Looking back and moving forward

(1)

Contents lists available atScienceDirect

Health

Policy

j o u r n a l h o m e p 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

Looking

back

and

moving

forward:

On

the

application

of

proportional

shortfall

in

healthcare

priority

setting

in

the

Netherlands

V.T.

Reckers-Droog

a,∗

_,

_N.J.A.

_van

_Exel

a,b

_,

_W.B.F.

_Brouwer

a

a_Erasmus_University_Rotterdam,_Erasmus_School_of_Health_Policy_&_Management,_Rotterdam,_The_Netherlands b_Erasmus_University_Rotterdam,_Erasmus_School_of_Economics,_Rotterdam,_The_Netherlands

a

r

t

i

c

l

e

i

n

f

o

Articlehistory:

Received22November2017

Receivedinrevisedform27March2018 Accepted1April2018 Keywords: Proportionalshortfall Equity Socialvalue Prioritysetting Decisionmaking Resourceallocation

a

b

s

t

r

a

c

t

Theincreasingdemandforhealthcareandtheresultingpressureonavailablebudgetsrenderpriority

settinginevitable.Ifsocietiesaimtoimprovehealthanddistributehealth(care)fairly,equity-efﬁciency

trade-offsarenecessary.IntheNetherlands,proportionalshortfall(PS)wasintroducedtoquantify

neces-sityofcare,allowingadirectequity-efﬁciencytrade-off.Thisstudydescribesthehistoryandapplication

ofPSintheNetherlandsandexaminesthetheoreticalandempiricalsupportforPSaswellasitscurrent

roleinhealthcaredecisionmaking.WereviewedtheinternationalliteratureonPSfrom2001onwards,

alongwithpubliclyaccessiblemeetingreportsfromtheDutchappraisalcommittee,Adviescommissie

Pakket(ACP),from2013to2016.Ourresultsindicatethatthereissupportforthedecisionmodelinwhich

necessityisquantiﬁedandincrementalcost-effectivenessratiosareevaluatedagainstassociated

mone-taryreferencevalues.Themodelenablesauniformframeworkforprioritysettingacrossallhealthcare

sectors.AlthoughconsensusabouttheapplicationofPShasnotyetbeenreachedandalternativewaysto

quantifynecessitywerefoundinACPreports,PShasincreasinglybeenappliedindecisionmakingsince

2015.However,empiricalsupportforPSislimitedanditmayinsufﬁcientlyreﬂectsocietalpreferences

regardingageandreducinglifetime-healthinequalities.Hence,furtherinvestigationintoreﬁningPS—or

explorationofanotherapproach—appearswarrantedforoperationalisingtheequity-efﬁciencytrade-off.

license(http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

The demandfor healthcare is rapidly increasingfor reasons thatincludeageingpopulationsandtheavailabilityofincreasingly advancedandexpensive(new)healthtechnologies.Ashealthcare resourcesremainscarce,theresultingpressureonavailable bud-getsrendershealthcareprioritysettinginevitable[1,2].Although politicallyand societally sensitive, theneedfor prioritisation is widelyrecognised andexplicitly addressingprioritysetting has becomeindispensablefordevelopingfairermethodsforresource allocationinhealthcare[3,4].

Economic evaluations of health technologies are often used toinformdecisionmakersregardinghow toallocatehealthcare resourcesinanoptimalwayforsociety.However,theoutcomes ofeconomicevaluationsonlypredictsuchdecisionstoamoderate extent[3,5,6].Oneexplanationforthisdisparityisthatdecision

∗ Correspondingauthorat:ErasmusUniversityRotterdam,ErasmusSchoolof HealthPolicy&Management,P.O.Box1738,3000DRRotterdam,TheNetherlands.

E-mailaddresses:reckers@eshpm.eur.nl(V.T.Reckers-Droog),

vanexel@ese.eur.nl(N.J.A.vanExel),brouwer@eshpm.eur.nl(W.B.F.Brouwer).

makers are not exclusively concerned with maximising health given available budgets, but also withdistributing health(care) equitably and fairly [3,5,7,8]. Hence, an optimal allocation of resourcesinvolvessettingprioritiesthatcontributetoboth efﬁ-ciencyandequityinthedistributionofhealth(care)[9].Recognising thatthesearebothimportantobjectivesofhealthcaresystems,it hasbeenadvocatedthatsocietalconcernsforequitybeexplicitly andtransparentlyincorporatedintothedecision-making frame-work[10–12].

In economicevaluations,thevalueof a healthtechnologyis commonlyexpressedintermsofanincrementalcostper quality-adjusted life-year (QALY) ratio (ICER) that is evaluated against somemonetarythresholdvalueperQALYgained[3,13–15].When theICERisbelow thisthreshold,a healthtechnologyis consid-eredcost-effectiveandeligibleforreimbursement[16].Theclassic approachintheeconomic-evaluationframeworkistovalueQALY gainsequally,i.e.toadheretotheprinciplethat“aQALYisaQALY isaQALY”,regardlessofbeneﬁciaryandhealthtechnology charac-teristics[17].However,thisapproachhasbeenhighlydebatedasit reliesontheassumptionofdistributiveneutrality[3].Inresponse tothisdebate,two generalapproacheshavebeensuggestedfor operationalisingtheequity-efﬁciencytrade-off[3,5].Oneofthese https://doi.org/10.1016/j.healthpol.2018.04.001

(2)

approachesappliesequity weightstoQALYgainsand evaluates theadjustedICERagainstafixedmonetarythresholdvalue,and theotherevaluatesan unadjustedICER againsta flexible mon-etarythreshold value[3,5,16]. Ideally, the operationalisation of theequity-efficiencytrade-offisbothnormativelyjustifiableand empiricallysupported.However,thisprovestobeneithereasynor straightforward[3,18].

In relation tothe operationalisation of theequity-efficiency trade-off,theseverityofillness(SOI)andfairinnings(FI)equity approacheshaveattractedmuchattentioninternationally. Accord-ing to the normative theories about distributive justice that underlie these approaches, priority should be given to those who areworse off in terms of health [11,19,20]. However,the approaches are based on different normative arguments with regardtowhomis consideredworseoff, and hencedifferwith regardtohow theyareoperationalised[3,19].Acommon oper-ationalisationofSOIaimstoequaliseabsolutehealthbenefitsin termsofcurrentand prospectivehealth, whileFIaimstodoso intermsof lifetimehealth [3,10,11,19].Assuch,FI also consid-ers past health [11,19]. Although both SOI and FI are tosome extentnormativelyjustifiableandempiricallysupported,neither oftheseapproachesappearstosatisfactorilyreflectsocietal prefer-encesforequity[3,5,7].Nonetheless,differentcountrieshaveeither implicitlyorexplicitlydevelopednormativeprinciplesor guide-linesthatinclude (aspectsof)SOI orFIfor informingallocation decisionsinhealthcare[3,5,7].Forexample,intheUnitedKingdom (UK),theNationalInstituteforHealthandCareExcellence(NICE) formalisedtheSOIapproachbylaunchingguidelinesfor prioritis-ingend-of-lifecare[21,22],andinNorway,theSOIapproach is currentlyformalisedintermsofabsoluteshortfall[23,24].Inan attempttobalancesocietalconcernsregardingSOIandFI[5], pro-portionalshortfall(PS)wasintroducedintheNetherlandsasan equityapproachthatcombinesaspectsofSOIandFI[3,5].Although consensusabouttheapplicationofPShasnotyetbeenreached[25], theapproachreceivedconsiderablesupportfrompoliticiansand policymakersandwasincorporatedintotheassessmentphaseof healthcareprioritysettingintheNetherlands[1,3]. Assuch,the Netherlandsisone ofthefirstcountriestoexplicatetheequity criterioninthiscontext[3,5].

Thisstudydescribesthehistoryand applicationofPSinthe Netherlandsandexaminesthetheoreticalandempiricalsupport forPSaswellasitscurrentroleinhealthcaredecisionmakingin theNetherlandsbyreviewingtheinternationalPSliteratureand publiclyaccessiblemeetingreportsfromtheDutchappraisal com-mittee,theAdviescommissie Pakket (ACP). Althoughthis study primarilyfocusesonhealthcareprioritysettingintheNetherlands, theresultsofthestudymayalsobeusefulforothercountries seek-ingtooperationalisetheequity-efﬁciencytrade-offforinforming allocationdecisionsinhealthcare.

2. Abriefhistoryofhealthcareprioritysettinginthe Netherlands

Thereport“Choicesinhealthcare”thatwaspresentedbythe DunningCommitteein1991[26]wasalandmarkpublicationon healthcareprioritysettingintheNetherlands.Inthisreport,four criteriaforprioritysettingwereformulated:necessity, effective-nessandefﬁciencyofcare,andpatients’individualresponsibility for(payingfor)care.Inthisreport,theDunningCommitteeused themetaphor of a funnel todescribe a criteria-based decision modelforevaluatingthecompositionofthepubliclyfunded health-insurancepackage.Basedonthishierarchicalmodel,technologies that(wouldsubsequently)passallcriteriaweretobeincludedin thebasicbeneﬁtspackage.Thereportwaspivotalforthe discus-siononprioritysetting, andin thefollowing years, thecriteria

wereputinto practice[2,6,27]. TheDutchHealthCareInstitute (ZiN)laterreformulatedthesecriteriaasnecessityofcare, effective-ness,cost-effectiveness,andnecessityofinsurance,respectively, andsupplementedthesewithafeasibilitycriterion[1,28].

Althoughnoneofthecriteriaweredefinedandoperationalised withoutdispute, this proved to beparticularly difficultfor the necessityofcarecriterion[2,29].TheDunningCommitteedefined necessityofcareascarethat isnecessaryforthepreventionof prematuredeathand/orforpatientswho—duetosomediseaseor condition—cannotfunctionnormallyinsociety[2,26].Thelatter partofthisdefinitionwasregardedasproblematic,asitwasunclear howtointerpretandquantify‘normal’functioning.Moreover,the term‘necessity’impliedanabsoluteratherthanarelativecut-off pointfordecisionmaking,whichwasamplifiedbytheDunning Committee’suseofafunnelmetaphor[2,26].Ifatechnologyfailed topass‘thesieveofnecessity’,thetechnologywouldnotbe incor-poratedintothepublichealth-insurancepackage,andassessment ofits(cost-)effectivenessandneedforinsurancewouldbe super-fluous[2].However,asthedegreetowhichhealthtechnologiesare necessaryvaries,itwassuggestedthatthiscriterionberegardedas neitherabsolutenorisolatedfromtheothercriteria[2,30,31].

In2001,Stolketal. [2] proposedadecision modelin which necessityofcarewasdefinedas‘burdenofillness’(BOI)and oper-ationalisedasarelativecriterionbyattachingahighernecessity scoretohealthtechnologiesthattargetdiseaseswithahigherBOI level.Stolketal.[2]describedBOIastheaveragedisease-related lossinqualityandlengthoflifeofpatients,relativetothe situa-tioninwhichthediseasehadbeenabsentandquantifiedBOIin termsofQALYsona0-1scale.Furthermore,theyproposed con-nectingthenecessityofcareand(cost-)effectivenesscriteriaby attachinga higher societalwillingness to pay(WTP) per QALY gainedtoahigherlevelofBOI.Specifically,theauthorssuggested dividingthecontinuous0-1 BOIscaleintosevencategoriesand evaluatingthe ICERof (new) health technologiesagainstseven associatedmonetarythresholdvaluesperQALYgained.The pro-posedcost-effectivenessthresholdvaluesperQALYgainedranged fromapproximatelyD4500toD45,000[32].Decidingontheexact cut-offpointsfortheBOIcategories,thecost-effectiveness thresh-old range, and the shape of their reciprocal relationship were regardedasmattersofpoliticalandsocietalconcern.

Theproposedmodelreceivedbroadsupportasitcontributed tothedevelopmentofatransparentandcoherentdecisionmodel forhealthcareprioritysettingintheNetherlandsbyexplicitly con-nectingthecriteria formulatedby theDunning Committeeand enabling a uniformand systematic quantiﬁcation of BOI across patientgroupsanddiseaseareas[23,27–29].Between2002and 2005,BOIwasfurtherformalisedasproportionalshortfall(PS)and deﬁnedasaprinciplethatisbasedonthenormativestandpoint thatpriorityinhealthcareshouldbegiventothosewho,dueto somediseaseandifleftuntreated,losethelargestproportionof theirQALYexpectancyinabsenceofthedisease[5,32,33].PSis measuredonascalefrom0(noQALYloss)to1(completelossof remainingQALY,i.e.immediatedeath),byapplying:

PS= Disease-relatedQALYloss

RemainingQALYexpectationinabsenceofthedisease (1) Forexample,adiseasethatresultsinthelossof30outof60 remainingQALYshasaPSlevelof0.5(30/60),andadiseasethat resultsinthelossof60outof80remainingQALYshasaPSlevel of0.75(60/80).TheremainingQALYexpectationinabsenceofthe diseasecanbecalculatedfromage-andgender-speciﬁcmortality data[25].Eq.(1)canberewrittenas:

PS=1− ExpectedQALYswithouttreatment

(3)

Table1

Maximumreferencevalues(inD)perQALYgained[25].

Burdenofillness MaximumreferencevalueperQALYgained 0.10–0.40 D20,000

0.41–0.70 D50,000 0.71–1.00 D80,000

ApplyingEq.(2)tothepreviousexample,thePSlevelof0.5is calculatedas1–(30/60),andthePSlevelof0.75iscalculatedas1– (20/80).PScanalsobecalculatedbyusingthenumberofexpected QALYs‘withcurrenttreatment’ratherthan‘withouttreatment’in theequations[34].ThismaybeamorelogicalcalculationofPS asitarguablyusesamorerelevantcomparatorandhenceagrees withtheeconomic-evaluationmethodology.However,itshouldbe notedthatcalculatingPSrelativetothecurrenttreatmentwilllikely leadtoadifferent,specificallylower,PSlevelforthesame benefi-ciariesand(new)healthtechnologies.Consequently,theoutcome ofareimbursementdecisionthatisinformedbyaPSlevelthatis calculatedrelativetothecurrenttreatmentmaybedifferentfor thesamebeneficiariesand(new)healthtechnologiesthanwhen thedecisionisinformedbyaPSlevelthatiscalculatedrelativeto havingnotreatment.Thedebateonthepreferredcomparatoris likelytocontinueinthecomingperiod.

While consensus concerning thedefinition and operational-isationof BOI graduallyincreased, its exact categories and the associatedcost-effectivenessthresholdrangeremainedasubject ofdiscussionforsometime.In2006,theCouncilforPublicHealth andSociety(RVZ)suggestedacontinuous,upward-slopingcurve withamaximumreimbursementofD80,000/QALY[29].This fig-urewassubstantiatedbytheWorldHealthOrganisation(WHO) ruleofthumbthatlessthanthreetimestheGDPpercapitaper disability-adjustedlife-year(DALY)avertedindicatedgoodvalue formoneyforahealthtechnology[35],bythefindingthatmost reimbursedhealthtechnologiesintheUKhadanICERof approx-imatelyD79,000/QALY[36],andbyestimationsofthevalueofa statisticallife[37,38].AlthoughthefigureofD80,000/QALYmay havebeensetsomewhatarbitrarily,itwasconsidered“reasonable” [29,39].Moreover,eventhoughD80,000/QALYwasnotofficially adoptedasthethresholdvalueatthattime,itwasinfluentialand providedthebasisforZiNtosetthreeBOIcategorieswitha maxi-mumreimbursementofD80,000/QALYforthehighestBOIcategory in2015[25].Table1presentsthesethreeBOIcategoriesandthe associatedmonetaryreferencevaluesandshowsthatahigherWTP perQALYgained,i.e.ahighernecessityscore,isattachedtohealth technologiesthattargetdiseases witha higherBOIlevel [2,25]. Forexample,theICERof ahealthtechnologythat targetsa dis-easewithaPSlevelof0.5isevaluatedagainstareferencevalueof D50,000/QALY,whiletheICERofahealthtechnologythattargets adiseasewithaPSlevelof0.75isevaluatedagainstareference valueofD80,000/QALY.Ahealthtechnologythattargetsadisease withaPSlevelbelow0.1,inprinciple,notconsideredfor reim-bursement.Hence,thiscategoryisnotincludedinthetable[25,29]. Whetheritisfeasible,inpractice,tonotreimburseahealth tech-nologythattargetsadiseasewithalowPSlevelremainsdoubtful [40].Forexample,episodicillnesseslikemigrainemaynotleadto ahighaveragePS,butdorepresentsubstantialshortfallduringthe episode.

GiventhemaximumreimbursementofD80,000/QALYforthe highestBOIcategoryandtheintentiontoassociateincreasing lev-elsofBOIwithincreasingmonetaryreferencevalues,ZiNsetthe twolowerthresholdsatD20,000andD50,000perQALY.Together thesemaybeseenasformingalogicalsetofvalues,giventhe end-pointofD80,000/QALYinrelationtothehighestBOI.Inrelationto theothertwovalues,ZiNalsoreferredtothethresholdvaluethatis

appliedtonationalimmunisationandpreventivecareprogrammes intheNetherlands(D20,000/QALYthreshold)andtoaDutchstudy onthesocietalWTPperQALYgained‘inothers’(D50,000/QALY) [25,35].ZiNadvisedreassessingthereferencevalueseveryﬁveto tenyears [25] andtonot usethemas strictcut-offvalues, but ratherasreferencesfortheDutchgovernmentwhenconducting pricenegotiationswithpharmaceuticalcompaniesandfortheACP whenrecommendingincorporationofhealthtechnologiesintothe publichealth-insurancepackage.

Themodel,inwhichBOIisquantiﬁedandtheICERsofhealth technologies are evaluated against associated reference values, enablesatransparentandcoherentdecision-makingframework. GiventhatthismodelisincreasinglyappliedintheNetherlands,the questionariseswhetherthereactuallyissufﬁcientsupportforthe operationalisationofBOIintermsofPStoexplicatetheequity cri-terion.Inthenexttwosectionswewilldiscussthetheoreticaland empiricalsupportforusingPStoinformprioritysettingin health-care.Inthesubsequentsectionwewillreviewthecurrentroleof PSinhealthcaredecisionmakingintheNetherlands.

3. Istheretheoreticalsupportforproportionalshortfall? Inordertooptimallyallocatehealthcareresourcesforsociety, ithasbeenadvocatedthatsocietalconcernsregardingequitybe incorporated in thedecision-making framework [10–12]. How-ever,whatsocietyconsiderstobeequitableandfairforpriority settingisanormativequestionthatdifferentpeopleindifferent contexts mayanswer differently. Hence,when operationalising the equity-efficiencytrade-off, an additional trade-off between differentsocietalconcernsregardingequitymustbemade. Con-sequently,increasingequalityinthedistributionofhealth(care) byapplyingoneequityapproachmayleadtoincreasinginequality inthecontextofapplyinganother[3,41].Ithasalsobeenargued that,whenoperationalisingtheequity–efficiencytrade-off, differ-entoperationalisationsareboundtofacecorrespondingdifficulties [3,16].For example,inthecontext ofcurativehealthcare, ques-tionsmayariseconcerningthehandlingofepisodicdiseasesand thequantificationofrelatedhealthbenefits[3,16].Inthecontext ofpreventivehealthcare,questionsmayariseconcerningthegroup ofbeneficiariesandthetimeframethatisregardedasrelevantfor estimating health benefits[3,16,18]. For example,should PSbe calculatedforalltreatedpersonsoronlythoseforwhomthe ill-nesswasprevented?AndshouldPSbecalculatedfromthetime ofthepreventivetreatmentorfromthetimethepreventedillness wouldotherwisehaveoccurred?Suchchoicescanhaveaprofound effectontheoutcomesofPScalculations[42].Otherquestionsmay, forexample,ariseconcerningtheuseofage-andgender-specific mortalitydataasareferencepointorthresholdforcalculatingPS [20,43,44].Theuseofsuchdifferentreferencepointsfor differ-ent(ageandgender)groupsimpliesthatthereisnotoneageor healthexpectancythatwouldserveasanormativereferencelevel for allgroups.Hence,this couldberegarded asincludingsome inequ(al)itiesinthecalculationofPS[20].Theseissuesillustrate thatnotonlyisthechoiceofanequityapproachnormative,but additionalnormative choices mustbemadewhen applyingthe chosenequityapproachinpractice[3].Inevitably,thesechoices have alargeimpactonPScalculationsand thereforemayhave distributionalconsequences[3,18].Althoughsomeinitialchoices weremadewhenoperationalisingPSintheNetherlands[1,2,31],it shouldbenotedthatthediscussionabouthowbesttosolvethese issuesisongoing(bothinthecontextofhealthcareprioritysetting intheNetherlandsandinternationally).

SOIandFIaretworenownedequityapproachesthatarebased ondifferentnormativeargumentsregardingwhomisconsidered worseoffintermsofhealth[3,19].Asdescribedearlier,SOI

(4)

com-Table2

Empiricalevidenceforsupportofproportionalshortfall(published2001–2017).

Study Year Country Design Modeofadministration N Sample SupportforPSa

Stolketal.[33] 2005 NL Rankingexercise Web-basedsurvey 65 Convenience ++ Olsen[50]b ₂₀₁₃ _Norway _{Pairwise-choice}_task _Web-based_survey ₅₀₃ _General_public_(age_and_gender) ₋₋

Brazieretal.[51] 2013 UK DCE Web-basedsurvey 3669 Generalpublic(ageandgender) −−/− VandeWeteringetal.[53] 2015 NL DCE Web-basedsurvey 1205 Generalpublic(age,gender,andeducationlevel) −− Bobinacetal.[54] 2015 NL WTP Web-basedsurvey 1320 Generalpublic(age,gender,andeducationlevel) − Rowenetal.[52]c ₂₀₁₆ _UK _DCE _Web-based_survey ₃₆₆₉ _General_public_(age_and_gender) ₊

Richardsonetal.[55]b ₂₀₁₇ _Australia _CSPC_task _Web-based_survey ₆₀₆ _General_public_(age) ₊

Notes:CSPC=constantsumpairedcomparison;DCE=discretechoiceexperiment;WTP=willingness-to-paystudy;PS=proportionalshortfall. a_Level_of_support_for_PS_indicated_by₋₋₌_no,₋₌_limited,₊₌_modest,₊₊₌_strong.

b _Olsen_[₅₀_]_and_Richardson_et_al._[₅₅_]_examined_support_for_PS_in_the_context_of_preferences_for_length_of_life. c _Rowen_et_al._[₅₂_]_examined_support_for_PS_relative_to_the_current_treatment.

monlyaimstoequalise healthbeneﬁtsin termsof current and prospectivehealth,andFIaimstodosointermsoflifetimehealth [3,10,11,19].Assuch,FIisconsistentwiththenotionthat,allelse equal,youngerpeopleshouldbeprioritisedoverolderpeopleas theyhavenotyetenjoyedafairshareoflifetimehealth[5,11].It shouldbenotedthattheroleofageismerelyindirectintheFI approachasitisappliedasaproxyforlifetimehealth[20,43,44]. Indeed,intheFIapproach,ageitselfisnotregardedasamorally relevantargumentforprioritysetting[44].

PSdoesnotoriginatefromauniquetheoryaboutdistributive justicebutwasdevelopedasanequityapproach thatcombines aspectsofSOIandFIbyprioritisingthosewho areworseoff in termsofaloweramountofprospectiveandlifetimehealth[3,5]. WhileSOIandFIaimtoequaliseabsolutehealthbenefits,PSaims toequaliserelativebenefitsbetweenpersonswithrespecttotheir potentialforhealth[5,33].IthasbeenarguedthatPSbalances soci-etalconcernsregardingSOIandFIandtreatsthetwoapproaches asequallyimportant[5].However,PSiscalculatedasthefraction ofdisease-relatedQALYlossrelativetotheremainingQALY expec-tationinabsenceofthediseaseratherthantothelifetime-QALY expectationfrombirth.Variousauthorshavediscussedtherelative natureofPSandthetheoreticalandempiricalrelevanceofusing alifetimeperspectiveforinformingallocationdecisionsin health-care[20,24,45–47].Here,wewouldliketopointoutthatPSmay beviewedasplacingmoreemphasisonrelativeprospective-health loss,i.e.theSOIcomponentofPS,thanonrelativelifetime-health loss,i.e.theFIcomponentofPS.ThisisillustratedbythefactthatPS doesnot,bydefinition,discriminatebetweenpeoplewithdifferent levelsof‘enjoyed’lifetimehealthashealthcarebeneficiariesofall agescouldpotentiallyexperiencethesamelevelofPS.For exam-ple,inthecaseofimmediatedeath,healthcarebeneficiarieswho are10and80yearsoldaregiventhesameweightinthe distribu-tionofhealthcare,asbothwillhaveaPSlevelof1.However,when thesamebeneficiarieslosetwooftheirremainingQALYs, more weightwillbegiventothe80yearolds,astheirPSlevelwillbe higherthanthatofthe10yearolds.Indeed,inallocationdecisions, PSmaymorefrequentlygiveahigherweighttoolderpatientsthan theFIapproachwould.Stolketal.[5]arguedthattheFIapproach “discriminatesagainsttheelderlymorestronglythanpolicy mak-ersseemtoprefer”andthatPScouldmitigatetheageismthatis impliedbytheFIapproach.Itwas,therefore,hypothesisedthatPS mightbebetteralignedwithdistributionalpreferencesofhealth policymakers.Shouldthishypothesisnotbesupportedby empir-icalevidence,theauthorssuggestedtoaddageweightsandadjust PSforage-relatedpreferences

AstrengthofPS,whichitshareswiththeSOIandFIapproaches, liesinitsquantiﬁcationofhealthlossesintermsofQALYs.This enablesthe application of PS across disease areas and patient populations.However,this strength comeswitha limitationas treatmentbeneﬁtsbeyondhealthandhealth-relatedqualityoflife (QOL)thatmaynotbecapturedbytheQALYareincreasingly

recog-nisedasbeingrelevant[48].Therefore,thecurrentapplicationof PS,i.e.itsquantificationintermsofQALYs, mayberegardedas appropriateforinformingdecisionsconcerningcurativeand pre-ventivetreatmentsbutlesssofordecisionsconcerningtreatments thatfocusonbroaderbenefits,forexamplerelatedtowellbeing [49].Iftheaimistogeneratesocialwelfarefromthepublic health-insurancepackage,theapplicationofanequityapproachthatis uniformlyapplicableandhencethatmodelsinformation concern-inghealth, QOL, and broaderwellbeing couldbe preferablefor informingdecisionsconcerningallhealthcaresectors.Westress thatthislimitationshouldnotbeattributedtoPS(ortoFIorSOI) asaprinciplebutrathertothewayinwhichPSiscurrently quanti-fiedandappliedindecision-makingpractice.Infact,PSdoesenable auniformdecisionmodelforprioritysettingacrossallhealthcare sectors,astheQALYinthePSequationcanbereplacedwith—or complementedby—anyother(generic)outcomemeasureofchoice.

4. Isthereempiricalsupportforproportionalshortfall? WeexaminedempiricalsupportforPSbyreviewingthe inter-national literature on PS in the context of healthcare priority setting.Weusedthesearchterms“proportionalshortfall”, “prefer-ence”,“elicitation”,“prioritysetting”,and“health”or“healthcare” inGoogleScholar.ThesearchwasperformedonOctober16,2017 andsupplementedwithahandsearch.Werestrictedthesearchto articleswritteninEnglishorDutch,publishedbetween2001,i.e. theyearinwhichPSwasintroducedintheNetherlands,and2017, andofwhichthefulltextwasavailable.Articleswereselectedfor reviewiftheaimofthestudywastoelicitpreferencesforPS rela-tivetoeitherpreferencesfornoequityweightingorforweighting onthebasisofanotherequityapproach,suchasSOIand/orFI.Our searchresultedin205studies,insevenofwhichpreferencesforPS wereelicited.Table2presentsanoverviewofthesesevenstudies andtheirresults.

Stolketal.[33]comparedsupportforSOI,FI,andPSbyasking respondentstoassignapriorityranktothetreatmentoftenhealth conditions.Stolketal.foundstrongevidenceforPSbeingconsistent withsocialpreferencesfor healthcareprioritysetting.Although preferencesforPSdominatedpreferencesforSOI,strongersupport wasfoundforFI.Theauthorsobtainedtheseresultsusingasmall conveniencesample intheNetherlands thatconsistedofhealth policymakers,researchers,andstudents.Consequently,theresults maybepronetobias,e.g.due torespondentssharingcommon opinions.

Olsen [50] examined support for PS in a sample that was representative,intermsofageandgender,ofthegeneraladult populationinNorwayintermsofageandgender.Olsenapplieda pairwise-choicetaskandaskedrespondentstoprioritisepatients basedontheirage,remaininglifetimehealthwithouttreatment, andincreaseinremaininglifetimehealthwithtreatment.Olsen

(5)

foundstrongsupportfortheFIapproach;however,hefoundno supportforPS.

Brazieretal.[51]examinedsupportforBOIoperationalisedin terms ofPSin asample that wasrepresentativeof thegeneral populationintheUKintermsofageandgenderbyperforming aweb-baseddiscretechoiceexperiment(DCE).Theirmainresults didnotsupportPS.However,whenrespondentswhoseemingly misunderstoodtheDCEtaskwereexcluded,somesupportforPS wasfound.

Rowenetal.[52]usedthedatafromBrazieretal.[51]toexamine supportforPSbyapplyingthenumberofexpectedQALYs‘with cur-renttreatment’ratherthan‘withouttreatment’inthePSequation. Rowenetal.concludedthat,althoughtheresultswerenotrobust againstdifferentversionsoftheDCEsurvey,therewassome mod-estsupportforBOIoperationalisedintermsofPSrelativetothe currenttreatment.

VandeWeteringetal.[53]examinedsupportforPSina sam-ple that was representative of the general adultpopulation in theNetherlandsintermsofage,gender,and educationlevelby conductingaweb-basedDCE.Theyfoundsubstantialpreference heterogeneityandsomecounterintuitiveresults,asrespondents werelesslikelytoprioritisepatientswithhigherlevelsofPS.

Bobinacet al.[54]examinedsocietalWTPfor QALYgainsin patientswithdifferentlevelsofPSinasamplethatwas represen-tativeofthegeneraladultpopulationintheNetherlandsinterms ofage,gender,andeducationlevelbyconductingaweb-based sur-vey.Theyfoundoccasional supportforPSasa predictorofthe WTPforQALYgains.SomesupportforPSwasfoundwhenQALY gainswererelativelysmall.However,thelevelofsupportdecreased whenQALYgainsincreasedinsize.Inaddition,supportforPSwas generallydominatedbyconcernsforthe(younger)ageofpatients. Richardsonetal.[55]examinedsupportforPSinasamplethat wasclosetobeingrepresentativeofthegeneraladultpopulationin Australiaintermsofage.Theyappliedconstant-sumpaired com-parisontasksandaskedrespondentstoprioritisepatientsbasedon theirgaininlifeyearsduetotreatment,age,yearstodeath with-outtreatment,andageatdeathwithandwithouttreatment.Their studyfoundsomesupportforPS;however,foundthatconcerns forPSweredominatedbyconcernsforthe(individual)SOIandFI approaches.Richardsonetal.furtherfoundthatPSinsufﬁciently reﬂectsrespondents’age-relatedpreferences.

Althougheach ofthesestudiesexaminedsocietalsupportby elicitingpreferencesforPS,itisimportanttonotethatthestudies differwithrespecttothesamples,methods,additionallyincluded variables,and/orequity approaches.In addition,Olsen[50] and Richardsonetal.[55]examinedpreferencesforequityinthe con-textoflengthof life,and hencedidnotpresent PSinterms of proportionalQALYshortfall.Consequently,adirectcomparisonof theresultspresentedinTable2isnotpossible.

5. Whatistheroleofproportionalshortfallinhealthcare decisionmaking?

Thenecessity (ofcare and ofinsurance), effectiveness, cost-effectiveness,andfeasibilitycriteriaareaddressedandquantiﬁed in the assessment phase of healthcare decision making in the Netherlands and subsequently assessed on social and ethical groundsintheappraisalphase.IftheoperationalisationofBOIin termsofPSisconsideredsuboptimalforexplicatingtheequity cri-terion,itseemsreasonabletoexpectthatthiswouldbeexplicitly discussedduringmeetingsoftheACPappraisalcommittee.

Toexamine thecurrent role of PSin theappraisal phase of healthcare decision intheNetherlands, we conducteda review ofpublicly accessibleACPmeeting reports thatwere published between January1, 2013 and December 31, 2016. The reports

Table3

SearchtermsusedforreviewingACPmeetingreports.

Domain Searchterm Priority-settingcriteria Necessityofcare

Necessityofinsurance Effectiveness;Effect Cost-effectiveness;Efﬁciency Feasibility

Equityconsiderations Severityofillness Fairinnings Burdenofillness Absoluteshortfall Proportionalshortfall

Treatmentbeneﬁts Therapeuticoutcome;Therapeuticvalue (Health-related)qualityoflife Quality-adjustedlife-year;QALY Wellbeing

Capability Lifesatisfaction Patientcharacteristics Age

Socio-economicstatus;SES Lifestyle

Culpability;Individualresponsibility Referencevalues Referencevalue(s)

(Monetary)threshold

include agendas, minutes, and documents, including decision reportsanddraftZiNreportsthatwerediscussedbytheACP.Table3 presentstheterms(andtheirdomains)addressinghealthcare pri-oritysettingthatweusedforsearchingthereports(intheDutch language,buttranslatedhereforclarity).Reportsthatdidnotallow adigitalsearch,includingACPreportsthatwerepublishedbefore January1,2013,wereexcludedfromthereview,asweresearch termsthatoccurredinthenamesofhealthorganisationsand gov-ernmentministries.Draftversionsofminuteswereincludedonly ifﬁnalminuteswerenotpublished.

Between2013and2016,179ACPreports werepublishedof whichtwowereexcludedfornotallowingadigitalsearch.Table4 presentsthefrequencywithwhichthesearchtermswere iden-tifiedintheremaining177reports.Thenecessityofcareandof insurance,effectiveness,cost-effectiveness(including thesearch termefficiency),andfeasibilitycriteriawereidentified1680,495, 8700,4423,and236times,respectively.Theeffectivenesscriterion wasmostfrequentlyfound,followedbythecost-effectivenessand necessityofcarecriteria.Thenecessityofinsuranceandfeasibility criteriawereidentifiedlessfrequently.

PSwasidentified14timesinatotalofsixreports,fourofwhich discussedtheoperationalisationofBOIintermsofPS.Inareport from2013,ZiNdescribedthedefinitionandcalculationofPS.Inthis report,ZiNstatedthat“PSwasdevelopedatatimewhenageism wasanimportantissueintheallocationofhealthcareresources” andthat“thereforeBOIiscalculatedinproportiontolife expecta-tion,whichensuresthatPSdoesnotdistinguishbetweenyounger andolderpeople”.However,ZiNalsostatedthat“recently,there are increasingindications that peopledo discriminate between agegroups”andthatpeople“valuehealthgainsinyounger peo-plemorethaninolderpeople”,which“arguesagainstPSandthe ruleofrescue,andinfavourofFI”.Intheaccompanyingminutes, anACPmemberstatedthat“thepassageaboutBOIisstillnotin agreementwithwhatwasdiscussedin previousmeetings”and thats/he“understand[s]thatapplyingthecapabilityapproachis outofreach”,butthat s/he“wouldliketoseethedenominator removedfromthepresented definitionofPS”.In a reportfrom 2015,ZiNstatedthat“becausewehavenotyetreachedconsensus aboutthequantificationofBOI,wewilltemporarily[...]quantify BOIintermsofDALYs”.Inthisreport,ZiNadditionallystatedthat “prioritywillbegiventosolvingthisissue”andthat“areporton thequantificationofBOIwillbeissuedthis summer”.In alater

(6)

Table 4 Frequency and mean frequency per report of terms addressing healthcare priority setting in ACP meeting reports, published between 2013 and 2016 (n = 177). Year Type N Necessity of care Necessity of insurance Effectiveness Cost- effectiveness Feasibility SOI FI BOI AS PS QOL Wellbeing Age SES Lifestyle Reference values 2013 Agenda 9 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Minutes a 9 10 2 53 28 1 0 0 6 0 1 10 1 2 1 2 0 Documents b 35 357 121 1820 572 20 0 8 191 0 5 339 28 323 22 137 6 2014 Agenda 7 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Minutes 7 6 2 50 46 1 0 0 0 0 0 6 0 0 1 1 0 Documents b 31 378 74 3626 1428 67 0 0 91 0 0 846 30 246 14 183 31 2015 Agenda 8 1 0 0 2 0 0 0 0 0 0 0 0 0 0 0 0 Minutes 8 10 4 104 84 1 0 1 11 0 0 46 4 9 0 1 12 Documents b 34 239 109 1781 1673 72 0 2 237 0 7 725 12 119 6 23 194 2016 Agenda 6 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Minutes 6 18 4 103 73 2 0 0 17 0 0 35 3 6 0 2 19 Documents b 17 659 179 1163 517 71 0 0 119 0 1 381 5 190 10 37 66 Frequency 1680 495 8700 4423 236 0 11 672 0 14 2388 83 895 54 386 328 Number of reports 177 112 64 121 120 55 0 4 68 0 6 89 28 79 17 43 26 Mean frequency per report 15.0 7.7 71.9 36.9 4.3 0 2.8 9.9 0 2.3 26.8 3.0 11.3 3.2 9.0 12.6 Notes : ACP = Adviescommissie Pakket (the healthcare appraisal committee in the Netherlands); AS = absolute shortfall; BOI = burden of illness; Cost-effectiveness = pooled result of “cost-effectiveness” and “efﬁciency”; Effective-ness = pooled result of “effect”, “effectiveness”, and “therapeutic outcome/value”; FI = fair innings; Lifestyle = pooled result of “lifestyle”, “individual responsibility”, and “culpability”; PS = proportional shortfall; QOL = quality of life, pooled result of “(health-related) quality of life”,“quality-adjusted life-year”, and “QALY”; Reference values = pooled result of “reference value(s)” and “(monetary) threshold”; SES = socio-economic status; SOI = severity of illness; Wellbeing = pooled result of “wellbeing”, “capability”, and “life satisfaction”. a Minutes for 2013 include one concept version. b Documents include meeting and decision reports.

reportfrom2015,ZiNstatedthat“thenextcomingmonthswillbe usedtoseehowtobetteraligntheequitycriterionPSwithcurrent socialpreferences”.Accordingtotheaccompanyingminutes,these statementsbyZiNwerenotdiscussedbyACPmembers.

ThereferenceofoneoftheACPmemberstoSen’scapability approach [41]may indicatea preference forquantifying health benefitsintermsofbroaderwellbeing,asforexampleisdoneby applyingtheICECAPmeasure[56].Wellbeing,includingtheterms capabilityandlifesatisfaction,wasidentified93timesin29reports from2013onwards,among whichthecapability approach was identified15timesinfivereports(notinTable4).Inthesereports, thecapabilityapproachwasdiscussedasanalternativeto quanti-fyinghealthbenefitsintermsofQALYs.Inareportfrom2013,ZiN statedthat“arecentdiscussioninvolvesthequestionofwhether thecapabilityapproachisbetteralignedwiththesocialbasisthat underliesmanagingthepublichealth-insurancepackage”andthat “applyingthisapproachmaybemoreappropriateforhealthcare sectorswhere‘healthgains’arenottheprimaryobjective,suchas long-termcareandmentalhealthcare”.Thesamereportstatedthat “changingthedesiredoutcomeofhealthcaredoesnotanswerthe questionofwhencareismorenecessaryforonepersonthanfor another”andthat“thecapabilitiesapproachcanalsobeapplied tocalculatelifetimecapabilities(fairinnings),prospective capa-bilities,ortherelativelossofcapabilities(proportionalshortfall)”, andso“applyingthecapabilityapproachwillnotsolvetheissueof prioritisationinhealthcare”.

TheACPmember’srequesttoremovethedenominatorfrom thePS equationmay indicate a preference for operationalising theequitycriterionintermsofabsoluteshortfall(AS)ratherthan proportional shortfall, and this may in turn indicate a prefer-enceregarding FI,age,and reducinglifetime-healthinequalities [23,50].ASwasnotidentifiedinanyoftheACPreportsandtheFI approachwasidentified11timesinfourreports.Incontrast,the SOIapproachwasidentified0times.However,concernsforSOIthat wereexpressedthroughconcernsforprospective-healthloss,the ruleofrescue,andfortheseverityof(symptomsof)adiseaseor con-ditionwereidentified3timesin1report,5timesin3reports,and 2614in92reports,respectively(notintable).Agewasidentified 895timesin79reports.Regardingageandotherpatient charac-teristics,agewasidentified16.6and 2.3timesmorefrequently thanSESandlifestyle(includingthesearchtermsculpabilityand individualresponsibility),respectively.

AlthoughtheoperationalisationofBOIintermsofPSwas occa-sionallydiscussedinsomereports,andinonereportfrom2014 anACPmemberstatedthat“BOIcannotbequantifiedinnumeric termsinthisspecificsituation”,theapplicationofBOIitselfwas notdiscussed.BOIwasidentified672timesin68reports,andfrom 2015onwards,increasinglyrelatedtothecorrespondingreference values(presentedinTable1),whichwereidentified328timesin 26reports.In2013,BOIwasmostfrequentlyexpressedin qualita-tiveterms,e.g.intermsof“low”or“high”BOI,andonlysometimes innumericterms,bydisabilityweightsusedforcalculatingDALYs. From2014onwards,BOIwaslessfrequentlyexpressedin qualita-tivetermsandwasmostlyquantifiedbydisabilityweightsorthe numberofDALYslost,whichattimeswerepresentedalongsidethe meanlifeexpectancyofpatientswithandwithoutthedisease.In threereports,BOIwasaddressedasarelativemeasure;however, thepresenteddisabilityweightsorDALYslostwerenotappliedas such.From2015onwards,BOIwasmostfrequentlyquantifiedin termsofPS,inatotalofsevenreports.Infourofthesereports,PS calculationswerepresentedalongsidedisabilityweights,number ofDALYslost,meanlifeexpectancy,oryearsoflifelostcalculations. Basedontheseresults,it appearsthattheapplicationofBOI wasnotpubliclydiscussedbyACPmembersbetween2013and 2016.Theoperationalisation ofBOIintermsofPS, andtherole ofPSinhealthcaredecisionmaking,wasinfrequentlydiscussed.

(7)

WhileBOIwasmostfrequentlyexpressedqualitativelyin2013,it wasincreasinglyquantifiedinlateryears,usuallyintermsof dis-abilityweightsorthenumberofDALYslost.From2015onwards, ICERswereincreasinglyrelatedtothemonetaryreferencevalues perQALYgainedthatweresetbyZiNthatyear[25]andBOIwas mostfrequentlyquantifiedintermsofPS.Inthiscontext,itneeds notingthattherewasachangeinACPmembersin2015andthis mayhavecontributedtotheincreasedapplicationofPSfromthen on.Inthereports,PScalculationswerefrequentlypresented along-sidedisabilityweights,numberofDALYlost,lifeexpectancy,and yearsoflifelosscalculations.Thisvarietymayreflectthatthereisno consensus(yet)abouttheapplicationofPSinhealthcaredecision makingintheNetherlands.

6. Discussion

Theimportanceofoperationalisingtheequity-efficiency trade-off for informing priority setting in healthcare is increasingly recognised.ThisstudydescribedthehistoryandapplicationofPSin theNetherlands,examinedthetheoreticalandempiricalsupport forPSasanoperationalisationoftheequity-efficiencytrade-off, andlookedintothecurrentroleofPSinhealthcaredecisionmaking. Overall,ourresultsindicategeneralsupportfor thedecision modelinwhichBOIisquantifiedandtheICERsofhealth technolo-giesareevaluatedagainstthereferencevaluesperQALYgained setbyZiN.Thismodelconnectsthecriteriaforhealthcaredecision makingthatwerepreviouslyformulatedbytheDunning Commit-teeandenablesauniformdecisionmodelforprioritysettingacross allhealthcaresectors.Consequently,themodelhasreceivedbroad supportinresearchandpolicycirclesandhasbeenincorporated intothehealthcaredecision-makingframeworkintheNetherlands. Althoughthemodelisincreasinglyappliedindecision-making practice,theresultsofourliteraturereviewsuggestthat theoreti-calsupportforPSismoderateatbest.InapplyingPS,atrade-off betweentheSOIand FI approachesis madeand, consequently, societalpreferencesforeitherofthetwoequityapproachesmay be insufficiently reflected when allocating resources in health-care.However,thismayberegardedasagenerallimitationthat comeswithapplyinganyequityapproachinpractice,asimproving equalityinthedistributionofhealth(care)byapplyingoneequity approachmayinevitablybeassociated withincreasing inequal-ityinthecontextofapplyinganother[3].Alimitationthatisnot restricted,butmaybemorespecific,toapplyingPS,isthatit miti-gatesageismbetweenpatientgroups,asbeneficiariesofallagescan experiencethesamelevelofPS.Theresultsofourstudysuggestthat thismayinadequatelyreflectsocietalpreferencesrelatingtoage andreducinglifetime-healthinequalitiesbetweenpatientgroups. Althoughtheresultsofourliteraturereviewsuggestthat empiri-calsupportforPSislimited,itshouldbenotedthatthatempirical evidenceregardingPSsofarisscarceandinconclusive,sothata rejectionofthePSapproachcanalsonotbeconcluded.Thesocietal concernregardingagethatiscurrentlyinsufficientlyreflectedby PScouldbeincorporatedbyadjustingPSforage.However,there isnoempiricalevidence(todate)tosupportthehypothesisthat thiswouldbetteralignwithsocietalpreferencesandhencefuture researchonthistopicwillbenecessary.Theresultsofourreviewof publiclyaccessibleACPreportssuggestthattheACPdidnotpublicly discussthedefinitionandoperationalisationofnecessityofcarein termsofBOIbetween2013and2016.Infact,BOIbecame increas-inglyquantified,andICERswereincreasinglyevaluatedagainstthe referencevaluesperQALYgainedsetbyZiNin2015.The opera-tionalisationofBOIintermsofPSwaspubliclydiscussedbythe ACP,althoughonlyonrareoccasions.Thismayindicatethatthe ACPsupportstheoperationalisationofBOIintermsofPS.However, thevarietyofBOIquantificationsinACPreportsdemonstratesthat

consensusabouttheoperationalisationandquantiﬁcationofBOI hasnotyetbeenreached.

Relativelyfewstudieshaveexaminedthetheoreticaland empir-ical supportfor PS, and to ourknowledge, noother study has examinedthecurrentroleofPSinhealthcaredecisionmakingin theNetherlands.Althoughthislimitsourabilitytocompareour resultswiththoseofothers,wewouldliketocomparetheresults ofourempiricalliteraturereviewtotheresultsofastudyconducted byNordandJohansen[57]andthepublicconsultationofNICEon thevalued-basedassessmentofhealthtechnologies[58–60].Nord andJohansen[57]examinedsupportforPS,relativetopreferences fornootherequityapproach,byconductinganempiricalliterature reviewthatbuiltonanearlierreviewbyShah[61]andincluded 20preferencestudiesthatwereconductedinninedifferent coun-triesbetween1991and2011.Undertheassumptionofastable healthconditionandnolossinlengthoflifeforpatients,Nordand JohansenfoundstrongsupportforPS,althoughthestrengthofthe supportvariedgreatlybetweentheincludedstudies.Regardingthe inclusioncriteriaforourliteraturereview,fiveofthestudiesthat NordandJohansenincludedwereconductedafterthe introduc-tionofPSintheNetherlands.Ofthesefivestudies,twoquantified healthbenefitsintermsofQOLandthreeintermsofQALYs. How-ever,noneof thethree latterstudieselicited preferencesforPS and,assuch,werenotincludedinourliteraturereview.Although theresultsofourreviewseem tobediscordantwithNordand Johansen’sresults,anditisworthmentioningthatintwoofthe fiveaforementionedstudiesthepublicwasfoundtobelesslikely toprioritisepatientswithhigherlevelsofSOI[45,62],adirect com-parisonofresultsisnotpossibleforreasonsthatarepreviously described.

NICEconductedapublicconsultationintheUKin2014onthe topicofvalue-basedassessmentof healthtechnologies[58–60]. NICEaskedthepublic,includingpatients,economists,academics, andmembersofthepharmaceuticalindustry,tenrelatedquestions. Oneof thequestionsconcernedtheextenttowhich thepublic regardedPSasanappropriateapproachforquantifyingBOI[58]. NICEreceivedreactionsfrom121individualsandorganisations, butnogeneralagreementemerged[59].Insummary,thepublic regardedPSasameasurethatisfeasibleandsuitablefor calculat-ingBOIintermsofhealthandQOLimpactincaseswhereadisease affectsolderpatients.However,asintheACP,therewereconcerns aboutPSnotbeingasuitablemeasureforcapturingbroaderaspects ofBOIthatarenotincludedintheQALY.Inaddition,therewere concernsaboutPSassigningalowerweighttotheBOIofyounger patientsthantoolderpatientsduetodifferencesinthePS denomi-nator,i.e.theremainingQALYexpectationinabsenceofthedisease. Moregenerally,therewereconcernsaboutthestrongrelianceon theQALYinhealthtechnologyassessmentandinthecalculationof BOI,resultinginapossibledoublecountingofbeneﬁtswhen set-tingprioritiesinhealthcare[59].Becauseofthelackofagreement thatemerged fromthepublicconsultation,NICEdecidedtonot changetheircurrenthealth-technologyappraisalandend-of-life guidelines[60].

Somelimitationsofourstudymustbementioned.Aﬁrst limi-tationconcernsthelackofasystematicreviewofthestudiesthat weusedtoexaminethetheoreticalandempiricalsupportforPS. However,asthenumber ofstudiesexaminingPSislimited,we believethatourreviewwascomprehensiveandthatourresults werenotinﬂuencedbythelackofasystematicsearch.Asecond limitationconcernstheuseofpubliclyaccessibleACPreportsto examinethecurrentroleofPSinhealthcaredecisionmakinginthe Netherlands.Inadditiontopublicmeetings,theACPheldclosed meetingsbetween2013and2016,andtheroleofPSmayhavebeen discussedmorefrequentlyinthese.However,thereportsofthese closedmeetingsarenotpubliclyaccessibleandthereforecouldnot beincludedinourreviewofACPreports.AlthoughtheroleofPS

(8)

mayhavebeendiscussedmorefrequentlyduringclosedmeetings, andtheinclusionofclosedmeetingreportsmighthavechangedthe reviewresults,itseemsreasonabletoexpectthatanydiscussionof PSinaclosedmeetingwouldhavebeenreflectedinapublic meet-ingwherethedecisionmakingactuallytookplace.Afinallimitation concernstheriskofdoublecountingsearchtermsduetoapossible overlapinACPmeetingreports.Thisriskwasreducedbyexcluding conceptversionsofminutesunlessafinalversionwasnotavailable; however,thisdistinction couldnotalwaysbemadeformeeting documents.For example,documentsconcerningthe reimburse-mentofaspecifictreatmentmayhavebeendiscussedatmorethan oneACPmeeting.Asaresult,ahigherrelativeimportancemayhave beenassignedtosomeofthesearchterms.However,asthemain objectiveofourreviewofACPreportswastoexaminetheroleof PSinhealthcaredecisionmaking,andPSwasinfrequently iden-tified,webelievethattheinfluenceofpossibledoublecounting onconclusionsdrawnfromthereviewislimited.Concerningthe frequencywithwhichthesearchtermswereidentifiedintheACP meetingreports,wewouldliketoadditionallypointoutthatthese frequenciesshouldbeconsideredinthebroadercontextof prior-itysettingintheNetherlands.Thisbroadercontextdeterminesthe agendaandtheprioritiesthataresetinACPmeetingsandhence influencesthefrequencywithwhichthesearchtermswere identi-fied.Apartfromtheselimitations,weconsideritastrengthofour studythatwehaveexaminedsupportforPSatthreedifferent lev-els,i.e.atatheoretical,empirical,anddecision-makinglevel.Toour knowledge,thisisthefirststudytoexaminesupportforPSinsuch anextensivemanner.

Ideally,theoperationalisationoftheequity-efficiencytrade-off arenormativelyjustifiableandempiricallysupported.Thevarious normativechoicesthatneedtobemadeinthiscontextindicatethat atrade-offoracompromisebetweendifferentsocietalconcerns regardingequityandfairnessneedstobemade.Consequently,the ‘perfect’explicationoftheequitycriterionmaynotexist,andPS, likeanyotherexplication,willhaveitsstrengthsandlimitations. Theresultsofourstudyindicatethatthedecisionmodelinwhich increasinglevelsof BOIare quantifiedand ICERsarerelated to theassociatedmonetaryreferencevaluesperQALYgainedis sup-portedandincreasinglyappliedindecision-makingpractice.The operationalisationofBOIinterms ofPSenablesauniform deci-sionmodel forprioritysetting acrossallhealthcaresectors that canbeappliedby replacingorcomplementing theQALYinthe PSequationwithabroader,wellbeing-related,genericoutcome measuresuchastheICECAP [56].The resultsofourstudyalso indicatethatPSinsufficientlyreflectssocietalpreferences regard-ingageandreducinglifetime-healthinequalitiesbetweenpatient groups.Futureresearchisneededtodevelopandexamine alter-nativeversionsofPS,suchasaversionofPSthatisadjustedfor wellbeing-andage-relatedpreferences.Thesecouldbecompared tothecurrentoperationalisationofPS,alsointermsofalignment withgeneralpublicpreferences.Therearedifferentpossibilitiesfor combiningPSandageinapreference-elicitationstudy.For exam-ple,itmaybeinterestingtoelicitpreferencesforaPSversionin whichthe denominatorof thePSequationconsistsof patients’ lifetime-QALYexpectation.Itmayalsobeinterestingtoelicit pref-erencesfor combinationsofPSand lifetimehealth (orage).For this,amatrixcombiningdifferentageandPSclassescouldbeused. Toincreasecomparabilitybetweenstudies,werecommendusing amorestandardisedapproachtoelicitingequityweights.Related tothis,wewouldliketomentionthatthevarietyofthewaysin whichBOIisquantifiedinACPreportsmay,understandably,reflect thepresentlackofconsensusabouttheapplicationofPS.However, thisvarietyisinitselfundesirableasithampersthetransparency andcomparabilityofBOIquantificationsfordifferentbeneficiaries and,subsequently,oftherelevantcost-effectivenessthresholdof (new)healthtechnologiesintheNetherlands.ZiNisexpectedto

publishareportonthestandardisationoftheBOIquantiﬁcation laterthisyeartosolvethisissue.

7. Conclusions

Theresultsofthisstudysuggestthatthereissupportforthe decisionmodelinwhichBOIisquantifiedandICERsareevaluated againstincreasingmonetaryreferencevalues.Although consen-susregardingtheapplicationofPShasnotyetbeenreached,BOI isincreasinglyquantifiedintermsofPSindecision-making prac-tice.Asany(generic)outcomemeasurecanbeincludedinthePS equation,PSenablesauniformdecision model forpriority set-tingacrossallhealthcaresectors.EmpiricalsupportforPSappears tobelimited,asPSmayinsufficientlyreflectsocietalpreferences regarding age and reducing lifetime-health inequalities.Hence, further investigationinto the refinementof PS—or exploration ofanotherapproach—appearswarrantedforoperationalisingthe equity-efficiencytrade-offinhealthcareprioritysetting.

Conﬂictofinterest

Theauthorshavenoconﬂictofinteresttodeclare. Acknowledgements

Thisstudyis partof a largerprojectexaminingthebroader societalbeneﬁtsofhealthcareandwasfundedbyaconsortiumof Pﬁzer,GlaxoSmithKline,AbbVie,Amgen,and AstraZenecainthe Netherlands.Theviewsexpressedinthisarticlearethoseofthe authors.WethankBertBoerforhishelpfulcommentsonaprevious versionofthearticle.

References

[1]CollegevoorZorgverzekeringen.Breedtegeneesmiddelenpakket.Amstelveen: CollegevoorZorgverzekeringen;2001.

[2]StolkEA, GoesE,KokE,VanBusschbach JJV.Uitwerkingcriteria noodza-kelijkheid, eigen rekening en verantwoording en lifestyle. College voor Zorgverzekeringen. Breedtegeneesmiddelenpakket,1.Amstelveen:College voorZorgverzekeringen;2001.p.54.

[3]VandeWeteringEJ,StolkEA,VanExelNJA,BrouwerWBF.Balancingequity andefﬁciencyintheDutchbasicbeneﬁtspackageusingtheprincipalof propor-tionalshortfall.TheEuropeanJournalofHealthEconomics2013;14(1):107–15. [4]SabikLM,LieRK.Prioritysettinginhealthcare:lessonsfromtheexperiences

ofeightcountries.InternationalJournalforEquityinHealth2008;7(1):4. [5]Stolk EA, Van Donselaar G, Brouwer WBF, Van Busschbach JJV.

Rec-onciliation of economic concerns and health policy. Pharmacoeconomics 2004;22(17):1097–107.

[6]FrankenM,NilssonF,SandmannF,DeBoerA,KoopmanschapM.Unravelling drugreimbursementoutcomes:acomparativestudyoftheroleof pharma-coeconomicevidenceinDutchandSwedishreimbursementdecisionmaking. Pharmacoeconomics2013;31(9):781–97.

[7]NordE.Cost-valueanalysisinhealthcare:makingsenseoutofQALYs. Cam-bridge:CambridgeUniversityPress;1999.

[8]Wouters S, Van Exel NJA, Baker R, Brouwer WBF. Priority to end of life treatments? Views of the public in the Netherlands. Value Health 2015;20(1):107–17.

[9]Maynard A. European health policy challenges. Health Economics 2005;14(S1):S255–63.

[10]NordE.Theperson-trade-offapproachtovaluinghealthcareprograms. Med-icalDecisionMaking1995;15:201–8.

[11]WilliamsA.Intergenerationalequity:anexplorationofthe‘fairinnings’ argu-ment.HealthEconomics1997;6(2):117–32.

[12]NordE,PintoJL,RichardsonJ,MenzelP,UbelP.Incorporatingsocietalconcerns forfairnessinnumericalvaluationsofhealthprogrammes.HealthEconomics 1999;8(1):25–39.

[13]ZorginstituutNederlandRichtlijnvoorhetuitvoerenvaneconomische evalu-atiesindegezondheidszorg.Diemen:ZorginstituutNederland;2016. [14]NationalInstituteforHealthandCareExcellence.Guidetothemethodsof

tech-nologyappraisal.London:NationalInstituteforHealthandCareExcellence; 2013.

[15]CanadianAgencyforDrugsandTechnologiesinHealth.Guidelinesforthe eco-nomicevaluationofhealthtechnologies:Canada,3rded.Ottawa:Canadian AgencyforDrugsandTechnologiesinHealth;2006.

(9)

[16]Bobinac A,VanExelNJA,RuttenFF,BrouwerWBF.Inquiry intothe rela-tionshipbetweenequityweightsandthevalueoftheQALY.ValueHealth 2012;15(8):1119–26.

[17]WeinsteinMC.AQALYisaQALYisaQALY—orisit?JournalofHealthEconomics 1988;7(3):289–90.

[18]Rappange DR,BrouwerWBF,RuttenFF, VanBaal PH. Lifestyle interven-tion: from cost savings to value for money. Journal of Public Health 2010;32(3):440–7.

[19]NordE.Concernsfortheworseoff:fairinningsversusseverity.SocialScience &Medicine2015;60(2):257–63.

[20]OttersenT.LifetimeQALYprioritarianisminprioritysetting.JournalofMedical Ethics2013;39(3):175–80.

[21]NationalInstituteforHealthandCareExcellence.Endoflifecareforadults. London:NationalInstituteforHealthandCareExcellence;2009.

[22]NationalInstituteforHealthandCareExcellence.Endoflifecareforinfants, childrenandyoungpeople.London:NationalInstituteforHealthandCare Excellence;2016.

[23]OttersenT,FørdeR,KakadM,KjellevoldA,MelbergHO,MoenA,etal.A newproposalforprioritysettinginNorway:openandfair.HealthPolicy 2016;120(3):246–51.

[24]NorwegianMinistryofHealthandCareServices.Principlesforprioritysetting inhealthcare:summaryofawhitepaperonprioritysettingintheNorwegian healthcaresector.Oslo:NorwegianMinistryofHealthandCareServices;2017. [25]ZorginstituutNederland.Kosteneffectiviteitindepraktijk.Diemen:

Zorginsti-tuutNederland;2015.

[26]CommissieKeuzenindezorg(DunningA.J.etal.).Kiezenendelen.DenHaag: MinisterieWelzijn,VolksgezondheidenCultuur;1991.

[27]CollegevoorZorgverzekeringen.Pakketbeheerindepraktijk3.Diemen: Col-legevoorZorgverzekeringen;2013.

[28]CollegevoorZorgverzekeringen.Pakketbeheerindepraktijk.Diemen:College voorZorgverzekeringen;2006.

[29]RaadvoordeVolksgezondheidenZorg.Zinnigeenduurzamezorg.Zoetermeer: RaadvoordeVolksgezondheidenZorg;2006.

[30]WetenschappelijkeRaadvoorhetRegeringsbeleid.Volksgezondheidszorg.Den Haag:SduUitgevers;1997.

[31]CommissieCriteriaGeneesmiddelenkeuze(VanWinzumC.etal.).Verdeling doorverdunning.Geldermalsen:StichtingGezondheidszorgenPublicaties; 1994.

[32]CollegevoorZorgverzekeringen.Vervolgonderzoekbreedte geneesmiddelen-pakket.Amstelveen:CollegevoorZorgverzekeringen;1997.

[33]Stolk EA, Pickee SJ,Ament AHJA, Van Busschbach JJV. Equity in health care prioritisation: an empirical inquiry into social value. Health Policy 2005;74(3):343–55.

[34]ZiN.TemplatePharmacoeconomicdossier.Diemen:ZorginstituutNederland; 2016.

[35]WorldHealthOrganisation.ReportoftheCommissiononMacroeconomicsand Health.Jakarta:WorldHealthOrganisation;2002.Availableat:http://apps. who.int/iris/bitstream/10665/127591/1/WP%20Report%20of%20CMH%20-%20SHP.pdf.[Accessed10November2017].

[36]DevlinN,ParkinD.DoesNICEhaveacost-effectivenessthresholdandwhat otherfactorsinﬂuenceitsdecisions?Abinarychoiceanalysis.Health Eco-nomics2004;13(5):437–52.

[37]DayB.Ameta-analysisofwage–riskestimatesofthevalueofstatisticallife.In: CentreforSocialandEconomicResearchontheGlobalEnvironment.London: UniversityCollegeLondon;1999.

[38]ViscusiWK,AldyJE.Thevalueofastatisticallife:acriticalreviewofmarket estimatesthroughouttheworld.NBERWorkingPaperSeries,WorkingPaper 9487.Cambridge,MA:NationalBureauofEconomicResearch;2003. [39]BrouwerWBF.Debasisvanhetpakket:Urgenteuitdagingenvoordeopzet

eninzetvaneconomischeevaluatiesindezorg.Rotterdam:InstituutBeleid& ManagementGezondheidszorg;2009.

[40]CollegevoorZorgverzekeringen.Uitvoeringstoetslageziektelast.Diemen: Col-legevoorZorgverzekeringen;2012.

[41]SenA.Inequalityreexamined.Oxford:OxfordUniversityPress;1992.

[42]RappangeDR,BrouwerWBF.Theevaluationofpreventivelifestyle interven-tionsintheNetherlands.HealthEconomicsPolicyandLaw2012;7(2):243–61. [43]BognarG.Fairinnings.Bioethics2015;29(4):251–61.

[44]BognarG.Prioritysettingandage.In:Prioritizationinmedicine:an inter-nationaldialogue.Switzerland:SpringerInternationalPublishing;2016.p. 163–77.

[45]DolanP,TsuchiyaA.Healthprioritiesandpublicpreferences:therelative importanceofpasthealthexperienceandfuturehealthprospects.Journalof HealthEconomics2005;24(4):703–14.

[46]OfﬁceofHealthEconomics.Clarifyingmeaningsofabsoluteandproportional shortfallwithexamples;2018.Availableathttps://www.nice.org.uk/Media/ Default/About/what-we-do/NICE-guidance/NICE-technology-appraisals/OHE-Note-on-proportional-versus-absolute-shortfall.pdf. [Accessed 06 February 2018].

[47]S.Altmann,Againstproportionalshortfallasapriority-settingprinciple, Jour-nalofMedicalEthics,PublishedOnlineFirst:10January2018.doi:10.1136/ medethics-2017-104488.

[48]Brazier J,Tsuchiya A.Improving cross-sectorcomparisons: goingbeyond the health-related QALY. Applied Health Economics and Health Policy 2015;13(6):557–65.

[49]BobinacA,VanExelNJA,RuttenFF,BrouwerWBF.ThevalueofaQALY: indi-vidualwillingnesstopayforhealthgainsunderrisk.Pharmacoeconomics 2014;32(1):75–86.

[50]OlsenJA.Prioritypreferences:endoflifedoesnotmatter,buttotallifedoes. ValueHealth2013;16(6):1063–6.

[51]BrazierJ,RowenD,MukuriaC,WhyteS,KeetharuthA,RiseA,etal.Eliciting soci-etalpreferencesforburdenofillness,therapeuticimprovementandendoflife forvaluebasedpricing:areportofthemainsurvey;2013.Availableat:http:// eprints.whiterose.ac.uk/99493/1/Eliciting%20societal%20preferences%20-%20011.pdf.[Accessed10November2017].

[52]RowenD,BrazierJ,MukuriaC,KeetharuthA,RisaHoleA,TsuchiyaA,etal. ElicitingsocietalpreferencesforweightingQALYsforburdenofillnessandend oflife.MedicalDecisionMaking2016;36(2):210–22.

[53]VandeWeteringEJ,VanExelNJA,BobinacA,BrouwerWBF.ValuingQALYsin relationtoequityconsiderationsusingadiscretechoiceexperiment. Pharma-coeconomics2015;33(12):1289–300.

[54]BobinacA,VandeWeteringEJ,VanExelNJA,BrouwerWBF.Equity-dependent socialwillingnesstopayforaQALY;2018.Availableat:https://repub.eur.nl/ pub/94502.[Accessed11October2017].

[55]RichardsonJ,IezziA,MaxwellA,ChenG.Doestheuseoftheproportional shortfallhelpaligntheprioritisationofhealthserviceswithpublicpreferences. TheEuropeanJournalofHealthEconomics2017:1–10.

[56]CoastJ,SmithR,LorgellyP.Shouldthecapabilityapproachbeappliedinhealth economics?HealthEconomics2008;17(6):667–70.

[57]NordE,JohansenR.Concernsforseverityinprioritysettinginhealthcare: areviewoftrade-offdatainpreferencestudiesandimplicationsforsocietal willingnesstopayforaQALY.HealthPolicy2014;116(2):281–8.

[58]NationalInstituteforHealthandCareExcellence.Consultationpaper:value basedassessmentofhealthtechnologies.NationalInstituteforHealthand CareExcellence;2018.Availableat:https://www.nice.org.uk/Media/Default/ About/what-we-do/NICE-guidance/NICE-technology-appraisals/VBA-TA-Methods-Guide-for-Consultation.pdf.[Accessed22February2017]. [59]NationalInstituteforHealthandCareExcellence.ResponsestoTAMethods

AddendumPublicConsultationbetween27thMarch2014and1stJuly2014; 2014.Availableat: https://www.nice.org.uk/Media/Default/About/what-we-do/NICE-guidance/NICE-technology-appraisals/VBA-Consultation-Comments. pdf.[Accessed22February2017].

[60]NationalInstituteforHealthandCareExcellence.Valuebasedassessment. Min-utesofthePublicBoardMeetingof17September2014,item14/086.London: NationalInstituteforHealthandCareExcellence;2014.

[61]ShahK.Severityofillnessandprioritysettinginhealthcare:areviewofthe literature.HealthPolicy2009;93:77–84.

[62]LancsarE,WildmanJ,DonaldsonC,RyanM,BakerR.Derivingdistributional weightsforQALYsthroughdiscretechoiceexperiments.JournalofHealth Eco-nomics2011;30(2):466–78.