BariumenemavcolonoscopyinrectalbleedingfalsepositivesTheseerrorsreducethesensitivityandspecificityofthebariumenemaTheseintrinsicbiasesinthetestevaluationsjustifiedthisprospectivetrialinordertocompa ID: 950975
Download Pdf The PPT/PDF document "Gut19882911881193Prospectivecomparisonof..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Gut,1988,29,1188-1193Prospectivecomparisonofdoublecontrastbariumenemaplusflexiblesigmoidoscopyvcolonoscopyinrectalbleeding:bariumenemavcolonoscopyinrectalbleedingEJANIRVINE,JOANNEO'CONNOR,RAFROST,PSHORVON,SSOMERS,GWSTEVENSON,ANDRHHUNTFromtheDepartmentsofGastroenterologyandRadiologyMcMasterUniversityMedicalCentre,Hamilton,Ontario,CanadaSUMMARYRectalbleedingoftenheraldsseriouscolonicdisease.Theliteraturesuggeststhatcolonoscopyissuperiortobariumenemaplussigmoidoscopy,althoughnogoodcomparativestudiesexist.Seventyonepatientswithovertrectalbleedinghadprospectivelyflexiblesigmoidoscopy,doublecontrastbariumenemaandcolonoscopycompletedindependently.Againstthegoldstandard,thesensitivityandspecificityofcolonoscopywere0-69and078respectivelyforaspectrumofcoloniclesions,whileforcombinedflexiblesigmoidoscopyanddoublecontrastbariumenemathesevalueswere0-80and056,respectively.Whenassessingadenomaorcarcinoma,colonoscopywasmoresensitiveat0O82v0O73,whileflexiblesigmoidoscopyplusdoublecontrastbariumenemawassuperiorfordetectingdiverticulardisease.Thepositivepredictivevalueforcolonoscopywas0O87against0O81forflexiblesigmoidoscopyanddoublecontrastbariumenema.Thisstudyconfirmsthatcolonoscopyshouldbeafirstlineinvestigationinsubjectslikelytorequirebiopsyortherapeuticintervention.Rectalbleedingisfrequentlyassociatedwithseriouscolonicpathology.1-3Patientsreferredtothegastro-enterologistwiththissymptomarediagnosedtohaveanaldiseasealoneinapproximately20%,25%haveclinicallyimportantlesionssuchasadenoma,adeno-carcinoma,diverticulardisease,inflammatoryboweldiseaseorangiodysplasia,andanother25%havecombinedanalandimportantdisease.Thirtypercentwillhavenocauseofbleedingidentified.2'Traditionally,theinvestigationofrectalbleedinghasbeenbysigmoidoscopyandbariumenemafollowedbycolonoscopyifsymptomspersisteddespiteanegativeexaminationorifthebariumenemawastechnicallyinadequate.10Specificindica-Addressforcorrespondence:DrEJanIrvine,DivisionofGastroenterology,Rm4W8-HSC,McMasterUniversity,1200MainStW,Hamilton,Ontario,Canada,L8N3Z5.Receivedforpublication7April1988.tionsforcolonoscopywerebiopsyofaparticularlesionorpolypectomy.Morerecently,physiciansarepromotingcolono-scopyastheinitialinvestigationforpatientswithrectalbleeding.Evidencesupportingthisrecom-mendation,however,comesfromretrospectivestudies'1014comparinghighqualityendoscopywithaverageradiology`'91'inselectedpatientpopula-tions.51618.21Prospectivestudiescomparingflexiblesigmoido-scopyordoublecontrastbariumenemavcolono-scopyhavealsobeenbiasedbyinequalityofexpertiseofradiologyandendoscopy.Forexample,fourprospectivecomparativestudiesusedcolonoscopyasthegoldstandardofdiagnosis.651622Suchadesignwoulddictatethatcolonoscopicfalsepositivesbeinterpretedasradiologicfalsenegativesandcolono-scopicfalsenegativesbeinterpretedasbariumstudy1188 Bariumenemavcolonoscopyinrectalbleedingfalsepositives.Theseerrorsreducethesensitivityandspecificityofthebariumenema.Theseintrinsicbiasesinthetestevaluationsjusti-fiedthisprospectivetrialinordertocomparethediagnosticpropertiesofcombinedflexiblesigmoido-scopyplusdoublecontrastbariumenemaversuscolonoscopyinacohortofpatientswithovertrectalbleeding.MethodsPATIENTSBetweenAugust1985,andDecember1986,allpatientsreferredtoMcMasterUniversityMedicalCentreforinvestigationofrectalbleedingbydoublecontrastbariumenemaorcolonoscopywereapproached.Subjectswereconsideredeligibleiftheyhadpassedredbloodperrectumwithinthepreviousthreemonthsandinasubsetwhowerehospitalised,hadnoposturalhypotension.20mmortransfusionrequirementofmorethan2unitsofpackedredbloodcells.Thosewhohadahistoryofmelaenastoolsalone,occultbleedingoracontraindicationtoeitherprocedurewereexcluded.Thecohortwhichcon-sentedunderwentflexiblesigmoidoscopyimmedi-atelybeforedoublecontrastbariumenema,followedwithintwoweeksbycolonoscopy.Eachprocedurewascarriedoutbyadifferentconsultantendoscopistorradiologistafterrapidcoloniclavagepreparation(Golytely,BraintreeLaboratories).Procedureswerecompletedinastandardfashionwithoutsedationforthe
flexiblesigmoidoscopyanddoublecontrastbariumenemabutcolonoscopywasprecededbyintravenousadministrationofameperedine/diazemulscombination.2324Eachexaminingphysicianwasblindedtoallothercolonicimagingprocedureresults.Atcolonoscopy,however,anunblindedobserverwaspresentanda'stagedwith-drawal'wasundertaken.Thisrequiredtheendo-scopisttoinsertthecolonoscopemaximally,thenwithdrawby10cmincrements(roughlyonesegmentofcolon-forexample,descendingtosigmoid).Missedpathologywasdefinedonlyafteralesionhadbeenpassedbyatleastonefullcolonicsegment.Theendoscopistwastheninformedandtheinstrumentreinsertedtocircumventtheneedforsubsequentrepeatexamination.Onlytheblindedexaminationresultswererecordedandaresearchassistantwaspresenttopreventbreechofconductor'hinting'atresults.Incaseswheretherewasdiscordanceofdiagnosisbetweenthecolonoscopyandothertestresults,subjectswereaskedtoundergoimmediaterepeatbariumenemawhichfocussedonthecontro-versialsegmentofbowel.Biopsywasnotpermittedatflexiblesigmoidoscopybecausedoublecontrastenemafollowedimmediately,andbiopsyorpoly-Table1Diagnosticfrequenciesin71studyparticipantsDescriptorDXIDX2DX3DX4Cancer5---Polyp-5mm12374Ischaemia1-AVM2-1-Endometriosis-1-IBD42-.-DD18114-Anal121763Volvulus1---Normal9---Total7138113AVM=angiodysplasia;IBD=inflammatoryboweldisease;DD=diverticulardisease.pectomywasdoneatcolonoscopyifclinicallyindi-cated.Theextentandqualityofexamination,anysideeffectsandcomplicationswererecordedforeachprocedure.Furthermore,bariumenemaswererereadinbatches,inablindedfashionbytwoobservers,threemonthsaftertheinitialprocedure.Atthistime,athirdinterpretationwasmadebyatleasttwoobserversinthelightofresultsofthetwoendoscopicprocedurestohighlightanysourcesoferrorinx-rayinterpretation.Aneconomicassessmentofcostsandeffectsaswellasassessmentofpatientpreferenceforpro-cedureswasalsoundertakenandisreportedelse-where.25ANALYSISOFDATAFinaldiagnosesforeachsubjectweregivenprioritybythelesion'spotentialgravity,withneoplasmsrankingmostandanallesionsleastserious,asillustratedbydiagnosis1inTable1.Uptothreeadditionaldiagnosesweremade,applyingidenticalrankingrulessuchas-forexample,anindividualwithacancer,diverticulosisandhaemorrhoidswouldbeclassedthusfordiagnosesone,two,andthreerespectively.Forthosewhocompletedthestudyprotocol,allhadflexiblesigmoidoscopy,doublecontrastenema,colonoscopyplusarepeatinterpretationofbariumx-rayexaminationatatimeremotefromtheinitialreading.Thegoldstandardconsistedofthemaximumdiagnosticdataavailableforeachpatientandanylesion(withtheexceptionofhaemorrhoidsandanalfissures)wasrequiredtobereportedonatleastthreetestresultsorbeupheldbyhistology.Eachdiagnostictestwascomparedwiththegoldstandardandtwo-by-twocontingencytableswereconstructed,asshownintheFigure,topermitcalculationofthe1189 Irvine,O'Connor,Frost,Shorvon,Somers,Stevenson,andHuntdiagnosticproperties,sensitivity,specificity,positiveandnegativepredictivevaluesandlikelihoodratiosforpositiveandnegativetests.ThisstudywasapprovedbythelocalhumanresearchreviewcommitteeinJune1985.ResultsPATIENTCHARACTERISTICSOf315eligiblesubjects,89gaveinformedconsentand71completedthestudy.Onepatienthadamyocardialinfarctionimmediatelybeforetheflexiblesigmoidoscopy,doublecontrastbariumenemacom-bination,possiblyasaresultofthebowelpreparationandtwosubjectsdroppedoutbeforeflexiblesig-moidoscopy.Anadditional15patientsrefusedtoundertakecolonoscopybecauseofpainatbariumenema(11patients),reluctancetoundertakeanyadditionaltests(threepatients),orhadleftthecountry(onepatient).Characteristicsofpartici-pants,non-participants,anddropoutsweresimilarwithrespecttoage,gender,andfinaldiagnosis.Themeanparticipant'sagewas53.5(16.7)(SD)and54%werewomen.Ofeligiblecandidates,onethirdwerereferredbyfamilyphysiciansandtwothirdsbygastroenterologistsorgastrointestinalsurgeons.In32subjectswhohadlesionsidentifiedandbiopsiedatcolonoscopy,histologywasavailable.Threesubjectsunderwentrepeatbariumenemabecauseofpolypsmissedatcolonoscopydespitethepresenceofanunblindedobserver.Twopatientswererequiredt
oundergorepeatcolonoscopybecauseofincompleteexaminationsorequivocalfindings.DIAGNOSESMultiplediagnoseswerecommon,occurringin38ofthe71participants(54%)andreflectingthehighprevalenceofdiverticulardisease(46%)andanallesions(54%)concurrentwithotherpathology(Table1).Non-participantsanddropoutshadfewersecondaryandtertiarydiagnosesasonly25of244subjectshadtwoproceduresandtheremainderhadsolelycolonoscopyorbariumenema.Moreover,ahigherproportionofnon-participantswascatego-risedas'normal'suggestingthatfurtherinvestigationwasnotpursuedsubsequenttoanormalexaminationorthatbleedinghadstopped.TESTPROPERTIESThecomputationalformulaeforthetestpropertiesareillustratedintheFigure.Thesensitivitytodetectallcoloniclesions,excludingdiminutivepolyps(mminsize,washighestat0.80forcombinedflexiblesigmoidoscopyandbariumenema.Colonoscopyhadasensitivityof0.69,whileflexiblesigmoidoscopyanddoublecontrastbariumenemaalonewerelowat0-58PositiveTestNegativeGoldStandardPositivea+cNegativea+bctdb+dSensitivity=a/a+cSpecificity=d/b+dppV=a/a+bLikelihoodratio(positive)=sensitivity/1-specificityLikelihoodratio(negative)=1-sensitivity/specificityGoldStandard=maximumdiagnosticdataTest=FS,DCBE,CSFig.1Contingencytablesandcalculationoftestproperties.and0-50respectively(Table2).ThevaluesforeachtestpropertyfellminimallywhendiminutivepolypswereincludedintheanalysisandarerepresentedinbracketsinTable3.Colonoscopyyieldedthehighestspecificity0-78,positivepredictivevalue0.87,andlikelihoodratioforapositivetest3.14(Table2).Diseasespecificsensitivitiesforeachofthediag-nostictestsisdepictedinTable4.Colonoscopywasmoresensitiveforthediagnosisofneoplasia-thatis,polyporcancer,andangiodysplasiaincontrastwithcombinedsigmoidoscopyandbariumenemawhichbetteredcolonoscopyfortheevaluationofdiverticu-lardisease.Bothdiagnosticapproacheswerecom-parableintheremainingdiseasecategories.EXAMINATIONCOMPLETENESSANDCOMPLICATIONSThecaecumorterminalileumwasreachedin70(99%)atbariumenemaandin59(83%)atcolono-scopy.Thehepaticflexureorascendingcolonwasreachedinanadditionalnine(total96%)atcolono-scopy.Threesubjectswereexaminedendoscopicallytothesigmoidcolonbecauseofastenosingcar-cinomaintwo(bariumenemaalsofailedinone)andTable2Testproperties:alldiagnosesTestSensSpecPPVNPVLR(+)LR(-)FS0-580.670-820.351-760-63DCBE0.500670870-351-520-75FS/BE0.800-560-810-561820.36CS0690-780870-54314040Sens=sensitivity;Spec=speciflicity;PPV=positivepredictivevalue;NPV=negativepredictivevalue;LR(+)=likelihoodratio(positiveresult);LR(-)=likelihoodratio(negativeresult).abcd11II1190 BariumenemavcolonoscopyinrectalbleedingTable3Testproperties:alldiagnosesTestSensitivityPositivepredictivevalueFS0-58(0-57)0-82(0-81)DCBE0-50(0.43)0-87(0.84)FS/DCBE0-80(0.77)0-81(0.79)CS0.69(0.67)0.87(0.86)FS=flexiblesigmoidoscopy;DCBE=doublecontrastbariumenema;CS=colonoscopy;()=includesdiminutivepolypsmm.equipmentfailureinone.Themeaninsertionforflexiblesigmoidoscopywas54cmwhile70%ofsubjectshadinsertionto60cm.Suboptimalexaminations,definedasatleastonesegmentofbowelpoorlyseenoccurredin30%ofpatientsatcolonoscopyand18%atsigmoidoscopyplusbariumenema.Inadequatebowelcleansingwasthemostcommoncauseofpoorexaminationforallthreetests.Onemajorcomplicationwasencounteredforeachmethodofinvestigation.Onepatienthadamyco-cardialinfarctionafterbowelpreparationforthesigmoidoscopybariumenemacombination.Asecondpatientwhohada4cmvillousadenomainthetransversecolonexperiencedhaemorrhagerequiringabloodtransfusionandhospitalisationovernightafterpiecemealpolypectomy.Minorcomplications-thatis,vasovagal,overdistension,tachycardia,andphlebitisoccurredinsixpatientsforbothinvestiga-tions.DiscussionRectalbleeding,asymptompresentinabout15%ofthegeneralpopulation'2predictsleftsidedcolonicneoplasiain8%to25%ofindividualswhenmixedinthestoolorpresentinthetoiletwater.'26Gastro-enterologistsarenobetteratpredictingneoplasiathangeneralpractitionersinthesesubjectsbasedTable4SensitivitybydiseasecategoryTestCSDCBE+FSDCBEFSNeoplasm0-93(0.82)0-79(0.73)0-77(0.47)0.64(0.49)Cancer1-00-830-830-
67Adenoma0-96(0-77)0-71(060)0-58(0.32)0-33(0.32)DD0-500-940.840.53Anal0-750-810-110-78AVM1-00-3300-33IBD0-830-830-330-67Misc0-290-290-290-14Overall0-670-770-430-56DD=diverticulardisease;AVM=angiodysplasia;IBD=inflammatoryboweldisease;Misc=miscellaneous;(l)=includingdiminutivepolypsmm.uponhistoryandproctoscopy.27Thus,colonicinvestigationismandatoryinmostpatients.Inourstudypopulation,afinaldiagnosisofadenomaorcarcinomaoccurredin34%ofpatients,yieldingapretestprobabilityordiseaseprevalenceofneoplasiaof0-34.Thisfigureislikelyanoverestimatebecauseonly25%ofalleligiblesubjectshadneo-plasia.Somepolypsorcancerscouldhavebeenmissedinnon-participants,however,whohadonlyonediagnostictestplacingthetrueprevalencefiguresomewherebetween25%and34%.Previousreportssuggestthatofsubjectswithrectalbleeding,20%haveanallesionsalone,25%haveclinicallyimportantlesions,25%clinicallyimportantplusanallesionsand30%havenocauseidentified.2"9Inourstudyonly13%ofpatientswereconsidered'normal',withnosourceofbleedingidentified;37%hadconcommitantanalandclinicallyimportantdiseaseand;33%hadatleastoneclinicallyimportantbleedinglesion.Anallesions(haemorrhoids,fissures,tears)anddiverticulardiseasewerepreva-lentin54%and46%ofpatientsrespectively.Thesewerethemostclinicallyimportantdiagnosesinonly17%,however,(12/71)and25%(18/71)respectively,andmostpatientshadothersourcesofbleeding.Thisemphasisesthedangerofassumingthathaemmorr-hoidsordiverticulosisisresponsibleforbleedingwithoutundertakingfullcolonicinvestigation.Furthermore,thediagnosticreturngreaterthananticipatedimpliesdeliveryofhighqualitydiagnosticcolonicimaging,radiologicandendoscopy,atourinstitution.Poorqualityradiologycannotbecom-paredwithhighqualityendoscopyorviceversaashasbeenthepracticeinthepreviousliterature.Thus,generalisationofourresultsisappropriateonlyafterconsiderationoflocalresources,expertise,andpopulationcharacteristics.Thevalueofthediagnostictestpropertiessensi-tivity,specificityandpredictivevalues,illustratedintheFigure,isbecomingincreasinglyfamiliartoclinicians.Lessfamiliar,perhaps,isthelikelihoodratiowhichgivesthe'odds'thataparticularresultwouldhavecomefromapatientwithasopposedtoapatientwithoutthedisease.Likelihoodratios,usefulforapplicationinindividualpatients,arenotdiseaseprevalencedependentasarepredictivevalues.28Colonoscopyhadthehighestlikelihoodratioof3.14reflectingitshighsensitivityandspecificity.Thesensitivityofcombinedflexiblesigmoidoscopyplusdoublecontrastbariumenemaforalllesionswasgreatestat0.78.Thisislargelybecauseofthebettersensitivityofthebariumenemafordiverticulardisease,andthehighprevalenceinourpopulation.Second,asflexiblesigmoidoscopyandbariumenemawereperformedfirst,mildinflammationoranalfissurescouldhavehealedbeforecolonoscopy.1191 1192Irvine,O'Connor,Frost,Shorvon,Somers,Stevenson,andHuntThird,mostindividualswhoundergocolonoscopyhavehadpreviousproctosigmoidoscopyandexam-inersmaybelesspreciseinreportingperianaldisease,animportantproportionofthediagnoses.Thisissupportedbythehighersensitivityofflexiblesigmoidoscopyforanallesionsof078inTable4.Finally,thesensitivityofflexiblesigmoidoscopyorbariumenemaindividuallyislow.Theseresultsre-establish,thatdoublecontrastbariumenemacannotstandaloneagainstcolonoscopybecauseofchangesinimageperceptioninthepresenceofdiverticulardiseaseandtheeffectsofredundantsigmoidloopsandresidualstool.29Whenweexaminedourbariumenemaerrors,mostoccurredintherectumandsigmoidcolon.Fordiagnosisofadenoma,carcinoma,andangio-dysplasia,colonoscopyisclearlymoresensitivethansigmoidoscopyandbariumx-ray.ThisissupportedbyWilliamsetalwhocomparedafaecaloccultbloodtest,rigidsigmoidoscopy,flexiblesigmoidoscopy,doublecontrastbariumenemaandcolonoscopyin330patientswithaprevioushistoryofadenomas.3'Theirresultsshowedasensitivityof092forcolono-scopy,071fordoublecontrastbariumenemaand054forflexiblesigmoidoscopyforcancerorpolypdetection.Onlylesions.7mmwereassessed,how-ever,anddataforcombinedsigmoidoscopyandenemacouldnotbeextractedfromtheirresults.Whenweadjustedourowndat
aforcanceroradenomadetectionandexcludedpolyps(mm,thesensitivityofcolonoscopyincreasedfrom0-82to0-93andforflexiblesigmoidoscopyandbariumenemafrom0-73to0-79.RichteretalandHowardetalhavepreviouslyestablishedthatangiodysplasiaisbestdiagnosedatcolonoscopy.3233Althoughtherewereonlythreecasesinourstudy,asexpectedwiththisflatmucosallesion,noneweredetectedbybariumenema.Incontrast,bariumenemaaloneorwithflexiblesigmoidoscopyissuperiorfordetectingdiverticulardiseasewithasensitivityof091,afindingpreviouslyreportedqualitativelyinatleastthreestudies.81622Thus,intheolderpatient,onlywhenneoplasmorangiodysplasiahasbeenruledoutoractivebleedingisdirectlyobservedatflexiblesigmoidoscopyshouldtheclinicianbeconfidentofthediagnosisofdiverticulosisoranaldiseaseasthesourceofbloodloss.Otherlesionssuchasinflammatoryboweldiseaseappearequallywelldemonstratedbyeitherdiagnosticapproachinthisstudyalthoughournumbersweresmall.Complicationsandsideeffectsofendoscopyandradiologywerecomparableinbothincidenceandseverity.Thehaemorrhageaftercolonoscopywassecondarytoatherapeuticintervention,anaddi-tionalbenefitofcolonoscopy.Whilecomplicationshavebeenwellrecognisedfordiagnosticandthera-peuticcolonoscopy,33theyappeartohavebeenunder-reportedfordoublecontrastbariumenema.3536Thus,furtherscrutinyisjustifiedforcomplicationsandsideeffectsofthebariumenema.Finally,wehaveshownthatexaminationofthebowelismorefrequentlyoptimalandcompletewiththesignmoidoscopy,bariumenemacombination.Elderlyordebilitatedsubjectsarelesscapableofmanoeveringonthex-raytabletogivehighqualitybariumstudies.3"38Thisgroupwasexcludedbydesignfromourstudytopermitunbiasedevaluationofalltests.Initialcolonoscopyistheinvestigationofchoiceinsuchindividualssupplementedbydoublecontrastbariumenemaiftheexaminationisincom-pleteorsuboptimal.Applyingtheresultsofthisstudyandareviewoftheliterature,wehavederivedalogicalapproachtotheinvestigationofthepatientwithrectalbleeding.Indicationsforcolonoscopytobetheprimaryinvesti-gationwouldbethepresenceofanyhighriskfactorforcolonicneoplasia;age.70years,asthequalityofbariumenemafalls;sickordebilitatedpatientsinwhomhighqualitydoublecontrastenemaisdifficulttoobtain;lackoflocallyavailablehighqualityradiology;orpersistentsymptomsdespiteanegativedoublecontrastbariumenemaandflexiblesigmoido-scopy.Combinedflexiblesigmoidoscopyanddoublecon-trastbariumeilemaistheappropriatechoiceinindividualswhohavenoriskfactorsforneoplasia;underage70andphysicallyfit;accesstogoodqualityradiology;andfinallyindividualswithpersistentsymptomsdespitenegativecolonoscopy.ThisresearchwassupportedbyHealthandWelfareCanadaNHRDPGrant#6606-2974andwaspre-sentedattheAmericanGastroenterologicalAssocia-tionmeetinginChicago,May1987.DrEJIrvinewassupportedbytheOntarioMinistryofHealthwhilecompletingthiswork.TheauthorswouldliketoacknowledgeandthankDrsRRossmanandRGoodacre,consultantgastroenterologistsatMcMasterUniversityMedicalCentreforcontribut-ingpatientstoandcompletingdataformsforthisstudy.References1DentOF,GoulstonKJ,ZubrzyckiJ,ChapuisPH.Bowelsymptomsinanapparentlywellpopulation.DisColonRectum1986;29:243-7.2ChapuisPH,GoulstonKJ,DentOF,TaitAD.Predic-tivevalueofrectalbleedinginscreeningforrectalandsigmoidpolyps.BrMedJ1985;290:1546-8.3StanilandJR,DitchburnJ,deDombalFT.Clinical Bariumenemavcolonoscopyinrectalbleeding1193presentationofdiseasesofthelargebowel-Adetailedstudyof642patients.Gastroenterology1976;70:22-8.4ThoeniRF,VenbruxAC.Thevalueofcolonoscopyanddouble-contrastbariumenemaexaminationintheevaluationofpatientswithsubacuteandchroniclowerintestinalbleeding.Radiology1983;140:603-7.5TedescoFJ,WayeJD,RasinJB.Colonoscopicevalua-tionofrectalbleeding:astudyof304patients.AnnInternMed1978;89:907-9.6KnutsonCO,MaxMH.Valueofcolonoscopyinpatientswithrectalbloodlossunexplainedbyrigidproctosigmoidoscopyandbariumcontrastenemaexaminations.AmJSurg1979;139:84-7.7WinawerSJ,LeidnerSD,BoyleC,KurtzRC.Compari-sonofflexiblesigmoidoscopywithotherdiagnostictechniquesinthediagnosisofrectocolonneoplasia.DigDisSci1979;24:277-81.8B
atL,PinesA,RabauM,NivY,ShemeshE.Colono-scopicfindingsinpatientswithhemorrhoids,rectalbleedingandnormalrectoscopy.IsraelJMedSci1985;21:139-41.9RexDK,LehmanGA,LappasJC,MillerRE.Sensi-tivityofdoublecontrastbariumstudyforleftcolonpolyps.Radiology1986;158:69-72.10WolffWI,ShinyaH,GeffenA,OzoklayS,DebeerRR.Comparisonofcolonoscopyandthecontrastenemain500patientswithcolorectaldisease.AmJSurg1975;129:181-6.11KaudeJV,HartyRF.Sensitivityofsinglecontrastbariumenemawithregardtocolorectaldiseaseasdiagnosedbycolonoscopy.EurJRadiol1982;2:290-2.12OttDJ,GelfanDW,WuWC,KerrRM.Sensitivityofdoublecontrastbariumenema:Emphasisonpolypdetection.AJR1980;135:327-30.13LeineckeJL,DoddsWJ,HoganWJ,StewartET.Acomparisonofcolonoscopyandroentganographyfordetectingpolypoidlesionsofthecolon.GastrointestRadiol1977;2:125-8.14ThoeniRF,VenbruxAC.Thevalueofcolonoscopyanddoublecontrastbariumenemaexaminationintheevaluationofpatientswithsub-acuteandchroniclowerintestinalbleeding.Radiology1983;140:603-7.15SpencerRJ,WolffBG,ReadyRI.ComparisonoftherigidsigmoidoscopeinconjunctionwithcolonX-rayfordetectionoflesionsofthecolonandrectum.DisColonRectum1983;26:653-55.16GilbertsenVA,WilliamsSE,ChumanL,McHughR.Colonoscopyinthedetectionofthecarcinomaoftheintestine.SurgGynecolObstet1979;149:877-88.17WayeJD.Colonoscopyinrectalbleeding.SAfrJSurg1976;14:143-9.18ShinyaH,CwernM,WolfG.Colonoscopicdiagnosisandmanagementofrectalbleeding.SurgClinNAm1982;62:897-903.19LeicesterRJ,HawleyPR,PollettWG,NichollsRJ.Flexiblefiberopticsigmoidoscopyasanoutpatientpro-cedure.Lancet1982;i:34-5.20WilliamsCB,MacraeFA,BartramCI.Aperspectivestudyofdiagnosticmethodsinadenomafollow-up.Endoscopy1982;14:74-8.21WinthropJD,BalfeDM,ShackelfordGD,McAlisterWH,RosenblumJL,SiegelMJ.Ulcerativeandgranu-lomatouscolitisinchildren.Comparisonofdoubleandsinglecontraststudies.Radiology1985;154:657-60.22TedescoFJ,GottfriedEB,CorlessJK,BrownsteinRE.Prospectiveevaluationofhospitalizedpatientswithnon-activelowerintestinalbleeding-Timingandroleofbariumenemaincolonoscopy.GastrointestEndosc1984;30:281-3.23LauferI.Doublecontrastenema:Technicalaspects.In:LauferI,ed.Doublecontrastgastrointestinalradiology.Philadelphia:Saunders,1979.24WayeJD.Colonoscopyintubationtechniqueswithoutfluoroscopy.In:HuntRH,WayJD,eds.Colonoscopy.London:ChapmanandHall,1981:147-78.25IrvineEJ,HuntRH,StevensonGW,FrostRA,O'ConnorJ.Acomparisonofcostsofperformingdoublecontrastbariumenemaplusflexiblesigmoido-scopyvscolonoscopytoexaminethelargebowel.Gastroenterology1987;92:1448.26SilmanAJ,MitchellP,NichollsRJ,etal.Self-reporteddarkredbleedingasamarkercomparablewithoccultbolldtestinginscreeningforlargebowelneoplasms.BrJSurg1983;70:721-4.27GoulstonKJ,CookI,DentOF.Howimportantisrecfalbleedinginthediagnosisofpolypsandbowelcancer?Lancet1986;ii:261-5.28SackettDL,HaynesRB,TugwellPX.Clinicalepi-demiology:abasicscienceforclinicalmedicine.Boston,Mass:LittleBrown,1985:59-138.29BakerSR,AltermanDD.Falsenegativebariumenemainpatientswithsigmoidcancerandcoexistentdiverticula.GastrointestRadiol1985;10:171-3.30DoddGD.Theaircontrastbariumenema/indicationsandvalidity.ProgClinBiolRes1985;186:133-49.31RichterJ,HedbergSE,AthanasoulisCA,SchapiroRH.Angiodysplasia-Clinicalpresentationandcolono-scopicdiagnosis.DigDisSci1984;29:481-5.32HowardOM,BuchananJD,HuntRH.Angiodysplasiaofthecolon.Lancet1982;ii:16-9.33GhaziA,GrossmanM.Complicationsofcolonoscopyandpolypectomy.SurgClinofNorthAm1982;62:889-96.34MacraeFA,TanKG,WilliamsCB.Towardssafercolonoscopy:Areportonthecomplicationsof5,000diagnosticortherapeuticcolonoscopies.Gut1983;24:376-83.35MaselH,MaselJP,CaseyKV.AsurveyofcolonexaminationtechniqueinAustraliaandNewZealandwithareviewofcomplications.AustralasRadiol1971;15:140-7.36SeamanWB,WellsWJ.Complicationsofthebariumenema.Gastroenterology1965;48:728-37.37WolfEL,FragerD,BeneventanoT.Feasibilityofdouble-contrastbariumenemaintheelderly.AJR1986;145:47-8.38OttDJ,ChenYM,GelfandDW,WuWC,MunitzHA.Singlecontrastversusdoublecontrastbariumenemainthedetectionofcolonicpolyps.AJR1986;146:993-6