ContentslistsavailableatScienceDirect
Health
Policy
jo u rn al h om ep a g e :w w w . e l s e v i e r . c o m / l o c a t e / h e a l t h p o l
How
does
participating
in
a
deliberative
citizens
panel
on
healthcare
priority
setting
influence
the
views
of
participants?
Vivian
Reckers-Droog
a,∗,
Maarten
Jansen
b,
Leon
Bijlmakers
b,
Rob
Baltussen
b,
Werner
Brouwer
a,c,
Job
van
Exel
a,caErasmusSchoolofHealthPolicy&Management,ErasmusUniversityRotterdam,theNetherlands bRadboudInstituteforHealthSciences,Radboudumc,Nijmegen,theNetherlands
cErasmusSchoolofEconomics,ErasmusUniversityRotterdam,theNetherlands
a
r
t
i
c
l
e
i
n
f
o
Articlehistory:
Received28January2019
Receivedinrevisedform14October2019 Accepted28November2019 Keywords: Healthcare Resourceallocation Prioritysetting Societalviews Publicdeliberation Qmethodology
a
b
s
t
r
a
c
t
Adeliberativecitizenspanelwasheldtoobtaininsightintocriteriaconsideredrelevantforhealthcare prioritysettingintheNetherlands.Ouraimwastoexaminewhetherandhowpanelparticipation influ-encedparticipants’viewsonthistopic.Participants(n=24)deliberatedoneightreimbursementcases inSeptemberandOctober,2017.UsingQmethodology,weidentifiedthreedistinctviewpointsbefore (T0)andafter(T1)panelparticipation.AtT0,viewpoint1emphasisedthataccesstohealthcareisaright andthatprioritisationshouldbebasedsolelyonpatients’needs.Viewpoint2acknowledgedscarcity ofresourcesandemphasisedtheimportanceoftreatment-relatedhealthgains.Viewpoint3focusedon helpingthoseinneed,favouringyoungerpatients,patientswithafamily,andtreatingdiseasesthat heav-ilyburdenthefamiliesofpatients.AtT1,viewpoint1hadbecomelessopposedtoprioritisationandmore considerateofcosts.Viewpoint2supportedout-of-pocketpaymentsmorestrongly.Anewviewpoint3 emergedthatemphasisedtheimportanceofcost-effectivenessandthatprioritisationshouldconsider patientcharacteristics,suchastheirage.Participants’viewspartlyremainedstable,specifically regard-ingequalaccessandprioritisationbasedonneedandhealthgains.Notablechangesconcernedincreased supportforprioritisation,considerationofcosts,andcost-effectiveness.Furtherresearchintotheeffects ofdeliberativemethodsisrequiredtobetterunderstandhowtheymaycontributetothelegitimacyof andpublicsupportforallocationdecisionsinhealthcare.
©2019TheAuthor(s).PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCC BY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/4.0/).
1. Introduction
Priority setting in the allocation of healthcare resources is
inevitableduetotheincreasingdemandforhealthcareand
result-ingpressure onlimitedbudgets. Different principleshavebeen
proposedforinformingallocationdecisions,includingthe
princi-plesofmaximisinghealthandprioritisingthosewhoareworseoff
intermsofhealth[1,2].Theproposedprinciplestosomeextent
allreflectasharedunderstandingofdistributivejustice;however,
noneaddressescompletelythecomplexandvalue-ladenproblems
thatarisefromtheneedtosetpriorities[1,3–5].Forexample,some
havearguedthattheseprinciplesinsufficientlyreflectpublicviews
andpreferencesconcerningtheallocationofscarceresources[6–9].
Aconsiderablepartofthepublicevenopposesprioritysetting
alto-∗ Correspondingauthorat:ErasmusUniversityRotterdam,ErasmusSchoolof HealthPolicy&Management,P.O.Box1738,3000DRRotterdam,theNetherlands.
E-mailaddress:reckers@eshpm.eur.nl(V.Reckers-Droog).
getherandconsidersaccesstohealthcarearighttowhichpatients
areentitledwithoutexceptionorrestriction[10–12].Thosewho
dosupportprioritysettingholddifferent,sometimesconflicting,
viewsaboutthecriteriathatshouldbetakenintoaccountwhen
set-tingpriorities[6–9].Thisheterogeneityofpublicviewsmaypartly
explainwhytheoutcomesofallocationdecisionsattimesleadto
publicdebateandcontroversy[12].
Inatimewhenthepublicdemandsgreatertransparencyand
accountability from their governments and increasingly seeks
opportunitiestoactivelyparticipateinshapingthepoliciesthat
affecttheirlives[13],ithasbeenarguedthatallocationdecisionsin
healthcarecouldbeimprovedbyconsideringpreferencesfromthe
publicthatareevidence-informedandelicitedbymeansofrational
democraticdeliberations[3,13–15].Suchdeliberativemethodsaim
tomeetthedemandforafair,legitimate,andpublicallytransparent
wayofdecisionmakingandmayincreasesupportfortheoutcomes
ofsuchdecisionsastheyaremoreinformed[3,15–17].Examples
of deliberative methods include deliberative focus groups,
citi-zensjuries,andcitizenspanels[17–20]thatallsharethefollowing
https://doi.org/10.1016/j.healthpol.2019.11.011
0168-8510/©2019TheAuthor(s).PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/ 4.0/).
characteristics:(i)theformationofasmallgroupofcitizenswho
representalargerpopulationbasedonpredefinedcharacteristics,
(ii)oneormoremeetingsabouttheissueofinterest,(iii)the
prepa-rationanddisseminationofbackgroundinformationconcerning
theissueofinterest,(iv)theinvolvementofexpertstoeitherinform
thecitizensoranswertheirquestionsabouttheissueofinterest,
and(v)theformulationofasetofrecommendationsorproposals
basedontheparticipants’deliberations[17].
Deliberativemethodsareincreasinglyappliedtoinform
allo-cation decisions in healthcare, even though they are generally
moretime-consuming,labour-intensive,andexpensivethan
non-deliberative methods (e.g. preference elicitation by means of
surveys) [21], and very little is known about their effect. For
example,empiricalevidenceconcerningtheireffectonallocation
decisionsandtheviewsandpreferencesofparticipantsisscarce
[17–22].
Intheautumnof2017,adeliberativecitizenspanelwasheld
toobtain insight into participants’ views and preferences
con-cerninghealthcareprioritysettingandidentifythecriteriathey
consideredrelevantfordecisionsconcerningthecompositionof
thebasicbenefitspackageofthehealth-insurancesysteminthe
Netherlands[23,24].Healthinsuranceismandatoryforall
inhab-itantsoftheNetherlandsandthebasicbenefitspackagecoversa
broadrangeofcurativeandpreventivetreatmentstoprotect
cit-izensagainstcatastrophichealthcarespending.Althoughinsome
countriesdeliberativecitizenspanelsaremorefrequentlyapplied,
e.g.thecitizenscouncilappliedbytheNationalInstituteforHealth
andCareExcellenceintheUnitedKingdom(UK)[25],thispanelwas
thefirsttobeappliedintheNetherlandsinthecontextofhealthcare
prioritysetting.Adetaileddescriptionoftheapplieddeliberative
approachandresultsofthepanelcanbefoundinBijlmakersetal.
[24].Theaimofthecurrentstudywastoexaminewhetherand
howpanelparticipationinfluencedparticipants’viewson
health-careprioritysetting.Tomeetthisaim,weusedQmethodologyto
investigatetheviewsamongparticipantsbeforeandafterthey
par-ticipatedinthepanel.Thismethodologyisincreasinglyappliedin
healthservicesresearch[7,26]andtoidentifyanddescribepublic
viewsonhealthcareprioritysetting[e.g.6–9].Inthecurrentstudy,
weextendedpreviousapplicationsofthismethodologyto
exam-inechangesinparticipants’viewsovertime.TheapplicationofQ
methodologyenabledustocombineaspectsofquantitativeand
qualitativemethodstosystematicallyexaminewhetherandhow
viewschangedatthegrouplevelaswellastheextenttowhich
individualparticipantsstillidentifiedwiththeirinitialviewpoints
afterthepanel.Withthisstudy,weaimtocontributetotheexisting
literatureontheeffectofapplyingdeliberativemethodsfor
inform-ingallocationdecisionsinhealthcare.Theapproachandresultsof
thisstudymaybeofinteresttopublicauthoritiesand
organisa-tionsinthehealthcaresectoraswellasinothersectorsthatapply,
orconsiderapplying,deliberativemethodsinthecontextof
pol-icydevelopmentandevaluation.Furthermore,theresultsofthis
studyprovideinsightintothepossibleadditionalvalueofapplying
deliberativemethodsinthecontextofhealthcareprioritysetting.
2. Methods
2.1. Citizenspanel
Twenty-fourcitizenswererecruitedforpanelparticipationby
Motivaction;anindependentresearchandconsultancyagencyin
theNetherlands.Thesamplingwasaimedatcomposingavaried,
yetbalanced,panelregardingage,gender,geographicalspread,and
eight‘mentalitygroups’.Eachofthesegroupsrepresentsa
differ-entsetofsharedvaluesregardingwork,leisure,andpoliticsand
hasadistinctlifestyleandconsumptionpattern[24,27].Formore
informationontherecruitmentofparticipantsandadescriptionof
the‘mentalitygroups’,werefertheinterestedreadertoBijlmakers
etal.[24]andMotivaction[27].
ThepanelmetduringthreefullweekendsbetweenSeptember
16andOctober29,2017.Twoexperiencedmoderators,whowere
employedbyMotivaction,leadthepanel’sdeliberationsoneight
reimbursement cases: dental (orthodontic) braces for children,
medicinesforpatientswithAlzheimer’sdisease,forpatientswith
heartburn(pyrosis),andforchildrenwithAttentionDeficit
Hyper-activityDisorder,theorphandrugeculizumab forpatientswith
atypicalHemolyticUremicSyndrome(aHUS), atotalbody scan,
bariatricsurgeryandpreventionforpatientswithobesity,andahip
replacementforelderlypatients[24].Thesecasesconcernabroad
range ofhealth technologiesand patient populationsand were
selectedtorepresentthevarietyofcriteria,arguments,dilemmas,
andsocietalvaluesthatthepanelcoulddeemrelevantforsetting
priorities[23,24].Thefirstfourcaseswerediscussedduringthefirst
weekendandthelatterfourduringthesecondweekend.Eachcase
wasintroducedwithashortvideoinwhichinformationwas
pro-videdabouttheprevalence,symptoms,andcourseofthediseaseas
wellastheavailabletreatmentoptions.Afterwatchingthevideo,
participantsreadwrittencasedescriptionsindividuallyand
delib-eratedontheminsmallgroups,followedbyplenarydeliberations.
Duringthethirdweekend,participantswereaskedtoprioritisethe
eightcasesforreimbursementanddiscussthetrade-offbetween
thecriteriatheydeemedrelevant forsettingthesepriorities.In
three separateplenarysessions that wereheld during the
sec-ondandthirdweekend,participantsweregiventheopportunity
todiscusstheirquestions aboutmedical,ethical, and economic
aspectsofhealthcareprioritysettingwiththreeexpertsonthese
topicswhoalsohadexpertiseonthereimbursementprocessinthe
Netherlands.Theseexpertswereinstructedtoanswerparticipants’
questionsbasedontheirprofessionalknowledgeandexperience,
butnotdivulgetheirpersonalviewsonthistopic.More
informa-tionontheselectionofthereimbursementcasesandadetailed
overviewoftheprogrammeofthepanelcanbefoundinBijlmakers
etal.[24].
OnSeptember4,i.e.twoweeksbeforethepanelcommenced,
aninformationmeetingwasheldduringwhichtheparticipants
receivedgeneralinformationaboutthetopic,objective,and
pro-cedureofthepanel.Theprovidedinformationwaskeptsparseto
avoidinfluencingtheparticipantsbeforethestartofthepanel[24].
Afterthefirstweekend,theparticipantsreceivedabrochurewith
informationaboutincreasinghealthcareexpenditures,the
organi-sationandfinancialstructureoftheDutchhealthcaresystem,and
howhealthcareprioritiesarecurrentlysetintheNetherlands.This
informationwasprovidedtofacilitatemorein-depthdeliberations
duringthesecondandthirdweekendofthepanel[24].
2.2. Approach
Ourstudywasconductedinthreeconsecutivestepscommonto
Qmethodologystudies[26].First,wedevelopedacomprehensive
setofstatementsrelatingtothetopicofhealthcarepriority
set-tingintheNetherlands.Second,wecollecteddatabyadministering
thesamestatement-rankingexercisetwice:beforetheparticipants
receivedtheinformationpackageduringtheinformationmeeting
anddirectlyafterthefinalpanelmeeting.Third,weanalysedthe
collecteddatatoexaminepossiblechangesinparticipants’views
duringthecourseofthepanel.Wedescribethestepsinmoredetail
below.
2.3. Statementset
Wedevelopedastructuredstatementsetthatwasbroadly
all issues that participants could deem relevant for healthcare
prioritysettingintheNetherlands.Forthis,weadoptedthe
con-ceptualframeworkofthemostrecentlyconductedQmethodology
studyonhealthcare prioritysettingin theNetherlands[9].This
studyfocused specificallyonprioritising end-of-lifecare inthe
Netherlands[9]anditsframeworkdistinguished20characteristics
thatarecategorisedintosixdomains:characteristicsofthepatient,
characteristicsoftheillness,characteristicsofthetreatment,health
effectsoftreatment,broadereffectsoftreatment,andmoral
princi-ples.Tobetteralignthisframeworkwithour—moregeneral—topic
ofinterest,weadditionallyinspectedtheframeworkofaQ
method-ology studythat focused more generallyon healthcare priority
settingintenEuropeancountries,amongwhichtheNetherlands
[7].Afterconsideringtherelevanceofthecharacteristicsincluded
in these two frameworks for the current study, we removed
statementsconcerning‘priorhealthconsumption/previoushealth
profile’, ‘distributionof fixedhealth gains/threshold effect’,and
‘capacitytobenefit’fromthefirstframework[9]andincluded
state-mentsconcerning‘rarityofthedisease’,‘costs/budgetimpactofthe
treatment’,and‘supplier-induceddemand’fromthesecond
frame-work[7].Wethenselected25statementsfromthefirstframework
[9]andonestatementfromthesecondframework[7],and
supple-mentedthesewithtwostatementsfromrelatedQmethodology
studiesthatwereconductedintheUK[6,8].Inordertoachievea
balancedstatementsetthatcoveredallissuesofinteresttothis
study,weformulatedsevenadditionalstatementsbasedon
crite-riaandconsiderationsthatpolicymakersintheNetherlandsdeem
relevantinallocationdecisionsthatwerenotyetreflectedinthe
statementset[23,24,28,29].Finally,wetranslatedthestatements
intotheDutchlanguage.Becausethesetwasbasedonfourprevious
carefullydesignedandpilotedstudies,nopilottestwasconducted.
Table3intheresultssectionincludesthefinalsetof35
state-mentsandtheirorigin.The20characteristicsinsixdomainsand
theassociatedstatementnumbersarepresentedinAppendixA.
2.4. Datacollection
All24participantsinthecitizenspanelalsoparticipatedinthis
study.Thissamplesizewassufficientforthepurposeofthis
analy-sis[26,30].Theparticipantscompletedthefirststatement-ranking
exerciseduringtheinformationmeetingonSeptember4,2017(T0)
andtheseconddirectlyafterthefinalpanelmeetingonOctober29,
2017(T1).Beforeperformingtheexercise,participantsreceivedan
oralgroupinstructiononhowtoperformtheexercisefromone
oftheresearchers(MJ).Theyreceivedacopyoftheseinstructions
onpaper(seeAppendixB),forreference.Thisresearcherremained
presentduringtheexerciseincaseparticipantshadanyquestions
abouttheprocedure.Subsequently,participantsreceivedasetof
the35statementsprintedoncards,asortinggrid(seeAppendix
C),andaresponsesheet.Participantsfirstreadallstatementsand
divided them intothree piles (‘agree’, ‘disagree’,and ‘neutral’).
Then,theyre-readthestatementsinthe‘agree’pile,selectedthe
twotheyagreedwithmost,andplacedthemincolumn9ofthe
sortinggrid,followedbyplacingthenextthreestatementsthey
thenagreedwithmostincolumn8andsoonuntiltheyfinished
thispile.Next,theyfollowedthesameprocedureforthe‘disagree’
pile,startingwithcolumn1,andfinallyplacedthestatementsinthe
‘neutral’pileintheremainingopenspotsinthemiddleofthegrid.
Afterfinishingtheexercise,participantsusedtheresponsesheet
toexplaininwritingtheirmotivationforplacingthestatementsin
theextremepositionsofthegrid,i.e.columns1and9.Thecolumns
werepresentedtoparticipantsasbeingfrom1to9onthesorting
gridtoavoidimposingconnotationsofnegative,neutralorpositive
tocolumnsofthegrid;however,werecodedthecolumnsto-4to
+4fortheanalysisofthedataandinterpretationoftheviewpoints.
2.5. Dataanalysis
We conducted a principal component analysis followed by
oblimin rotation to identifygroups of participants withhighly
(Pearson)correlatedstatementrankingsatbothtimepoints
sep-arately.Thistypeofobliquerotationmethodistypicallyusedto
allowforanon-orthogonalrotation.Weselectedthebestnumber
offactorsfromallpossiblefactorsolutionsthatweresupported
bythedatabyapplyingthecriteria:(i)eigenvaluesoffactors>1
and(ii)aminimumoftwonon-confounded‘exemplars’perfactor.
Exemplarsarethoseparticipantswith(i)afactorloadingabove
thesignificancethresholdof0.33(p<0.05;calculatedas1.96/√35,
where35isthenumberofstatements)and(ii)forwhomthesquare
of theloadingfor a factoris largerthanthesumofthesquare
loadingsforallotherfactors[30,31].Basedoninspectionsofthe
correlationsbetweenfactorsandtheinterpretationofthefactors
ineachfactorsolution,weselectedthefactorsolutionthatleadto
themostintelligiblereductionofthedata.Subsequently,we
com-putedfactorarraysforeachfactor.Thesearraysrepresenthowa
participantwithacorrelationof1withafactorwouldhaveranked
thestatements.Weusedthefactorarrays,includingthe
charac-terisinganddistinguishingstatements,forinterpretingthefactors
asviewpoints.Characterisingstatementsarethosethatholdthe
positions-4,-3,+3,and+4inthefactorarrays,andassuch
repre-sentthestatementsthatparticipantswithaspecificviewpointleast
andmostagreedwith.Distinguishingstatementsarethosewitha
statisticallysignificantlydifferentpositioninafactorarrayfrom
theirpositioninthearrayofatleastoneotherfactor(p<0.05;
cal-culatedbasedontheabsolutedifferenceinz-scoresofstatements
betweenthefactorarrays).Weusedtheverbatimquotesof
exem-plarsthatweobtainedfromtheresponsesheetstohelpdescribe
theviewpointsinthewordingoftheparticipants.
Weexaminedchangesinviewpointsinmultipleways.Atthe
leveloftheviewpoints,weexaminedthecorrelationsandthemain
similaritiesanddifferencesbetweentheviewpointsatT0andT1.
Attheleveloftheparticipants,weexaminedtheextenttowhich
participantsassociatedthemselveswiththeinitialviewpoints,i.e.
theviewpointsidentifiedatT0,aftertheyparticipatedinthepanel
(atT1).Forthis,wecombinedthedataofT1withthefactorarraysof
T0andcalculatedthemean(SD)differenceincorrelationwiththe
initialviewpointsbetweenT0andT1.Furthermore,weexamined
thetransitionsbetweenviewpointsmadebyexemplarsovertime
andtheextenttowhichtheviewsofparticipantsconvergedafter
thepanel.Wedidthisbyexaminingthemean(SD)correlations
ofthestatementrankingsbetweenparticipantsatT0andT1and
applyinganF-testforsmallsamplesizestoexaminethedifference
intheassociatedvariances.
We used Cohen’s classification system for interpreting the
obtainedcorrelationcoefficients[30].Inlinewiththissystem,we
interpretedcorrelationsbelow0.30aslow,between0.30and0.50
asmoderate,andabove0.50ashigh[32].Weusedthe‘qmethod’
packageinRstudio1.0.143(Rstudio,Inc.,Boston,MA,USA)for
con-ductingtheanalyses[31].
2.6. Ethics
TheCommitteeonResearchInvolvingHumanSubjectsofthe
RadboudUniversityMedicalCenterreviewedandwaivedethical
approvalforthisstudy(reference2017-3444).
3. Results
Tables1and2presentthesocio-demographiccharacteristicsof
participantsandtheirfactorloadingswiththeviewpointsatT0and
Table1
Panelcharacteristics(n=24).a
n(%) Mean(SD) Min Max
Age(Years) 44.5(17.4) 20.0 72.0
Sex(Female) 12(50.0) Educationlevelb
Middle 9(37.5) High 15(62.5)
aParticipantsweredistributedequallyacrosstheeightmentalitygroupsand,
therefore,thischaracteristicisomittedfromthetable.
b Middle=middlevocationalandsecondaryschool,High=highervocationaland
academiceducation.
Table3presentsthefactorarrays,includingthecharacterising
anddistinguishingstatements.Below,wedescribetheviewpoints
beforeandafterthepanelanddiscussthechangesinviewpoints.
Wepresentthenumbersofthemostrelevantstatementswithin
parenthesiswithahashtag(#),followedbytheirpositioninafactor
array,e.g.(#1,+4).Distinguishingstatementsarepresentedwithan
additionalasterisk,e.g.(#1,+4*).Verbatimquotesofexemplarsare
presentedwithinquotationmarks,followedbytheiridentification
number,e.g.(id2).
3.1. Viewpointsbeforethepanel
AtT0,weidentifiedthreefactorsthattogetherexplained61.6%
ofthevarianceinthestatementrankings.Thecorrelationsbetween
thefactorswerelowtomoderate(=0.14for 1vs.2,=−0.01
for1vs. 3,=−0.30for2vs. 3).The factorshad12,eight, and
threeexemplars,respectively.Factor3hadtwopositiveexemplars
(id9andid10)andonenegativeexemplar(id8)andwas,therefore,
interpretedasbeingbipolar.Oneparticipant(id6)wasstatistically
significantlyassociatedwithfactor1;however,didnotmeetthe
secondcriterionforbeingidentifiedasanexemplar.
3.1.1. Viewpoint1
People withviewpoint 1 considered accessto healthcare as
a rightand believed thateveryoneshouldhave equalaccessto
healthcare.Accordingtopeoplewiththisviewaccessshouldsolely
bebasedonpatients’needforcareandnotontheirpersonal
char-acteristics,suchastheirgender,age,ethnicity(#16,+3;#18,−3),
lifestyle(#19,−4*;#28,−4*), orsocio-economicstatus(#13,−3).
“Everyonehasarighttohealthcare[and]personalcharacteristics
arenotimportantatall”(id7).Asprioritisationinhealthcareshould
bebasedonpatients’needforcare,“thereshouldbeno
discrimi-nation[betweenpatients]”(id24).Peoplewiththisviewbelieved
thathealthcarecostsshouldplaynoroleinprioritysettingas“you
cannotregardalifeinaneconomicway”(id4).Ifthereisawayof
helpingpatients,itismorallywrongtodenythemthistreatment
(#14,+3*).Peopleholdingthisviewdidnotbelievethatatreatment
shouldreceivelesspriorityifthetotalcostsoftreatingadisease(for
allpatients)arehigh(#31,−3).Rather,ifatreatmentiscostlyin
relationtoitsbenefits,butistheonlytreatmentavailable,itshould
stillbeprovided(#21,+3*).Peoplewiththisviewalsobelievedthat
patients’choicefortreatmentshouldbesupported,evenifitisvery
costlyinrelationtoitsbenefits(#11,+2*).“Everyonehasarightto
healthcare;evenwhenthereisnoorlittletreatmentbenefityou
cannotdenytreatment[topatients]!”(id13).Theyemphasisedthat
youcannotputapriceonlife(#17,+4)andifitispossibletosave
alife,everyeffortshouldbemadetodoso(#29,+4).“Regardlessof
money,ifitispossible,alifehastobesaved”(id18).
3.1.2. Viewpoint2
Peoplewithviewpoint2believedthateveryonehasarightto
healthcare,butthatthisdoesnotmeanthateverythingcanalways
bereimbursed(#25,+3*).“Everyoneisinsuredandhas[...]aright
tohealthcare,butnot everythingcanalwaysbecoveredbythe
[publichealth]insurance” (id15).As“healthcarecostskeep
ris-ing,thereshouldberestrictingmeasures”(id21).Peoplewiththis
viewemphasisedtheimportanceoftheeffectivenessoftreatments.
Table2
FactorloadingsatT0andT1(n=24).
ViewsatT0 ViewsatT1 id 1 2 3 1 2 3 1 0.74* 0.02 −0.28 0.85* −0.17 0.10 2 0.72* 0.00 0.19 0.32 0.37 −0.33 3 0.50 0.55* −0.06 0.16 0.67* −0.35 4 0.58* 0.46 0.25 0.80* 0.14 0.27 5 0.94* −0.20 −0.01 0.66* 0.14 −0.39 6 0.37 0.28 0.28 0.03 0.69* −0.23 7 0.77* 0.21 −0.15 0.42 0.47 0.25 8 0.31 0.28 −0.52* 0.21 0.49* 0.10 9 −0.13 0.03 0.71* 0.12 0.14 0.54* 10 −0.01 −0.01 0.82* 0.14 −0.13 −0.22 11 −0.16 0.89* 0.10 0.13 0.73* 0.02 12 −0.09 0.79* −0.18 −0.38 0.85* 0.01 13 0.81* −0.04 0.21 0.62* −0.03 −0.37 14 0.62* −0.31 0.39 0.30 0.11 −0.73* 15 0.20 0.76* 0.02 0.59* 0.42 0.13 16 0.91* −0.01 −0.17 0.60* 0.36 −0.01 17 0.75* −0.13 0.04 0.40 0.05 0.63* 18 0.84* 0.06 0.00 0.86* −0.08 −0.22 19 0.47 0.63* 0.07 0.18 0.83* −0.03 20 0.25 0.48* 0.11 −0.02 0.78* −0.03 21 −0.28 0.78* −0.25 −0.13 0.73* 0.38 22 0.78* −0.01 −0.14 0.85* −0.05 0.03 23 0.05 0.68* 0.01 0.15 0.54 0.53 24 0.72* 0.22 0.13 0.47 0.26 −0.42 Explainedvariance(%) 33.6 19.2 8.8 22.6 22.5 11.1 Exemplarsa(n) 12 8 3 8 8 3
aThefactorloadingsofexemplarsareindicatedwithanasterisk(*).Theseloadingsmeetthefollowingtwocriteria:(i)theloadingisabovethesignificancethresholdof
0.33(p<0.05,calculatedas1.96/√35,where35isthenumberofstatements)and(ii)thesquareoftheloadingforafactorislargerthanthesumofthesquareloadingsfor allotherfactors[28,29].
Table3
FactorarraysatT0andT1.
ViewsatT0 ViewsatT1
Statement 1 2 3 1 2 3
1 Youngerpeopleshouldbegivenpriorityoverolderpeople,becausetheyhaven’thadtheirfairshareofhealthyet.a −2* −3 +1 −2 −3 +2*
2 Children’shealthshouldbegivenpriorityoveradults’health.a −2* −2 +4 −1* −3* +2*
3 Individualresponsibilityshouldnotbetakenintoaccountbecausepeopledon’talwayshavecontrolovertheirwayof living.a
+1* 0* −4* +1* −2 −2 4 Thehealthsystemshouldbeaboutlookingafterthosepatientsingreatestneed.a 0 0 −1 +1 0 +1
5 Priorityshouldbegiventothosetreatmentsthatgeneratethemosthealth.a 0* +2* −2* +1 0 +3*
6 Priorityshouldbegiventorestoringhealthtoalevelthatissufficientforpeopletoparticipateintheirusualactivities.a +1 +2 +1 +1 +2 0*
7 Priorityshouldbegiventopreventivehealthcare.c +1 +4* 0 +2 +1 0*
8 Patientswithafamilyshouldbeprioritisedbecausetheirtreatmentswillbenefitothersaswellasthepatient themselves.a
−1 −2 +3* −4 −4 0*
9 Treatmentsthatareverycostlyinrelationtotheirhealthbenefitsshouldnotbereimbursed.a −2 +1* −1 −2* −1* +3*
10 Thehealthsystemshouldrestrictitselftotreatmentsthathaveproventobringabouthealthgains.e −1* +1* −4* 0 +1 +3*
11 Weshouldsupportpatients’choicefortreatment,evenifitisverycostlyinrelationtoitshealthbenefits.a +2* −2 −2 0* −2 −2
12 It’simportanttorespectthewishesofpatientswhofeeltheyshouldtakeeveryopportunitytoextendtheirlife.a +2* −3* 0* +1* −2 −2
13 Poorerpeopleshouldbegivenprioritybecausetheydon’thavethesameopportunitiesinlife.a −3 −4 +1* −4* −4* −1*
14 Ifthereisawayofhelpingpatients,itismorallywrongtodenythemthistreatment.a +3* 0 0 +4* 0* −3*
15 Ifyouchoosetospendalotofmoneyonaspecificpatientgroup,youhavetorealisethattherewillbelessmoneyleft forotherpatientgroups.e
+1 +1 +2 +2 +3 +1 16 Accesstohealthcareshouldbebasedonneedforcare,notonpatientcharacteristics,suchastheirgender,age,or
ethnicity.a
+3 +2 0* +4 +3 −3*
17 Youcan’tputapriceonlife.a +4 +1* +4 +2* 0* −4*
18 Priorityshouldbegiventoyoungerpeople,becausetheymaybenefitfromtreatmentforlonger.a −3 −4 +2* −3 −3 +2*
19 Peoplewholiveahealthylifeshouldbeprioritisedoverpeoplewithanunhealthylifestyle.a −4* +1 +1 −1* 0 +1
20 Peoplewithasevereconditionshouldbetreatedwithpriorityoverpeoplewithanon-severecondition.a +2* −1 −1 +2 −1* +2
21 Ifatreatmentiscostlyinrelationtoitshealthbenefits,buttheonlytreatmentavailable,itshouldstillbeprovided.a +3* −1 0 +3* 0** −2*
22 Thereisnosenseinsavinglivesifthequalityofthoseliveswillbereallybad.a 0* +2* −3* 0* +2* +4*
23 Thereisnopointinprovidingtreatmentsthatdonotgenerateconsiderablehealthbenefits.e 0* +3* −2* 0* +1* +4*
24 Treatmentofillnessesthatputahighburdenonpatients’familiesshouldreceivepriority.a −1 −1 +3* −2 −1 −1
25 Everyonehasarighttohealthcare,butthisdoesn’tmeanthateverythingcanalwaysbereimbursed.a 0 +3* 0 0 +3* +1
26 Attheendoflifeitismoreimportanttoprovideadeathwithdignitythantreatmentsthatwillonlyextendlifefora shortperiodoftime.a
0* +4* −3* 0* +4* +1*
27 Thehealthsystemshouldbeaboutgettingthegreatesthealthbenefitoverallforthepopulation.a +1* +3* −1* −1 +2* 0
28 Peoplewhoareillthroughnofaultoftheirownshouldreceivepriorityoverpeoplewhoinsomewayareresponsible fortheirownillness.a
−4* 0* +2* −2 +1* −1 29 Ifitispossibletosavealife,everyeffortshouldbemadetodoso.a +4 −3* +3 +3* −1 0
30 Peoplecanpayforinexpensivetreatmentsoutofpocket.e −1 −1 −1 0 +4* 0
31 Ifthetotalcostsoftreatingadisease(forallpatients)arehigh,thistreatmentshouldreceivelesspriority.e −3 −1 −2 −1 −1 0
32 Atreatmentmaycostmoreifitisnotonlybeneficialforthepatientbutalsoforsociety.e −1* 0* +2* −3 +1* −3
33 Priorityshouldbegiventopeoplewithrarediseases,evenwhenthesediseasesdonotnecessarilycausemorehealth damagethanmorecommonones.b
0* −2 −3 −3 −2 −1 34 Medicaltestsfortheearlydetectionofdiseasesthatoftenleadtounnecessarytreatmentsshouldnotbereimbursed.e −2* 0 0 −1* +2* −4*
35 Ifatreatmentistheonlyavailabletreatmentforadisease,itshouldbereimbursed.d +2* 0 +1 +3* 0* −2* aStatementfromWoutersetal.[9].
bStatementfromVanExeletal.[7]. c StatementfromBakeretal.[6]. d StatementfromMcHughetal.[8].
eBasedonconsiderationsthattheDutchHealthCareInstitutedeemedrelevantinthereimbursementcases.
*Distinguishingstatement,i.e.statementwithastatisticallysignificantlydifferentpositioninafactorarrayfromtheirpositioninthearrayofatleastoneotherfactor
(p<0.05).
Thehealth systemshouldbeaboutgettingthegreatest benefit
overallforsociety(#27,+3*)andthereisnopointinproviding
treat-mentsthatdonotgenerateconsiderablehealthbenefits(#23,+3*).
Accordingly,theysupportprioritisationbasedontreatment
charac-teristics,suchasthetypeandsizeofhealthgainsfromtreatment,
butlikeviewpoint1,theyopposeprioritisationbasedonpatient
characteristics,suchastheirage(#18,−4).Theyfurther
empha-sisedthat,attheendoflife,itismoreimportanttoprovideadeath
withdignitythantreatmentsthatmayextendlifeonlyforashort
periodoftime(#26,+4*).Theyneitherbelievedthat,ifitispossible
tosavealife,everyeffortshouldbemadetodoso(#29,−3*)nor
thatitisimportanttorespectthewishesofpatientswhofeelthey
shouldtakeeveryopportunitytoextendtheirlife(#12,−3*).They
dobelievethatpriorityshouldbegiventopreventivehealthcare
(#7,+4*),because“thiscansavealotofmoney”(id12).
3.1.3. Viewpoint3
Peoplewithviewpoint3werepositivelyorientedtowards
pri-oritisationbasedonpatientcharacteristics,suchastheirage.They
believedthatchildrenshouldbegivenpriorityoveradults(#2,+4),
becausetheymaybenefitfromtreatmentlonger(#18,+2*).
“Chil-drenholdthefutureand,if[...]achoicehastobemade,thechild
is thefirst one entitledtoreceivingcare” (id9). However, they
opposed prioritisation based onlifestyle (#3,−4*). People
hold-ing this view alsofoundthat broader treatmenteffects should
betakenintoconsideration.Theybelievedthattreatmentof
ill-nesses that put a high burden on families of patients should
receive priority(#24,+3*), because treating these patients
ben-efits them aswellas others(#8,+3*). Consequently,treatments
that are beneficial for both the patient and society should be
allowed tocost more (#32,+2*). Although beingpositively
ori-entedtowardsprioritisationinhealthcare,theyemphasisedthat
you cannot put a price on life (#17,+4) and that, if it is
pos-sible to save a life, every effort should be made to do so
(#29,+3).They believedthere is asense insavinglives, even if
thequalityofthoseliveswillbereallybad(#22,−3*),andin
pro-vidingtreatmentsthatdonotgenerateconsiderablehealthgains
Incontrast,peoplewhoopposedthisviewpointwereinfavourof
prioritysettingbasedonlifestyle(#3,−4*).Theyalsobelievedthat
priorityshouldbegiventothosetreatmentsthatgeneratethemost
health#5,−2*)andthatthehealthsystemshouldrestrictitselfto
treatmentsthathaveproventobringabouthealthgains(#10,−4*).
“Ifthereisevidencethatatreatmentiseffective,itshouldalways
bereimbursed”(id8).
3.2. Viewpointsafterthepanel
AtT1,weidentifiedthreefactorsthattogetherexplained56.3
%ofthevariance.Thecorrelationsbetweenviewpointswereagain
lowtomoderate(=0.30for1vs.2,=−0.18for1vs.3,=0.06for
2vs.3).Thefactorshadeight,eight,andthreeexemplars,
respec-tively.Factor3hadtwopositiveexemplars(id9andid17)andone
negativeexemplar(id14)andwas,therefore,interpretedasbeing
bipolar.Fourparticipants(id2,id7,id23, andid24)were‘mixed
loaders’astheywerestatisticallysignificantlyassociatedwithmore
thanonefactor.Theydidnotmeetthesecondcriterionforbeing
identifiedasexemplars.Oneparticipant(id10)wasa‘nullloader’
ass/hewasnotstatisticallysignificantlyassociatedwithanyofthe
factors.
Factors1and2atT1stronglyresembledfactors1and2atT0,
with=0.84and=0.78,andhencecanberegardedasslightly
differentmanifestationsoftheircorrespondingviewpointsatT0.
Therefore,wedescribeonlythemainsimilaritiesanddifferences
betweentheseviewpointsatT0andT1.Thecorrelationbetween
factors3atT0andT1was=0.32and,therefore,weregardand
describefactor3atT1asanewlyemergedviewpoint.
3.2.1. Viewpoint1
Beforethepanel,peoplewithviewpoint1emphasisedequal
access to care and that all treatments should be available for
patients.Likepeoplewiththisviewbeforethepanel,peoplewith
viewpoint1atT1believedthatitismorallywrongtodenypatients
treatment, if there is a way of helping them (#14,+4*) or if a
treatmentistheonlyoneavailable(#35,+3*).Theyalsobelieved
thataccesstocareshouldbebasedonneedand notonpatient
characteristics,suchastheirgender,age,ethnicity(#16,+4),oror
socio-economicstatus(#13,−4*).However,peoplewiththisview
lesswerestronglyopposedtoprioritisationbasedonlifestylethan
thosewithviewpoint1atT0(#19,−1*;#28,−2)andmorestrongly
opposedtoprioritisation basedoncharacteristicsof theillness,
suchasitsrarity(#33,−3).Theywerenotablymoreconsiderateof
treatmentcosts.Theybelievedlessstronglythatyoucannotputa
priceonlife(#17,+2*)andthattreatmentshouldalwaysbe
sup-ported,evenifit isverycostlyinrelationtoitshealthbenefits
(#11,0*).Theyalsobelievedlessstronglythatatreatmentmaycost
moreifitisnotonlybeneficialforapatientbutalsoforsociety
(#32,−3).
3.2.2. Viewpoint2
Before the panel, people with viewpoint 2 believed that
everyonehasan equal right tohealthcare and emphasised the
importanceof treatmenteffectiveness and efficiency. Like
peo-plewiththis viewbeforethepanel,peoplewithviewpoint2at
T1believedthateveryonehasarighttohealthcare,butthatthis
doesnotmeanthateverythingcanalwaysbereimbursed(#25,+3*).
“Theresimplyisalimitedbudget[and]choiceshavetobemade”
(id19).Peoplewiththisviewbelievedthataccesstocareshould
bebasedonneedforcareandnotonpatientcharacteristics,such
astheirgender,age,ethnicity(#16,+3;+18,−3),orsocio-economic
status(#13,−4*).However,peoplewiththisviewwerelessstrongly
opposedtoprioritisationbasedonlifestyle(#3,−2).Theybelieved
morestronglythanthosewithviewpoint2atT0thatinexpensive
treatmentscanbepaidoutofpocket(#30,+4)as“itisrelatively
cheap”(id6)and“doesnotreallyaffectpatients’disposableincome”
(id12).Theyalsobelievedmorestronglythatmedicaltestsforthe
earlydetectionofdiseasesthat oftenleadtounnecessary
treat-ments,shouldnotbereimbursed(#34,+2*)andthatifyouchoose
tospendalotofmoneyonaspecificpatientgroup,youhaveto
realisetherewillbelessmoneyleftforothergroups(#15,+3).For
peoplewiththisviewpoint,itwas“moreimportantthatpatients
candiewithdignity”(id20)thantoextendlifeforashortperiodof
time(#26,+4*).
3.2.3. Viewpoint3
Peoplewithviewpoint3atT1believedthatprioritisationshould
bebased onthehealth effectof treatmentand patient
charac-teristicssuchastheirgender,age,andethnicity(#1,+2*;#2,+2*;
#13,−1*;#16,−3*;#18,+2*),andlifestyle(#3,−2;#19,+1;#28,−1).
“Peopledohavecontrolovertheirlives,theycannotliverecklessly
and stillbenefit”(id9). Accordingtopeoplewiththis view,the
healthsystemshouldrestrictitselftotreatmentsthathaveproven
tobringabouthealthgains(#10,+3*).Theyconsideredtreatments
thatgeneratethemosthealthtobethemostimportant(#5,+3*)and
believedthereisneitherapointinprovidingtreatmentsthatdonot
generatesignificanthealthgains(#23,+4*),norinsavinglivesifthe
qualityofthoseliveswillbereallybad(#22,+4*).Peoplewiththis
viewdidnotagreewiththestatementsthatyoucannotputaprice
onlife(#17,−4*)andthatitismorallywrongtodenypatients
treat-ment(#14,−3).Theybelievedthattreatmentsthatareverycostly
inrelationtotheirhealthgainshouldnotbereimbursed(#9,+3*).
Nonetheless,theydisagreedthatmedicaltestsfortheearly
detec-tionofdiseases,thatoftenleadtounnecessarytreatments,should
notbereimbursed(#34,−4*).
Incontrast,peoplewhoopposedthisviewpointbelievedthat
“costsarenottheonlythingthatmatters”(id14).Ifatreatmentis
theonlyavailabletreatmentforadiseaseitshouldbereimbursed
andifitisnotonlybeneficialforthepatientbutalsoforsocietyit
maycostmore(#32,−3;#35,−2*).“Ifcostsneedtobetakeninto
account,peoplecanpayforinexpensivetreatmentsthemselvesin
ordertoreimburseexpensivetreatments[frompublicfunding]”
(id14).
3.3. Associationwithinitialviewpoints
The mean (SD) correlation between participants’ statement
rankingsatT0 andT1was0.57(0.17),rangingfrom0.19to0.78
(seeAppendixD).For18participantsthecorrelationbetweenT0
andT1wasstrong,forthreemoderate,andforanotherthreelow.
Althoughnoneoftheparticipantsrankedthestatementsinexactly
thesameway,theserelativelyhighcorrelationsindicatethatthe
viewsofmostparticipantswerelargelysimilarbeforeandafterthe
panel.
Table4presentstheextenttowhich participantsassociated
themselveswiththeinitialviewpoints,i.e.theviewpointsfrom
beforethepanel(atT0),aftertheyparticipated inthepanel(at
T1).Theseresultsshowthatmostparticipants(n=19)correlated
lessstronglywiththeinitialviewpoint1atT1,withamean(SD)
decreaseincorrelationof0.08(0.21).Oftheparticipants,17
cor-related morestronglywiththeinitialviewpoint 2atT1 witha
mean(SD)increaseincorrelationof0.07(0.21)and18correlated
lessstronglywiththeinitialviewpoint3atT1,withamean(SD)
decreaseincorrelationof0.15(0.22).
Table5presentsthetransitionsbetweenviewpointsmadeby
exemplarsovertime.Theseresultsshowthatofthe12exemplars
withviewpoint1atT0,sevenmadenotransitionandstilladhered
tothisviewpoint,twochangedtheirviewtoviewpoint3,andthree
werenolongerassociatedwithoneoftheviewpointsatT1.Ofthe
eightexemplarswithviewpoint2,sixmadenotransitionandstill
Table4
Factorloadingsontheinitialviewpoints(i.e.theviewsidentifiedatT0)before(atT0)andafter(atT1)thepanelandthedifferenceinfactorloadingsbetweenthetwotime
points(n=24).
View1atT0 View2atT0 View3atT0
id T0a T1 T1-T0 T0a T1 T1-T0 T0a T1 T1-T0 1 0.74 0.73 −0.02 0.02 0.24 0.23 −0.28 −0.01 0.27 2 0.72 0.42 −0.30 0.00 0.12 0.12 0.19 0.04 −0.15 3 0.50 0.35 −0.15 0.55 0.57 0.02 −0.06 −0.11 −0.05 4 0.58 0.49 −0.09 0.46 0.28 −0.18 0.25 0.01 −0.25 5 0.94 0.80 −0.14 −0.20 0.22 0.42 −0.01 −0.10 −0.09 6 0.37 0.29 −0.08 0.28 0.49 0.22 0.28 −0.16 −0.44 7 0.77 0.37 −0.41 0.21 0.51 0.30 −0.15 −0.42 −0.27 8 0.31 0.28 −0.03 0.28 0.44 0.17 −0.52 −0.18 0.34 9 −0.13 −0.09 0.04 0.03 0.11 0.08 0.71 0.11 −0.59 10 −0.01 0.11 0.12 −0.01 −0.21 −0.20 0.82 0.77 −0.05 11 −0.16 0.26 0.42 0.89 0.59 −0.30 0.10 −0.04 −0.13 12 −0.09 −0.28 −0.19 0.79 0.46 −0.33 −0.18 −0.25 −0.07 13 0.81 0.73 −0.08 −0.04 0.03 0.07 0.21 0.14 −0.06 14 0.62 0.54 −0.08 −0.31 −0.13 0.18 0.39 0.40 0.01 15 0.20 0.59 0.39 0.76 0.53 −0.23 0.02 −0.23 −0.24 16 0.91 0.56 −0.34 −0.01 0.28 0.29 −0.17 −0.31 −0.14 17 0.75 0.19 −0.55 −0.13 0.14 0.28 0.04 −0.22 −0.26 18 0.84 0.71 −0.13 0.06 0.08 0.02 0.00 0.04 0.04 19 0.47 0.31 −0.16 0.63 0.71 0.08 0.07 −0.26 −0.33 20 0.25 0.19 −0.06 0.48 0.70 0.22 0.11 −0.27 −0.38 21 −0.28 −0.15 0.13 0.78 0.59 −0.19 −0.25 −0.24 0.01 22 0.78 0.68 −0.11 −0.01 0.24 0.24 −0.14 −0.12 0.02 23 0.05 0.04 −0.02 0.68 0.79 0.11 0.01 −0.28 −0.29 24 0.72 0.57 −0.16 0.22 0.19 −0.02 0.13 −0.34 −0.47 Mean(SD)difference NA NA −0.08(0.21) NA NA 0.07(0.21) NA NA −0.15(0.22) NA,NotApplicable.
aThesefactorloadingscorrespondwiththefactorloadingsatT
0presentedinTable2.
Table5
Transitionmatrixofexemplars’views.
ViewsatT1 1 2 3 Nodistinct viewpoint Total ViewsatT0 1 7 NA 2 3 12 2 1 6 NA 1 8 3 NA 1 1 1 3 Nodistinct viewpoint NA 1 NA NA 1 Total 8 8 3 5 24
NA,NotApplicable.
andonewasnolongerassociatedwithoneoftheviewpointsatT1.
Ofthethreeexemplarswithviewpoint3,nonestilladheredtothis viewpointatT1.Oneexemplarchangedhis/herviewtoviewpoint
2,onechangedhis/herviewtothenewviewpoint3,andonewas nolongerassociatedwithoneoftheviewpointsatT1.
3.4. Convergencebetweenviews
AtT0,themean(SD)correlationbetweenparticipants’
state-mentrankingswas0.32(0.28),rangingfrom−0.43to0.79.AtT1,
this was0.32(0.25),rangingfrom−0.26to0.73. SeeAppendix Eforthecorrelationmatricesofparticipants’rankingsatT0 and
T1.Thedifferenceinvariancedecreasedmarginallybetweenthe
rankingsatbothtimepoints(p<0.001),indicatingsomemodest convergencebetweentheviewsofparticipantsovertime. 4. Discussion
Inthisstudy,weexaminedwhetherandhowparticipationina deliberativepanelinfluencedtheviewsofparticipantson health-careprioritysetting.Ourmainfindingisthatparticipants’views beforeand afterthepanelpartly remainedstable. Therewasa strongresemblance betweentwo of thethree views identified
beforeandafterthepanel,whilethethirdviewwasdistinctly differ-entatbothtimepointsand18participantsshowedhighcorrelation betweentheirviewsatT0andT1.Equalaccesstohealthcare,
priori-tisationbasedonpatients’needs,andtherelevanceofthesizeand typeoftreatmentbenefitsremainedimportantduringthecourse ofthepanel.Weobservedtwonotablechanges.Firstly,support forprioritisationinhealthcaregenerallyseemstohaveincreased afterpanelparticipation.Secondly,participantsbecamemore con-siderateofhealthcarecostsandofcost-effectivenessasarelevant criterionforsettingprioritiesinhealthcare.
To ourknowledge, this study is one of the few toexamine changesinviewsonhealthcareprioritysettingthrough deliber-ationandthefirsttodosointheNetherlands.Thislimitsusinour abilitytocompareourresultswiththoseofotherstudies.However, wecancompareourresultstotwootherstudiesthatexaminedthe effectofdeliberationonviewsinthecontextofhealthcarepriority settingandtwoQmethodologystudiesthatexaminedviewsonthis topicintheNetherlands.Dolanetal.examinedtheeffectof delib-erationonviewsinasampleof60patientsintheUK[21].They
observedatrendtowardstreatingdifferentpatientgroupsmore
equallyandparticipantswhowereinitiallyunwillingtoprioritise
betweenpatientgroupsremainedsoafterdeliberation.Abelson
etal.examinedtheeffectofdeliberationinasampleof46
partici-pantsinCanada,byusingacontrolleddesign[17].Theyfoundthat
participants’viewsbecamemoresusceptibletochangewhenmore
deliberationwasintroduced.Participantswhochangedtheirview
didsoinasimilardirection,indicatingthatdeliberationmaylead
toincreasedconsensusamongparticipants.Likeinthesestudies,
wefoundviewsopposingprioritysettingthatremainedrelatively
stableandthatdeliberationcanleadtochangesinviewpointsas
wellastoconvergencebetweenthem.VanExeletal.andWouters
etal.appliedQmethodologytoexamineviewsonhealthcare
prior-itysettingintheNetherlands.Likeinthesestudies,wefoundthat
membersofthepublic—beforedeliberation—generallyholdaview
onprioritysettingthatemphasisestheimportanceofequalaccess
health-careresourcesandarewillingtoacceptcertaincriteriaforsetting
priorities[7,9].
Beforediscussing the main strengths and limitations of our
study,wewouldliketoreflectonthebipolarnatureofviewpoints3
atT0andT1.Previousliteratureshowsthattherearedifferentways
todealwiththecomputationandinterpretationofbipolarfactors.
Somehavearguedthat negativeexemplarsshouldbeexcluded
fromthecomputationofthefactorarrayasthisleadstoaclearer,or
purer,interpretationofthepositivepoleoftheviewpoint[e.g.33].
Othershavearguedthatnegativeexemplarsshouldbeincludedin
thecomputationofthefactorarray.Excludingthemwouldleadto
anunbalancedinterpretationofthefactor,asitnolongerfully
rep-resentstheviewsoftheparticipantswhodefinethefactor(albeit
ondifferentsidesofthepole)[e.g.26].Here,wefollowedthe
lat-terargumentandchosetoretainthenegativeexemplarsinthe
computationofthefactorarraysandtheinterpretationofthe
bipo-larfactors3atT0 andT1.Inordertoexploretheimplicationsof
thischoice,wealsoinspected asolutionexcludingthenegative
exemplars.AtT0,thecorrelationbetweenfactors3withand
with-outnegativeexemplarswas0.97,andhencethesefactorsseem
toportraythesameview.AtT1,thecorrelationbetweenfactors3
withandwithoutnegativeexemplarswas0.65andthe
position-ingofsomestatementschangedconsiderably.Morespecifically,
comparedtotheinterpretationpresentedintheResultssection,
theviewpointwouldagreelessstronglythatpersonal
character-isticsshouldbetakenintoaccountinhealthcareprioritysetting
(#1,+1*;#2,+1*;#13,−4*;#16,−1*;#18,+1*),andmorestrongly
thatindividualresponsibilityisrelevant(#3,−3*)andinexpensive
treatmentscanbepaidoutofpocket(#30,+4).Although
exclud-ingthenegativeexemplarleadstoaslightlydifferentviewpoint
3atT1,itremainsanewviewascomparedtoviewpoint3atT0
(excludingthenegativeexemplar;=−0.03)and,therefore,does
notaffectthemainfindingofourstudy.
Themain strengthofourstudyliesintherepeateduseofQ
methodologytoexamineindepthwhetherandhowdeliberation
influencesviewsonhealthcareprioritysetting.Toourknowledge,
thisapproachhasnotbeenappliedbefore,neitherinnoroutside
thefieldofhealthcare.Despitethisstrength,somelimitationsneed
tobediscussed.Firstly,although we speakofthe ‘influence’of
deliberationonviews,nocausalconclusionscanbedrawninthe
absenceofacontrolgroup.Secondly,thereimbursementcasesmay
haveprimedtheneedforsettingprioritiesandtherelative
impor-tanceofcertaincharacteristicsafterthepanel.We donotethat
thecases werecarefullyselectedtorepresentallissues
partici-pantsmayhavedeemedrelevantforsettingprioritiesinabroad
rangeofhealthtechnologiesandpatientpopulations.Inthatsense,
theywerealignedwiththebroadconsiderationsrepresentedin
thestatementset.Therefore,insofarthecasesinfluencedthe
state-mentrankingsafterthepanel,wethinkthisinfluenceisrelevant
inthecontextofthisstudy.Finally,lower-educatedpeoplearenot
representedinthepanel.However,thisisonlyproblematicifthey
differfromhigher-educatedpeoplewithrespecttotheir
suscepti-bilityfordeliberation.Thiswedonotknowandwouldbearelevant
topicforfurtherresearch.
Ourstudycontributestothelimitedliteratureontheeffectof
deliberativemethodsbygivinginsightintowhetherandhow
delib-erationinfluencesviewsonhealthcareprioritysetting.Basedon
ourresults,somequestionscanberaisedregardingtheapplication
ofdeliberativemethodsinthecontextofhealthcareprioritysetting.
Forexample,ifthepurposeistoinformallocationdecisions,
ques-tionscanberaisedabouttheextenttowhichparticipants’views
overtimestillrepresenttheactualviewsofthepublic.Ifthelatteris
desiredinapanel,onecouldarguethatthetimeanyoneparticipates
insucha panelshouldberestrictedand thatpanelparticipants
shouldregularlybereplacedbyothermembersofthepublic.
How-ever,ifchangesinviews,asobservedhere,areinterpretedasthe
effectoflearningandthepurposeisthatbetterinformedandmore
consideredviewsarerepresentedinapanel,itcanalsobeargued
thatpanelmembersshouldparticipateinapanelforalongerperiod
oftime.Inthiscase,onecouldalsoargueagainsttheapplicationofa
deliberativecitizenspanelandinfavourofbetterinformation
pro-visiontothepublicandmorepublicdebate,throughwhichasimilar
learningeffectperhapscanbeachievedinmembersofthepublic
atlarge.Notwithstanding,itisimportanttonotethatitisunlikely
thatanyone oftheseapproaches willlead topublicconsensus
aboutallocationdecisions.Therecurrentfindingintheliterature
thatviews onprioritysetting inhealthcare differ andcan
con-flict,togetherwiththecurrentfindingthatviewsremaindiverse
andonlymoderatelyconvergeafterdeliberation,suggeststhatany
allocationdecisionwillprobablystillbemetwithoppositionfrom
somegroupinsociety.Still,insightintothediversityofviewsis
importanttobeabletounderstandtheoppositionthatallocation
decisionscanbringaboutandhowtheoutcomesofdecisions,ifso
desired,canbebetteralignedwithsocietalpreferences.
We appreciate that, based onthe design and resultsof the
currentstudyitremainsunclearwhyexactlyparticipants’views
changedandtheextenttowhichtheirviewschangedunderthe
influenceof,forexample,theotherparticipants,information
pro-vided,andexpertsconsulted.If changesdo notresultfromthe
deliberations,butratherfromexternalinfluences(e.g.from
stake-holders,suchasexperts,patients,andindustry),adeliberatepanel
may have limited additional value as these views usually are
alreadyrepresentedinallocationdecisions.Thecrucialquestionin
thiscontextisthepurposeofapplyingdeliberativepanels.Isitfor
policymakerstoconsultcitizensorgivethemavoteinallocation
decisions,strengthentheappraisalofavailableevidence,increase
thelegitimacyofdecisions,orrathertopredictorincrease
pub-licsupportfortheoutcomesofsuchdecisions?Regardlessofthe
purpose,itisimportantthatcitizenscontributeinawaythatis
complementarytootherstakeholders.Althoughansweringthese
questionsliesoutsidethescopeofthispaper,theyarerelatedtothe
issuethatpanelparticipantsmayexperience(moderate)changesin
theirviewpointsovertime.Moreover,theyemphasisethatfurther
researchisindispensableforapplyingthesemethodsinawaythat
contributestothelegitimacyofandpublicsupportforallocation
decisionsinhealthcare.
5. Conclusions
Our study showed that participants’ views partly remained
stableover thecourseof thepanel,specificallyregarding equal
accesstohealthcare,prioritisationbasedonpatients’needs,and
theimportanceofthesizeandtypeoftreatmentbenefits.Notable
changes after deliberationconcernedthe increasedsupport for
prioritisation, consideration of costs, and relevance of a
cost-effectiveness criterion in allocation decisions. Considering the
increasinginterestindeliberativemethodsamongpolicymakersin
healthcareandthelimitedempiricalevidenceconcerningtheeffect
ofdeliberativemethodsonparticipants’viewsandpreferences,
fur-therresearchisrequiredtobetterunderstandhowdeliberative
methodscancontributetothelegitimacyofandpublicsupport
fortheoutcomesofallocationdecisionsinhealthcare.
DeclarationofCompetingInterest
Theauthorshavenoconflictsofinteresttodeclare.
Acknowledgements
The NetherlandsOrganisation for Scientific Research (NWO)
roleinthestudydesign,datacollectionandanalysis,preparation
ofthemanuscript,anddecisiontopublish.Theviewsexpressedin
thisarticlearethoseoftheauthors.
AppendicesSupplementarydata
Supplementarymaterialrelatedtothisarticlecanbefound,in
theonlineversion,atdoi:https://doi.org/10.1016/j.healthpol.2019.
11.011.
References
[1]OttersenT,MæstadO,NorheimOF.LifetimeQALYprioritarianismin prior-itysetting:quantificationoftheinherenttrade-off.CostEffectivenessand ResourceAllocation2014;12(1):2.
[2]Hernæs UJ,Johansson KA,OttersenT,NorheimOF.Distribution-weighted cost-effectiveness analysisusinglifetimehealth loss.Pharmacoeconomics 2017;35(9):965–74.
[3]FleckLM.Justcaring:indefenseoftheroleofdemocraticdeliberationinhealth carerationingandpriority-setting.In:DenExterA,BuijsenM,editors.Rationing healthcare.Hardchoicesandunavoidabletrade-offs.Antwerpen:Maklu;2012. p.25–45.
[4]vandeWeteringEJ,StolkEA,vanExelNJA,BrouwerWBF.Balancingequity and efficiencyintheDutchbasicbenefitspackageusingtheprincipleof proportionalshortfall.EuropeanJournalof HealthEconomics 2013;14(1): 107–15.
[5]Reckers-DroogVT,vanExelNJA,BrouwerWBF.Lookingbackandmoving for-ward:ontheapplicationofproportionalshortfallinhealthcareprioritysetting intheNetherlands.HealthPolicy2008;122(6):621–9.
[6]BakerR,WildmanJ,MasonH,DonaldsonC.Q-ingforhealth—anewapproachto elicitingthepublic’sviewsonhealthcareresourceallocation.HealthEconomics 2014;23(3):283–97.
[7]vanExelNJA,BakerR,MasonH,DonaldsonC,BrouwerWBF.Publicviewson principlesforhealthcareprioritysetting:findingsofaEuropeancross-country studyusingQmethodology.SocialScience&Medicine2015;126:128–37.
[8]McHughN,BakerRM,MasonH,WilliamsonL,VanExelNJA,DeogaonkarR,etal. Extendinglifeforpeoplewithaterminalillness:amoralrightandanexpensive death?Exploringsocietalperspectives.BMCMedicalEthics2015;16(1):1–15.
[9]Wouters S, van Exel NJA, Baker R, Brouwer WBF. Priority to end of life treatments? Views of the public in the Netherlands. Value Health 2017;20(1):107–17.
[10]MasonH,vanExelNJA,BakerR,BrouwerWBF,DonaldsonC,EuroVaQTeam. Fromrepresentingviewstorepresentativenessofviews:illustratinganew (Q2S)approachinthecontextofhealthcareprioritysettinginnineEuropean countries.SocialScience&Medicine2016;166:205–13.
[11]MasonH,CollinsM,McHughN,GodwinJ,vanExelNJA,DonaldsonC,etal. Is“endoflife”aspecialcase?ConnectingQwithsurveymethodstomeasure societalsupportforviewsonthevalueoflifeextendingtreatments.Health Economics2018:1–13.
[12]Reckers-DroogVT,vanExelNJA,BrouwerWBF.Whoshouldreceivetreatment? Anempiricalenquiryintotherelationshipbetweensocietalviewsand prefer-encesconcerninghealthcareprioritysetting.PLoSOne2018;13(6):e0198761.
[13]CaddyJ.Evaluatingpublicparticipationinpolicy-making.Paris:Publications del’OCDE;2005.
[14]DanielsN.Justhealth:meetinghealthneedsfairly.Cambridge:Cambridge UniversityPress;2008.
[15]JansenMP,HeldermanJK,BoerB,BaltussenR.Fairprocessesforpriority set-ting:puttingtheoryintopractice:commenton ¨expandedHTA:enhancing fairnessandlegitimacy¨.InternationalJournalofHealthPolicyandManagement 2017;6(1):43.
[16]AbelsonJ,ForestPG,EylesJ,SmithP,MartinE,GauvinFP.Obtainingpublicinput forhealth-systemsdecision-making:pastexperiencesandfutureprospects. CanadianPublicAdministration2002;45(1):70–97.
[17]AbelsonJ,EylesJ,McLeodCB,CollinsP,McMullanC,ForestPG.Does deliber-ationmakeadifference?Resultsfromacitizenspanelstudyofhealthgoals prioritysetting.HealthPolicy2003;66(1):95–106.
[18]DegelingC,CarterSM,RychetnikL.Whichpublicandwhydeliberate?—A scop-ingreviewofpublicdeliberationinpublichealthandhealthpolicyresearch. SocialScience&Medicine2015;131:114–21.
[19]AbelsonJ,WagnerF,DeJeanD,BoesveldS,GauvinFP,BeanS,etal. Pub-licandpatientinvolvementinhealthtechnologyassessment:aframework foraction.InternationalJournalofTechnologyAssessmentinHealthCare 2016;32(4):256–64.
[20]MittonC,SmithN,PeacockS,EvoyB,AbelsonJ.Publicparticipationinhealth careprioritysetting:ascopingreview.HealthPolicy2009;91(3):219–28.
[21]DolanP,CooksonR,FergusonB.Effectofdiscussionanddeliberationon thepublic’sviewsofprioritysettinginhealthcare:focusgroupstudy.BMJ 1999;318(7188):916–9.
[22]StreetJ,DuszynskiK,KrawczykS,Braunack-MayerA.Theuseofcitizens’ juriesinhealthpolicydecision-making:asystematicreview.SocialScience &Medicine2014;109:1–9.
[23]BaltussenR,BijlmakersL,JansenMP,BurgerforumProjectTeam.Draagvlak voorlastigekeuzes.EindrapportvanhetBurgerforum‘Keuzesindezorg’. Avail-ableathttp://hdl.handle.net/1765/114121.[Accessed2July2018].
[24]BijlmakersL,BaltussenR,JansenMP,BoerB,HeldermanJK,GroenewoudS, etal.Increasingthelegitimacyoftoughchoicesinhealthcarereimbursement: approachandresultsofacitizenforumintheNetherlands.ValueHealth2019,
http://dx.doi.org/10.1016/j.jval.2019.07.015.
[25]NationalInstituteforHealthandClinicalExcellence,Socialvaluejudgements: principlesforthedevelopmentofNICEguidance–Secondedition.Available at https://www.nice.org.uk/Media/Default/About/what-we-do/Research-and- development/Social-Value-Judgements-principles-for-the-development-of-NICE-guidance.pdf.[Accessed22May2019].
[26]WattsS,StennerP.DoingQmethodologyresearch:theory,methodand inter-pretation.London:SagePublicationsLtd.;2012.
[27]Motivaction,TheMentalitygroups.Availableathttps://www.motivaction.nl/ en/mentality/the-mentality-groups.[Accessed22February2018].
[28]ZorginstituutNederland.Packageadviceinpractice:deliberationsforarriving atafairpackage.Diemen:ZorginstituutNederland;2017.
[29]Kleinhout-VliekT,deBontA,BoerB.Thebarenecessities?Arealistreviewof necessityargumentationsusedinhealthcarecoveragedecisions.HealthPolicy 2017;121(7):731–44.
[30]BrownSR.Politicalsubjectivity:applicationsofQmethodologyinpolitical science.NewHaven,Connecticut:YaleUniversityPress;1980.
[31]ZabalaA.qmethod:apackagetoexplorehumanperspectivesusingQ method-ology.TheRJournal2014;6(2):163–73.
[32]CohenJ.Statisticalpoweranalysisforthebehavioralsciences,2nded.New Jersey:LawrenceErlbaum;1988.
[33]HackertMQ,BrouwerWBF,HoefmanRJ,vanExelNJA.Viewsofolder peo-pleintheNetherlandsonwellbeing:aQ-methodologystudy.SocialScience& Medicine2019:112535.