meantwww.dajiba.com

RESEARCHARTICLEOpenAccessComparisonofpatientpreferencesforfecalimmunochemicaltestorcolonoscopyusingtheanalytichierarchyprocessYinghuiXu1*,BarceyTLevy1,2,JeanetteMDaly1,GeorgeRBergus1andJeffreyCDunkelberg3AbstractBackground:Inaverage-riskindividualsaged50to75years,thereisnodifferenceinlife-yearsgainedwhencomparingcolonoscopyevery10yearsvs.
annualfecalimmunochemicaltesting(FIT)forcolorectalcancerscreening.
Littleisknownaboutthepreferencesofpatientswhentheyhaveexperiencedbothtests.
Methods:Thestudywasconductedwith954patientsfromtheUniversityofIowaHospitalandClinicsduring2010to2011.
PatientsscheduledforacolonoscopywereaskedtocompleteaFITbeforethecolonoscopypreparation.
Followingbothtests,patientscompletedaquestionnairewhichwasbasedonananalytichierarchyprocess(AHP)decision-makingmodel.
Results:IntheAHPanalysis,thetestaccuracywasgiventhehighestpriority(0.
457),followedbycomplications(0.
321),andtestpreparation(0.
223).
Patientspreferredcolonoscopy(0.
599)comparedwithFIT(0.
401)whenconsideringaccuracy;preferredFIT(0.
589)comparedwithcolonoscopy(0.
411)whenconsideringavoidingcomplications;andpreferredFIT(0.
650)comparedwithcolonoscopy(0.
350)whenconsideringtestpreparation.
Theoverallaggregatedprioritieswere0.
517forFIT,and0.
483forcolonoscopy,indicatingpatientsslightlypreferredFITovercolonoscopy.
Patients'preferencesweresignificantlydifferentbeforeandafterprovisionofdetailedinformationontestfeatures(p<0.
0001).
Conclusions:AHPanalysisshowedthatpatientsslightlypreferredFITovercolonoscopy.
Theinformationprovidedtopatientsstronglyaffectedpatientpreference.
Patients'testpreferencesshouldbeconsideredwhenorderingacolorectalcancerscreeningtest.
Keywords:Colorectalcancerscreening,AnalyticHierarchyProcess(AHP),Patientpreference,Colonoscopy,FecalimmunochemicaltestBackgroundColorectalcancer(CRC)isthethirdleadingcauseofcan-cerdeathforbothmenandwomenintheUnitedStates[1].
Screeningforcolorectalcancerleadstoapproximatelya50%mortalityreduction[2].
TheAmericanCollegeofPhysicians(ACP)recentlyreviewedmultipleguidelinesforcolorectalcancerandrecommendedthatallaverage-riskadults50yearsofageoroldershouldbeofferedanyofthefollowingscreeningmethods:fecaloccultbloodtestingan-nually,sigmoidoscopyeveryfiveyears,orcolonoscopyevery10years[3].
AlthoughCRCscreeningrateshavegraduallyincreasedinthepast10years,one-thirdofadults50andolderstillarenotup-to-datewithCRCscreenings[4].
IndividualpreferencesforacertainscreeningtesthavebeenfoundtoinfluenceuptakeinaCRC-screeningprogram[5,6].
Incorporatingpatients'preferencesintotheclinicaldecision-makingprocessmaybeawaytoimprovescreeningrates.
Colonoscopyisthemostpopularscreeningtestforcolo-rectalcancerintheUnitedStates[7],beingagoldstand-ardforearlydetectionandpreventionofcolorectalcancer,eventhoughnostudieshaveshownittobesuper-iortoasensitivefecaloccultbloodtestinanaverage-riskpopulationtoreducemorbidityandmortalityfromCRC.
*Correspondence:yinghui-xu@uiowa.
edu1DepartmentofFamilyMedicine,RoyJ.
andLucilleA.
CarverCollegeofMedicine,UniversityofIowa,IowaCity,IA52242,USAFulllistofauthorinformationisavailableattheendofthearticle2015Xuetal.
;licenseeBioMedCentral.
ThisisanOpenAccessarticledistributedunderthetermsoftheCreativeCommonsAttributionLicense(http://creativecommons.
org/licenses/by/4.
0),whichpermitsunrestricteduse,distribution,andreproductioninanymedium,providedtheoriginalworkisproperlycredited.
TheCreativeCommonsPublicDomainDedicationwaiver(http://creativecommons.
org/publicdomain/zero/1.
0/)appliestothedatamadeavailableinthisarticle,unlessotherwisestated.
Xuetal.
BMCHealthServicesResearch(2015)15:175DOI10.
1186/s12913-015-0841-0Thenewerfecalimmunochemicaltest(FIT)ismoresensitive[8,9],andcost-effectivethantheguaiactest[10],butisunderusedintheUnitedStates[11].
Thereisnodifferenceinlife-yearsgainedwhencomparingascreeningstrategywithcolonoscopyevery10yearsvs.
anannualsensitivefecaloccultbloodtest,suchasaFIT[12].
However,in2011,theCentersforDiseaseControlandPrevention(CDC)reportedthatonly11.
8%ofpatientsaged50–75completedfecaloccultbloodtestswithinthepreviousyear[13].
IfwearetoimproveCRCscreeningrates,itwouldbehelpfultoobtaininsightintothepatientpreferencesbetweencolonos-copyandFIT.
PreviousstudieshavedescribedvariationsinpatientpreferencesforspecificCRCscreeningtests.
Thesestud-iessuggestindividualswhovalueaccuracythemostaremorelikelytoselectcolonoscopy[14],whereasotherspreferredfecaloccultbloodtestingbecauseofitslowercomplicationrateandsimplerprocedure[6,15].
Re-cently,severalstudies[16-19]haveappliedamethodcalledtheAnalyticHierarchyProcess(AHP)toevaluatepatientpreferencesforCRCscreeningtests.
TheuniquecontributionofthispaperisthatwecomparedpatientpreferencesforFITvs.
colonoscopyafterpatientshadcompletedbothtests.
Thepurposesofthisstudywereto:1)analyzepatientpreferencesforcolorectalcancerscreeningbetweenFITandcolonoscopy;2)identifythetestfeaturesthatareimportantinthedecisionmaking;and3)assesswhethertestpreferencesareassociatedwithpatients'knowledgeofthetests.
Itwasanticipatedthatthein-formationfromthisstudywouldimprovephysician-patientdiscussionsonscreeningoptionsandenhancecompliancewithcolorectalcancerscreening.
MethodsRecruitmentThestudyandmethodswereapprovedbytheUniversityofIowaInstitutionalReviewBoardandallparticipantsprovidedwritteninformedconsent.
Thestudysamplewascomprisedofpatientssched-uledforacolonoscopyattheUniversityofIowaHospi-talsandClinics.
Patientswereeligibleforthestudyiftheymetthefollowingcriteria:1)adultsage40to75years,2)scheduledforscreeningorsurveillancecol-onoscopy,3)colonoscopyscheduledwithin10daysto6weeks,4)validaddressandtelephonenumber,and5)Englishspeaking.
Patientswereexcludediftheymetanyofthefollowing:1)familialpolyposissyndromes,ulcera-tivecolitis,orCrohn'sdisease;2)personalhistoryofcoloncancer;3)activerectalbleeding;4)changeinbowelhabits;or5)pencil-likestoolsinthepasttwomonths.
BetweenJanuary22,2010andNovember22,2011,re-searchstaffintheDepartmentofFamilyMedicineattheUniversityofIowaidentifiedeachpatientwithascheduledcolonoscopyfromtheelectronicmedicalrecordsystem,Epic.
Arecruitmentmailing(includingacoverletter,twoinformedconsents,asingle-sampleFITkit,andapostage-paidreturnenvelope)wassentto2336potentialpartici-pants.
ParticipantswereaskedtocompletetheFITkitathomeuptothedaybeforetheystartedtheircolonoscopypreparationandmailittothestudyteam.
IftheinformedconsentandFITwerereturned,andthecolonoscopywascompleted,subjectsweremailedafollow-upquestion-nairethedayaftercompletingthecolonoscopy.
Non-responderswereremindedbyfollow-uptelephonecallsandsecondmailingswhennecessary.
Follow-upquestionnaireThefollow-upquestionnairewasdevelopedbyagroupcomprisingpracticingfamilyphysicians,includinganex-pertinissuesconcerningmedicaldecision-making(GB).
Theinformationandquestionsprovidedinthequestion-nairewerediscussedandrevisednumeroustimesafterdeliberation.
TheinformationwasbasedonthecurrentCRCscreeningguidelines,literature,andthewebsiteoftheCentersforDiseaseControlandPrevention.
Partici-pantswereawarethatiftheycompletedaFITandtheresultwerepositive,thenacolonoscopywouldberec-ommended(iftheychoseFITasascreeninginthe"real"world).
Thefinalquestionnairedevelopedbythere-searchteamconsistedof2multiple-choicequestionsaskingpatientpreferencebetweenFITorcolonoscopy;6questionsrelatedtotheAHPmodel;3questionsonpreferenceswithvariousassumptionsforout-of-pocketcosts;4questionsonpersonalandfamilyhistory,and7questionsondemographics.
Patientswerefirstaskedwhichtesttheypreferredbasedontheexperienceofhavinghadbothtests(FITandcolonoscopy).
Thisques-tionwasamultiple-choiceformattoallowforpatientstochooseonetest.
ParticipantsthenansweredtheremainingquestionsbasedontheAHPmodel.
Thethreemaincriteria(testaccuracy;testcomplications;andtestpreparation,frequency,andprocedure)werede-scribedindetailforeachtest.
Participantsthencom-paredtherelativeimportancebetweeneachpossiblepairofthethreetestfeatures.
Theywereaskedtoidentifythemoreimportanttestfeatureorwhethertheywereequallyimportantwhenselectingacolorectalcancerscreeningtest.
Anexampleofquestionsfollows:Thinkingabouteachpairoffeatures,whichfeatureismostimportanttoyouCheckonebox.
AccuracyofthetestAvoidingcomplicationsBothareequallyimportantXuetal.
BMCHealthServicesResearch(2015)15:175Page2of9Iftheywerenotequallyimportant,theparticipantswererequestedtochooseoneanswerfromthefollowing3options:HowmuchmoreimportantNotverymuchSomewhatMuchmoreTheparticipantswerethenprovidedwithside-by-sidecomparisoninformationbetweenFITandcolonoscopyforeachofthethreetestfeatures(testaccuracy;testcomplications;andtestpreparation,frequencyandpro-cedure).
Participantsrepeatedthecomparisonprocesstodeterminetheirpreferenceswithrespecttoeachtestfea-ture.
Anexamplequestionfollows:Pleaseindicatewhichtestyoupreferbasedonthein-formationontheaccuracyofthetestweprovidedabove.
Checkonebox.
FITeveryyearColonoscopyevery10yearsEithertestisfine(nopreferenceforeitherone)Iftheypreferredonetestovertheother,theparticipantswererequestedtochooseoneanswerfromthe3optionsof"notverymuch","somewhat",or"muchmore".
AfterthequestionsonAHPmodel,weonceagainaskedparticipantswhichteststheypreferredafterread-ingthedetailedinformationoneachtestprovidedinthequestionnaire.
Patientpersonalandfamilyhistoryofdi-gestivediseaseanddemographicinformationsuchasage,gender,education,insurance,andincomewerealsocollected.
AnalyticHierarchyProcess(AHP)AHPisawidelyuseddecision-makingmethoddevel-opedbySaaty[20,21]inthe1970stoassistpeopleinmakingcomplexdecisions.
IntheAHPmodel,theprob-lemisdecomposedintoahierarchyofgoal,criteriaandalternatives,withthegoalplacedatthetop,criteriaplacedattheintermediatelevelandalternativesatthebottom.
Theelementsineachhierarchicalgrouparecomparedaspairswithrespecttotheirimportanceinmakingthedecision.
Thesecomparisonsareusedtoob-taintheproportionalweightsoftheimportanceofthecriteriaandtherelativeimportanceofthealternativesintermsofeachindividualdecisioncriterion.
Inthelaststepoftheprocess,finalprioritiesarecalculatedacrossthehierarchyforeachofthedecisionalternatives.
Thealternativethatattainsthehighestfinalpriorityisthoughttobethemostsuitabledecision.
ThemodelusedforthisstudyisshowninFigure1.
Thegoalofthedecision,shownonthetop,definedas"preferredtest",wastodeterminethepatients'preferencesforcolonoscopyorFITforcolorectalcancerscreening.
Wefocusedonthreetestfeaturesrelatedtodecision-makingincolorectalcancerscreening:1)testaccuracy;2)frequencyofcomplications;and3)preparation,frequencyandprocedureasthedecisioncriteriaforthisproblem.
Thealternativeswerecomprisedoftwooftherecom-mendedscreeningtests,colonoscopyevery10yearsoranannualFITtest.
Therelativeimportanceofthreefeatureswascomparedbyconstructinga3*3matrixofthethreetestfeaturesforthesecondlevel.
Thetwoalternativeswerecomparedwithrespecttoeachofthetestfeaturesaswell,leadingtothree2*2matricesforthethirdlevel.
Thefinalstepwastosynthesizetheresultstoobtainthefinalprioritiesofthetwoscreeningmethods.
Wemulti-pliedeachalternativebythepriorityofitscriterionandaddedtheresultingweightsforeachalternativetocalcu-lateitsfinalpriority.
AHPcomputationsanddataanalysisThecollecteddatawereanalyzedwithSASsoftwareandthecodeswerewrittenspecificallyfortheAHPapplica-tion(YX).
Thecomparisonofimportancewasconvertedtoanumericalvalueusingascaleof1to4,where1meantequalimportance(ornopreference),2"notverymuch",3"somewhat",and4"muchmoreimportant".
Thegeometricmeanmethodwasusedtocalculateaneigenvectorforeachpatient,andthevaluesoftheweightsFigure1ThestructureoftheAnalyticHierarchyProcessmodel.
Xuetal.
BMCHealthServicesResearch(2015)15:175Page3of9werenormalizedsotheysummedto1.
Aconsistencyra-tio(CR)wascalculatedforeachpatientatthesecondlevel.
IftheCRvaluewasgreaterthan0.
15,thematrixwasconsideredinconsistentandthepatient'sjudgmentswereexcluded[18,22].
WedidnotcalculatetheCRatthethirdlevel,becauseforthematricesofsize2*2theCRisnotapplicable,butweexcludedpatients'judgmentsiftheywereexcludedatthesecondlevel.
Tables1and2showsacaseoftheinputdataandhowtheeigenvectorforlevel2andfinalprioritieswerecalculated.
Inthiscase,thehighestweightcriterionwas"accuracy"(0.
625),followedby"complications"(0.
239),and"preparation"(0.
136).
TheCRwaslow(0.
0559)anditwasconsideredtobeaconsist-entjudgment.
Thefinalpriorityofthegivenalternativewasobtainedfromtheweightofthealternativemultipliedbythepriorityofeachcriterion.
Thefinalprioritiesofthealternativeswere"FIT"(0.
418)and"Colonoscopy"(0.
582)forthisparticipant,indicatingthatthispatientpreferredcolonoscopyevery10yearsmorethanaFITannually.
Whileparticipantsactedasseparateindividuals,theaggregationofindividualprioritieswascalculatedwithanarithmeticmeanforsynthesizingindividualdecisionsintoagroupdecision[23].
DemographiccharacteristicsofparticipantswhowereincludedandexcludedintheAHPanalysisweresum-marized.
At-testwasusedtocomparethemeansofthecontinuousvariables,andthechi-squaretestwasusedtocomparethepercentagesbetweentwogroups.
Toevaluatethechangesofthepatients'preferencesbe-tweenFITandcolonoscopybeforeandaftertheyreadthedetailedinformationinthequestionnaire,theSASFREQprocedurewiththeagreementtestforsymmetrywasusedforthecomparison.
Allanalyseswereper-formedusingSAS,version9.
3(SASInstituteInc,Cary,NorthCarolina).
ResultsAsshowninFigure2,ofthepotential2336patients,1140patientsreturnedaconsentformandaFIT(49%).
Atotalof1090patientscompletedacolonoscopy;however,136wereexcludedfromanalysisforhavingadiagnosticcol-onoscopyand/ornotreturningfollow-upquestionnaire,leaving954patientsinthestudyforanalysis.
Ofthe954patients,667patients(667/954=70%)wereincludedinAHPanalysiswithconsistencyratios(CR)≤0.
15.
Pa-tientcharacteristicsofincludedandexcludedpatientsintheAHPmodelareshowninTable3.
Theagewassignificantlyyoungerintheincludedpatientsthanthatintheexcludedpatient(56.
7vs58.
0,p=0.
014).
Therewerenosignificantdifferencesbetweenincludedandexcludedpatientsforotherdemographiccharacteristicsincludinggender,maritalstatus,race,ethnicity,educa-tionlevel,insurancestatus,incomeandlocation.
Ofthe667patientsincludedintheAHPanalysis,theaveragetimeforreturningthefollow-upquestionnaireaftercol-onoscopywas16.
9(SD18.
3)days.
Ofthese667pa-tients,59%werefemale,71%weremarried,93%werewhite,andonly1%wereHispanic.
Eighty-onepercenthadgraduatedfromcollegeorhigher.
Approximately65%ofpatientsreportedannualhouseholdincomesof$40,000ormore,andmostpatientsreportedhavingsometypeofhealthinsurance.
AlmosttwothirdswerefromurbanareasinIowa.
Table4displaystheaverageprioritiesforeachcriterionatlevel2andtheaverageweightsforthetwoalternativeswithrespecttoeachcriterionatlevel3.
Thefinalcolumnshowstheaverageofindividualfinalpriorities.
Theaver-ageprioritiesforthesecondlevelcriteriaofaccuracy,complications,andtestpreparationwere0.
457,0.
321,and0.
223,respectively,indicatingthatparticipantsconsideredaccuracyasthemostimportant,followedbycomplica-tions,andtestpreparation.
Participantspreferredcolonos-copy(0.
599)comparedwithFIT(0.
401)withrespecttoaccuracy;theypreferredFIT(0.
589)comparedwithcolon-oscopy(0.
411)withrespecttoavoidingcomplications;andpreferredFIT(0.
650)comparedtocolonoscopy(0.
350)withrespecttotestpreparation.
Theaverageofin-dividualfinalprioritiesforFITwas0.
517,andforcolonos-copywas0.
483,indicatingparticipantshadaslightpreferenceforFITovercolonoscopyoverall.
Patients'preferencesfromthemultiplechoicequestionsaskeddirectlybeforeandaftertheprovisionofdetailedin-formationaredisplayedinTable5.
Ofthe382subjectswhoinitiallypreferredFIT,222(58.
1%)continuedtopre-ferFITafterinformationaboutthetest,but139(36.
4%)changedtheirpreferencesfromFITtocolonoscopy.
Amongthe130subjectsinitiallypreferringcolonoscopy,114(87.
7%)continuedtoprefercolonoscopyafterinfor-mationaboutthetest,whileonly7(5.
4%)changedtheirpreferencesfromcolonoscopytoFIT.
Overall,391(61.
5%)subjectskepttheirpreferencesthesame.
Thekappacoeffi-cientwas0.
40(95%CI,0.
34,0.
45),indicatingfairagree-mentbeforeandaftertheprovisionofinformation.
ThetestforsymmetryindicatedthatpatientschangedtheirTable1Acaseofthepairwisecomparisonandprioritycalculation(consistencyratio(CR)=0.
0559)AccuracyComplicationPreparationGeometricmeansNormalizedprioritiesAccuracy1341343p2:2892.
289/(2.
289+0.
874+0.
5)=0.
625Complication1/3120.
8740.
239Preparation1/41/210.
5000.
136Xuetal.
BMCHealthServicesResearch(2015)15:175Page4of9preferencessignificantlyfollowingprovisionofdetailedin-formation(p<0.
0001).
Patients'preferenceprioritiesthroughtheAHPmodelandfromthedirectly-askedquestionaftertheprovisionofinformationwerealsocompared.
Thekappacoefficientwas0.
57(95%CI,0.
51,0.
62),indicatingmoderateagreementbetweenthetwoformatsforobtainingpreferences.
Thetestforsym-metryindicatedthatthereweresignificantdifferencesbetweenthetwoformatsforpreferences(p<0.
0001).
Preferenceswerealsoelicitedunderhypotheticalout-of-pocketcosts.
Undertheassumptionthesameout-of-pocketcostsforbothtests,43%preferredtobescreenedbyFIT,colonoscopywaspreferredby42%,and15%felteithertestwouldbefine.
Whenassuminganout-of-pocketcostof$40forFITand$4000forcolonoscopy,preferenceschangedsignificantly(p<0.
0001bythetestforsymmetry):77%preferredtobescreenedbyFIT,col-onoscopywaspreferredby17%,and6%felteithertestwouldbefine.
Therewasnosignificantassociationbetweenpatients'preferenceprioriesandtheirrace,maritalstatus,insur-ance,orrural/urbanresidency.
FemalesweremorelikelytopreferannualFITthanmales,54%vs43%,re-spectively,p=0.
011.
About28%patientswithaneduca-tionofhighschoolorlessindicatedequalprioritiesbetweenannualFITandcolonoscopycomparedto11%forthosewithcollegeorhighereducation,p<0.
01.
Thepreferenceprioritiesalsovariedbyhouseholdincome.
Ashouseholdincomesincreased,patientstendedtoprefercolonoscopy.
Thepercentpreferringcolonoscopyforthosehouseholdincomelessthan$40,000,$40,000to$80,000,andgreaterthan$80,000were34%,46%,and51%,respectively,p<0.
01.
DiscussionToourknowledge,thisisthefirststudytomeasureprefer-encesforCRCscreeningmethodswithpatientswhohadimmediatepriorexperiencewithboththecolonoscopyandTable2CalculationstoobtainthefinalprioritiesAccuracy(0.
625)Complication(0.
239)Preparation(0.
136)FinalprioritiesFIT0.
250.
6670.
750.
418Colonoscopy0.
750.
3330.
250.
582FinalpriorityofFIT=0.
625*0.
25+0.
239*0.
667+0.
136*0.
75=0.
418.
FinalpriorityofColonoscopy=0.
625*0.
75+0.
239*0.
333+0.
136*0.
25=0.
582.
Figure2Participantenrollmentflowchart.
Xuetal.
BMCHealthServicesResearch(2015)15:175Page5of9afecalimmunochemicaltest.
Decidingamongthescreen-ingoptionswasacomplexdecisionproblembecauseeachoptionpresentedadvantagesanddisadvantages.
Inthepresentstudy,weusedalargecohortofpatientswhowerescheduledforcolonoscopy.
WedemonstratedintheAHPmodelthatpatientsvalueaccuracymorehighlythantestcomplicationsandpreparation.
Overall,patientsslightlypreferredannualFITovercolonoscopy.
Inthisstudy,thedecision-makingprocessinthecolon-oscopyscreeningchoiceswasconductedwithaquestion-nairebasedonAHP.
Thedifficultyandconfusionforpatientstomakeachoiceconsistentwiththeirvaluesandpreferencesforcoloncancerscreeninghasbeenrecog-nized[24].
AHPhasthedistinctadvantageinthatitde-composesacomplexdecisionproblemintoahierarchicalstructureofthecriteria,whichprovidespatientswithabetterfocusonspecificcriteriaandsub-criteriawhende-terminingpriorities.
Itallowspatientstomakeadequatejudgmentsinapairwisefashion,andthepairwisecompar-isonsarestraightforwardandconvenient.
Itprovidesamechanismtocheckinconsistenciesamongthedifferentpairwisecomparisonsbycomputingtheconsistencyratio.
InourstudyusingtheAHPmethod,wenotonlyquanti-fiedtherelativeimportanceofspecifictestfeatures,butalsocomputedrelativeweightsforthealternativesinre-gardtoeachofthetestfeatures.
Notsurprisingly,thereweresomediscrepanciesinthepreferencesthroughtheAHPmodelandfromthedirectly-askedquestion.
Patientsansweredthedirectly-askedquestionsintuitively.
Itisdif-ficultforpatientstomakeconsistentdecisionswhenfacedwithunfamiliarproblemsinvolvingtrade-offsbetweentheadvantagesanddisadvantages.
TheAHPwasdesignedtohelpthemmakemoreinformedchoicesinastep-by-stepmanner.
Ontheotherhand,investigatorsneedsufficientknowledgeandunderstandingoftheproblembeingexam-inedinordertosuccessfullystructuretheAHPmodelandincludealloftheelementsinthemodelwithaccuracy.
ThevalidityoftheAHPprocesshasbeenextensivelytestedinthewell-designedmodels[25].
Ourfindingsareconsistentwithotherstudies[5,26-30]thatindicateafairlyclosepreferencebetweencolonoscopyandFITasoptionsforcolorectalscreening.
However,otherstudiesusedqualitativeratingsandrankingsurveyTable3Demographicsintheincludedandexcludedsubjects(n=954)VariableIncludedsubjects(n=667)Excludedsubjects(n=287)p-valuesAgeinyears(SD)56.
7(7.
4)58.
0(7.
4)0.
0135Gender0.
3476Female399(59.
9%)162(56.
6%)Male267(40.
1%)124(43.
4%)MaritalStatus0.
6326Single166(24.
9%)65(22.
8%)Married468(70.
3%)203(71.
2%)Widowed32(4.
8%)17(6.
0%)Race0.
9115White617(93.
1%)268(94.
4%)Black17(2.
6%)6(2.
1%)Asian17(2.
6%)7(2.
5%)AmericanIndian3(0.
5%)1(0.
4%)Others9(1.
4%)2(0.
7%)Ethnicity0.
6120Hispanic7(1.
1%)4(1.
4%)Educationlevel0.
1134Highschoolorless112(16.
8%)64(22.
3%)Somecollegeorhigher552(82.
8%)221(77.
0%)InsurancestatusPrivate446(67.
2%)186(65.
5%)0.
6161Medicaid/IowaCare129(19.
4%)59(20.
8%)0.
6338Medicare110(16.
6%)53(18.
7%)0.
4334None12(1.
8%)7(2.
4%)0.
5165Annualincome0.
7372<$40,000204(30.
6%)88(30.
7%)$40,000to<$80,000169(25.
3%)80(27.
8%)≥$80,000266(39.
9%)105(36.
6%)Unreported28(4.
2%)14(4.
9%)Rurality0.
2249Rural217(32.
5%)105(36.
6%)Urban450(67.
5%)182(63.
4%)Table4AveragelocalandfinalprioritiesbasedonAHPmodel(n=667)AccuracyComplicationsPreparationLocalprioritiesFinalprioritiesAlternatives0.
4570.
3210.
223Colonoscopy0.
5990.
4110.
3500.
483*FIT0.
4010.
5890.
6500.
517**Thefinalprioritieswereaveragesbasedontheindividualfinalpriorities.
Table5Subjects'preferenceforcoloncancerscreeningbeforeandafterinformationprovided(n=636)BeforeinformationprovidedAfterinformationprovidedFITeveryyearColonoscopyevery10yearsEithertestisfineFITeveryyear22213921382(60.
1%)Colonoscopyevery10years71149130(20.
4%)Eithertestisfine274255124(19.
5%)256(40.
3%)295(46.
4%)85(13.
4%)636Xuetal.
BMCHealthServicesResearch(2015)15:175Page6of9methodstodeterminetheproportionofpatientswhopre-fereachoftheCRCtests.
Afewstudies[16-18]haveusedtheAHPmethodtostudypreferencesandprioritiesre-gardingcolorectalcancerscreening.
Dolanetal.
[17,18]indicatedthatpatientprioritiesvariedwidelyandcannotbepredictedusingdemographicfactors,numeracy,orlit-eracyskills.
Katsumuraandcolleagues[16]foundthatsubjectsgavehigherprioritytocolonoscopythantoFOBT.
However,thisstudywasconductedinJapanamonginter-netusers,inagroupofsubjectswhomayhaveattainedhigherlevelsofeducationandannualhouseholdincome.
Thesestudiesillustratedtheimportanceofidentifyingpa-tients'preferencesinchoosingamongcurrentlyrecom-mendedcolorectalcancerscreeningoptions.
However,inthesestudies,subjectshadnothadtheexperienceofcom-pletingbothacolonoscopyandFIT,andourstudyisuniqueinthatourquestionnaireswereconductedaftersubjectshadexperiencedbothtestsandthuspatientscouldfactortheirexperienceintotheirknowledgeofthetests.
Previousstudiesconsistentlyreportedthatcolonoscopyandthestoolbloodtestwerethemostoften-preferredop-tionsbypatients[5,30].
Thenumberofcolonoscopyproce-dureshasrisensteadilyintheUnitedStates[13]becauseofitsadvantages,suchasaccuracy,sensitivity,allowingdiagnosisandtherapyinonesession,andalongerintervalbetweentestsifnoabnormalfindings.
ThenumberofFOBTsperformedannuallydecreasedgraduallyfrom21.
1%to11.
8%between2002and2010[13].
Primarycarephysicianstendtorecommendcolonoscopyoverotherscreeningtestsregardlessofapatient'spreference.
Hawleyetal.
[5]found49%ofthepatientspreferredtomakemed-icaldecisionsthemselves,only10%ofthepatientsfavoredhavingaphysicianmaketheirmedicaldecisions.
ArecentstudybyInadomiandcolleagues[31]testedthepatient-centeredapproachin1000patientswhowererandomizedintothreearms:FOBTonly,colonoscopyonly,orachoiceofeithertest.
Participantswhowereofferedonlycolonos-copycompletedthescreeningatasignificantlylowerrate(38%)thanthosewhowereofferedFOBT(67%)orgivenachoiceofeithertest(69%).
Thisstudyindicatesthatnotpermittingapatienttochoosetheircolorectalcancerscreeningtestmaycontributetonon-complianceandpo-tentiallyascreeningfailure.
Ontheotherhand,presentingmultipleoptionsforCRCscreeningmayincreasepatientconfusionandasaresult,patientsmayfailtobescreened[24].
Thepsychologyliteraturehasnotedthattoomanychoicesconstituteabarriertodecision-makingandleadpeoplesimplytomakenochoiceatall[32].
OurfindingsofthefairlyclosepreferencebetweencolonoscopyandFIThaveimplicationsforcliniciansthattheycouldoffereitherofthetestsandpresenttheadvantagesanddisadvantagesofeachwhendiscussingCRCscreening.
PatientsshouldbemadeawareoftheadvantagesofFIT(samplecollectedintheprivacyofone'shome,nodietaryorotherrestrictions,notestpreparation,nocomplications)andthatnostudieshaveshowncolonoscopyisbetterthanasensitivefecaloccultbloodtestforaverageriskindividualstoreducemor-bidityandmortalityfromCRC.
Adecisionmodelingana-lysisshowednodifferenceinlife-yearsgainedbetweenastrategyofannualfecalimmunochemicaltestingandcolon-oscopyevery10years[12].
Inaddition,manyEuropeancountries,GreatBritain,andAustraliahaveaprogramofbiennialfecaloccultbloodtestingandreservecolonoscopyforthosewithpositivefecaloccultbloodtests[33-36].
Ourresultsconfirmedthattestpreferenceswereasso-ciatedwiththeimportanceofthetestfeaturesdeter-minedbythepatients.
OurresultsusingAHPanalysismethodwereconsistentwiththosestudiesthatfoundpatientsvaluedaccuracyasmoreimportant[27,37,38],andtestpreparationastheleastimportant.
Inourstudy,whenpatientsconsideredtheaccuracy,theyweremoreinclinedtoselectcolonoscopy;whentheyconsideredtestpreparationortestcomplications,theyfavoredFIT.
Itisimportantforphysicianstoidentifyaparticularpa-tient'svaluesregardingthesetestfeatures,andtoadoptashareddecision-makingapproachwhendiscussingscreeningoptions.
Earlierstudies[27,30]reportedtheassociationbetweenrace/ethnicity,educationlevel,andincome.
Patientsofnon-Latinoethnicity,thosewithhigherhouseholdin-comesandhighereducationweremorelikelytoprefercolonoscopy,butdidnotfindassociationbetweentestpreferencesandgender.
Wefoundthatpatientprioritiesdifferedbygender,educationlevel,andhouseholdincome.
MorewomenpreferredFITovercolonoscopyandthosewithhigherhouseholdincomestendtoprefercolonos-copy.
Ourresultsindicatethatphysiciansshouldalsoconsiderindividualpatient'sdemographiccharacteristicsduringtheshareddecision-makingprocess.
Inourstudy,wefoundthat30%patientsprovidedin-consistentresponsesontheAHPquestions.
Onepossiblereasonthatwehadarelativelyhighpercentageofpatientswithinconsistentresponseswasbecauseallpatientscom-pletedthequestionnaireonpaperandmailedthemtous.
Wewerenotabletohavetheopportunitytoprovidefeedbackandcorrectsubstantiallyinconsistentjudg-ments.
Thishighlightstheneedtodevelopclinicallyfeas-ibledecision-supporttoolsthatcanprovidefeedbackforinconsistentresponsestofacilitatethisdecision-makingprocess.
Thecomparisonbetweenpatients'preferencesbeforeandaftertheinformationwasprovidedshowedthesig-nificantimpactthattheinformationprovidedhadonapatient'spreferredscreeningmethod.
Inthepresentstudy,theprovisionofthescreeninginformationchan-gedthepreferredscreeningmethodfor38%ofthesub-jects.
Thepercentofchangingpreferencewaslowinthegroupwhichchosecolonoscopyinitially(12%)comparedXuetal.
BMCHealthServicesResearch(2015)15:175Page7of9tothegroupwhichinitiallychoseannualFIT(42%).
Thepossibleexplanationcouldbethatpatientsmaylearnthatcolonoscopyprovidesmoreaccurateresultswithalongertestintervalandhenceswitchtheirpreferencestocolonoscopy.
Althoughpatientshadexperiencewithbothtests,andthisexperienceprovidedthemwithanunderstandingofthetestingprocedures,itislikelythatsomepatientsstilllacksufficientknowledgeofthetestaccuracy,thetestinginterval,andpossiblecomplica-tions.
Previousstudieshaveshownthatpatientsshouldbewellinformedabouttestinformationandbeinvolvedinthedecision-makingprocess[16].
Theprovisionofin-formationininfluencinganinitialchoicehighlightstheimportanceofeducatingpatientswiththestrengthsandlimitationsofeachscreeningoption.
Thisstudyhasseverallimitations.
WeimplementedourstudyataMidwesternacademicmedicalcenterusinganadvancedelectronicmedicalrecordsystem;ourfindingsmaynotapplytolessstructuredsettings.
Oursamplewaslimitedtomostlywell-educated,whitepa-tientswithinsurancewhowerescheduledforcolonos-copyscreeningorsurveillance.
Futureresearchinamoreraciallydiverseareashouldbecompleted,sincepreferencesmayvarybyeducationalandsocioeconomiclevels.
Althoughweaskedsomequestionsconcerningcost,ourstudydidnotincludecostamongthetestfea-turesintheAHPmodel.
Ourresultsfromthetwohypo-theticalout-of-pocketcostquestionswereconsistentwithtwopreviousstudies[39,40]onCRCscreeningpreferencesthatindicatedthattheout-of-pocketcostscanaffectpreferences.
Patientswhovalueout-of-pocketexpensesasanimportantfactorintheirdecision-makingmighthavebeenmoreinclinedtochoosetheFIT.
OurstudyofferedtheFITatnocosttopatients,andthismighthaveinfluencedpatientpreferencestowardFIT.
Weexcludedthe30%ofsubjectswhoseresponseswereinconsistent,andsubjectsweresignificantlyolderinthegroupwhoseresponseswereexcluded.
ConclusionsInsummary,wehavefoundthataverage-riskpatientswhoexperiencedbothcolonoscopyandFITtestsdidnothaveadistinctpreferencebetweenFITandcolonoscopyforscreening.
Weconfirmedthatpatientsprioritizedaccuracyoverotherfeaturesandtheirpreferencescanbelinkedtotestfeatures.
Primarycarephysiciansshoulddiscusswithpatientstherisksandbenefitsofscreeningtestsforcolo-rectalcancer,eventhoughitisnotpossibletodiscussallaspectsofscreeninginabusyclinicsetting.
Thediscus-sionscouldfocusoncomparingthemaintestfeaturesofFITandcolonoscopy.
AdditionalresearchtodeterminetheroleoftheAHPmodelonthehealthcaredecisionmakingisnecessary.
AbbreviationsACP:AmericanCollegeofPhysicians;AHP:AnalyticHierarchyProcess;CDC:CentersforDiseaseControlandPrevention;CR:Consistencyratio;CRC:Colorectalcancer;FIT:Fecalimmunochemicaltest.
CompetinginterestsTheauthorsdeclarethattheyhavenocompetinginterests.
Authors'contributionsYXwasresponsibleforstudyconcept,acquisitionofdata,statisticalanalysis,andthewritingofinitialandsubsequentmanuscriptdraftsandnumerousrevisions.
BLwasresponsibleforstudyconceptanddesign,acquisitionofdata,obtainingfunding,andstudysupervision.
JMDwasresponsibleforacquisitionofdata,criticalrevisionforintellectualcontent,andstudysupervision.
GBandJCDwereresponsibleforstudyconcept,andcriticalrevisionforintellectualcontent.
Allauthorscontributedtotheinterpretationofdataandreadandapprovedthefinalversionofthemanuscript.
AcknowledgementsFinancialsupportforthisstudywasprovidedbytheNationalCancerInstitute–1RC1CA144907-01(PrincipalInvestigator,BarceyT.
Levy).
Thefundingagreementensuredtheauthors'independenceindesigningthestudy,interpretingthedata,writing,andpublishingthereport.
Authordetails1DepartmentofFamilyMedicine,RoyJ.
andLucilleA.
CarverCollegeofMedicine,UniversityofIowa,IowaCity,IA52242,USA.
2DepartmentofEpidemiology,CollegeofPublicHealth,UniversityofIowa,IowaCity,IA52242,USA.
3DepartmentofInternalMedicine,RoyJ.
andLucilleA.
CarverCollegeofMedicine,UniversityofIowa,IowaCity,IA52242,USA.
Received:30April2014Accepted:10April2015References1.
AmericanCancerSociety.
Cancerfacts&figures2013.
Atlanta:AmericanCancerSociety;2013.
2.
EdwardsBK,WardE,KohlerBA,EhemanC,ZauberAG,AndersonRN,etal.
Annualreporttothenationonthestatusofcancer,1975–2006,featuringcolorectalcancertrendsandimpactofinterventions(riskfactors,screening,andtreatment)toreducefuturerates.
Cancer.
2010;116(3):544–73.
3.
QaseemA,DenbergTD,HopkinsJrRH,HumphreyLL,LevineJ,SweetDE,etal.
Screeningforcolorectalcancer:aguidancestatementfromtheAmericanCollegeofPhysicians.
AnnInternMed.
2012;156(5):378–86.
4.
RimSH,JosephDA,SteeleCB,ThompsonTD,SeeffLC,CentersforDiseaseControlandPrevention(CDC).
Colorectalcancerscreening-UnitedStates,2002,2004,2006,and2008.
MMWRSurveillSumm.
2011;60Suppl:42–6.
5.
HawleyST,McQueenA,BartholomewLK,GreisingerAJ,CoanSP,MyersR,etal.
Preferencesforcolorectalcancerscreeningtestsandscreeningtestuseinalargemultispecialtyprimarycarepractice.
Cancer.
2012;118(10):2726–34.
6.
WolfRL,BaschCE,BrouseCH,ShmuklerC,SheaS.
Patientpreferencesandadherencetocolorectalcancerscreeninginanurbanpopulation.
AmJPublicHealth.
2006;96(5):809–11.
7.
MeissnerHI,BreenN,KlabundeCN,VernonSW.
PatternsofcolorectalcancerscreeninguptakeamongmenandwomenintheUnitedStates.
CancerEpidemiolBiomarkersPrev.
2006;15(2):389–94.
8.
AllisonJE,SakodaLC,LevinTR,TuckerJP,TekawaIS,CuffT,etal.
Screeningforcolorectalneoplasmswithnewfecaloccultbloodtests:updateonperformancecharacteristics.
JNatlCancerInst.
2007;99(19):1462–70.
9.
vanRossumLG,vanRijnAF,LaheijRJ,vanOijenMG,FockensP,vanKriekenHH,etal.
Randomcomparisonofguaiacandimmunochemicalfecaloccultbloodtestsforcolorectalcancerinascreeningpopulation.
Gastroenterology.
2008;135(1):82–90.
10.
TelfordJJ,LevyAR,SambrookJC,ZouD,EnnsRA.
Thecost-effectivenessofscreeningforcolorectalcancer.
CMAJ.
2010;182(12):1307–13.
11.
AllisonJE.
FIT:avaluablebutunderutilizedscreeningtestforcolorectalcancer-it'stimeforachange.
AmJGastroenterol.
2010;105(9):2026–8.
12.
ZauberAG,Lansdorp-VogelaarI,KnudsenAB,WilschutJ,vanBallegooijenM,KuntzKM.
Evaluatingteststrategiesforcolorectalcancerscreening:adecisionanalysisfortheU.
S.
PreventiveServicesTaskForce.
AnnInternMed.
2008;149(9):659–69.
Xuetal.
BMCHealthServicesResearch(2015)15:175Page8of913.
CentersforDiseaseControlandPrevention(CDC).
Vitalsigns:colorectalcancerscreening,incidence,andmortality–UnitedStates,2002–2010.
MMWRMorbMortalWklyRep.
2011;60(26):884–9.
14.
Schroy3rdPC,EmmonsKM,PetersE,GlickJT,RobinsonPA,LydotesMA,etal.
Aid-assisteddecisionmakingandcolorectalcancerscreening:arandomizedcontrolledtrial.
AmJPrevMed.
2012;43(6):573–83.
15.
AlmogR,EzraG,LaviI,RennertG,HagoelL.
Thepublicprefersfecaloccultbloodtestovercolonoscopyforcolorectalcancerscreening.
EurJCancerPrev.
2008;17(5):430–7.
16.
KatsumuraY,YasunagaH,ImamuraT,OheK,OyamaH.
Relationshipbetweenriskinformationontotalcolonoscopyandpatientpreferencesforcolorectalcancerscreeningoptions:analysisusingtheanalytichierarchyprocess.
BMCHealthServRes.
2008;8:106.
doi:10.
1186/1472-6963-8-106.
17.
DolanJG.
Patientprioritiesincolorectalcancerscreeningdecisions.
HealthExpect.
2005;8(4):334–44.
18.
DolanJG,BoohakerE,AllisonJ,ImperialeTF.
Patients'preferencesandprioritiesregardingcolorectalcancerscreening.
MedDecisMaking.
2013;33(1):59–70.
19.
HummelJM,SteutenLG,Groothuis-OudshoornCJ,MulderN,IjzermanMJ.
Preferencesforcolorectalcancerscreeningtechniquesandintentiontoattend:amulti-criteriadecisionanalysis.
ApplHealthEconHealthPolicy.
2013;11(5):499–507.
20.
SaatyTL.
Ascalingmethodforprioritiesinhierarchicalstructures.
JMathPsychol.
1977;15(3):234–81.
21.
SaatyTL.
Theanalytichierarchyprocess:planning,prioritysetting,resourceallocation.
NewYork:McGraw;1980.
22.
SatoY.
Comparisonbetweenmultiple-choiceandanalytichierarchyprocess:measuringhumanperception.
IntTransOperRes.
2004;11(1):77–86.
23.
FormanE,PeniwatiK.
AggregatingindividualjudgmentsandprioritieswiththeAnalyticHierarchyProcess.
EurJOperRes.
1998;108(1):165–9.
24.
JonesRM,VernonSW,WoolfSH.
IsdiscussionofcolorectalcancerscreeningoptionsassociatedwithheightenedpatientconfusionCancerEpidemiolBiomarkersPrev.
2010;19(11):2821–5.
25.
WhitakerR.
ValidationexamplesoftheAnalyticHierarchyProcessandAnalyticNetworkProcess.
MathComputModel.
2007;46(7–8):840–59.
26.
LingBS,MoskowitzMA,WachsD,PearsonB,SchroyPC.
Attitudestowardcolorectalcancerscreeningtests.
JGenInternMed.
2001;16(12):822–30.
27.
DeBourcyAC,LichtenbergerS,FeltonS,ButterfieldKT,AhnenDJ,DenbergTD.
Community-basedpreferencesforstoolcardsversuscolonoscopyincolorectalcancerscreening.
JGenInternMed.
2008;23(2):169–74.
28.
HawleyST,VolkRJ,KrishnamurthyP,Jibaja-WeissM,VernonSW,KneuperS.
Preferencesforcolorectalcancerscreeningamongracially/ethnicallydiverseprimarycarepatients.
MedCare.
2008;46(9Suppl1):S10–6.
29.
LeardLE,SavidesTJ,GaniatsTG.
Patientpreferencesforcolorectalcancerscreening.
JFamPract.
1997;45(3):211–8.
30.
PowellAA,BurgessDJ,VernonSW,GriffinJM,GrillJP,NoorbaloochiS,etal.
ColorectalcancerscreeningmodepreferencesamongUSveterans.
PrevMed.
2009;49(5):442–8.
31.
InadomiJM,VijanS,JanzNK,FagerlinA,ThomasJP,LinYV,etal.
Adherencetocolorectalcancerscreening:arandomizedclinicaltrialofcompetingstrategies.
ArchInternMed.
2012;172(7):575–82.
32.
SchwartzB.
Theparadoxofchoice:whylessismore.
NewYork:Ecco;2004.
33.
GrazziniG,VenturaL,ZappaM,CiattoS,ConfortiniM,RapiS,etal.
Influenceofseasonalvariationsinambienttemperaturesonperformanceofimmunochemicalfaecaloccultbloodtestforcolorectalcancerscreening:observationalstudyfromtheFlorencedistrict.
Gut.
2010;59(11):1511–5.
34.
DeutekomM,vanRijnAF,DekkerE,BlaauwgeersH,StronksK,FockensP,etal.
UptakeoffaecaloccultbloodtestcolorectalcancerscreeningbydifferentethnicgroupsintheNetherlands.
EurJPublicHealth.
2009;19(4):400–2.
35.
HalloranSP,LaunoyG,ZappaM,InternationalAgencyforResearchonCancer.
Europeanguidelinesforqualityassuranceincolorectalcancerscreeninganddiagnosis.
FirstEdition–Faecaloccultbloodtesting.
Endoscopy.
2012;44Suppl3:SE65–87.
36.
YoungGP.
Population-basedscreeningforcolorectalcancer:Australianresearchandimplementation.
JGastroenterolHepatol.
2009;24Suppl3:S33–42.
37.
MarshallDA,JohnsonFR,PhillipsKA,MarshallJK,ThabaneL,KulinNA.
Measuringpatientpreferencesforcolorectalcancerscreeningusingachoice-formatsurvey.
ValueHealth.
2007;10(5):415–30.
38.
ShokarNK,CarlsonCA,WellerSC.
Informeddecisionmakingchangestestpreferencesforcolorectalcancerscreeninginadiversepopulation.
AnnFamMed.
2010;8(2):141–50.
39.
GriffithJM,LewisCL,BrennerAR,PignoneMP.
Theeffectofofferingdifferentnumbersofcolorectalcancerscreeningtestoptionsinadecisionaid:apilotrandomizedtrial.
BMCMedInformDecisMak.
2008;8:4-6947-8-4.
40.
PignoneM,BucholtzD,HarrisR.
Patientpreferencesforcoloncancerscreening.
JGenInternMed.
1999;14(7):432–7.
SubmityournextmanuscripttoBioMedCentralandtakefulladvantageof:ConvenientonlinesubmissionThoroughpeerreviewNospaceconstraintsorcolorgurechargesImmediatepublicationonacceptanceInclusioninPubMed,CAS,ScopusandGoogleScholarResearchwhichisfreelyavailableforredistributionSubmityourmanuscriptatwww.
biomedcentral.
com/submitXuetal.
BMCHealthServicesResearch(2015)15:175Page9of9