Benefits and Harms of Mammography Screening

download Benefits and Harms of Mammography Screening

of 9

description

Autores: Magnus Loberg et al.

Transcript of Benefits and Harms of Mammography Screening

  • 25/9/2015 www.medscape.com/viewarticle/849292_print

    http://www.medscape.com/viewarticle/849292_print 1/9

    www.medscape.com

    AbstractandIntroductionAbstract

    Mammographyscreeningforbreastcanceriswidelyavailableinmanycountries.Initiallypraisedasauniversalachievementtoimprovewomen'shealthandtoreducetheburdenofbreastcancer,thebenefitsandharmsofmammographyscreeninghavebeendebatedheatedlyinthepastyears.Thisreviewdiscussesthebenefitsandharmsofmammographyscreeninginlightoffindingsfromrandomizedtrialsandfrommorerecentobservationalstudiesperformedintheeraofmoderndiagnosticsandtreatment.Themainbenefitofmammographyscreeningisreductionofbreastcancerrelateddeath.Relativereductionsvaryfromabout15to25%inrandomizedtrialstomorerecentestimatesof13to17%inmetaanalysesofobservationalstudies.UsingUKpopulationdataof2007,for1,000womeninvitedtobiennialmammographyscreeningfor20yearsfromage50,2to3womenarepreventedfromdyingofbreastcancer.Allcausemortalityisunchanged.Overdiagnosisofbreastcanceristhemainharmofmammographyscreening.BasedonrecentestimatesfromtheUnitedStates,therelativeamountofoverdiagnosis(includingductalcarcinomainsituandinvasivecancer)is31%.Thisresultsin15womenoverdiagnosedforevery1,000womeninvitedtobiennialmammographyscreeningfor20yearsfromage50.Womenshouldbeunpassionatelyinformedaboutthebenefitsandharmsofmammographyscreeningusingabsoluteeffectsizesinacomprehensiblefashion.Inaneraoflimitedhealthcareresources,screeningservicesneedtobescrutinizedandcomparedwitheachotherwithregardtoeffectiveness,costeffectivenessandharms.

    Introduction

    Theverb'toscreen'isdefinedas'tosiftbypassingthroughascreen'.[1]'To'sift'derivesfromanoldDutchword('zeef')a'utensilconsistingofacircularframewithafinelymeshedorperforatedbottom,usedtoseparatethecoarserfromthefinerparticlesofanyloosematerial'.[1]

    Thedefinitionsofscreeningvaryamongdifferentcultures,settings,andtimeperiods.[2,3]Ingeneral,alldefinitionsofscreeningincludeanidentificationofdiseaseordiseaseprecursoramongpresumptivelyhealthyindividuals.Therearemainlytwodifferentapproachesofcancerscreening:preventionofdiseasebyfindingandremovingpremalignantprecursorsofcancerandearlydetectionofcancerwherethegoalistotreattheinvasivecancerinanearlycurablestage.[4]In1968,theWorldHealthOrganizationsuggested10principlesthatshouldbefulfilledbeforeimplementingscreeninginapopulation().[5]Someoftheprinciplesregardknowledgeaboutbiologicdevelopmentofcancer(principles4and7).

    Table1.TheWorldHealthOrganization's10principlesofscreening

    1. Theconditionsoughtshouldbeanimportanthealthproblem

    2. Thereshouldbeanacceptedtreatmentforpatientswithrecognizeddisease

    3. Facilitiesfordiagnosisandtreatmentshouldbeavailable

    4. Thereshouldbearecognizablelatentorearlysymptomaticstage

    5. Thereshouldbeasuitabletestorexamination

    6. Thetestshouldbeacceptabletothepopulation

    7. Thenaturalhistoryofthecondition,includingdevelopmentfromlatenttodeclareddisease,shouldbeadequatelyunderstood

    8. Thereshouldbeanagreedpolicyonwhomtotreataspatients

    9. Thecostofcasefindings(includingdiagnosisandtreatmentofpatientsdiagnosed)shouldbeeconomicallybalancedinrelationtopossibleexpenditureonmedicalcareasawhole

    10. Casefindingshouldbeacontinuingprocessandnota'onceandforall'project

    Screeningforbreastcancerwithmammographyaimsatdetectingbreastcanceratanearly,curablestage.Forearlydetectionbyscreeningtobebeneficial,weanticipateacontinuous,lineargrowthpatternoftumors,andthatbreastcancerhasnotspreadatthetimewhentumorsaredetectableatmammography.Thus,iftheassumptionsoftumorgrowtharenotcorrectorifgrowthoftumorsisheterogenic,screeningmammographymightnotbeanadequatetooltoreducetheburdenofbreastcancer.[6]

    TheideaofearlydetectionstartedintheUSintheearly20thcenturywitheducationalmasscampaignswherethemessageof'donotdelay'seekingmedicalhelpforavarietyofcancersignsandsymptomswascentral.[7]However,noneoftheseearlycampaignshadaneffectonthemortalityofbreastcancer.[8]In1963thefirstrandomizedtrialofmammographyscreeningwaslaunchedwithintheHealthInsurancePlaninNewYork,[8]andseveralothertrialsfollowed.[9]MostofthetrialswereperformedbeforewidespreaduseofantiestrogensandmodernchemotherapywiththeexceptionoftheCanadianNationalBreastScreeningStudyandtheagetrial.[10,11]

    Incontrasttoothercancerscreeningtools,mammographyscreeningwasevaluatedinrandomizedtrialsbeforeitwaswidelyrecommendedandimplemented.Nevertheless,therehasbeenacontinuousdiscussionofmammographyscreening,whichstartedinfullin2000afteraCochranereviewoftherandomizedtrialsindicatedlittleeffectofscreening.[12]Morerecently,theeffectofmammographyscreeningoutsidetheexperimentalsetting,inthemodernerawithimprovementsinawareness,diagnostics,andtreatment,hasbeendiscussed.[13,14]

    Themammographydebatehasnotonlybeenaboutthebeneficialeffectsofmammographyscreening,butmorerecentlyalsotheharms.Inthelast10yearsincreasingawarenessofoverdiagnosisinmammographyscreeninghasemerged.Overdiagnosisisdefinedasthedetectionoftumorsatscreeningthatmightneverhaveprogressedtobecomesymptomaticorlifethreateningintheabsenceofscreening.Thisisadirectharmofscreeningbecausemarkerstodistinguishtheoverdiagnosedtumorsfromthepotentiallifethreateningtumorsarelackingand,thus,alltumorsaretreated.Womenwithoverdiagnosedtumorsonlyexperiencetheharmsandsideeffectsoftreatment,withoutanybenefit.Inthisreviewwediscussthebenefitsandharmsofmammographyscreeningandgiveanoverviewofthefindingsfromrandomizedtrialsandfrommorerecentobservationalstudiesfromtheeraofmoderndiagnosticsandtreatment.Weaimatpresentingthebenefitsandharmsper1,000womeninvitedtomammographyscreeningwhostartedscreeningatage50yearsandwerescreenedeverysecondyearuntilage69yearsscreeningofthisagegrouphasbeenshowntoachievemostofthebenefitwithlessharm.[15,16]

    ScreeningMammographyAttendanceRates

    Mammographyscreeningisrecommended(andinEuropeofferedthroughorganizedprograms)inmostWesterncountries.However,inSwitzerlandanindependentpanelofexperts(theSwissMedicalBoard)reviewedtheevidenceonmammographyscreeningandconcludedthatharmsoutweighedthebenefitsandrecommendedagainstmammographyscreening[17]thatis,thatscreeningprogramsshouldnotbeimplementedinareaswheresuchprogramsdonotexistandthattheongoingprogramsshouldbephasedout.Whenscreeningisrecommended,theeligibleagerangediffersindifferentcountriesfrom40to74years.[4,18,19]Therecommendedintervalbetweentwoscreensvariesfrom1to3years.[18]Mammographyscreeningiswellacceptedonaverage,morethanhalfofeligiblewomenattendscreeningmammography.Inmostcountries,attendanceratesarehigherthan70%.Womenaged50to69yearshavethehighestattendancerate.[18,19]Theattendanceratevariesbetweencountries(19.4%to88.9%),andindifferentagegroups.Mostwomenwhohaveparticipatedoncecontinuetoparticipate.

    FalsePositiveTests

    Aswitheverydiagnostictestthesensitivityandspecificityofmammographyscreeningarenotperfectvariouslevelsofsensitivityandspecificityfordetectingbreastcancerhavebeenpublished.[20,21]Theriskofexperiencingafalsepositivemammogramforwomenundergoingbiennialscreeningfromage50to69yearsinEuropeisabout20%,[21]andtheriskofexperiencingabiopsyduetoafalsepositivetestis3%.[21]BasedondatafromtheUK,2.3%ofallwomenwithafalsepositivetesthadalumpectomy,representing76outof100,000womenscreenedinonescreeninground.[22]TheriskisevenhigherintheUS,wherethe10yearfalsepositiverateis30%,and50%ofallwomenwillexperienceafalsepositivemammogramatonetime.[23,24]Thechallengeswithafalsepositivetest,apartfromthemonetarycosts,areimpairedpsychologicalwellbeingandchangesinhealthbehavioramong

    BenefitsandHarmsofMammographyScreeningMagnusLbergMetteLiseLousdalMichaelBretthauerMetteKalagerBreastCancerRes.201517(63)

  • 25/9/2015 www.medscape.com/viewarticle/849292_print

    http://www.medscape.com/viewarticle/849292_print 2/9

    womenwiththefalsepositivetest.After6months,only64%ofthoserecalledduetoafalsepositivetestweredeclaredcancerfreeafter1yearapproximately90%weredeclaredcancerfree,andonlyafter2yearswereallthosewhowereinfactfreeofcancerdeclaredcancerfree.[25]Researchhasshownthatfalsepositiveresultsnegativelyinfluencewomen'spsychologicalwellbeingduringtheperiodimmediatelyafterthetests,andarecentstudyshowedthatwomenwithfalsepositivefindingsexperiencepsychologicalharmforatleast3yearsafterscreening.[26]Womenwithfalsepositivefindingshadhigheruseofhealthcareservices55%ofwomenwhoexperiencedapositiverecallreturnedtotheoutpatientclinicinthefirstyearafterscreening,someuptoeighttimes,[27]andreportedlowerqualityoflifethanthosewithout.[27,28]Somewomenmayalsohavealteredhealthbehaviorandtrustinthehealthcaresystem.[28]

    FalseNegativeTests

    Intervalcancersarecancersdetectedafteranormalscreeningmammogramandbeforethenextscheduledmammogram.Intervalcancerseitherwereoverlookedatthelastmammogramorarerapidlygrowingcancersthatbecomeapparentinthescreeninginterval.[29]Inareinterpretationofintervalcancers,around35%wereoverlooked,[30]while65%werenotvisibleatthelatestmammogramandappearedintheintervalbetweenscreeningmammograms.Ofallbreastcancersdetectedamongwomenwhoparticipateinscreening,28to33%areintervalcancers,[20]andthisproportionseemstobestableinthedifferentscreeningrounds.[29]Useofdigitalmammographyisincreasing,anddetectionratesofductalcarcinomainsitu(DCIS)andinvasivecancersarehigher.Whetherthiswilldecreasetheproportionofintervalcancersisunknown,buttherateofmissedcancersseemstobesimilartothatofanalogue,screenfilmmammography.[31]Onemightanticipate,therefore,thattheproportionofintervalcancerswithdigitalmammographywillbecomparabletothatwithanaloguescreenfilmmammography.However,theincreasingdetectionrateswithdigitalmammographymightincreasetheamountofoverdiagnosis.

    Womendiagnosedwithintervalcancerdonotbenefitfromearlydetection,butcouldbefalselyreassuredbytheirlastnormalmammogramanddelayseekingmedicalcare.However,thismightnotseemtobethecaseaswomenwithintervalcancerdonothavepoorerprognosisthanwomenwhochosenottoutilizemammographyscreening.[29]

    For1,000womeninvitedtomammographyscreeningeverysecondyearfor20yearsfromage50,200willexperienceafalsepositivemammogram,30willundergoabiopsyduetoafalsepositivemammogram,and3willbediagnosedwithintervalcancer[32,33](Figure1).

    Figure1.

    Summaryofbenefitsandharmswhen1,000womenarescreenedeverysecondyearsfor20yearsstartingatage50.Numberofwomenwithfalsepositivemammogramsandfalsepositivebiopsiesarebasedonareview[32].NumberofintervalcancersarebasedonreportednumberofintervalcancerintheNationalHealthServicebreastscreeningprogramme[33].Thenumbersofoverdiagnosedandpreventedbreastcancerdeathsareestimatedbasedon31%overdiagnosis[19]and13to17%reductioninmortalityfrombreastcancer[35].Theserelativenumbersareappliedtotheobservedincidenceofinvasivebreastcancer(womenaged50to69years)andmortality(womenaged55to74years)intheUKin2007[32]thisresultedin15overdiagnosedwomenand2to3preventedbreastcancerdeathsper1,000women.Nodeathsarepreventedoverall[9].

    OverdiagnosisMammographyscreeninginevitablyentailsincreasedbreastcancerincidence[36]duetoearlierdetectionofcancersthatwouldotherwisehavebeendiagnosedlaterinlifeandduetodiagnosisofcancersthatwouldnothavebeenidentifiedclinicallyinsomeone'sremaininglifetime.Thelattercategoryiscommonlyreferredtoasoverdiagnosis.Theoretically,overdiagnosiscanoccurbecausethetumorlackspotentialtoprogresstoaclinicalstage,orevenregresses,[37]orbecausethewomandiesfromothercausesbeforethebreastcancersurfacesclinically.Inreality,thesethreealternativescannotbereliablydisentangled.Inanyofthethreescenariostheindividualwomanwouldbediagnosedandtreatedwithnopossiblesurvivalbenefit.Hence,overdiagnosisrepresentsasubstantialethicaldilemmaandburdensthepatientandthehealthcaresystem.Treatmentforbreastcancerincludessurgery,radiotherapy,chemotherapy,andantiestrogentreatment.Riskofdeathfromcardiovasculardiseaseisincreasedinwomentreatedwithradiotherapy,[38]andadjuvanttreatmentmaybecardiotoxic(forexample,taxanes,anthracyclines,ortrastuzumab).[39]Itispossiblethatovertreatmentcausesincreasedmortalitybyothercausesbesidesbreastcancer.Thismayexplainwhythereisnoreductioninmeasurableoverallmortalitywithscreeningmammography[9](Figure2).

  • 25/9/2015 www.medscape.com/viewarticle/849292_print

    http://www.medscape.com/viewarticle/849292_print 3/9

    Figure2.

    Scenariosfordifferentoutcomesofscreeningmammography.(A)Screeningisineffective.(B)Screeningiseffective.(C)Screeningleadstooverdiagnosis.(D)Screeningleadstooverdiagnosisthatcausesdeathfromsideeffectsoftreatment.

    Overdiagnosisdoesapplytobothcarcinomainsituandinvasivecancerthelifetimeriskofprogressionofcarcinomainsitutoinvasivebreastcancerisunknown,butprobablylessthan50%[40]andtheleadtimeislongerforinsituthaninvasivecancers.Thus,itislogicalandintuitivethatcarcinomainsitucanbeoverdiagnosed.However,pathologicalverifiedinvasivecancerscanalsobeoverdiagnosed.Thiscontradictswhatmostcliniciansweretaughtinmedicalschool,andcanbehardtounderstandforbothcliniciansandthepublic.Onewayoflookingatthischallengeisbyusingthe'icebergmodel':[40]thedevelopmentofcancerisalengthyandcomplexprocess,whereunrepairedgeneticinstabilityandchangesintumormicroenvironmentcouldleadtodistinct,heterogeneoussubpopulationsofabnormalcells.Cancercanbeenvisionedasanicebergofdisease,wherethevisibletipabovethewaterlinecomprisesthemostaggressivelesionsthosethatproducesymptomsandclinicaldisease.Themajorityofourbodyofknowledgeconcerningthenaturalhistoryofmalignanciescomesfromobservationsfromthese'topoftheiceberg',symptomaticlesionsabovethewaterline.[40]Underneaththewater'ssurface,however,theremightbemultiple,indolentcancersubpopulationsofcells.Thesesubpopulationswilllooklikecancertothepathologistifdetectedthroughscreening.[40]Earlydetection(suchasmammographyscreening)divesunderthesurfaceandpicksupsilentlesions.Thenaturalhistoryoftheseasymptomaticlesionshasnotbeenstudiedandisthereforeessentiallyunknown,butmanyofthesemaybeindolentovertimeandnevergeneratesymptomsordiseasewithoutscreening.

    EstimatesofOverdiagnosis

    Preciseestimationofoverdiagnosisisacomplicatedanddifficulttask.Thereisnoperfectanalysisthatwouldbeuniversallyapplicabletothisproblem.Consequently,recentstudiesshowalargevariationintheestimatedoverdiagnosisofbreastcancer,fromnoneto54%.[41]Instudiesbasedonstatisticalmodelingtoadjustforleadtime,estimatesofoverdiagnosisareconsistentlybelow5%.[42,43]Incontrast,observationalstudieshavepublishedhigherestimates,between22and54%,[37,41,42]dependingontheuseofthedenominator.[44]In,wepresenttheamountofoverdiagnosisandreductioninmortalityestimatedwithdifferentdenominators(incidence/deathfrombreastcancerindifferentagegroups).Itclearlyshowsthatdifferentdenominators(rows2to4in)resultindifferentamountsofoverdiagnosisandmortalityreduction.Thus,itisimportantthatbenefitsandharmsofmammographyscreeningarepresentedusingsimilardenominators(in).

    Table2.Differentpercentagesofoverdiagnosisandmortalityreductionbasedonthenumberofcancersoverdiagnosedanddeathsavoidedfrombreastcancerusingdifferentdenominators(incidence/deathfrombreastcancerindifferentagegroups)inNorwayin2010

    Age(years) Expectednumberofcancers Percentageofoverdiagnosis(n=714.4) Expectednumberofbreastcancerdeaths Percentageofmortalityreduction(n=53.7)

    5099 2,208 19.4 693 7.8

    5079 1,571 27.3 506 10.6

    5069 942 45.5 334 16.1

    TheexpectednumberofbreastcancersandbreastcancerdeathsisestimatedastheobservedincidenceandmortalityratesinNorwayfrom1980to1984multipliedbytheNorwegianfemalepopulationin2010[4547].Thenumberofoverdiagnosedcancers(714.4cancers)isbasedonstudiesbyFalkandcoworkers[45]andKalagerandcoworkers[44]andthenumberofreducedbreastcancers(53.7avoideddeathsfrombreastcancer)isestimatedbyreducingthenumberofexpected(358)breastcancerdeathsby15%intheagegroup55to74years(3580.15=53.7).

    Table2.Differentpercentagesofoverdiagnosisandmortalityreductionbasedonthenumberofcancersoverdiagnosedanddeathsavoidedfrombreastcancerusingdifferentdenominators(incidence/deathfrombreastcancerindifferentagegroups)inNorwayin2010

    Age(years) Expectednumberofcancers Percentageofoverdiagnosis(n=714.4) Expectednumberofbreastcancerdeaths Percentageofmortalityreduction(n=53.7)

    5099 2,208 19.4 693 7.8

    5079 1,571 27.3 506 10.6

    5069 942 45.5 334 16.1

    TheexpectednumberofbreastcancersandbreastcancerdeathsisestimatedastheobservedincidenceandmortalityratesinNorwayfrom1980to1984multipliedbytheNorwegianfemalepopulationin2010[4547].Thenumberofoverdiagnosedcancers(714.4cancers)isbasedonstudiesbyFalkandcoworkers[45]andKalagerandcoworkers[44]andthenumberofreducedbreastcancers(53.7avoideddeathsfrombreastcancer)isestimatedbyreducingthenumberofexpected(358)breastcancerdeathsby15%intheagegroup55to74years(3580.15=53.7).

    Table2.Differentpercentagesofoverdiagnosisandmortalityreductionbasedonthenumberofcancersoverdiagnosedanddeathsavoidedfrombreastcancerusingdifferentdenominators(incidence/deathfrombreastcancerindifferentagegroups)inNorwayin2010

  • 25/9/2015 www.medscape.com/viewarticle/849292_print

    http://www.medscape.com/viewarticle/849292_print 4/9

    Age(years) Expectednumberofcancers Percentageofoverdiagnosis(n=714.4) Expectednumberofbreastcancerdeaths Percentageofmortalityreduction(n=53.7)

    5099 2,208 19.4 693 7.8

    5079 1,571 27.3 506 10.6

    5069 942 45.5 334 16.1

    TheexpectednumberofbreastcancersandbreastcancerdeathsisestimatedastheobservedincidenceandmortalityratesinNorwayfrom1980to1984multipliedbytheNorwegianfemalepopulationin2010[4547].Thenumberofoverdiagnosedcancers(714.4cancers)isbasedonstudiesbyFalkandcoworkers[45]andKalagerandcoworkers[44]andthenumberofreducedbreastcancers(53.7avoideddeathsfrombreastcancer)isestimatedbyreducingthenumberofexpected(358)breastcancerdeathsby15%intheagegroup55to74years(3580.15=53.7).

    Overdiagnosismightbeunderestimatedinthestatisticalmodelingstudiesbecausetheytestedonlyoneassumptionatatime,basedeitheronassumptionsfortheriskofprogressionfromcarcinomainsitutoinvasivecancer,[42]oronsojourntimewithadjustmentforleadtime.[42,43]Instatisticalmodelsbasedonsojourntimeandleadtime,overdiagnosishasbeendisregardedintheestimationofleadtime,sincetheassumptionofgrowthhasbeenbasedonaprogressivedisease.This,however,isnotthecaseforoverdiagnosiswherethediseaseisnonprogressiveorperhapsevenregressive.[37]Thus,whenusingtheseestimates,overdiagnosisislikelytobeunderestimated.[48]

    Sincewedonothaveanydirect,biologicalevidenceofnonprogressionorregressionofbreastcancer,assumptionscannoteasilybetested,andrepresentonlya'guess'.Evidencefromobservationalstudiesismoreconvincing.Thedifferenceintheestimatesfromobservationalstudies(22to54%)mightbeduetodifferentassumptionsofexpectedchangesinbreastcancerincidenceduetochangesinbreastcancerriskfactors,differentfollowuptimeafterintroductionofscreening,anddifferencesinaccountingforleadtime.After25yearsoffollowup,theCanadianNationalBreastScreeningStudy,[10]comparingphysicalbreastexaminationwithcombinedphysicalbreastexaminationandannualmammographyinwomenaged40to59years,foundanexcessofinvasivecancerinthescreeningarm,resultingin22%overdiagnosis.Whenthenumberofbreastcancersdetectedatscreeningisusedasthedenominator(asintheCanadianstudy),theamountofoverdiagnosisobservedinthepreviousrandomizedtrialsisstrikinglysimilar(22to24%)[10,49]andinlinewiththe30%reportedintheCochranereviewofscreeningforbreastcancerwithmammography.[9]TheamountofoverdiagnosismightevenbehigherbecauseDCIS,whichaccountsforoneoutoffourbreastcancersdetectedatmammographyscreening,wasnotincludedintheseestimates.[10]IfDCISisaprecursorofinvasivebreastcancer,wewouldexpectadropinincidenceofinvasivebreastcancerafterdetectionandremovalofDCIS.Thereisnoevidenceforthis.Onthecontrary,incidencerateskeepincreasingincountrieswithmammographyscreening.[50]

    Giventheuncertaintyoftheestimatesfrommodelingandobservationalstudies,weusedthebestavailableestimateofoverdiagnosisfromobservationaldatafromaUSstudywhereDCISandinvasivecancerwereincluded,followupwasmorethan25yearsafterscreeningwasinitiatedandnoextensiveuntestableassumptionsweremade.[19]However,intheUSthereisnomammographyscreeningprogram,andtherateoffalsepositivesishigherthaninEuropeandAustralia.Thus,itmightbepossiblethattheamountofoverdiagnosisdiffersbetweentheUSandEuropeandAustralia.SincenoneoftheestimatesofoverdiagnosisfromEuropeorAustraliawerebasedonfollowupaslongasintheUSstudy,wechoosetousetheUSestimateof31%overdiagnosis(inlinewithwhatisobservedintherandomizedtrials).[19]Weestimatedthenumberofoverdiagnosedwomenbasedontheobservedincidenceofinvasivebreastcancerinwomenaged50to69yearsintheUKin2007.[19,34,49]For1,000womeninvitedtobiennialmammographyscreeningfor20yearsfromage50,15willbeoverdiagnosed(Figure1).Basedondifferentmetaanalysesandreviewsofbenefitsandharmsofmammographyscreening[9,22,32]andourbestestimate,[19,34,35]wepresentafigureshowingthedifferentestimatesofoverdiagnosisandpreventeddeathsfrombreastcancer(Figure3).

    Figure3.

    Differentestimatesofoverdiagnosedwomenandsavedlivesfrombreastcancerindifferentmetaanalysesandtrials.Euroscreen:estimatesderivedfromareviewofobservationalstudies,whereestimatesofmortalityreductionfromcasecontrolstudiesareincluded[32].UKIndependentreview:estimatesonrelativeeffectderivedfromrandomizedtrialsofmammographyscreeningandappliedtoUKnationalratesforwomenaged55to79years[22].UKObservational:estimatesbasedon31%overdiagnosis[19]and13to17%reductioninmortalityfrombreastcancer[35]andappliedtotheobservedincidenceofinvasivebreastcancer(womenaged50to69years)andmortality(womenaged55to74years)intheUKin2007[34]thisresultedin2to3preventeddeathsfrombreastcancer.Cochranereview:estimatesfromtherandomizedtrialsofmammographyscreening[9].TheCochranereviewdoesnotassumetheeffectofmammographyscreeningtolastfor20yearsasisassumedintheotherestimates,butrelatestowhatwasobservedintherandomizedtrials[9].

    Tobeabletodifferentiatebetweenpotentiallethalandnonlethalcancers,experimentalstudieshavetobeperformed,preferablyasaninterdisciplinarycooperationbetweenthebiomedicalandclinicalcommunities.First,however,onehastoacceptthatoverdiagnosisdoesoccur,andperhapsalsochangetheterminologyofnonlethalcancerto'IDLEtumor'(InDolentLesionsofEpithelialorigin),asrecentlysuggested.[6]

    BreastCancerMortalityAccordingtotherandomizedbreastcancerscreeningtrials,therelativereductioninmortalityfrombreastcancerrangesbetween15and25%[9,22,36,51]forwomenaged50to69years.Thedifferencesintheseestimatesareduetodifferencesininclusionofrandomizedtrialsinpooledestimates.Forthe25%estimatedreduction,mammographyscreeningversusnoscreeningiscomparedthus,theCanadiantrialwasnotincludedbecausetheycomparedphysicalbreastexaminationtocombinedphysicalbreastexaminationandannualmammography.[10,36]Forthe15%estimatedreduction,methodologicallimitationsinsomeoftherandomizedtrialswasaccountedfor[9]withoutthis'adjustment',a20%reductionwas

  • 25/9/2015 www.medscape.com/viewarticle/849292_print

    http://www.medscape.com/viewarticle/849292_print 5/9

    found.[9,22,52]Noneoftherandomizedtrialsshowedanyeffectoncancermortalityorallcausemortality.[9]Giventhenumberofwomenenrolledintherandomizedtrials(660,000)anda20%reductioninbreastcancermortality,a2%reductioninallcausemortalityshouldhavebeendetectable.[52]Theabsenceofareductioninallcausemortalityindicatesthatwomendieofotherdiseasesataboutthesametimeinlifewithandwithoutscreening.

    StudyDesigns

    Thereareanumberofmethodstoinvestigatetheeffectofmammographyscreeninginanonexperimentalsetting.Cohortstudies,casecontrolstudies,andtrendstudiesshowdifferentestimatesofmortalityreduction,rangingfromnoeffectto50%reductioninbreastcancermortality.[53,54]

    CohortStudies.Theoptimalnonexperimentaldesigntoinvestigatetheeffectofmammographyscreeningisacohortstudyofwomeninvitedandwomennotinvitedtomammographyscreeningwhohavesimilarbaselineriskforbreastcancerandbreastcancerdeathandsimilaropportunitiesforoptimalbreastcancertreatment.Onlyfewsuchstudiesexist,andtheestimatedeffectofmammographyscreeningonbreastcancermortalityvariesfrom10to25%reduction.[35]Apooledestimateofthesetrialsshowedareductioninbreastcancermortalityof13to17%.[35]

    CaseControlStudies.Incasecontrolstudies(sometimescalledcasereferentstudies)casesarewomenwhodieofbreastcancerandcontrolsarewomenwhoarealivestratifiedbywhethertheyhaveundergonescreeningmammographyornot.Thus,thesestudieswhenperformedinsettingswheremammographyscreeningisrecommendedorwherescreeningprogramsexistarecomparisonsofwomenwhoparticipateandwhodonotparticipateinmammographyscreening.Thevalidityofthesestudiesislowbecauseofhealthyscreeneeandselfselectionbias,aswomenwithbreastcancerarenoteligibletomammographyscreeningortobecontinuedtobescreened(selectionofthemosthealthy),andwomenwhochoosetoparticipateinmammographyscreening(selection)maydifferwithregardtoriskofdeathfromthosewhodonotparticipate.[55]Attemptstoadjustforthesebiaseshavebeendonebyadjustingfortherelativeriskinbreastcancermortalitybetweenthenonparticipantsandthenoninvitedcomparisongroup.[7,56]Theunderlyingassumptionoftheseadjustmentsisthatwedoknowtheriskofuninvitedwomen.Inrandomizedtrials,wecaneasilyfindtheriskofbreastcancerdeathforthosenotinvitedtomammographyscreening(thecontrolgroup).However,inobservationalstudieswhereeverybodyisinvitedorrecommendedtoundergomammographyscreening,wehavetomakeassumptionsonriskofdeathfrombreastcanceramongtheuninvitedwomen.Theseassumptionscannotbetestedandarethereforebasedon'bestguess'estimates.Incasecontrolstudies,a50%reductioninmortalityfrombreastcancerisfound,andsimilarreductionsarefoundincohortstudiesofparticipantsandnonparticipantsinmammographyscreening.[54,57]WhentherandomizedtrialfromMalmwasanalyzedasacasecontrolstudy,a58%reductioninmortalityfrombreastcancerwasfound,whereasthereal,observedreductioninthetrialwasonly4%(8%whentheresultswereadjustedfornoncomplianceandcontamination).[36]Thus,estimatesfromcasecontrolstudiessystematicallyoverestimatetheeffectofscreening.

    TrendStudies.Trendstudiesarestudiesofpopulationbasedbreastcancermortalityovertimeindifferentages(agestandardization)andgeographicareas.Dataonpopulationbasedbreastcancermortalityareeasytoretrieve,butastheyearlymortalityrateisnotreflectiveoftimeofdiagnosis,deathsfrombreastcancerdiagnosedbeforeinvitationinfluencesthemortalityratesomeyearsafterscreeningisimplemented.Further,whenalleligiblewomenareinvitedandascreeningprogramhasbeenrunningforsometime,themortalityrateisexpectedtoreachasteadystateandfurtherreductioncannotbeexpected.After7yearsoffollowupintheHealthInsurancePlanstudy,themortalityreductionwasnolongerapparent,[58]indicatingthatscreeninghasnoeffectifnolongeroffered.Foracontinuingprogram,however,themortalityeffectwillnotdisappear,butreachasteadystate.Thus,inthefirstyearsafterscreeninghasbeenintroducedandreachedfullcoverageintheareastudied,thecauseofchangeintrendsofbreastcancermortalitycanbedifficulttostudyandinterpret.MosttrendstudiesshowthatbreastcancermortalityhasdeclinedinmostEuropeancountriessincetheearlytomid1990s.Thedeclineinmortalityisevenhigheramongwomenyoungerthantheeligibleagerangeforscreeningandforsomecountriesareductionisobservedalsoforwomenolderthantheeligibleagerange.[59]Theinterpretationoftheseresultscouldbethatheightenedawarenessandimprovedtherapyratherthanmammographyscreeningareresponsiblefortheobservedreduction.[53,59,60]

    TumorStage

    Anotherbenefitofmammographyscreeningcouldbethatbreastcancersdetectedatscreeningaresmallerandthuslessadvancedthanthosedetectedclinically.Ingeneral,smallertumorsaremorelikelytoberesectedbylumpectomy,andwithlessnodepositivedisease,lessadjuvanttherapyisneeded.Basedontherandomizedmammographyscreeningtrials,however,thisisnotthecasescreeningwasassociatedwithanincreaseinthenumberofmastectomiesofabout20%.[9]Thereasonisthatmammographyincreasedboththenumberofwomendiagnosedwithinvasivebreastcancerandthenumberfoundtohavemultiplemicroscopiccancersdistributedthroughoutthebreast,forwhichmastectomyisrecommended.Further,intheNationalHealthServicebreastscreeningprogramintheUK,30%ofDCISand24%ofinvasivebreastcancersweretreatedwithmastectomy,soearlierdetectiondoesnotnecessarilymeanlessaggressivetreatment.[61]Asmentionedabove,anotherbenefitofmammographyscreeningcouldbelessaggressiveadjuvanttherapy,duetosmallerandlessaggressivetumors.AsseeninthestagedistributioninscreeningandnonscreeninggroupsinNorway,[41]screeningledtothediagnosisof58%morestageI(localizedcancer)and22%morestageII(regionalcancerorcancerinvolvingthelymphnodes)cancers,withoutanyreductioninadvancedstagedisease(stagesIIIandIV).Sinceallthesepatientsreceivesurgery(eithermastectomyorbreastconservingsurgerywithradiation)andmoststageIIpatientsarerecommendedtoreceiveadjuvantchemotherapy,screeningmayhaveledto58%morewomenundergoingbreastsurgeryand22%morewomenundergoingadjuvantchemotherapy.[41]Thus,screeningmammographydoesnotseemtoreducetheburdenofreceivingmoreaggressivetreatment.

    CauseofDeath

    Thenumberofwomensavedfrombreastcancerdeathmightbeoutweighedbydeathfromothercausesduetoharmsoftreatmenthowever,duetouncertaintyabouttheoverallnumberofwomensaved,wepresentdifferentestimatesofwomensavedfrombreastcancerindifferentmetaanalysesofrandomizedandobservationalstudiesofbreastcancer[19,22,32,34,35]

    (Figure3).Thenumberneededtobeinvitedtomammographytosaveorharmwomenishighlydependentontheunderlyingriskofbreastcancerordeathfrombreastcancer(Figures4and5,showingriskofbreastcanceranddeathfrombreastcancerintheUSandUK[49,62]).IntheestimatesshowninFigure1,weuseUKdatafrom2007formortalityfrombreastcancerinwomenaged55to74years,[34]andtherelativereductionof13to17%inbreastcancermortalitybasedonametaanalysisofobservationalstudies.[35]For1,000womeninvitedtomammographyscreeningeverysecondyearfor20twentyyearsfromage50,2to3womenarepreventedfromdyingfrombreastcancer(Figure1).

    Figure4.

  • 25/9/2015 www.medscape.com/viewarticle/849292_print

    http://www.medscape.com/viewarticle/849292_print 6/9

    Benefitandharmwithscreeningmammographyanduseofaspirinover10years[62].Shownarethe10yearriskofdeathfrombreastcancer(barsabove0)andthe10yearriskofthediagnosisofbreastcancer(barsbelow0)amongwomenaged40yearsand50years,withandwithoutmammographyscreening.Alsoshownarethe10yearriskofdeathfromcancer(barabove0)andthe10yearriskofmajorextracranialbleeding,definedasbleedingnecessitatingtransfusionorresultingindeath(barbelow0),associatedwiththeuseornonuseofaspirinasaprimarypreventivemeasure(onthebasisoffindingsfromrandomizedtrials).Ineachpair(noscreeningversusscreeningandnoaspirinversusaspirin),thedifferencebetweenthepercentagesrepresentedbythebarsshowstheabsolutebenefitorharmassociatedwithscreeningmammographyortheuseofaspirin.Backgrounddataarederivedfromtheliterature.

    Figure5.

    Twentyyearriskfordiagnosisof,anddeathfrom,breastandprostatecancerwithandwithoutscreeningintheUnitedKingdom[49].Displayedare20yearabsoluterisksforincidence(includingoverdiagnosis)andmortalitywithandwithoutscreening.Overdiagnosisissetto45%forprostatecancerand22%forbreastcancer,respectively(age50to69years).Mortalityreductionissettobe20%forbothcancers(age55and74years).Forprostatecancer,theestimatesarebasedontheobservedincidenceandmortalityin1998(beforeanywidespreaduseofprostatespecificantigen(PSA))andforbreastcancerin2007(latestdataavailable).

    InformationtoWomenScreeningdiffersfromclinicalpractice.Individualswhoundergoascreeningprocedureareinvitedtoparticipatewiththeimpliedexpectationthattheywillbenefit.Thiscontrastswithclinicalpractice,wherethepatientsapproachthemedicalpractitionerwithasymptomorcomplaintforhelp.[3]Thus,itisofutmostimportancethatinformationaboutbenefitsandharmsofmammographyscreeningisbalanced.However,theharmsofscreeninghavenotbeencommunicatedtothepublicaswellasthebenefits.[63,64]Withincreasingevidenceofoverdiagnosis,thisisofconcernandviolatestheindividual'spossibilitytomakeaninformedchoice.

    However,properinformationonrisksandbenefitsisnoteasy.Firstly,howdoclinicianscommunicatebenefitsandharms?Theuseofrelativerisksmaysuggestgreatereffectsthanexist,whereastheuseofabsoluterisks(orequivalents,suchasthenumberneededtoscreen)preventsthismisunderstanding.Theuseofrelativerisksshouldbeavoidedoremployedonlyincombinationwithmorecomprehensibleformsofcommunicatingrisk,suchasabsoluterisksornumbersneededtoscreen.[65]Secondly,manycannotinterpretnumbersaswellaswordsandhavedifficultyunderstandingnumericalexpressionsofrisk.[66]Inmedicalschools,coursesinstatisticsusuallydonotgofarenoughinteachingstatisticalorprobabilisticthinking,andfewteachstrategiesforeffectivecommunication.Hence,mostphysiciansarepoorlyequippedtodiscussriskfactorsinawaythatisreadilycomprehensibletotheirpatients.Thisdeficiencyputstheidealofinformedconsentinjeopardy.[65,67]

    Framingisthepresentationoflogicallyequivalentinformationindifferentforms.Positiveframingemphasizestheabsenceofdiseasenegativeframingemphasizesthepresenceofdisease[65](Figure6).Basedonthe20yearriskforawomanintheUKtodieofbreastcancer,theriskofdyingfrombreastcancerwithmammographyscreeningwouldbe15per1,000womenand17to18per1,000womenwithoutmammographyscreening.[49]Positiveframingwouldbethatthenumberofwomenthatwillnotdiefrombreastcancerrisesfrombetween982and983to985per1,000womenwiththeadditionofscreeningforbreastcancer.[34,35]AnexampleofpositiveframingisillustratedinFigure6.

  • 25/9/2015 www.medscape.com/viewarticle/849292_print

    http://www.medscape.com/viewarticle/849292_print 7/9

    Figure6.

    Positiveframing.Outof1,000womenaged50to69yearsinvitedeverysecondyear,781arealivewithscreeningandthesamenumberwithoutscreeningoverthecourseof20years.Correspondingly,985womenand982to983womenwithoutscreeningwillnotdieofbreastcanceraged55to74years.Negativeframing:outof1,000womenaged50to69yearsinvitedeverysecondyear,204womenwilldiewithscreeningandthesamenumberwithoutscreening.Correspondingly,15womenwithscreeningand17to18womenwithoutscreeningwilldieofbreastcancerbetween55and74yearsold.Numberofwomendyingamongwomenaged55to74yearsisbasedontheobservedmortalityratesinEnglandandWalesin2007[68].Thenumberofwomendyingovera20yearperiodisestimatedbysummingthemortalityratesfortheages55to74[68].

    Womenarenotonlyoverestimatingtheirriskofbreastcancer,butalsosubstantiallyoverestimatingthebenefitofmammographyscreening.[67,6971]Over50%ofallwomenaskedthoughtmammographyscreeningreducedtheriskofdyingfrombreastcancerbyatleast50%.[67,69]Further,womenwantedtohavebalancedinformationandsharethedecisionwiththeirphysician,[71]butmanyreportedtheywereneverprovidedinformationonfalsepositivesandsideeffects.[71]AreportfromNorway,wherewomenareinvitedwithaprescheduledtimeanddateofascreeningmammographyappointment,showedthatiftheinvitationletterincludedaninformationleafletaimedatenablingwomentomakeafreeandinformedchoice,theprescheduledappointmentunderminedtheoptionofnotparticipating.[72]Theauthorsconcludedthatthecurrentrecruitmentproceduresgaveprioritytoscreeninguptakeattheexpenseofinformedchoice.[72]Thus,theprincipleofinformedchoicemightbeinjeopardy.[72]

    ConclusionWomenshouldbecorrectlyinformedaboutthebenefitsandharmsofmammographyscreening(Figures1and2).AcomprehensiblewayofcommunicatinginformationonbenefitsandharmsofmammographyscreeningispresentedinFigure1:among1,000womenwhostartscreeningatage50andarescreenedfor20years,2to3willavoiddyingfrombreastcancerand200womenwillhaveatleastonefalsepositivetest,30willundergoabiopsy,3willbediagnosedwithanintervalcancer,andbreastcancerwillbeoverdiagnosedin15.

    Inaneraoflimitedresourcesforhealthcareandpreventiveservices,weneedtoscrutinizeoureffortsinscreeningandprevention.Oneoftheoverarchinggoalsofscreeningisthereductionofincidenceormortalityofdisease.Currently,wedorecommendsomescreeningservices(suchasmammography),whileothersaredebatedordiscouraged(suchasprostatespecificantigenscreeningforprostatecanceroraspirinforprimarypreventionofcardiovasculardiseaseandprematuredeath).However,asFigures4and5show,thesedifferencesinrecommendationsdooftennotreflectdifferencesineffectivenessorharmsbetweenthedifferenttests.[49,62]

    References

    1. ShorterOxfordEnglishDictionary.Oxford,UnitedKingdom:OxfordUniversityPress2010

    2. RaffleAE,GrayJAM.Screening:EvidenceandPractice.Oxford,UnitedKingdom:OxfordUniversityPress2007.

    3. HollandWW,StewartS.ScreeninginDiseasePrevention.Whatworks?Oxford,UnitedKingdom:TheNuttfieldTrust/RadcliffePublishingLtd2005.

    4. BretthauerM,KalagerM.Principles,effectivenessandcaveatsinscreeningforcancer.BrJSurg.2013100:5565.

    5. WilsonJMG,JungerG.PrinciplesandPracticeofScreeningforDisease.Geneva,Switzerland:WHO1968.

    6. EssermanL,ThompsonIM,ReidB,NelsonP,RansohoffDF,WelchHG,etal.Addressingoverdiagnosisandovertreatmentincancer:aprescriptionforchange.LancetOncol.201415:e23442.

    7. AronowitzRA.UnnaturalHistory:BreastCancerandAmericanSociety.NewYork:CambridgeUniversityPress2007.

    8. ReynoldsH.TheBigSqueeze:aSocialandPoliticalHistoryoftheControversialMammogram.NewYork:CornellUniversityPress2012.

    9. GtzscheP,JrgensenKJ.Screeningforbreastcancerwithmammography.CochraneDatabaseSystRev.20136,CD001877.

    10. MillerA,WallC,BainsC,SunP,ToT,NarodS.TwentyfiveyearfollowupoftheCanadiannationalbreastscreeningstudy.BMJ.2014348:g366.

    11. MossSM,CuckleH,EvansA,JohnsL,WallerM,BobrowL.Effectofmammographicscreeningfromage40yearsonbreastcancermortalityat10years'followup:arandomisedcontrolledtrial.Lancet.2006368:205360.

    12. GtzscheP,OlsenO.Isscreeningforbreastcancerwithmammographyjustified?Lancet.2000355:12934.

    13. JrgensenKJ,KeenJD,GtzcshePC.Ismammographicscreeningjustifiableconsideringitssubstantialoverdiagnosisrateandminoreffectonmortality?Radiology.2011260:6217.

    14. KopansDB,SmithRA,DuffySW.Mammographicscreeningand"overdiagnosis".Radiology.2011260:61620.

    15. MandelblattJS,CroninKA,BaileyS,BerryDA,deKoningHJ,DraismaG,etal.Effectsofmammographyscreeningunderdifferentscreeningschedules:modelestimatesofpotentialbenefitsandharms.AnnInternMed.2009151:73847.

  • 25/9/2015 www.medscape.com/viewarticle/849292_print

    http://www.medscape.com/viewarticle/849292_print 8/9

    16. NelsonHD,TyneK,NaikA,BougatsosC,ChanBK,HumphreyL,etal.Screeningforbreastcancer:anupdatefortheU.S.PreventiveServicesTaskForce.AnnInternMed.2009151:72737.

    17. SwissMedicalBoard.SystematischesMammographieScreening[http://www.medicalboard.ch/fileadmin/docs/public/mb/Fachberichte/20131215_Bericht_Mammographie_Final_rev.pdf]

    18. GiordanoL,vonKarsaL,TomatisM,MajekO,deWolfC,LancuckiL,etal.MammographicscreeningprogrammesinEurope:organization,coverageandparticipation.JMedScreen.201219Suppl1:7282.

    19. BleyerA,WelchH.Effectofthreedecadesofscreeningmammographyonbreastcancerincidence.NEnglJMed.2012367:19982005.

    20. TrnbergS,KemetliL,AscunceN,HofvindS,AnttilaA,SradourB,etal.ApooledanalysisofintervalcancerratesinsixEuropeancountries.EurJCancerPrev.201019:8793.

    21. HofvindS,PontiA,PatnickJ,AscunceN,NjorS,BroedersM,etal.FalsepositiveresultsinmammographicscreeningforbreastcancerinEurope:aliteraturereviewandsurveyofservicescreeningprogrammes.JMedScreen.201219Suppl1:5766.

    22. IndependentUKPanelonBreastCancerScreening.Thebenefitsandharmsofbreastcancerscreening:anindependentreview.Lancet.2012380:177886.

    23. ElmoreJG,BartonMB,MoceriVM,PolkS,ArenaPJ,FletcherSW.Tenyearriskoffalsepositivescreeningmammogramsandclinicalbreastexaminations.NEnglJMed.1998338:108996.

    24. HubbardRA,KerlikowskeK,FlowersCI,YankaskasBC,ZhuW,MigliorettiDL.Cumulativeprobabilityoffalsepositiverecallorbiopsyrecommendationafter10yearsofscreeningmammography:acohortstudy.AnnInternMed.2011155:48192.

    25. LidbrinkE,ElfvingJ,FrisellJ,JonssonE.Neglectedaspectsoffalsepositivefindingsofmammographyinbreastcancerscreening:analysisoffalsepositivecasesfromtheStockholmtrial.BMJ.1996312:2736.

    26. BrodersenJ,SiersmaVD.Longtermpsychosocialconsequencesoffalsepositivescreeningmammography.AnnFamMed.201311:10615.

    27. vanderSteegAFW,KeyzerDekkerCM,DeVriesJ,RoukemaJA.Effectofabnormalscreeningmammogramonqualityoflife.BrJSurg.201198:53742.

    28. BartonM.Increasedpatientconcernafterfalsepositive.JGenInternMed.200116:1506.

    29. KalagerM,TamimiR,BretthauerM,AdamiHO.Prognosisinwomenwithintervalbreastcancer:populationbasedobservationalcohortstudy.BMJ.2012345:e7536.

    30. HofvindS,SkaaneP,VitakB,WangH,ThoresenS,EriksenL,etal.Influenceofreviewdesignonpercentagesofmissedintervalbreastcancers:aretrospectstudyofintervalcancersinapopulationbasedscreeningprogram.Radiology.2005237:43743.

    31. HoffSR,AbrahamsenAL,SamsetJH,VigelandE,KleppO,HofvindS.Breastcancer:missedintervalandscreeningdetectedcanceratfullfielddigitalmammographyandscreenfilmmammographyresultsfromaretrospectivereview.Radiology.2012246:37886.

    32. PaciE,EUROSCREENWorkingGroup.SummaryoftheevidenceofbreastcancerservicescreeningoutcomesinEuropeandfirstestimateofthebenefitandharmbalancesheet.JMedScreen.201219Suppl1:513.

    33. NHSBreastScreeningProgramme.Nationalcollationofbreastintervalcancerdata:Screeningyears1stApril200331stMarch2005[http://www.cancerscreening.nhs.uk/breastscreen/publications/nhsbspoccasionalreport1203.pdf]

    34. SteliarovaFoucherE,O'CallaghanM,FerlayJ,MasuyerE,FormanD,ComberH,etal.EuropeanCancerObservatory:CancerIncidence,Mortality,PrevalenceandSurvivalinEurope.Version1.0(September2012).EuropeanNetworkofCancerRegistries,InternationalAgencyforResearchonCancer[http://eco.iarc.fr,http://eco.iarc.fr/EUREG/AnalysisG.aspx]

    35. IrvinVL,KaplanRM.Screeningmammography&breastcancermortality:metaanalysisofquasiexperimentalstudies.PLoSOne.20149:e98105.

    36. VainioH,BianchiniF.IARCHandbookofCancerPrevention.Volume7.BreastCancerScreening.Lyon,France:IARCPress2002.

    37. ZahlPH,MlenJ,WelchHG.Thenaturalhistoryofinvasivebreastcancersdetectedbyscreeningmammography.ArchInternMed.2008168:23116.

    38. DarbySC,EwertzM,McGaleP,BennetAM,BlomGoldmanU,BrnnumD,etal.Riskofischemicheartdiseaseinwomenafterradiotherapyforbreastcancer.NEnglJMed.2013368:98798.

    39. YehETH,BickfordC.Cardiovascularcomplicationsofcancertherapy:incidence,pathogenesis,diagnosis,andmanagement.JAmCollCardiol.200953:223147.

    40. KramerBS,CroswellJM.Cancerscreening:theclashofscienceandintuition.AnnuRevMed.200960:12537.

    41. KalagerM,AdamiHO,BretthauerM,TamimiRM.Overdiagnosisofinvasivebreastcancerduetomammographyscreening:resultsfromtheNorwegianScreeningProgram.AnnInternMed.2012156:4919.

    42. PulitiD,DuffySW,MiccinesiG,deKoningH,LyngeE,ZappaM.OverdiagnosisinmammographicscreeningforbreastcancerinEurope:aliteraturereview.JMedScreen.201219Suppl1:4256.

    43. deKoningHJ,DraismaG,FrancheboudJ,deBruijnA.Microsimulationmodelingestimatesbasedonobservedscreenandclinicaldata.BreastCancerRes.20068:202.

    44. KalagerM,LbergM,FnnebVM,BretthauerM.Failuretoaccountforselectionbias.IntJCancer.2013133:27513.

    45. FalkRS,HofvindS,SkaaneP,HaldorsenT.Overdiagnosisamongwomenattendingapopulationbasedmammographyscreeningprogram.IntJCancer.2013133:70512.

    46. StatisticsNorway.Population,bysexandage[https://www.ssb.no/statistikkbanken/selectvarval/Define.asp?subjectcode=&ProductId=&MainTable=FolkemEttAarig&nvl=&PLanguage=0&nyTmpVar=true&CMSSubjectArea=befolkning&KortNavnWeb=folkemengde&StatVariant=&checked=true

    47. NORDCANDatabase[http://wwwdep.iarc.fr/NORDCAN/english/frame.asp]

    48. ZahlPH,JrgensenKJ,GtzscheP.Leadtimemodelsshouldnotbeusedtoestimateoverdiagnosisincancerscreening.JGenInternMed.2014doi:10.007/s1160601428122.

    49. KalagerM,AdamiHO,BretthauerM.Toomuchmammography.BMJ.2014348:g1403.

    50. JrgensenKJ,GtzschePC.Overdiagnosisinpubliclyorganizedmammographyscreeningprogrammes:systematicreviewofincidencetrends.BMJ.2009339:b2589.

    51. HumphreyLL,HelfandM,ChanBK,WoolfSH.Breastcancerscreening:asummaryoftheevidencefortheU.S.PreventiveServicesTaskForce.AnnInternMed.2002137:34760.

    52. JniP,ZwahlenM.Itistimetoinitiateanotherbreastcancerscreeningtrial.AnnInternMed.2014160:8646.

  • 25/9/2015 www.medscape.com/viewarticle/849292_print

    http://www.medscape.com/viewarticle/849292_print 9/9

    AbbreviationDCIS:Ductalcarcinomainsitu

    AcknowledgementsThestudywassupportedbygrantsfromtheNorwegianCancerSociety(PhDscholarshipMagnusLberg,grantnumberHS0220090082),USNorwayFulbrightFoundationforEducationalExchange(FulbrightfellowshipMagnusLberg),andHelseSorOst(ResearchgrantMetteKalager,grantnumber2014106).

    BreastCancerRes.201517(63)2015BioMedCentral,Ltd.

    Copyrighttothisarticleisheldbytheauthor(s),licenseeBioMedCentralLtd.ThisisanOpenAccessarticle:verbatimcopyingandredistributionofthisarticlearepermittedinallmediaforanypurpose,providedthisnoticeispreservedalongwiththearticle'soriginalcitation.

    53. JrgensenKJ,ZahlPH,GtzschePC.BreastcancermortalityinorganizedmammographyscreeninginDenmark:comparativestudy.BMJ.2010340:c1241.

    54. NicksonC,MasonKE,EnglishDR,KavanaghAM.Mammographicscreeningandbreastcancermortality:acasecontrolstudyandmetaanalysis.CancerEpidemiolBiomarkersPrev.201221:147988.

    55. MorrisonAS.ScreeninginChronicDisease.2nded.NewYork:OxfordUniversityPress1992[MonographsinEpidemiologyandBiostatistics.Volume19].

    56. DuffySW,CuzickJ,TabarL,VitakB,ChenTHH,YenMF,etal.Correctingfornoncompliancebiasincasecontrolstudiestoevaluatecancerscreeningprogrammes.ApplStat.200251:23543.

    57. HofvindS,UrsinG,TretliS,SebudegrdS,MllerB.BreastcancermortalityinparticipantsoftheNorwegianBreastCancerScreeningProgram.Cancer.2013119:310612.

    58. ShapiroS.Screeningforbreastcancer:theHIPRandomizedControlledTrial.HealthInsurancePlan.JNatlCancerInstMonogr.199722:2730.

    59. AutierP,BoniolM,LaVecchiaC,VattenL,GavinA,HryC,etal.Disparitiesinbreastcancermortalitytrendsbetween30Europeancountries:retrospectivetrendanalysisofWHOmortalitydatabase.BMJ.2010341:c3620.

    60. AutierP,BoniolM,GavinA,VattenLJ.BreastcancermortalityinneighbouringEuropeancountrieswithdifferentlevelsofscreeningbutsimilaraccesstotreatment:trendanalysisofWHOmortalitydatabase.BMJ.2011343:d4411.

    61. NHSBreastScreeningProgrammeandAssociationofBreastSurgery.AnAuditofScreenDetectedBreastCancersfortheYearofScreeningApril2010toMarch2011[http://www.cancerscreening.nhs.uk/breastscreen/publications/baso20102011.pdf]

    62. SmithRA,KerlikowskeK,MigliorettiDL,KalagerM.Principles,effectivenessandcaveatsinscreeningforcancer.NEnglJMed.2012367:e31.

    63. JrgensenKJ,GtzschePC.Contentofinvitationforpubliclyfundedscreeningmammography.BMJ.2006332:53841.

    64. JrgensenKJ,GtzschePC.Informationinpractice.Presentationonwebsitesofpossiblebenefitsforbreastcancer:crosssectionalstudy.BMJ.2004328:148.

    65. ElmoreJG,GigerenzerG.Benignbreastdiseasetheriskofcommunicatingrisk.NEnglJMed.2005353:2979.

    66. DeyoRA,PatrickDL.HopeorHype:theObsessionwithMedicalAdvancesandtheHighCostofFalsePromises.NewYork:AMACOM2005.

    67. ChamotE,PernegerTV.Misconceptionsaboutefficacyofmammographyscreening:apublichealthdilemma.JEpidemiolCommunityHealth.200155:799803.

    68. OfficeforNationalStatistics.Mortalitystatistics:deathregisteredin2007[http://www.ons.gov.uk/ons/datasetsandtables/index.html?newquery=mortality+statistics%3Adeath+registered+in+2007&newoffset=150&pageSize=50&contenttype=Reference+table&contenttype=Dataset&contenttypeorig=%22Dataset%22+OR+contenttype_original%3A%22Reference+table%22&sortBy=none&sortDirection=none&applyFilters=true]

    69. DomenighettiG,D'AvanzoB,EggerM,BerrinoF,PernegerT,MosconiP,etal.Women'sperceptionofthebenefitsofmammographyscreening:populationbasedsurveyinfourcountries.IntJEpidemiol.200332:8168.

    70. McMenaminM,BarryH,LennonAM,PurcellH,BaumM,KeeganD,etal.AsurveyofbreastcancerawarenessandknowledgeinaWesternpopulation:lotsoflightbutlittleillumination.EurJCancer.200541:3937.

    71. DaveyHM,BarrattAL,DaveyE,ButowPN,RedmanS,HoussamiN,etal.Medicaltests:women'sreportedandpreferreddecisionmakingrolesandpreferencesforinformationonbenefits,sideeffectsandfalseresults.HealthExpect.20025:33040.

    72. sterlieW,SolbjrM,SkolbekkenJA,HofvindS,StnanAR,ForsmoS.Challengesofinformedchoiceinorganizedscreening.JMedEthics.200834:e5.