feature
Methods
for
exploring
and
eliciting
patient
preferences
in
the
medical
product
lifecycle:
a
literature
review
VikasSoekhai1,7,8,§,ChiaraWhichello1,7,§,BennettLevitan2,JorienVeldwijk1,7,CathyAnnePinto3,
BasDonkers4,7,IsabelleHuys5,ElinevanOverbeeke5,JuhaeriJuhaeri6and
EstherW.deBekker-Grob1,7,debekker-grob@eshpm.eur.nl
Preference
studies
are
becoming
increasingly
important
within
the
medical
product
decision-making
context.
Currently,
there
is
limited
understanding
of
the
range
of
methods
to
gain
insights
into
patient
preferences.
We
developed
a
compendium
and
taxonomy
of
preference
exploration
(qualitative)
and
elicitation
(quantitative)
methods
by
conducting
a
systematic
literature
review
to
identify
these
methods.
This
review
was
followed
by
analyzing
prior
preference
method
reviews,
to
cross-validate
our
results,
and
consulting
intercontinental
experts,
to
confirm
our
outcomes.
This
resulted
in
the
identification
of
32
unique
preference
methods.
The
developed
compendium
and
taxonomy
can
serve
as
an
important
resource
for
assessing
these
methods
and
helping
to
determine
which
are
most
appropriate
for
different
research
questions
at
varying
points
in
the
medical
product
lifecycle.
Introduction
Thereisanemergingconsensusthatthepatient perspective should be incorporated within decisions inthemedicalproductlifecycle(MPLC;seeGlossary)
[1–4],wherethemedicalproductlifecycleinthis studyisdefinedasthelifecyclesofdrugs,biologics andmedicaldevices.Broadlyencouragingthe in-volvementofpatientshas,therefore,become in-creasinglyimportant[5,6].Takingthepatientvoice intoconsiderationhasnot only becomeincreasingly importantforcompaniesthatdevelopnewmedical productsbutalsofortheauthoritiesthatassess, regulateanddecidewhichproductsareeffective, safe,well-toleratedandcost-effective[7–16].
Toincorporatethepatientvoice,patient preferencesneedtobeexplicitlyexploredor
elicitedthroughrevealed-orstated-preference methods.Inthispaper,preferenceexploration methodsaredefinedasqualitativemethodsthat collectdescriptivedatathroughparticipantor phenomenonobservation,examiningthe sub-jectiveexperiencesanddecisionsmadeby participants.Elicitationmethodsaredefinedas quantitativemethodscollectingquantifiable dataforhypothesistestingandotherstatistical analyses.Whereastheuseof revealed-prefer-encemethodsstillrepresentsamethodological challengeinhealth,manydifferentmethods existtoassessstatedpreferencesofpatients
[17,18].Anup-to-datecompendiumofdifferent stated-preferencemethodstoexploreorelicit patientpreferenceswithintheMPLCismissing.
Therehavebeenfewpublicationsonwhat methodscanbeusedtoassesspatient prefer-encesinascientificway,inthecontextofthe MPLCspecifically.In2001,Ryanetal.[19] pro-videdanoverviewofmethodsknownatthe timeforelicitingpublicpreferencesfor health-care.In2015,theMedicalDeviceInnovation Consortium(MDIC)developedanoverviewof differentpreferenceelicitationmethodsaspart oftheirframeworkonincorporationofpatient preferencesintoregulatoryassessmentsof medicaldevices[20].Althoughboth publica-tionsmadeusefulcontributions,thestudyfrom Ryanetal.[19]doesnotreflectmethods de-velopedsince2001,andthestudyfromthe MDIC[20]didnotincludepreference
explora-Features
PERSPECTIVE
PERSPECTIVE DrugDiscoveryTodayVolume24,Number7July2019
tionmethodsoruseasystematicapproachfor identifyingpreferenceelicitationmethods. Therefore,theaimofourstudywastodevelop anup-to-datecompendiumandtaxonomyof explorationandelicitationpreferencemethods withintheMPLCcontext.Thiswillbean im-portantsteptofurtherdrivetheincorporationof patientpreferencesforward,inadditiontothe studyofvanOverbeekeetal.[6],andin de-velopingguidanceonwhenandhowtoassess patientpreferencesscientificallyinthecontext ofdecision-makingintheMPLC.
Compendiumofpreferencemethods
Asystematicliteraturereviewwasconducted, followedbyananalysisofpriorreviewsbyRyan etal.[19]andfromtheMDIC[20]andexpert consultationswithinternationalpreference experts,toidentifyallpotentialpreference ex-plorationandelicitationmethodswithinthe contextoftheMPLC.Inthispaper,abroad defi-nitionofapreferencemethodwasused:any methodthatenabledustogaininsightintoa patient’srelativedesirabilityoracceptabilityof specifiedalternatives;orchoicesamong treat-mentalternativesoroutcomes;orotherattributes thatdifferamongalternativehealthinterventions
[7].Ultimately,208paperswereanalyzedduring thesystematicliteraturereviewtoidentify pref-erence exploration and elicitation methods within thecontextoftheMPLC.Moreinformationabout theapproachusedinthesystematicliterature
reviewisprovidedinAppendixA.1(see Supple-mentarymaterialonline).Analphabetical over-viewofallreviewedfull-textpapersislistedin AppendixB(seesupplementarymaterialonline).
Weidentified19differentmethods:five ex-plorationmethodsand14elicitationmethods, inthesystematicliteraturereview.Themost frequentlycitedexplorationmethodsincluded focusgroups(n=29,13.9%)and (semi-)struc-turedindividualinterviews(n=47,22.6%), whereasmostcitedelicitationmethodpapers includeddiscretechoiceexperiments(n=57, 27.4%)andthevisualanalogscale(n=12,5.8%). Contingentvaluation(n=11,5.3%),standard gamble(n=11,5.3%)andtimetrade-off(n=11, 5.3%)werealsofrequentlyincludedinthe an-alyzedpapers.Fourstudiesincludedbest–worst scalingtype1,2(n=4,2%).
Throughtheanalysisofthepreference methodreviewsofRyanetal.[19]andtheMDIC
[20],andaftercondensingseveralofthese methods,weidentified23preference explora-tionandelicitationmethods.Thisselection in-cludedninepreferenceexplorationand14 elicitationmethods.Fromthese23preference methods,13methodswerealsoidentifiedinour systematicliteraturereview(56%).Theexpert consultationsconfirmedthemethodsidentified inthesystematicliteraturereviewandinthe analysisofpriorpreferencemethodreviews. Also,consensuswasreachedonincludingfour additionalelicitationmethods.Theexpert
con-sultationsalsoresultedintheexclusionof methodsfocusingonscale-related(e.g.,Likert scales)ordecision-makingframework-related(e. g.,multicriteriadecisionanalysis)techniques, becausethesetechniqueswereregardedas inconsistentwithourdefinitionofapreference method.Asdescribedabove,weidentified19 methodsthroughthesystematicliterature re-view,the23methodsthroughtheanalysisof previouslyconductedreviewsandthefour ad-ditionalmethodsviaexpertconsultations.In total,32uniquepreferencemethodswere identified:tenexplorationand22elicitation methods.Table1 summarizesandbriefly describesthesemethods.
Taxonomyofpreferencemethods
Therearemanywaystogrouppreference methods.Inthisstudy,wegroupedthe identi-fiedmethodsaccordingtotheirmannerofdata collectionandthesimilaritiesintheirmethodof analysis.Thisgroupingwasnotintendedtobea formallexiconbutprimarilyservedasa taxon-omytoorganizeresultsandtodevelopa compendiumofpreferenceexplorationand elicitationmethods.Preferenceexploration methodscanbegroupedaccordingtothe numberofparticipantsthemethodutilizesin onesession(Fig.1).(Semi-)structuredindividual interviews,in-depthinterviewsandcomplaints proceduresuseinterviewswithoneparticipant (n=1)inasinglesettingorsession.TheDelphi
TABLE1
Overviewofidentifiedmethods
Method Description Refs
Explorationmethods
Citizens’juriesb Groupofindividualsdiscussingissuesonthebasisofevidenceprovidedbytwotrainedmoderators [24,25]
Complaints
proceduresb
Methodinwhichstakeholderscanregistercomplaintstobeinvestigatedbyexperts [26,27]
Conceptmappingb Methodthatutilizessmallgroupsofparticipantsrespondingtovarioustopicsorissues,whileensuringeach
respondentisgivenequalopportunitytoexpresstheiropinionsandaddressothergroupdynamicissues
[28,29]
Delphimethoda,b Structured,iterativeforecastingmethodinvolvingapanelofexpertswhoprovideanonymousresponsesto
questionnaireswiththeopportunitytorevisetheirresponseswhentheanonymoussummaryofresponse
fromthepriorroundisrevealed
[30,31]
Dyadicinterviewa,b Methodthatutilizestwoparticipantsinasingleinterview,respondingtoopen-endedquestionsaskedbyan
interviewertoidentifyhowaproduct,serviceoropportunityisperceived
[32,33]
Focusgroupa,b Methodthatutilizesagroupofinteractingindividualsthatprovideinformationaboutaspecificissueto
identifyhowaproduct,serviceoropportunityisperceived
[34,35]
Indepth–individual
interviewa,b
Interviewtechniquethatallowsforanintensivediscussionwithoneintervieweetoexploretheirperspectives
onaparticulartopicortheme,togainadeeperunderstandingofthisparticulartopicortheme.Oftenonlya
limitedamountofquestionsorthemesarepreparedbytheinterviewer,andtherestofthequestionsare
basedontheresponseoftheinterviewee
[36,37]
Nominalgroup
techniqueb
Methodthatutilizesagroupprocessthatinvolvesmakingdecisionsbyvoteandrankingresponsesgivenby
membersofthegroup
[38,39]
Publicmeetingsb Methodtogainpublicopinionsonparticularissuesbyallowinggeneralmembersofthepublictoattendand
voicetheirresponses
[40,41]
(Semi-)structured
individualinterviewa,b
Interviewtechniquethatallowsnewideastobebroughtupduringtheinterviewasaresultofwhatthe
intervieweesaysinasemi-structuredsetting,whereasinthestructuredsettingtheinterviewerstrictlysticks
toaninterviewguideanddoesnotaskquestionsbasedontheresponseoftheinterviewee
[42,43]
Features
PERSPECTIVE DrugDiscoveryTodayVolume24,Number7July2019
TABLE1(Continued)
Method Description Refs
Elicitationmethods
Adaptiveconjoint
analysisa
Methodsimilartoregularconjointanalysis,butwithadaptiveconjointchoicetasksbasedontheearlier
choicesmadewithinthesurvey,intheoryallowingthesurveytofocusattentiononthoseattributesorlevels
ofthoseattributesthathavethemostinfluenceonthechoicesofthatindividual.Unlikediscretechoice
experimentsthismethodisfoundedinthetheoryofconjointmeasurement(CM),whichismorefocusedon
thebehaviorofnumbersystemsinsteadofthebehaviorofhumanpreferences
[44,45,81]
Allocationofpointsb Methodthatinvolvesaskingrespondentstoratetheirconditionsonscales,whileknowingtheweightswhich
theyattachtodifferentcriteria,indicatingtherelativeimportanceofparticularareasoftheirlives
[46,47]
Analytichierarchy
processa,b
Methodinwhichrespondersassesstherelativeimportanceofpairsofattributes(treatmentendpoints,
properties,criteria,items,objects,etc.)towardachievingagoal,wheretheseresponsesareusedtocompute
aweightforeachattribute
[20,48]
Best–worstscaling
(types1,2,3)a,b
Involvesrespondentsansweringsurveysthatincludelistsofattributesorprofilesandbeingaskedtoindicate
thebest(ormostappealing/important)andtheworst(orleastappealing/important)ofthem.Thismethod
consistsofthreetypes:intype1asetofattributesisshowedthatmightnotreflectthecharacteristicsofany
particulartreatment,ofwhichtherespondentpicksthebestandworst.Type2involvesasituationinwhich
theattributescollectivelycharacterizeaparticularprofileandtherespondentchoosesthebestandworst.In
type3threeormoreprofilesareshownandtherespondentselectsthebestandworstprofiles
[20,49,50]
Constantsumscalingc Constantsumscalingconsistsofacomparativescalewhererespondentsareaskedtoallocateafixedamount
(orconstantsum)ofpoints,dollarsoranythingamongasetofobjectsaccordingtoacriterion
[51,52]
Contingentvaluationa,b Methodtodeterminethewillingnesstopay(WTP),whereindividualsarepresentedwithachoicebetween
nothavingthecommodityvaluedandhavingthecommoditybutforgoingacertainamountofmoney.The
moneybeingthattheyarewillingtoforgotohavethecommodityistheirWTPforthatcommodity.WTPcan
becalculateddirectlyusingathresholdorindirectlyusingadiscretechoiceexperimentforexample
[53,54]
Controlpreference
scalea
Thecontrolpreferencesscale(CPS)isamethodtodeterminethedegreeofcontrolapatientwantsregarding
medicaltreatment.Thepreferenceordersareanalyzedusingunfoldingtheorytodeterminethedistribution
ofpreferencesindifferentpopulationsandtheeffectofcovariatesonconsumerpreferences
[55–57]
Discretechoice
experimenta,b
Methodthatutilizesanattribute-basedmeasureofbenefit,duringwhichindividualsareofferedaseriesof
hypotheticalchoicesituations(i.e.,choicesets),fromwhichtheyareaskedtochoosebetweentwoormore
profiles.Therearenumerousvariantsofdiscretechoiceexperiments.Incontrasttoconjointanalysis,this
methodreliesonatheoryofthebehaviorofhumanpreferences[forexamplerandomutilitytheory(RUM)]
[58,59,60,81]
Measureofvalueb Methodusedtoidentifytheoptimalbundleofservicestobeprovidedgivenresourceconstraints.Individuals
areaskedtoallocateafixedamountofresourcesbetweendifferentservices.Theseallocationsareanalyzed
toidentifythetrade-offsindividualsmake
[61]
Outcomeprioritization
toola
Instrumentthatallowsparticipantstoprioritizeoutcomesmakinguseofaspecifictoolaccordingtothe
‘trade-off’principle,implyingthattheyarewillingtocompromiseonthelessimportantoutcomes
[62]
Persontrade-offa,b Anextensionofthetimetrade-off.Withpersontrade-offanindividualevaluatesthehealtheffectsof
interventionsusingpersons(insteadoftime)astheequilibratingmechanism
[63,64]
(Probabilistic)
thresholdtechniquea,b
Methodthatdeterminesthemaximalchangeinoneattributerespondentsarewillingtoaccepttoachievea
givenchangeinanotherattribute
[20,65]
Q-methodologyc Methodthatusesaspeciallydesignedresponsegridtopresentrespondentswithasetofstatementsand
askingthemtoorder,usuallybasedontheextenttowhichtheyagreewiththem
[66,67]
Qualitative
discriminantprocessb
Methodthatinvolvesascoringandrankingprocessbasedondecisionanalysistechnique,involvingthe
definitionofoptionsintermsofqualitativecategories,thenderivinganumericpointestimateandfinally
solvingamaximizationproblemwithgivenconstraints
[68]
Repertorygrid
methoda
Methodusedforelicitingpersonalconstructs(i.e.,whatpeoplethinkaboutagiventopic).Toidentify
preferencesoverlappingandratingtechniquesareused
[69,70]
Self-explicated
conjointc
Methodthatasksexplicitlyaboutthepreferenceforeachattributeratherthanthepreferenceofseveral [71]
Standardgamblea,b Methodinwhichrespondentsareaskedtochoosebetweenacertainoutcomeandagamblethatmight
resultineitherabetteroutcomewithaprobabilityPoraworseoutcomethantheoriginalwithaprobability
1-P
[72,73]
Startingknown
efficacya
Methodsimilarto(probabilistic)thresholdtechniques,butwithaspecificknownstartingpoint.Thismethod
isspecificallyusedwithinthecontextofthemedicalproductlifecycle
[74]
Swingweightingb Methodforsettingtheweightsinwhichadecision-relevantrangeisspecifiedforeachattribute,andthe
impactof‘swinging’theattributethroughthatentirerangeofvaluesisassignedaweightrelativetothe
impactofswingingtheattributewiththelargestweight
[19,20]
Testtrade-offc Methodthatcanberegardedasanextensionofthetimetrade-offthatisspecificallyusedtoevaluateanew
biomarkerbyusingrisks(insteadoftime)astheequilibratingmechanism
[75,76]
Timetrade-offa,b Methodthatpresentsindividualswithachoicebetweenlivingforaperiodinaspecified,butlessthan
perfect,stateversushavingahealthierlifeforaperiodoftime,wheretimeisvarieduntiltherespondentis
indifferenttothealternatives
[20,77,78]
Visualanalogscalea,b Aself-reportinginstrumentconsistingofalineofpredeterminedlengththatseparatesextremeboundaries
ofthephenomenonbeingmeasured
[79,80]
Intotal32uniquemethodswereidentified.
a
Identifiedinsystematicreview(19methods).
b
Identifiedthroughanalysisofpreviouspreferencemethodreviews(23methods).
c
Identifiedwithexpertconsultations(4methods).
Features
Preference
exploration
Individual methods
Group methods
Individual/group
methods
In-depth individual interview (Semi-)structured
individual interview Delphi method Public meetings
Concept mapping
Focus group
Dyadic interview
Nominal group technique
Complaints procedures Citizens’ juries
Drug Discovery Today
FIGURE1
Groupingofpreferenceexploration(qualitative)methodsintothreegroups:individual,groupandindividual/groupmethods.
Discrete choice experiment/best-worst scaling type 3 Best-worst scaling Type 1,2 Adaptive conjoint analysis Self-explicated conjoint Analytic hierarchy process Measure of value Allocation of points
Starting known efficacy Time trade-off Standard gamble
Person trade-off
Visual analog scale Constant sum scaling
Qualitative discriminant process
Repertory grid method
(Probabilistic) threshold technique Swing weighting Outcome prioritization tool Contingent valuation Control preferences scale Q-methodology Test trade-off
Preference
elicitation
Discrete-choice-based methods Ranking methods Indifferencemethods Rating methods
Drug Discovery Today
FIGURE2
Groupingofpreferenceelicitation(quantitative)methodsintofourgroups:discretechoicebased,ranking,indifferenceandratingmethods.
Features
method,focusgroups,dyadicinterviews,public meetings,nominalgrouptechniqueandcitizen juriestypicallydirectquestionstomorethan oneparticipant(n>1)inasinglesetting. Con-ceptmappingcanemployeitherindividualor groupsettingsfordatacollection(n1).
Preferenceelicitationmethodscanbe groupedintofourdistinctgroups(Fig.2),with methodsfromlefttorightbeingabletoanswer asmallersubsetofresearchquestions[adiscrete choiceexperiment(DCE)isforexampleableto providewillingness-to-pay(WTP)information andprobabilityscoreswhereascontingent val-uationprovidesWTPinformationonly].First, discretechoice-basedmethodstypically exam-inetheimportanceoftrade-offsbetween attri-butesandtheiralternativesthroughaseriesof
choicesetsthatpresent(hypothetical) alterna-tives.Second,ranking(orrelated)methodswere classifiedbasedontheuseofrankingexercises tocapturetheorderofalternativesorattributes withinapresentedset.Third,indifference techniquesaremethodsthatvarythevalueof oneattributeinoneofthealternativesuntilthe participantisindifferent,orhasnopreference, betweenalternatives.Finally,rating(orrelated) methodsaremethodsbasedontheirutilization ofcomparativeratingapproaches,often allow-ingparticipantstoexpressthestrengthoftheir preferencesalongalabeledscale.
Trendsintheuseofpreferencemethods
Withthesystematicliteraturereview,spanning 37yearsofliterature,weobservedanoverall
upwardstrendinthenumberofMPLCpatient preferencestudiesperyear.Themeannumber ofpreferencestudiesincreasedfrom1.1peryear to6.5peryearto20.3peryear.Thisisforthe periods1980–2000,2001–2010and2011–2016, respectively(AppendixC,seesupplementary materialonline).Wealsoobservedthatour includedpapersoriginatedfromalloverthe world,coveringfivedifferentcontinents(Table2
).Themajority(73%)ofpaperswerefromNorth America(n=90)andEurope(n=62).
Analyzingtheseparateuseofpreference explorationandelicitationmethodsovertime, weobservedatrendofpreferenceexploration methodsbeingusedmorefrequentlyinrecent years.Wedidnotconsidertheperiod1980– 2005becausethisperiodonlyincludedafew
PERSPECTIVE DrugDiscoveryTodayVolume24,Number7July2019
TABLE2
Backgroundinformationofidentifiedpatientpreferencemethodsinthesystematicreviewfocusingonthemedicalproductlifecycle
Method Frequency Continentsoforigin Studynumbers
n=19 n=208a (%) Continents(frequency)a n=208
Explorationmethods
Delphimethod 3 (1.4) Asia(2),NorthAmerica(1) 24,107,308
Dyadicinterview 1 (0.5) Africa(1) 269
Focusgroup 29 (13.9) Africa(1),Asia(2),Australia/Oceania
(3),Europe(15),NorthAmerica(8)
2,14,17,18,43,45,71,72,84,97,109,116,119,121, 211,220,222,236,253,269,282,283,286,290,294, 300,308,313,317 Indepth–individual interview 9 (4.3) Asia(1),Australia/Oceania(1),
Europe(3),NorthAmerica(4)
32,41,108,147,173,191,193,211,316
(Semi-)structured
individualinterview
47 (22.6) Africa(2),Asia(6),Australia/Oceania
(6),Europe(18),NorthAmerica(15)
2,9,17,18,21,30,41,43,57,58,65,67,87,94,100,101, 120,129,141,153,162,164,184,193,198,205,211, 215,217,222,226,229,230,232,239,267,268,269, 272,280,284,285,286,302,306,310,323 Elicitationmethods Adaptiveconjoint analysis 3 (1.4) NorthAmerica(3) 88,89,243 Analytichierarchy process 1 (0.5) Europe(1) 221
Best-worstscaling(types
1,2,3)
4 (1.9) Asia(1),Australia/Oceania(1),North
America(2)
133,180,189,300
Contingentvaluation 11 (5.3) Asia(2),Australia/Oceania(1),North
America(2)
29,35,144,148,155,166,167,180,199,244,298
Controlpreferencescale 3 (1.4) Asia(1),NorthAmerica(2) 147,175,316
Discretechoice
experiment
57 (27.4) Africa(1),Asia(7),Australia/Oceania
(6),Europe(15),NorthAmerica(28)
19,25,26,34,42,48,57,66,73,79,80,90,100,101, 109,114,117,119,122,133,134,154,155,160,161, 163,166,179,180,184,192,194,200,212,213,215, 218,219,222,227,229,234,238,239,243,246,247, 249,257,264,266,272,281,309,311,312,313 Outcomeprioritization tool 1 (0.5) Europe(1) 304
Persontrade-off 1 (0.5) Europe(1) 274
Repertorygridmethod 1 (0.5) Europe(1) 255
Standardgamble 11 (5.3) Asia(1),Australia/Oceania(1),
Europe(2),NorthAmerica(7)
34,42,155,180,195,200,209,219,237,277,312
Startingknownefficacy 1 (0.5) NorthAmerica(1) 201
(Probabilistic)threshold
technique
2 (1.0) NorthAmerica(2) 42,172
Timetrade-off 11 (5.3) Australia/Oceania(1),Europe(2),
NorthAmerica(8)
33,34,78,155,180,200,209,219,237,277,318
Visualanalogscale 12 (5.8) Asia(2),Europe(3),NorthAmerica
(7)
93,115,168,171,178,195,208,223,278,281,287,314
a
Includedcountriespercontinent:Africa–Kenya,SouthAfrica;Asia–China,Iran,Japan,Malaysia,Singapore,SouthKorea,Taiwan,Thailand,Turkey;Australia/Oceania–Australia;Europe
–France,Germany,UK,Hungary,TheNetherlands,Norway,Spain;NorthAmerica–Canada,USA.
Features
datapointstocomputerepresentative percen-tages.Fortheperiod2002–2006,33.3%ofthe papersusedapreferenceexplorationmethodto gaininsightsintopatientpreferences (com-putedasthefrequencyofanexplorationor elicitationmethodineachindividualpaper). Thisincreasedto48.8%intheperiod2007–2011 andto45.8%for2012–2016.Amongpreference explorationmethods,theproportionofstudies thatusedfocusgroupsincreasedfrom23%in theperiod2002–2006to35%intheperiod 2012–2016.Theproportionof(semi-)structured individualinterviewsremainedmoreorless constantwith55%intheperiod2002–2006and 52%intheperiod2012–2016,whereasin-depth individualinterviewsdecreasedfrom23%in 2002–2006to8%in2012–2016.Overtime,we alsoobservedmorediversitywithinthegroupof preferenceexplorationmethods.TheDelphi methodanddyadicinterviewsbeganappearing in2007.
Amongpreferenceelicitationmethods,we observedthatthenumberofpapersthatmade useofadiscretechoiceexperimentincreased from38%in2002–2006to58%in2012–2016. Papersthatincludedavisualanalogscale de-creasedfrom16%to3%,andcontingent valu-ationshowedasimilartrend(17%to9%). Standardgambleandtimetrade-offshowedan upwardtrend,from5%and4%in2002–2006to 9%and6%in2012–2016,respectively.Overall, weobservedthat,overtime,amorediverse groupofpreferenceelicitationmethodswas used.
Comparisonofsources
Theresultsofthisstudywerepartlyinlinewith theresultsfoundbyRyanetal.(2001)andthe MDIC(2015)[19,20].Fifty-sixpercent(13outof 23)ofmethodsreportedbyRyanetal.[19]and/ ortheMDIC[20]wereidentifiedinour sys-tematicliteraturereview.Thedifferencesarise because:(i)thesearchinthisstudyfocused specificallyonmethodstoobtainpatient pre-ferencesfordrugsandmedicaldevices,whereas Ryanetal.[19]focusedonpublicviewsonthe provisionofhealthcare;(ii)MDIC[20]excluded preferenceexplorationmethods;and(iii)the MDIC[20]effortdidnotuseasystematic ap-proachforidentifyingmethods.Thetaxonomy ofpreferencemethodsproposedinthisstudyis alsoinlinewithresultsfromMt-Isaetal.[21], Zhangetal.[22]andGonzalezetal.[23],in whichelicitationmethodsweregroupedby rating,rankingandtrade-off(whichincluded choice-basedmethods)techniques,although manyotherwaystogroupthesemethodsare possible.
Resultsfromourstudy’ssystematicliterature review(19preferencemethodsidentified) showedthatmostreviewedpapersusedfocus groups,(semi-)structuredindividualinterviews, discretechoiceexperimentsorthevisualanalog scaletogaininsightsintopatientpreferences. MostofthesestudieswereconductedinNorth AmericaorEurope.Wealsoshowedthatthe meannumberofpatientpreferencestudiesfor drugsandmedicaldevicesincreasedovertime. Furthermore,thisstudyshowedthat,for pref-erenceexplorationandelicitationmethods,a morediversemixofmethods(explorationand elicitationmethods)wasusedovertimeto exploreorelicitpreferences.
Concludingremarks
Inthisstudywedevelopedanup-to-date compendiumandtaxonomyofpreference ex-plorationandelicitationmethodsinthecontext oftheMPLC.Thesystematicreview(19 meth-ods),analysisofpriorconductedpreference methodreviews(23methods)andexpert con-sultations(fourmethods)contributedtothis compendium.Intotal,32uniquemethodswere identified.Preferenceexplorationmethodswere groupedinthreemaingroups,whereasthe preferenceelicitationmethodsweregroupedin fourmaingroups.Becausechoosingwhich methodtousewilldependontheMPLCphase andwhatthemeasuredpreferencesarebeing usedfor,futureresearchmightfocuson deter-miningwhichmethodsaremostappropriateto exploreorelicitpatientpreferences,andunder whatcircumstances,throughoutthedifferent phasesintheMPLC.Inaddition,itmightbeof interestforfutureresearchtofocusonthe specificcombinationsofpreferenceexploration andelicitationmethodsusedinmixed-method studies,andthereasoningbehindsuchstudy designs.
Acknowledgments
TheauthorswouldliketothankJ.Bridges(Johns HopkinsBloombergSchoolofPublicHealth),B. Craig(InternationalAcademyofHealth PreferenceResearch),J.M.Gonzalez(Duke University),T.Hammad(EMDSerono),B.Hauber (RTIHealthSolutions),R.Hermann
(AstraZeneca),R.Johnson(DukeUniversity),U. Kihlbom(UppsalaUniversity),A.Mohamed (Bayer),V.Patadia(Sanofi),S.Russo(European InstituteofOncology)andA.Stiggelbout (LeidenUniversityMedicalCenter)fortheir valuableinputduringthewritingofthispaper. ThisworkhasreceivedsupportfromtheEU/ EFPIAInnovativeMedicinesInitiative[2]Joint UndertakingPREFERgrantno.115966.Thistext
anditscontentsreflecttheauthors’andthe PREFERproject’sviewandnottheviewofIMI, theEuropeanUnionorEFPIA.
AppendicesA–C. Supplementarydata
Supplementarymaterialrelatedtothisarticle canbefound,intheonlineversion,atdoi:
https://doi.org/10.1016/j.drudis.2019.05.001.
References
1 Hoos,A.etal.(2015)Partneringwithpatientsinthe
developmentandlifecycleofmedicines:acallfor
action.Ther.Innov.Regul.Sci49,929–939
2 Anderson,M.andMcCleary,K.(2016)Onthepathtoa
scienceofpatientinput.Sci.Transl.Med.8,336ps11
3 Smith,M.Y.etal.(2016)Patientengagementata
tippingpoint–theneedforculturalchangeacross
patient,sponsor,andregulatorstakeholders:
insightsrromtheDIAconference,``Patient
EngagementinBenefitRiskAssessmentThroughout
theLifeCycleofMedicalProducts”.Ther.Innov.Reg.
Sci.50,546–553
4 deBekker-Grob,E.W.etal.(2017)Givingpatients’
preferencesavoiceinmedicaltreatmentlifecycle:The
PREFERPublic–PrivateProject.Patient10,263–266
5 Anderson,R.M.andFunnell,M.M.(2005)Patient
empowerment:reflectionsonthechallengeof
fosteringtheadoptionofanewparadigm.Patient
Educ.Counsel.57,153–157
6 vanOverbeeke,E.etal.(2019)Factorsandsituations
influencinguseofpatientpreferencestudiesalongthe
medicalproductlifecycle:aliteraturereview.Drug
Discov.Today24,57–68
7 FDA(2016)GuidanceforIndustry,FoodandDrug
AdministrationStaffandOtherStakeholders.Patient
PreferenceInformation–VoluntarySubmission,Review
inPremarketApprovalApplications,Humanitarian
DeviceExemptionApplications,andDeNovoRequests,
andInclusioninDecisionSummariesandDevice
Labeling.PublicReport2016.2016
8 FDA(2017)PublicWorkshopOnPatient-FocusedDrug
Development:Guidance1CollectingComprehensiveAnd
RepresentativeInput.PublicReport2017a.2017
9 FDA(2017)TheVoiceofthePatient:ASeriesofReports
fromFDA’sPatient-FocusedDrugDevelopmentInitiative.
PublicReport2017b.2017
10 FDA(2017)PDUFAReauthorizationPerformanceGoals
AndProceduresFiscalYears2018Through2022.Public
Report2017c.2017
11 Kievit,W.etal.(2017)Takingpatientheterogeneityand
preferencesintoaccountinhealthtechnology
assessments.Int.J.Technol.Assess.HealthCare33,562–
569
12 Abelson,J.etal.(2016)Publicandpatientinvolvement
inhealthtechnologyassessment:aframeworkfor
action.Int.J.Technol.Assess.HealthCare32,256–264
13 Lowe,M.M.etal.(2016)Increasingpatientinvolvement
indrugdevelopment.ValueHealth19,869–878
14 Stewart,K.D.etal.(2016)Preferenceforpharmaceutical
formulationandtreatmentprocessattributes.Patient
PreferAdherence10,1385–1399
15 Minion,L.E.etal.(2016)Endpointsinclinicaltrials:what
dopatientsconsiderimportant?Asurveyofthe
ovariancancernationalalliance.Gynecol.Oncol.140,
193–198
16 Bloom,D.etal.(2018)Therulesofengagement:CTTI
recommendationsforsuccessfulcollaborations
Features
betweensponsorsandpatientgroupsaroundclinical
trials.Ther.Innov.Regul.Sci52,206–213
17Beshears,J.etal.(2008)Howarepreferencesrevealed?
J.PublicEcon.92,1787–1794
18Lambooij,M.S.etal.(2015)Consistencybetweenstated
andrevealedpreferences:adiscretechoiceexperiment
andabehaviouralexperimentonvaccination
behaviourcompared.BMCMed.Res.Methodol.15,19
19Ryan,M.etal.(2001)Elicitingpublicpreferencesfor
healthcare:asystematicreviewoftechniques.Health
Technol.Assess.5,1–186
20MedicalDeviceInnovationConsortium(MDIC)(2015)
PatientCenteredBenefit–RiskProjectReport:AFramework
forIncorporatingInformationonPatientPreferences
RegardingBenefitandRiskintoRegulatoryAssessmentsof
NewMedicalTechnology.PublicReport2015.2015
21Mt-Isa,S.etal.(2013)IMI-PROTECTBenefit–RiskGroup
RECOMMENDATIONSREPORT:Recommendationsforthe
MethodologyandVisualisationTechniquestobeUsedin
theAssessmentofBenefitandRiskofMedicines.IMI
PROTECTRecommendationsReport2013.2013
22Zhang,W.etal.(2015)Whendrugdiscoverymeetsweb
search:learningtorankforligand-basedvirtual
screening.J.Cheminform.7,5
23Gonzalez,J.M.etal.(2017)Comparingpreferencesfor
outcomesofpsoriasistreatmentsamongpatientsand
dermatologistsintheU.K.:resultsfroma
discrete-choiceexperiment.Br.J.Dermatol.176,777–785
24Mitton,C.etal.(2009)Publicparticipationinhealth
careprioritysetting:ascopingreview.HealthPolicy91,
219–228
25Street,J.etal.(2014)Theuseofcitizens’juriesinhealth
policydecision-making:asystematicreview.Soc.Sci.
Med.109.http://dx.doi.org/10.1016/j.
socscimed.2014.03.005
26Bourne,T.etal.(2015)Theimpactofcomplaints
proceduresonthewelfare,healthandclinicalpractise
of7926doctorsintheUK:across-sectionalsurvey.BMJ
Open2015,5
27Wensing,M.andGrol,R.(1998)Whatcanpatientsdoto
improvehealthcare?HealthExpect.1,37–49
28Burke,J.G.etal.(2005)Anintroductiontoconcept
mappingasaparticipatorypublichealthresearch
method.Qual.HealthRes.15,1392–1410
29Trochim,W.andKane,M.(2005)Conceptmapping:an
introductiontostructuredconceptualizationinhealth
care.Int.J.QualityHealthCare17,187–191
30Boulkedid,R.etal.(2005)Usingandreportingthe
Delphimethodforselectinghealthcarequality
indicators:asystematicreview.PLoSOne6,e20476
31deMeyrick,J.(2003)TheDelphimethodandhealth
research.HealthEducation103,7–16
32Eisikovits,Z.andKoren,C.(2010)Approachestoand
outcomesofdyadicinterviewanalysis.Qual.Health
Res.20,1642–1655
33Morgan,D.L.etal.(2013)Introducingdyadicinterviews
asamethodforcollectingqualitativedata.Qual.Health
Res.23,1276–1284
34Basch,C.E.(1987)Focusgroupinterview:an
underutilizedresearchtechniqueforimprovingtheory
andpracticeinhealtheducation.HealthEducation
Quarterly14,411–448
35Schulze,B.andAngermeyer,M.C.(2003)Subjective
experiencesofstigma.Afocusgroupstudyof
schizophrenicpatients,theirrelativesandmental
healthprofessionals.Soc.Sci.Med.56,299–312
36Harris,M.A.etal.(2000)Validationofastructured
interviewfortheassessmentofdiabetes
self-management.Diab.Care23,1301
37Williams,J.B.(1988)Astructuredinterviewguidefor
theHamiltonDepressionRatingScale.Arch.Gen.
Psychiatry45,742–747
38Allen,J.etal.(2004)Buildingconsensusinhealthcare:
aguidetousingthenominalgrouptechnique.Br.J.
Commun.Nurs.9,110–114
39Gallagher,M.etal.(1993)Thenominalgroup
technique:aresearchtoolforgeneralpractice?Fam.
Pract.10,76–81
40Ham,C.(1997)Prioritysettinginhealthcare:learning
frominternationalexperience.HealthPolicy42,49–66
41McComas,K.A.(2001)Theoryandpracticeofpublic
meetings.Commun.Theor.11,36–55
42Barriball,L.K.andWhile,A.(1994)Collectingdatausing
asemi-structuredinterview:adiscussionpaper.J.Adv.
Nurs.19,328–335
43Whiting,L.S.(2008)Semi-structuredinterviews:
guidancefornoviceresearchers.Nurs.Standard22,
35–40
44Beusterien,K.M.etal.(2005)Understandingpatient
preferencesforHIVmedicationsusingadaptive
conjointanalysis:feasibilityassessment.ValueHealth8,
453–461
45Fraenkel,L.etal.(2001)Understandingpatient
preferencesforthetreatmentoflupusnephritiswith
adaptiveconjointanalysis.Med.Care39,1203–1216
46Haywood,K.L.etal.(2003)Patientcenteredassessment
ofankylosingspondylitis-specifichealthrelatedquality
oflife:evaluationofthePatientGeneratedIndex.J.
Rheumatol30,764
47Schwappach,D.L.B.andStrasmann,T.J.(2006)Quick
anddirtynumbers?Thereliabilityofa
stated-preferencetechniqueforthemeasurementof
preferencesforresourceallocation.J.HealthEcon.25,
432–448
48Liberatore,M.J.andNydick,R.L.(2006)Theanalytic
hierarchyprocessinmedicalandhealthcare
decision-making:aliteraturereview.Eur.J.Operation.Res.189,
194–207
49Flynn,T.N.(2010)Valuingcitizenandpatient
preferencesinhealth:recentdevelopmentsinthree
typesofbest–worstscaling.ExpertRev.Pharmacoecon.
Outcome.Res.10,259–267
50Flynn,T.N.etal.(2007)Best–worstscaling:whatitcan
doforhealthcareresearchandhowtodoit.J.Health
Econ.26,171–189
51Mai,R.andHoffmann,S.(2012)Tasteloversversus
nutritionfactseekers:howhealthconsciousness
andself-efficacydeterminethewayconsumers
choosefoodproducts.J.ConsumerBehaviour11,
316–328
52Skedgel,C.D.etal.(2015)Choosingvs.allocating:
discretechoiceexperimentsandconstant-sumpaired
comparisonsfortheelicitationofsocietalpreferences.
HealthExpect.18,1227–1240
53Bärnighausen,T.etal.(2007)Willingnesstopayfor
socialhealthinsuranceamonginformalsectorworkers
inWuhan,China:acontingentvaluationstudy.BMC
HealthService.Res.7,114
54Cunningham,S.J.andHunt,N.P.(2000)Relationship
betweenutilityvaluesandwillingnesstopayin
patientsundergoingorthognathictreatment.
Commun.DentalHealth17,92–96
55Degner,L.F.etal.(1997)TheControlPreferencesScale.
Can.J.Nurs.Res.29,21–43
56McPherson,C.J.etal.(2001)Effectivemethodsof
givinginformationincancer:asystematicliterature
reviewofrandomizedcontrolledtrials.J.PublicHealth
23,227–234
57Henrikson,N.B.etal.(2011)Measuringdecisional
controlpreferencesinmennewlydiagnosedwith
prostatecancer.J.Psychosoc.Oncol.29,606–618
58Bridges,J.F.P.etal.(2011)Conjointanalysisapplications
inhealth–achecklist:areportoftheISPORGood
ResearchPracticesforConjointAnalysisTaskForce.
ValueHealth14,403–413
59Soekhai,V.etal.(2019)Discretechoiceexperimentsin
healtheconomics:past,presentandfuture.
PharmacoEconomics37,201–226
60Lancsar,E.andLouviere,J.(2008)Conductingdiscrete
choiceexperimentstoinformhealthcare
decision-making.PharmacoEconomics26,661–677
61Weinstein,M.C.etal.(2003)Principlesofgoodpractice
fordecisionanalyticmodelinginhealth-care
evaluation:reportoftheISPORTaskForceonGood
ResearchPractices–modelingstudies.ValueHealth6,
9–17
62Case,S.M.etal.(2015)Olderadults’recognitionof
trade-offsinhealthcaredecision-making.J.Am.Geriatr.
Soc.63,1658–1662
63Green,C.(2001)Onthesocietalvalueofhealthcare:
whatdoweknowaboutthepersontrade-off
technique?HealthEcon.10,233–243
64Nord,E.(1995)Theperson-trade-offapproachto
valuinghealthcareprograms.Med.DecisionMaking15,
201–208
65Kopec,J.A.etal.(2007)Probabilisticthreshold
techniqueshowedthatpatients’preferencesfor
specifictrade-offsbetweenpainreliefandeachside
effectoftreatmentinosteoarthritisvaried.J.Clin.
Epidemiol.60,929–938
66Cross,R.M.(2005)Exploringattitudes:thecaseforQ
methodology.HealthEducationRes.20,206–213
67vanExel,J.etal.(2007)Careforabreak?An
investigationofinformalcaregivers’attitudestoward
respitecareusingQ-methodology.HealthPolicy83,
332–342
68Mullen,P.M.(1999)Publicinvolvementinhealthcare
prioritysetting:anoverviewofmethodsforeliciting
values.HealthExpect.2,222–234
69Davis,C.J.etal.(2006)Communicationchallengesin
requirementselicitationandtheuseoftherepertory
gridtechnique.J.Comput.Inform.Syst.46,78–86
70Rowe,G.etal.(2005)Assessingpatients’preferences
fortreatmentsforanginausingamodifiedrepertory
gridmethod.Soc.Sci.Med.60,2585–2595
71Riquelme,H.andRickards,T.(1992)Hybridconjoint
analysis:anestimationprobeinnewventuredecisions.
J.BusinessVenturing7,505–518
72Gafni,A.(1994)Thestandardgamblemethod:whatis
beingmeasuredandhowitisinterpreted.Health
ServicesRes.29,207–224
73Morimoto,T.andFukui,T.(2002)Utilitiesmeasuredby
ratingscale,timetrade-off,andstandardgamble:
reviewandreferenceforhealthcareprofessionals.J.
Epidemiol.12,160–178
74Man-Son-Hing,M.etal.(1996)Warfarinforatrial
fibrillation.Thepatient’sperspective.Arch.Intern.Med
156,1841–1848
75Baker,S.G.andKramer,B.S.(2014)Evaluatingsurrogate
endpoints,prognosticmarkers,andpredictivemarkers:
somesimplethemes.Clin.Trials12,299–308
76Baker,S.G.etal.(2012)Evaluatinganewmarkerforrisk
predictionusingthetesttradeoff:anupdate.Int.J.
Biostat.8,1–17
77Arnesen,T.andTrommald,M.(2005)AreQALYsbased
ontimetrade-offcomparable?—Asystematicreview
ofTTOmethodologies.HealthEcon.14,39–53
PERSPECTIVE DrugDiscoveryTodayVolume24,Number7July2019
Features
78Brazier,J.etal.(1999)Areviewoftheuseofhealth
statusmeasuresineconomicevaluation.J.Health
ServicesRes.Policy4,174–184
79Holdgate,A.etal.(2003)Comparisonofaverbal
numericratingscalewiththevisualanaloguescale
forthemeasurementofacutepain.Emerg.Med.15,
441–446
80Räsänen,P.etal.(2006)Useofquality-adjustedlife
yearsfortheestimationofeffectivenessofhealthcare:
asystematicliteraturereview.Int.J.Technol.Assess.
HealthCare22,235–241
81Louviere,J.etal.(2010)Discretechoiceexperiments
arenotconjointanalysis.J.ChoiceModel.3,57–72
VikasSoekhai1,7,8,§
ChiaraWhichello1,7,§
BennettLevitan2 JorienVeldwijk1,7
CathyAnnePinto3
BasDonkers4,7 IsabelleHuys5 ElinevanOverbeeke5
JuhaeriJuhaeri6
EstherW.deBekker-Grob1,7,*
1
ErasmusSchoolofHealthPolicy&Management, Erasmus University Rotterdam, P.O. Box 1738, Rotterdam,3000DRtheNetherlands
2
JanssenResearch&Development,1125 Trenton-Harbourton Road, PO Box 200, Titusville, NJ, 08560,USA
3Merck,Sharpe&Dome,2000GallopingHillRd.,
Kenilworth,NJ,07033,USA
4DepartmentofBusinessEconomics,Erasmus
Uni-versityRotterdam,P.O.Box1738,Rotterdam,3000 DRtheNetherlands
5Clinical Pharmacology and Pharmacotherapy,
UniversityofLeuven,Herestraat49Box521, Leu-ven,3000Belgium
6
Sanofi, 55 Corporate Drive Bridgewater, NJ, 08807,USA
7
ErasmusChoiceModellingCentre,Erasmus Uni-versityRotterdam,P.O.Box1738,Rotterdam,3000 DRtheNetherlands
8
Department of Public Health, Erasmus MC – University Medical Centre, Dr. Molewaterplein 40,3000CARotterdam
*Correspondingauthor.
§Jointfirstauthor.
GLOSSARY
Medicalproductlifecycle(MPLC)termto describethelifecyclesofdrugsandmedical devices
Patientpreferencesrelativedesirabilityor acceptabilityofspecifiedalternatives;or choicesamongtreatmentalternativesor outcomes;orotherattributesthatdiffer amongalternativehealthinterventions[7]
Preferenceelicitationmethod quantitativemethodscollecting quantifiabledataforhypothesistesting andotherstatisticalanalyses
Preferenceexplorationmethod
qualitativemethodsthatcollectdescriptive datathroughparticipantorphenomenon observation,examiningthesubjective experiencesanddecisionsmadeby participants
Quality-adjustedlifeyear(QALY)a measureofthestateofhealthofaperson orgroupinwhichthebenefits,intermsof lengthoflife,areadjustedtoreflectthe qualityoflife[77]
Features