetechniques.Formostpatients,bleedinghasceased,andtherapeuticoptionsareappliedtopreventrecurrence.Theriskforrecurrentbleeding,andthereforetheneedforpreventiveintervention,dependsonthecharacteristicsofthelesion,themagnitudeoftheinitialhemorrhage,andthespecificpatient.Forexample,althoughtheriskforrecurrentdiverticularhemorrhageisrelativelylow,electivecolonicresectionmaystillbeappropriateinapatientwithsignificantcoronarydiseasewhohasalreadysufferedamajorhemorrhage.Fortheabout15%ofpatientswhocontinuetobleed,therapyismoreurgent.Inpatientswithhemodynamicinstability,anappropriategoalistoinstitutetherapywithin2hoursofpresentation.Thisdependsonthedevelopmentofinstitution-specificprotocolsforthemultidisciplinarymanagementofthesepatients.[2]Theavailabilityofanendoscopisttrainedintechniquesofhemostasisandofanappropriatesupportstaffiscritical.Likewise,angiographicexpertisemustbeimmediatelyaccessible.Despitethevarietyofrelativelynewmodalitiesfornonoperativecontrolofbleeding,theearlyinvolvementofthesurgicalteamremainsessential.TraditionalserieshavedemonstratedthatthemorbidityandmortalityofoperationforGIbleedingincreasesignificantlyinpatientswhohavelostmorethan6unitsofblood.Thisincreaseisparticularlymarkedinelderlypatientsandthosewithmajorcomorbidconditions,suggestingthatinterventioninthesepatientsneedstobeearlierthaninyoung,healthypatientswhomightotherwisebebetteroperativecandidates.Althoughimprovementsinsupportivecareanddirectedtherapy,particularlyendoscopic,mayhavemoderatedthisapproachtoadegree,surgicaltherapymustalwaysbeaseriousconsiderationinthecontextofbloodlossofthismagnitude.EmailtoColleaguePrintVersionCopyright©2008ElsevierInc.Allrightsreserved.-www.mdconsult.com