[PDF] [PDF] Value of Genetic Testing for Hereditary Colorectal Cancer in a - UAB

Population prevalence of familial cancer and common - Nature

breast-ovarian cancer and hereditary nonpolyposis colon cancer Results: criteria for hereditary nonpolyposis colorectal cancer (HNPCC) were adapted from the Amsterdam II criteria11 and revised Widowed/divorced/separated 29 0 11 2

Comparison of motivations and concerns for genetic testing in

available are hereditary non-polyposis colorectal cancer degree relatives of individuals with breast, ovarian, or colon cancer Single, divorced or widowed

[PDF] Predictive Genetic Testing and Counselling for Hereditary - CORE

Hereditary Non-Polyposis Colorectal Cancer (HNPCC) A prospective follow-up study of acceptance and psychosocial consequences Katja Aktan-Collan

DNA Alterations in Lynch Syndrome

Keywords Lynch syndrome • Hereditary nonpolyposis colorectal cancer • Hereditary and T Boveri's (and M Boveri's, his wife) description of atypical chromo-

[PDF] Randomized trial of a decision aid for individuals considering

BACKGROUND Despite the potential benefits of genetic testing for hereditary nonpolyposis colorectal cancer (HNPCC) risk, individuals can find the genetic

[PDF] Value of Genetic Testing for Hereditary Colorectal Cancer in a - UAB

(no false negatives, results disclosed to primary care physi- cian) Only 41 (95 or hereditary nonpolyposis colorectal cancer (HNPCC), has empirical Divorced 48 (14) Separated 6 (2) Never married 33 (9) Living with partner 12 (3)

[PDF] Hérédité - Attestation porte fort _désignation heritiers

[PDF] Hérédité du diabète - France

[PDF] heredite liee au sexe - Tir À L'Arc

[PDF] Hereinspaziert - GKD Recklinghausen

[PDF] Hereinspaziert ‒ der Flur

[PDF] Hérépian les "Goldwing" vont débarquer

[PDF] Hérésies évangéliques - Michelle d`Astier de la Vigerie - La Religion Et La Spiritualité

[PDF] Here`s - Alliance Française de Portland - France

[PDF] Here`s - Centre des niveaux de compétence linguistique canadiens - Gestion De Projet

[PDF] Here`s - Inside Scoop SF - Anciens Et Réunions

[PDF] Here`s a picture of a public toilet in HOUSTON - Anciens Et Réunions

[PDF] Here`s to you - Anciens Et Réunions

[PDF] Herforder Psychiatrietage 2015

[PDF] Héria, petite fille d`Egypte - Figurines

[PDF] Héricourt Notre Ville - NOVEMBRE DECEMBRE 2011 - France

Value ofGenetic Testing

for HereditaryColorectal Cancer in aProbability-Based USOnline Sample Sara J.Knight, PhD,Ateesha F.Mohamed, MA,Deborah A.Marshall, PhD, Uri Ladabaum,MD, MS,Kathryn A.Phillips, PhD,Judith M.E. Walsh,MD, MPH

Background. Whilechoices aboutgenetic testingare

increasingly commonfor patientsand families,and public opinion surveyssuggest publicinterest ingenomics, itis not knownhow adultsfrom thegeneral populationvalue genetic testingfor heritableconditions. Wesought to understand ina USsample therelative valueof thechar- acteristics ofgenetic teststo identifyrisk ofhereditary colorectal cancer,among thefirst genomicapplications with evidenceto supportits translationto clinicalsettings.

Methods

.AWeb-enabled choice-formatconjoint survey was conductedwith adultsage 50years andolder from a probability-basedUS pan el.Participan tswereaskedto makease ries ofchoices between 2hypotheticalb loodtests thatd ifferedinriskoffa lse- negativ etest, privacy,andcost.

Randomparam eterslogitmodelswere usedtoest imate

preferences,thedollar valueo fgeneticinform ation,and

intenttohav eg enetictestin g.Results. Ato talof355individualscompleted choice-formatquestio ns.Costandprivacyweremo rehigh lyvalue dthanreducing thechance

of afa lse-negativeresult.Most(9 7%[95%confid enceinter- val(C I)],95%-99%) wouldhavegenet ictestingto reduce theri skofdy ingof colorectalca ncerin thebestscenario (nofa lsenegativ es,resultsdisclosedto primarycarephys i- cian).Only41 %(95%CI ,25%-57 %)wou ldhave genetic testinginth eworst case(20 %falsenega tives,re sultsdis- closedtoinsu ran cecompany).Conclusions. Giventhe characteristicsandlevelsincl udedin thechoice,iffa lse- negativetestresu ltsare unlikelyandresu ltsare shared witha prim arycarephysi cian,the majoritywouldhave genetictestin g.Asgenomicservic esbe comewidelyava il- able,prim arycareprofe ssionalswi llneedtobeincreasingly knowledgeableaboutgeneticte stingdecis ions.Keywo rds: genetics;oncology ;internalmedicine ;discretechoice ;util- ityass essment.(MedDe cisMaking20 15;35:734- 744) P ersonalized medicine - healthcare targetedto the characteristicsof individuals,including genetics - has developedrapidly duringthe past decade. Patientsand membersof thepublic saythey would betested toprevent diseasein themselvesor family members. 1,2

Some personalizedmedicine ap-

plications areused todetermine optimaltreatments and areonly relevantto individualswith specific conditions. Othersare usedto identifyat-risk indivi- duals whomight benefitfrom moreintensive screen- ing orprophylactic treatmentand arerelevant to larger segmentsof theasymptomatic population.

Genetic testingused inrisk assessmentfor heredi-

tary colorectalcancer, specificallyLynch syndromeorhe reditarynonpoly posiscolorectalca ncer(HNPCC), has empiricalevidencefor its clinicalval idityandutil- ity. 3,4

Evidence-basedguidelin es forh ereditarycolo-

rectalca ncerhavebe enwidely disseminated, 5-8 and cost-effectivenessevalua tionsha vebee nconducted. 9- 11 The purposeofgenetic te sting inthiscasei stoi den- tifythos ewhowould benefit frommore frequentand intensives creeningorfromproph ylactic surger y, suchas cole ctomy. 12

While themutationsas sociated

withhe reditarycolorectalcancer arenotco mmon, manyp eoplewillhavequ estionsabou tcolorectalcan- cer,a high lyprevalentdise ase,andwhethertheyor relatedfamily memb erswouldbenefitfromgenetic testing.

Many viewgenetics as''a verygood orgood

thing,'' 13 but patientsand thepublic alsohave signif- icant concernsabout genetictesting, includingcost, accuracy, andpotential fordiscrimination basedon genetics. 14-17

Individualsfacingt hesedecisionsma ke

complextradeoffs amongthesefactor s.Beyon dstudies

The Author(s)2015

Reprints andpermission:

DOI: 10.1177/0272989X14565820

734MEDICAL DECISIONMAKING/AUGUST 2015

ORIGINAL ARTICLE

at US DEPT OF VETERAN AFFAIRS on March 28, 2016mdm.sagepub.comDownloaded from of attitudest owardtesting,there islittle quantitative informationonh owindividuals atriskforhealthcondi- tionso rmemberso fthege neralp ublicweigh costswith benefits. 18

This informationiscrit icalfor healthprofes-

sionals whowillneed tobe preparedfor di scuss ions about geneticte stingwiththeirp atients.Without knowledgeof tradeoffs,it isnot possibleto fullyunder- standt hevalueof genetic testing orthefactorsas soci- ated withitsadoption and utilization. 19

In thisst udy,

we soughtto examinet herela tivevalue of specificchar- acteristicsofg enetictesting forhere ditarycolorectal cancer inaprobability-bas eds ampleofadults 50years and olderfromthe general USpopulation,agroupfor whomro utinecolorectalc ancers creeningisrelevant.

METHODS

We useda choice-formatconjoint surveyto mea-

sample ofadults fromthe USpopulation. Choice- format conjoint,also knownas choice-basedconjoint (CBC), isa formof conjointanalysis. Overthe past decade, thistype ofsurvey hasbeen usedincreas- ingly toquantify preferencesfor characteristicsof health careand policy. 20-23

These surveyssimulate

clinical andpolicy decisionmaking andprovide a systematicmethod ofeliciting tradeoffsto quantify the relativeimportance thatindividuals placeon treatment characteristicsor outcomes.This approach composed ofsets ofcharacteristics (e.g.,efficacy, vention isa functionof thesecharacteristics. 24

Survey

The constructionof choicealternatives usedin the

survey wasbased on7 focusgroups ofclinical experts, average-riskcommunity members,and patients atrisk forhereditary cancer(total n= 42). (n= 8),1 groupof geneticcounselors (n= 3),and 2 groups ofphysicians (n= 12).The focusgroups were moderatedby 2experienced qualitative researchers usinga structuredguide andwere recorded usinga digitalaudio recorder.Also, each group wasobserved byseveral ofthe investigators who tookhandwritten notes.A whiteboardwas used torecord thespecific characteristicsand levels of thesecharacteristics identifiedby focusgroup ing. Aftereach group,a digitalphotograph wastaken to recordthe materialfrom thewhiteboard.

Two genetictesting scenarioswere presentedto

the groupsfor discussion.The questionsincluded in thestructured guidewere thesame foreach group, but theperspective wasdifferent. Forexample, while being tested,physicians wereasked howlikely they an open-endeddiscussion ofgenetic testingand 2) a highlystructured discussionof thespecific charac- teristics thatfocus groupmembers indicatedhad influenced orwould influencetheir decisionsabout genetic testing. included ahighly structureddiscussion ofpotential levels orcategories foreach characteristic(e.g., privacy: primarycare doctor,genetics healthprofes- sionals, orinsurance companieswill receivegenetic test results).During thestructured discussion,we asked focusgroup membersto identifya rangeof rel- evant levelsor categoriesfor eachcharacteristic. For Received 21April 2013from HealthServices Researchand Develop- ment Program,Birmingham VeteransAffairs MedicalCenter, Birming- ham, Alabama,USA (SJK);Division ofPreventive Medicine, Department ofMedicine, Universityof Alabamaat Birmingham,Bir- mingham, Alabama,USA (SJK);CHEOR, SpecialtyMedicine, Bayer Healthcare Pharmaceuticals,Inc., Whippany,New Jersey,USA (AFM); Departmentof CommunityHealth Sciences,University ofCal- gary, Calgary,Alberta, Canada(DAM); Departmentof Medicine,Stan- ford University,Palo Alto,California, USA(UL); Departmentof Clinical Pharmacy andInstitute forHealth PolicyStudies, Universityof California at SanFrancisco, SanFrancisco, California,USA (KAP);and Depart- ments ofMedicine andEpidemiology andBiostatistics, Universityof California atSan Francisco,San Francisco,California, USA(JMEW). Uri Ladabaum,MD, MS,declares apotential perceivedconflict ofinter- est withExact SciencesCorporation (Madison,WI) andGiven Imaging Ltd. (Duluth,GA)/Covidien Plc(Dublin, Ireland),for whomhe hasserved as consultant,and MaunaKea Technologies,Inc. (Suwanee,GA), for whom hehas servedon anadvisory board.This workwas supported with agrant fromthe NationalCancer Institute(P01CA130818-1A1) (Kathryn A.Phillips, principalinvestigator; JudithM. E.Walsh and Sara J.Knight, co-leadersof thePreferences Project).This articlerep- resents theperspective ofthe authorsand doesnot necessarilyrepre- sent theviews ofthe Departmentof VeteransAffairs orthe US government. Thefunding agreementensured theauthors' indepen- Address correspondenceto SaraJ. Knight,PhD, TheUniversity ofAla- bama atBirmingham, Divisionof PreventiveMedicine, MedicalTowers Building, 171711th AvenueSouth, Suite650, Birmingham,AL 35205, USA; telephone:(205) 934-5685;e-mail: sjknight@uab.edu.

GENETIC TESTINGFOR HEREDITARYCOLORECTAL CANCER

ORIGINAL ARTICLE735

at US DEPT OF VETERAN AFFAIRS on March 28, 2016mdm.sagepub.comDownloaded from characteristics suchas sensitivity,specificity, false positive, andfalse negative,we provideddefinitions tration appropriatefor thoseof lowernumeracy. Def- initions andsupporting graphicswere displayedon a whiteboardthat wasvisible toall inthe room.We then moderateda discussionof thelevels andcatego- ries leadingto refinementof therelevant andimpor- tant levelsand categories.Finally, weasked the members voteon thehighest, lowest,and intermedi- ate levelsor categoriesthat weremost meaningfulto decisions aboutgenetic testing.

Three experiencedqualitative investigatorscoded

the verifiedtranscripts ofthe recordingsand the notes usinga contentanalysis approach. 25

First, the

3 codersreviewed thetranscripts anddefined atotal

of 10unique codingcategories. Second,after thecat- egories wereclearly definedand acoding manual developed, theinvestigators thenconducted anini- tial roundof codingeach transcript.Discrepancies in codingwere resolvedthrough discussion,and the codingmanual wasrefined. Afinal roundof cod- ing thetranscripts wasthen conductedagain with discrepancies beingresolved throughdiscussion.

Analysis ofthe transcriptsgeneratedarange oftest

characteristics importantto decisionsabout genetic testing forcancer risk.While thediscussions ofthe patient andcommunity membergroups weighed heavily inthe considerations,input fromall groups was consideredin selectingthe characteristicsand levels tobe usedin thechoice task.Among these, ations: 1)frequency ofmentions and2) conceptual distinctness. Thisallowed usto constructchoice alternatives thatincluded asmany relevantcharac- ber toreduce thecognitive burdenof thechoice task.

In addition,we consideredthe potentialof selected

characteristics toprovide informationon therelative importance ofone testcharacteristic comparedwith another. Forexample, whileconcerns aboutprivacy are wellknown, noquantitative informationis avail- able onthe importanceof privacycompared with other characteristics,such asaccuracy.

The finalselection oflevels andcategories was

based onboth thefocus groupdiscussions andthe current literature.For example,cost levelswere based onwhat focusgroup memberssaid theywould pay forgenetic testingand whatwas publishedin the literature ontypical copaymentsfor genetictests to identify Lynchsyndrome. Inaddition, categoriesor levels wereselected toprovide contraststhat had

not beenexamined inprevious studiesand thatwould revealnew informationabout thecharacteris- tics. Forexample, privacycategories werebased on

counselors andother geneticspecialists, andinsur- ance companiesas potentialtest resultrecipients in addition tothe personbeing tested.While previous surveys havesuggested thatthe publicvalues pri- mary carephysicians assources ofinformation about genetic testing,the relativevalue ofinvolving pri- mary carecompared withother healthprofessionals, such asgenetic specialists,also valuedfor their knowledge, isnot known.

The finalchoice alternativesincluded 3genetic

test characteristics:accuracy (false-negativeresults), privacy (whoother thanthe personbeing testedhas access tothe results),and cost(personal costnot cov- ered byinsurance). 25,26

While testaccuracy canbe

described inseveral ways,we selectedone dimen- sion ofaccuracy toinclude amongthe choicealterna- tives toexamine comparisonsbetween distinct characteristics. Amongthe dimensionsof accuracy, we selectedchance ofa false-negativetest result becauseourfocus groupsand priorstudieshaveindi- cated thatboth patientsand communitymembers see as importanthaving informationabout canceror a predispositionfor cancerso thatsomething canbe done andsee asconcerning havinga cancerrisk or a geneticmutation evenwhen testresults are normal.

25,27,28

Each characteristicwas associatedwith 3of 4lev-

egory wasused ineach choicetask alternative.The fractional factorialdesign usedto createthe survey versions wasgenerated usingSAS version9.2 (SAS

Institute, Cary,NC). Tocreate testprofiles forthe

choice questions,we employeda D-optimalalgo- rithm toconstruct afractional factorialmain-effects experimental designin SASversion 9.2,resulting in

36 choicepairs.

29-31

The finalexperimental design

consisted of4 surveyversions, eachcontaining 9 choice questions.Each respondentwas randomly assigned to1 ofthe 4versions, andthe 9choice ques- tions wererandomized ineach surveyversion.

Each choicetask included2 hypotheticalgenetic

test alternatives(blood testA, bloodtest B)and a no-testoption. Theno-test optionwas includedso we couldestimate predictedtest uptake.Including the no-testoption wasalso importantin makingthe choice scenariomore realisticto participants,as in therealworld,''notest'' isanoptiontheycanchoose.

Each setof alternatives(blood testA, bloodtest B)

was presentedwith ahypothetical levelof colorectal cancer risk(i.e., 10%,25%, 50%)(Figure 1)to

KNIGHT ANDOTHERS

736

MEDICAL DECISIONMAKING/AUGUST 2015

at US DEPT OF VETERAN AFFAIRS on March 28, 2016mdm.sagepub.comDownloaded from provide abaseline contextfor eachchoice task.The level ofrisk wasvaried throughoutthe choicetasks in thequestionnaire. Inthe fullexperimental design (36 choicequestions), eachof the3 baselinerisk lev- els occurredexactly 33%of thetime (24/72).In addi- tion, ineach ofthe 4versions, eachbaseline risk occurred 33%of thetime (6/18).

The surveyincluded severalquestions inaddition

to thechoice tasks.Because genetictest resultsfor colorectal cancerhave consequentrisk reductionrec- ommendations, weincluded aseparate questionon choice ofrisk reductionstrategy givenhereditary colorectal cancerin thefamily. Inthis question,we asked participantsto chooseeither colectomyto eliminate therisk orcolonoscopy toreduce therisk of dyingof colorectalcancer. Ifcolectomy was selected, theparticipant wasasked aboutthe maxi- noscopy wasselected, theparticipant wasasked to identify themaximum thathe orshe wouldpay for colonoscopies overa lifetime.Several independent evant personcharacteristics (e.g.,cancer history)that were mentionedin thefocus groups,but thatcannot be manipulatedexperimentally.

Because understandingnumerical probabilitiesis

often cognitivelychallenging, weused pictographs to providegraphical riskinformation. 32

Plain-

language definitionsand pictureswere usedto

define conceptsused inthe choicetasks, suchas false-negative testresult, geneticshealth professio-nals, colectomy,and colonoscopy.

We testedand refinedthe initialversion ofthe sur-

vey usingstructured interviewswith 10community- dwelling adultsage50yearsandolder.Weaskedpar- ticipants to''think aloud''as theycompleted thesur- whether theyunderstood definitionsand instruc- tions andaccepted thehypothetical contextof the survey. Surveyquestions andthe choicetask were refined afterinitial feedbackfrom thefirst 5partici- pants. Usingthe finalversion ofthe survey,we con- firmed thatparticipants sawthe characteristicsand levels asrelevant andof concernand werewilling

On thebasis ofthe InternationalSociety for

Pharmacoeconomics andOutcomes Researchrecom-

mendations forgood researchpractice, weused a standardalgorithm toconstruct afractional-facto- rial experimentaldesign. Thistype ofdesign is advantageous becauseit maximizesthe statistical information obtainablefrom choicetasks usingfewer questions thanrequired ina fullfactorial design(i.e., all combinationsof characteristicsand levelsor cate- gories). 33,34

The finalsurvey consistedof 4versions,

each including9 choicetasks. Participantswere ran- domly assignedto 1of the4 versions.

Each surveywas designedto takeapproximately

20 minutesto complete.In additionto choicetasks,

the surveyincluded awritten introductionto the Table 1Domains, Characteristics,and Levelsfor theChoice Tasks a

DomainCharacteristicLevels

Risk b Chance thatyou willget colorectalcancer 10 outof 100(10%)

25 outof 100(25%)

50 outof 100(50%)

Accuracy Chanceof afalse-negative testresult (thetest resultsays people DO NOTHAVE thegene whenpeople actuallyDO HAVEit) 0 outof 10times (0%)

1 outof 10times (10%)

2 outof 10times (20%)

Privacy Inaddition toyou, whoelse seesthe testresults Your primarycare doctor

Your geneticshealth professionals

Your lifeinsurance andhealth

insurance companies Cost Personalcost toyou notcovered byinsurance $250 $500 $1000 or$1500 c and pretestedusing structuredinterviews ofcommunity members. b. Riskof colorectalcancer giventhe presenceof thegenetic mutation. c. Halfthe participantssaw $1000,and halfthe participantssaw $1500.

GENETIC TESTINGFOR HEREDITARYCOLORECTAL CANCER

ORIGINAL ARTICLE737

at US DEPT OF VETERAN AFFAIRS on March 28, 2016mdm.sagepub.comDownloaded from tions aboutdemographics, personaland familyhis- tory ofcancer, experiencewith genetictesting, and comprehension ofrisk information.

California atSan FranciscoCommittee onHuman

Research Participationand theRTI International

Office ofResearch Protectionand Ethics(Research

TrianglePark,NC). Participantswererequiredto pro-

ment ''Yes,I agreeto participate.''Procedures

Knowledge Networksadministered theWeb-

enabled surveyto membersof theironline survey panel inApril 2010.Using acombined random-digit dial samplefrom theUS landlinepopulation andan address-based samplefrom theUS PostalService National AddressFile, thepanel isa representative, probability-based sampleof theUS population.

Probability-based samplesare preferredto non-

probability basedsamples becauseof theirgreater

Figure 1Choice questionexample.

KNIGHT ANDOTHERS

738

MEDICAL DECISIONMAKING/AUGUST 2015

at US DEPT OF VETERAN AFFAIRS on March 28, 2016mdm.sagepub.comDownloaded from representativeness ofthe USpopulation. 35

Knowl-

edge Networkspanel memberswho donot already have Internetservice areprovided withWeb access.

Participants wereeligible forthe surveyif they

were 50years orolder anda USresident. Ofthe 650 members ofthe KnowledgeNetworks panelasked to participate inthe surveyin April2010, 451were determined tobe eligibleand completedthe survey (70% participationrate). Theconjoint analysis included the355 surveysthat metdata qualitystand- ards neededfor analyticmodel assumptions.Those with incompleteresponses tothe questionon value of riskreduction withcolectomy orcolonoscopy (n= 5)and thosewho selectedthe samegenetic test alternative indiscriminately(test A,test B,no test) for allnine choices( n= 91)were excludedfrom the analysis asis standard. 36

These responsepatterns

may indicatethat thesurvey respondentswere not paying attentionto thechoice questions.Survey pre- testing demonstratedthat participantssimilar to those inthe onlinepanel understoodthe instructions for thechoice questions,making poorcomprehen- sion ofthe choicetask anunlikely explanationfor these responses.However, becausethese responses inflate errorin themodel estimates,it iscommon practice toomit suchparticipants inthe analysis.

Analysis

We estimatedrandom-parameters logitmodels

view, NY)to obtainpreference parametersfrom the choice-format questions. 37,38

This analysisyields rel-

ative preferenceweights forall levelsof eachcharac- teristic includedin thesurvey andestimates ofthe relative importanceof eachcharacteristic overthe range oflevels includedin thesurvey. Theobserved pattern ofanswers tothe choicequestions ina con- joint surveyreveals therelative importanceof each characteristic toparticipants, therate atwhich participants arewilling toaccept tradeoffsamong characteristics andlevels, andthe relativevalue of different combinationsof levels.

The finalmodel includedall ofthe characteristics

shown inTable 1 - colorectalcancer risk,accuracy, privacy, andthe personalcost ofthe testnot covered by insurance.We usedeffects codingin themodel to estimate thereference levelsas thesum ofthe restof the preferenceweights foreach attribute.To estimate the standarderrors, weused thevariance-covariance matrix. Oncewe obtainedthe standarderror, weesti- mated thezscore asthe beta/standarderror andcal- culated thePvalue.We obtainedthe participant-specificvalue of avoiding colorectalcancer riskby combiningthe spe- cific colorectalcancer riskinformation shownto that participant (i.e.,10%, 25%,50%) andthat partici- pant's responseto thequestions aboutwillingness to undergocolectomy orcolonoscopy toeliminate or reducethe riskof havingcolorectal cancer.In the analysis, thebaseline riskvariable wasinteracted with therisk reductionlevel chosenin thequestions about willingnessto undergocolectomy orcolono- scopy. Thisvalue wasthen usedto adjustthe esti- mate forrisk ofa false-negativetest result.This was done toprovide individual-levelinformation onthe value oftest accuracyor avoidingthe consequences of afalse-negative test.

To providea commonmetric forthe valueof

genetic testalternatives, wecalculated themonetary ferencein thevalue ofaparticularg ene tictesting alternativeandtheval ueoftheno-tes tal ternative dividedby thevaluep erdollarof cost. Theoverall valueof genetic testingcompar edwithnotesti ng wascalculated byset tingthetest characteristicsquotesdbs_dbs7.pdfusesText_13