Simulation
and
education
Cost-effectiveness
of
extracorporeal
cardiopulmonary
resuscitation
after
in-hospital
cardiac
arrest:
A
Markov
decision
model
Benjamin
Y.
Gravesteijn
a,b,*
,
Marc
Schluep
a,
Daphne
C.
Voormolen
b,
Anna
C.
van
der
Burgh
c,d,
Dinı´s
Dos
Reis
Miranda
e,
Sanne
E.
Hoeks
a,
Henrik
Endeman
ea
DepartmentofAnaesthesiology,ErasmusMedicalCentre,Rotterdam,TheNetherlands
b
DepartmentofPublicHealth,ErasmusMedicalCentre,Rotterdam,TheNetherlands
cDepartmentofInternalMedicine,ErasmusMedicalCentre,Rotterdam,TheNetherlands dDepartmentofEpidemiology,ErasmusMedicalCentre,Rotterdam,TheNetherlands
eDepartmentofIntensiveCareMedicine,ErasmusMedicalCentre,Rotterdam,TheNetherlands
Abstract
Background:Thisstudyaimedtoestimatethecost-effectivenessofextracorporealcardiopulmonaryresuscitation(ECPR)forin-hospitalcardiacarrest treatment.
Methods:AdecisiontreeandMarkovmodelwereconstructedbasedoncurrentliterature.Themodelwasconditionalonage,CharlsonComorbidity Index(CCI)andsex.Threetreatmentstrategieswereconsidered:ECPRforpatientswithanAge-CombinedCharlsonComorbidityIndex(ACCI)below differentthresholds(2–4),ECPRforeveryone(EALL),andECPRfornoone(NE).Cost-effectivenesswasassessedwithcostsperquality-of-life adjustedlifeyears(QALY).
Measurementsandmainresults:TreatingeligiblepatientswithanACCIbelow2pointscosts8394(95%CI:4922–14,911)europerextraQALYper IHCApatient;treatingeligiblepatientswithanACCIbelow3costs8825(95%CI:5192–15,777)europerextraQALYperIHCApatient;treatingeligible patientswithanACCIbelow4costs9311(95%CI:5478–16,690)europerextraQALYperIHCApatient;treatingeveryeligiblepatientwithECPRcosts 10,818(95%CI:6357–19,400)europerextraQALYperIHCApatient.ForWTPthresholdsof0–9500euro,NEhasthehighestprobabilityofbeingthe mostcost-effectivestrategy.ForWTPthresholdsbetween9500and12,500,treatingeligiblepatientswithanACCIbelow4hasthehighestprobabilityof beingthemostcost-effectivestrategy.ForWTPthresholdsof12,500orhigher,EALLwasfoundtohavethehighestprobabilityofbeingthemost cost-effectivestrategy.
Conclusions:GiventhatconventionalWTPthresholdsinEuropeandNorth-Americaliebetween50,000–100,000euroorU.S.dollars,ECPRcanbe consideredacost-effectivetreatmentafterin-hospitalcardiacarrestfromahealthcareperspective.Moreresearchisnecessarytovalidatethe effectivenessofECPR,withafocusonthelong-termeffectsofcomplicationsofECPR.
Keywords:Extracorporealmembraneoxygenation,Extracorporeallifesupport,In-hospitalcardiacarrest,Cost-effectiveness,Decisionmodel, Intensivecare
Introduction
* Correspondingauthorat:DepartmentofPublicHealth,P.O.Box2040,3000CA,Rotterdam,TheNetherlands. E-mailaddress:b.gravesteijn@erasmusmc.nl(B.Y. Gravesteijn).
https://doi.org/10.1016/j.resuscitation.2019.08.024
Received6May2019;Receivedinrevisedform9August2019;Accepted14August2019 Availableonlinexxx
Available
online
at
www.sciencedirect.com
Resuscitation
Cardiacarrest,cardiopulmonaryarrest,orcirculatoryarrestistheloss of effective blood circulation, which inevitably leads to death if cardiopulmonaryresuscitation(CPR)isnotstarted.Cardiacarrestis usuallydividedbasedonlocationintoout-of-hospitalcardiacarrest (OHCA)andin-hospitalcardiacarrest(IHCA).OHCAisdescribedto occur around19–104timesper 100,000population per year and resultsin10%survivalathospitaldischarge.1TheincidenceofIHCAis
1–6eventsper1000hospitaladmissions2–4andrecentmeta-analyses
showedapooledsurvivaltodischargeof15%(rangingfrom3%to 40%)andaone-yearsurvivalof13%(rangingfrom4%to69%).5,6 Patient-specificfactorsassociatedwithsurvivalareage,7,8 comor-bidities9–12andpresenceofshockablerhythm.13
ApossibleadvantageforpatientssufferingIHCAversusOHCAis thathospitalsareequippedwithadvancedlifesupportteams,who couldemployextracorporealcardiopulmonaryresuscitation(ECPR) using veno-arterial extracorporeal membrane oxygenation (VA-ECMO).Thistechniquehasseenanincreaseinuseoverthelast decades.14,15 Bytakingovercardiacandrespiratory function,
VA-ECMOensuresoxygenation and circulation.16 Although evidence
fromrandomizedcontrolledtrialsislacking,17observationalstudies haverepeatedlyshownanincreaseinsurvivalafterECPRcompared toconventionalCPR.18–20Furthermore,theAmericanHeart associa-tion recommends thein-hospital use of ECPR in patients with a reversiblecauseofCA(e.g.:acutecoronarysyndrome).
When assessing whether or not to implement ECPR, cost-effectivenessshouldbetakenintoaccount. Ethicalandeconomic considerations are of increasing importance in decision making pertaining to intensive care allocation.21 Financial resources are
limitedandhealthcareshouldbefocusedmoreontherapiesthatdo notonlyextendlife,butratherofferareasonablehealth-relatedquality oflife(HRQoL).Thisstudywasdesignedtoprovidecost-effectiveness evidenceforinternationalcomparisonandtoprovideanoverviewof currentknowledgeoftheeconomicaspectsofECPR.
Twosmallobservationalstudies(USandAustralia)haveshown indicationsofcost-effectivenessofECPRforbothOHCAandIHCA.22,23
Therearehoweverseveralcaveats.Becauseoflowsamplesizeand estimatespertainingtolocalsituationsthesestudiesarenotlikelytobe generalizabletoallsettings.Furthermore,forthein-hospitaland out-of-hospitalsetting,effectivenessshouldbeassessedseparately.
Theprimaryaimofthisstudywasthereforetoassessthe cost-effectiveness of ECPR treatment after IHCA based on current literature.Byusingallavailableevidence,thismodellingapproach
wouldensureahighgeneralizabilityofourresults.Forthispurpose,a decisiontreeandMarkovmodelweredeveloped.Bothmodelsare frequently usedinhealth-economicevaluations,becausetheyare abletocalculatequalityoflifeadjustedlifeyears(QALY).24,25The secondaryaimwastoassessinwhichpatientgroupECPRismost likelytobecost-effective.
Methods
This cost-effectiveness evaluation is reported according to the CHEERSreportingguidelines.26WesearchedPubMedforrelevant
studiestoinformonallparametersusedforthemodels.Weusedthe searchterms“in-hospitalcardiacarrest”and“extracorporeal cardio-pulmonaryresuscitation”incombinationwiththespecificparameterof interest.Furthermore,wefoundliteratureusingthereferencelistof alreadyfoundstudies.
Decisiontree
Athree-strategydecisiontreewascreated,whichencompassesthe in-hospital phase. This type of model uses known absolute and relativeriskstocalculatetheprobabilityofanoutcome.Thedecision tree calculates the probability of dying before discharge. The strategiesconsideredwereECPRfornoone(NE),ECPRforevery eligiblepatient(EALL)andECPRforeligiblepatientswithan Age-CombinedCharlsonComorbidityIndex(ACCI)scorebelowacertain threshold(EACCI_lo).ThethresholdsfortheACCIanalysedranged fromtwotofour:patientswithanACCIabovethethresholddidnot receive ECPR. The ACCI thresholds have been based on best availableECPRguidelinestoexcludepatientswithaterminalillness, comorbiditiesthatformacontraindicationforICUadmissionorfor intravascularcannulation.27Furthermorepatients>75yearsofage
aregenerallynotconsideredeligible.TheACCIscoreisdescribedin Table1,Supplement1.
TheACCIthresholdcanbeillustratedbythefollowingexample:a patientof50yearsoldwithmoderaterenaldisease(GFR<40mL/min/ 1.73m2)willhaveanACCIof3.Ifthepatientwouldsufferamyocardial infarctionthescorewillriseto4.
The decision tree consists of multiple nodes with probability estimates foundin literature(Fig.1 andTable 1). Thefirstnode representspatientswithaDo-Not-Resuscitate(DNR)status.Thisis
Fig.1–Decisiontreeofthein-hospitalphaseofthemodel.Fortheassumedprobabilities(P),oddsratio’s(OR),relative risks(RR),andbetas,seeTable1.DNR=do-not-resuscitate;CPR=cardiopulmonaryresuscitation;ROSC=returnof spontaneouscirculation.
anagreementbetweenapatientandahealthcareprofessionalnotto attemptcardiopulmonaryresuscitationincaseofcardiacarrest.Since aDNRstatusismoreoftenagreeduponbypatientswithhigherage,28
we assumed higher probabilities for higher aged patients. We assumedthatforpatientswhosufferedcardiacarrestwithaDNR status, no CPR would be attempted and death iscertain. When patientsdidnothaveaDNRstatus,CPRwouldbeattempted.The nextnoderepresentstheprobabilityofhavingacontra-indicationfor ECPR.Havingacontra-indication,e.g.refractorycardiacdiseaseor metastaticcancer,wasassumedtoincreasetheriskofdyingafter CPR.IfCPRwasstartedandnocontra-indicationwaspresent,the nextnoderepresentstheprobabilityofhavingreturnofspontaneous circulation(ROSC)within20minaftercardiacarrest.29IfROSCwould
notbeachievedwithin20min,ECPR couldbe startedandcould increasetheremainingsurvivalprobability.18Theprobabilityofhaving acomplicationofECPRandtheprobabilityofsubsequentdeathare alsotakenintoaccount.30–32Theseprobabilitieswerecalculatedfrom
theELSOdatabase.33Theextraprobabilityofmortality,giventhatthe
patienthadacomplicationwas:themortalityrateofpatientswitha complicationminustheoverallmortalityrate.Finally,themortalityrate after CPR increases with increasing Age-Combined Charlson ComorbidityIndex(ACCI).9,10
TheprevalenceofDNRstatusbelow75yearswasassumedto be around 5% (range 2–10%), based on experience in our hospital: the Erasmus Medical Center, Rotterdam. The probabilityofhavingacontra-indicationforECPRwasalsobased onexperienceinourhospital,whereweimplementedECPRin
2016.Weassumedthat20%(range10–30%)ofthepatientshave the contra-indications described by Makdisi et al. Since the described contra-indications (e.g. refractory cardiac diseaseor metastatic cancer)aresevere conditions,therisk ofdyingwas assumed to double (OR: 2.0, with a minimum of 1.4, and a maximum2.9).
Markovmodel
Forthecalculationoflong-termoutcomes,aMarkovmodelwasused. AMarkovmodelusesstatesandtransitionprobabilitiestocalculate long-termoutcomes.24Weproposeamodelconsistingoftwostates:
an alive state (with decreased HRQoL) and a dead state (the absorbingstate).Markovmodelscanbeusedtocalculatethetime spentineachstate.Therefore,QALYscanbecalculated,makingthis typeofmodelusefulforcost-effectivenessanalysis.25Eachindividual probabilityofdyingattheendofthedecisiontreedescribedaboveis usedasinputinthesubsequentMarkovmodel.Themodelsimulated 20yearsoffollow-upandthemodelcycleswereoneyearlong.The dataonageandsexspecificmortalityrateswereprovidedbyStatistics Netherlands(CBS).34WedidnotassumealastingeffectofIHCAon long-termsurvival.35Theamountoflife-yearswerethenmultipliedby
thesex-specificutilityscoreafterIHCAtoobtainQALYsformenand women36(Table1).
Asanexample,considerapatientwitha100%chanceofsurviving thein-hospitalphase:theMarkovmodelwillcalculatetheamountof lifeyearsthispatientwillspendafterdischarge.Forapatientwith0% Table1–Assumedestimatesandtheirdistributionsforthedecisiontreeintheprobabilisticsensitivityanalysis. Abbr. Parameter Median(IQR) Distribution Source
Decisiontree
P1 ProbabilityofhavingDNRstatusif<75
years
0.05(0.02–0.10) Beta(5,95) Clinicalinsight
P2 ProbabilityofECPRcontraindication 0.19(0.11–0.32) Beta(10,40) Clinicalinsight
P3 TheprobabilityofdyingafterCPR 0.85(0.83–0.87) Beta(850,150) ZhuandZhang5
P4 TheprobabilityofhavingROSCwithin
20min
0.38(0.35–0.41) Beta(338,556) Khanetal.29
P5 Probabilityofcomplication 0.38(0.29–0.47) Beta(38,62) Sheuetal.,Mulleretal.and
Sakamotoetal.30–32
P6 Probabilityofdyingbecauseof
complication
0.2(0.1–0.32) Beta(10,40) Clinicalinsight
RR1 Therelativeriskofdying,ECPRvsnon
ECPR
0.43(0.3–0.62) Lognormal( 0.85,
0.19)
Chenetal.18 OR1 ORofdyingwhencontraindicationfor
ECPR
2.00(1.40–2.93) Lognormal(0.69,
0.2)
Clinicalinsight
OR2 ORofhavingDNRstatusif75-84
years,comparedto<75years
1.71(1.23–2.32) Lognormal(0.53,
0.16)
Cooketal.28 OR3 ORofhavingDNRstatusif>85years,
comparedto<75years
2.98(2.38–3.75) Lognormal(1.09,
0.12)
Cooketal.28 Beta1 ThelogoddsincreaseindyingperACCI
increase
0.09(0.03–0.14) Log-Lognormal
(0.09,0.03)
Hirlekaretal.9 Costsandutilities
In-hospitalincrementalcostofECPR aftercardiacarrest
51756.66 (31377.83– 73978.21) Normal(51997, 10767) OudeLansink-Hartgring etal.37
Utilityscoreformen 0.79(0.69–0.87) Triangle(a=0.66,
b=0.89,c=0.82)
Israelssonetal.36
Utilityscoreforwomen 0.74(0.62–0.81) Triangle(a=0.58,
b=0.82,c=0.81)
Israelssonetal.36
ECPR=extracorporealcardiopulmonaryresuscitation;CPR=cardiopulmonaryresuscitation;ROSC=returnofspontaneouscirculation;ACCI=Age-Combined CharlsonComorbidityIndex;DNR=donotresuscitate.
chanceof survivingthe in-hospital phase,the Markov modelwill estimate0lifeyearsafterdischarge.Forchancesbetween0%and 100%,themodelcalculatestheaveragelifeyearsthatpatientswith thesamecharacteristicswillspendafterdischarge.
Cost-effectivenessanalysis
ThetotalcostsofECPRwerecalculatedbasedonhowmanypatients received ECPR following the decision tree outcomes: a patient receivedECPRaccordingtothetreatmentstrategyiftheydidnothave aDNRstatus,nocontra-indication,andnoROSCwithin20min(Fig.1
andTable1).
Onlydirect additionalcosts ofECPRtreatmentwere taken into account,takingahealthcare’sperspective.Theaverageadditionalcosts ofECPRdescribedintheliteraturewereusedinthemodel.Adetailed descriptionoftheitemsincludedinthetotalcostshasbeendescribedby Lansink-Hartgring et al.37 A discount rate of 4% was applied, the
appropriaterateforcost-effectivenessanalysesintheNetherlands.38To
assesscost-effectivenessofthestrategies,incremental cost-effective-nessratios(ICER)werecalculated,whereNEservesasthereference category.TheICERinformsabouthowmanyextraeuroperQALYa strategycosts,comparedtoNE.TheincrementalcostsandQALYswere plottedandthecost-effectivenessacceptabilitycurveswerecalculated anddrawntoobtainthemostcost-effectivestrategy.
ImportanttotakeintoaccountisthatthecalculatedcostsforECPR arenotablylowerthanthecostsofECMO.Thisisduetothemodel structure,inwhichcostsarecalculatedforanaveragepatientwho suffersIHCA, therebyincluding alsopatients whodo notreceive ECPR.
Probabilisticsensitivityanalysis
To take the uncertainty of our model parametersinto account, a probabilisticsensitivityanalysis(PSA)wasperformed.APSArepeats the model a large number of times with different (but probable) parameters. The type of distributions that were used were beta distributionsforprobabilities,log-normaldistributionsfortheoddsratios andrelative risks, and log–log-normal distribution forthe log-odds increaseinmortalityforanACCIpointincrease.Thecharacteristicsof thedistributionswereadjustedsothatthemedianandinterquartile rangewereidenticaltotheestimateand95%confidenceinterval.The type andcharacteristicsof the distributionsofthe parametersare describedinTable1.Fromthesedistributions,1000randomsamples weredrawn,resultingin1000replicatesofthemodel.Additionally,a representative cohort of 1000 patients was randomly sampled (Table2).10,39Afterrunningthe1000replicatesofthemodelinthis
cohort, outcomes were calculated 1000 times. We calculated the QALYsandcostsperstrategy.Themedianwastakenasthemost probableestimateofthemodel.The2.5thand97.5thpercentilewere calculated,whichindicatedthebordersofthe95%credibilityinterval. To estimate whether the conclusions were affected by the parametersthatwerenotfoundinliterature,linearregressionwas performed.Asthedependentvariable,theICERoftheEALLstrategy periterationwasused. Aspredictors,thestandardizedparameter valueswereused.Thecoefficientsofthemodelcouldthereforebe interpreted as “with one standarddeviation (SD)increase in the parameter,theICERfortheEALLstrategyincreaseswithx”.
AllanalyseswereperformedusingR(RCoreTeam(2013).R:A languageandenvironmentforstatisticalcomputing.RFoundationfor StatisticalComputing,Vienna,Austria).FortheMarkovmodel,the
“dampack” packagewasused.40 Thecodeofthemodelisonline availableinAppendix2,fortransparencyandreproducibility.41
Results
In the decision tree, survival rates between 9% and 13% were observedfortheNEstrategy,andbetween30%and35%fortheEALL strategy (Fig. 1,Supplement 1). After applying aMarkov model, expectedlifeyearsafterCPRperpatientfortheNEstrategyranged from0.79to2.48andfortheEALLstrategyfrom2.57to6.55years (Fig.2,Supplement1).
TheexpectedcostsperICHApatientfortreatingeligiblepatients belowanACCIof2pointswithECPRare3975(95%CI:2418–5780) euro, and increasedto 23,272(95%CI: 14,159–33,838) eurofor treating alleligible patients(Table 3). Theassociated QALYsfor treatingnopatientswithECPRare1.2(95%CI:1.0–1.5);fortreating eligiblepatientsbelowanACCIof2points1.7(95%CI:1.4–2.0);for treatingeligiblepatientsbelowanACCIof3points2.1(95%CI:1.7– 2.6);fortreatingeligiblepatientsbelowanACCIof4points2.6(95% CI:2.0–3.2);andfortreatingalleligiblepatients3.4(95%CI:2.4–4.2).
Table2–Patientcharacteristicsofthesimulated cohort,basedonliterature.10,39
Characteristic N=1000 Age(mean(sd)) 65.49(15.71) Male(%) 578(57.80) CCI(%) 0 373(37.30) 1 230(23.00) 2 183(18.30) 3 107(10.70) 4 43(4.30) 5 40(4.00) 6 15(1.50) 7 4(0.40) 8 5(0.50)
CCI=CharlsonComorbidityIndex.
Table3–Thehealtheconomicevaluationforeach strategy.
Strategy Costsa QALY ICERb
NE – 1.2(1.0–1.5) –
ACCI<2 3975(2418–5780) 1.7(1.4–2.0) 8394(4922–14,911) ACCI<3 8066(4909–11,731) 2.1(1.7–2.6) 8825(5192–15,777) ACCI<4 12,942(7881–18,829) 2.6(2.0–3.2) 9311(5478–16,690) EALL 23,272(14,159–33,838) 3.4(2.4–4.2) 10,818(6357–19,400) ThestrategiesarenobodyECPR(NE),treatingeveryonewithan Age-CombinedCharlsonComorbidityIndex(ACCI)of2,3or4orless,and treatingeveryonewithECPR(EALL).Therangesindicate95%credibility intervals(CI).
aInEuro,onlydirectadditionalECPRcosts.
bTheincrementalcost-effectivenessratio(ICER)iscalculatedwiththe
mostconservativemethod(NE:nobodyECPR)asthereferencemethod.It representsthecostsperextraQALY.
ComparedtotreatingNE,theexpectedincrementalcostsperextra QALY(ICER)fortreatingeligiblepatientswithanACCIbelow2points is8394(95%CI: 4922–14,911)europerextraQALY; fortreating eligiblepatientswithanACCIbelow3,theICERis8825(95%CI: 5192–15,777) euro per extraQALYcompared to NE;for treating eligiblepatientswithanACCIbelow4,theICERis9311(95%CI: 5478–16,690)europerextraQALY;fortreatingalleligiblepatients,the ICERwas 10,818(95% CI: 6357–19,400) euro per extraQALY.
Table 3 displays an overview of the economic evaluation. The consideredstrategiesarecomparableintermsofmeanICER,butthe incrementalcostsandincrementalQALYsvarysignificantlybetween theconsideredstrategies(Fig.3,Supplement1).
Thecost-effectivenessacceptabilitycurvesdepictedinFig.2show thatforWTPthresholdsof0–9500euro,NEhasthehighestprobability ofbeingthemostcost-effectivestrategy.ForWTPthresholdsbetween 9500and12,500,treatingeligiblepatientswithanACCIbelow4hasthe highestprobabilityofbeingthemostcost-effectivestrategy.ForWTP thresholdsof12,500orhigher,EALLwasfoundtohavethehighest probabilityofbeingthemostcost-effectivestrategy.
Theonlyparameterthatwasfoundtoinfluencethecost-effectiveness significantlywastherelativeriskofdyingofECPR(effectofoneunit increaseoftheparameterontheICERwas 255( 481to 28)euros perincrementalQALY),seeTable2,Supplement2.
Discussion
In thisstudy we found that the expectedcostsper IHCA patient oftreating eacheligibleIHCApatientwithECPRareapproximately23,000euro.A patient was eligible when no contraindications waspresent,andinwhom ROSCcannotbeachievedwithin20minaftercardiacarrest.PerQALY
increase,theassociatedcostswerearound15,000.The Willingess-To-Pay tresholds in Europe and North-America are between 50,000– 100,000 euro perincrementalQALY.Withinthisrange,performing ECPRineveryeligibleIHCApatient,islikelytobecosts-effective.
TheuseofECMOhassteadilyincreasedfrom2007onwards.14
Positive results fromobservationalstudies andincreasing clinical applicabilityledtotheinclusionofECPRintheAdvancedLifeSupport Guidelines by the European Resuscitation Counsil.42 However, ECPRiscostlyandlabour-intensiveandcarefuleconomicevaluation wasstilllacking.
Because ECPR was found to be cost-effective, this study substantiatesitsincreasedimplementationandinclusionaspossible treatmentintheguidelines.Theallocationofintensivecaretreatments shouldbecriticallyevaluated,especiallywhenfinancialresourcesare limited.21ThedifferenceinsurvivalprobabilityafterECPRseemsto
be sufficient to render the therapy cost-effective. Because we performed an analysis takingall uncertainties of parameters into account,webelievethatwereliablyestimatedtheaveragecostper IHCApatientwheneveryeligiblepatientistreatedwithECPR:around 11,000europerextraQALY.
Ourcost-effectiveness analysisbasedonliterature supportsfindings ofempiricalstudies.Firstly,ourstudyconfirmstheresultsofarecent smallretrospectivestudyintheUnitedStatesthatsuggestedthatECPR afterIHCAiscost-effective,consideringonlyin-hospitalcosts.22This studysuggestedthatthecostsperextraQALYsavedisaround56,000 U.S.dollars.Thisestimateislargerthanourestimateof11,000euros, buthealthcareexpendituresintheUnitedStatestendtobehigherthan inEurope.43Nevertheless,itisreassuringthatbothstudiesconclude
thatECPRafterIHCAiscost-effective,sincetheybothassessprimarily in-hospital costs. Secondly, our study confirms the results of Dennis et al. Fig.2–Cost-effectivenessacceptabilitycurves.Forgivenwillingnesstopay(WTP)thresholds,theprobabilityofbeing themostcost-effectivestrategyisplotted.ThestrategiesarenobodyECPR(NE),treatingeveryonewithan Age-CombinedCharlsonComorbidityIndex(ACCI)of2,3or4orless(thr2,thr3,thr4respectively),andtreatingeveryone withECPR(EALL).ThedottedlinesindicatestheWTPthresholdsof9500and12,500.
ThisstudyshowedthatforIHCA,15,000euros(25,000AUD)perextra QALYwasexpected,whichissimilartoourestimate.23
Theresultsofourstudyare alsosimilartoresultsofthe cost-effectivenessofamobileECPRteam.44Thisteamisabletotreat patientswithECPRinmultiplecentres,anditsapplicationwasfound tobepotentiallycost-effective.Theapplicationcouldbenefitcentres thatdonothavetheresourcesforECPRorlackexperiencewithits application.CentresthatoftenuseECPRrelyonperfusionistsforaid ininitiationandmaintenanceoftreatment,whichenhancesthecosts. Therefore,itcouldwellbethatECPRismostlycost-effectivewhen thereisnoneedfortheseextracosts.Thishypothesis, however, warrantsfurtherinvestigation.
TherangeofcostsofECMOfoundin theliteratureislarge.45
Mostlybecausestudiesinconsistentlyreporttheirresults,thereareno factorsdescribedthatexplainthisvariation.Weusedastructured Dutch studyas input for our cost-effectivenessanalysis, sinceit describesclearlytheincrementalcostsforECPR.37Thisstudyfound
thatthemajorityofthecostsarecomposedofnursingdays.Beingable toshorten thelengthof ICUstay would thereforeenhance cost-effectivenessofECPRafterIHCA.
WedidnotfindthattreatingasubgroupofIHCApatientswith ECPRbasedonAge-CombinedCharlsonComorbidityIndexaffected cost-effectiveness. Sinceothers described that cost-effectiveness dependsonpatientcharacteristics,44weconsiderthistobeattributed
totwofactors.First,theeffectofcomorbidityonsurvivalofCPRis uncertain.10,46 More research into this relationship is necessary. Second,ifthereisaneffectofcomorbidity,thiseffectismorelikelyto besignificantinacohortwithahighprevalenceofcomorbidities.The prevalenceinourrepresentativecohort,however,waslow.10,39
Thisstudyhasseverallimitations.Unfortunately,notallinformation neededforthemodel couldbefoundintheliterature.The lackof evidencehadtwoconsequences.First,itwasnecessarytobasesome oftheparametersonclinicalknowledge;e.g.,fortheprobabilityof havingacontraindicationforECPR.However,asensitivityanalysis showedthattheseparameterswerenotlikelytoinfluencetheoverall cost-effectiveness of ECPR. Second, cost-effectiveness might be somewhatoverestimated.Evidencefromrandomizedcontrolledtrials was unfortunatelyabsent at this moment.17 Observationalstudies couldhaveoverestimatedtheeffectofECPRonsurvivalbecauseof confoundingbias.18,19AnoverestimatedeffectofECPRwouldresultin
anoverestimatedcost-effectiveness.Additionally,wewerenotableto modellong-termeffectsofcomplicationsofECPR:theextrahealthcare costs andlowerquality oflifeaftermajorcomplicationsofECPR(stroke, acutekidneyinjury)coulddecreaseoverallcost-effectiveness.
Althoughwedidnottakenon-directcostsofECPRintoaccount, westillbelievethisstudyprovidesavalideconomicevaluation.Other identifiablecostsarecostsofrehabilitation,futurehealthcarecosts andnon-medicalcostssuchaslossofparticipationinworkinglife. However,thesecostsaremoreinterestingfromasocietalperspective thanahealthcareperspective.Othercoststhatarenottakeninto accountarethecostsofimplementation.Theseexpensesarelarge andcouldexplainthestagnatingincreaseintheuseofECPR.47,48 Therefore,webelievethatourfindingsaremostapplicabletolarge hospitalsinwesterncountries,whichoftendohaveaccesstothese resourcestoovercomethefirstbarriertoanapparentcost-effective therapy.
Webelievefuturestudiesshouldhavethreegoals.First,toidentify patientswhocould benefitmostfrom ECPR. Second,randomized controlledtrialsarenecessary,asindicatedintheadvancedlifesupport guidelines.42Fortunately,fiveongoingrandomizedcontrolledtrialswill
hopefullyfillthis knowledgegap in theupcoming years.20Third,the
long-termeffectsofcomplications ofECPR shouldbeinvestigated,since they could decrease the cost-effectiveness of the intervention. The knowledgegainedfromfurtherresearchcouldimproveimplementation andcost-effectivenessofthiscostlyandlabour-intensiveintervention.
Conclusion
Forin-hospitalcardiacarrestpatients,extracorporeal cardiopulmo-nary was demonstrated to be cost-effective from a healthcare perspective giventhat conventional WTPthresholds lie between 50,000–100,000euroorU.S.dollars.Moreresearchisnecessaryto validatetheeffectivenessofECPR,withafocusonthelong-term effectsofcomplicationsofECPR.
Conflict
of
interest
DRMreceivedspeakingfeesfromXeniosGmbH.Theauthorsdeclare thattheydidnotreceiveanyotherfinancialsupportforthisstudyand thattheXeniosGmbH had noroleinthecommencement,development, interpretation,orreportingofthiscost-effectivenessanalysis.
Acknowledgements
WewouldliketokindlythankprofessorMyriamHuninkandJeremy Goldhaber-Fiebert, who assisted with the development of this decisionmodel.
Appendix
A.
Supplementary
data
Supplementarymaterialrelatedtothisarticlecanbefound,intheonline version,atdoi:https://doi.org/10.1016/j.resuscitation.2019.08.024.
REFERENCES
1.GräsnerJ-T,LeferingR,KosterRW,etal.EuReCaONE-27Nations, ONEEurope,ONERegistry:aprospectiveonemonthanalysisof out-of-hospitalcardiacarrestoutcomesin27countriesinEurope. Resuscitation2016;105:188–95Availablefrom:https://doi.org/10. 1016/j.resuscitation.2016.06.0040300-9572/.[Cited11January 2019].
2.SkogvollE,IsernE,SangoltGK,GisvoldSE.In-hospital cardiopulmonaryresuscitation:5years’incidenceandsurvival accordingtotheUtsteintemplate.ActaAnaesthesiolScand 1999;43:177–84..Availablefrom:http://www.embase.com/search/ results?subaction=viewrecord&from=export&id=L29059312. 3.HodgettsTJ,KenwardG,VlackonikolisI,etal.Incidence,locationand
reasonsforavoidablein-hospitalcardiacarrestinadistrictgeneral hospital.Resuscitation2002;54:115–23Availablefrom:www.elsevier. com/locate/resuscitation.[Cited22March2018].
4.SandroniC,NolanJ,CavallaroF,etal.In-hospitalcardiacarrest: incidence,prognosisandpossiblemeasurestoimprovesurvival. IntensiveCareMed2007;33:237–45.
5.ZhuA,ZhangJ.Meta-analysisofoutcomesofthe2005and2010 cardiopulmonaryresuscitationguidelinesforadultswithin-hospital cardiacarrest.AmJEmergMed2016;34:1133–9..Availablefrom:
http://www.embase.com/search/results?subaction= viewrecord&from=export&id=L610622958.
6.SchluepM,GravesteijnBY,StolkerRJ,EndemanH,HoeksSE. One-yearsurvivalafterin-hospitalcardiacarrest:asystematicreviewand meta-analysis.Resuscitation2018;132:90–100Availablefrom:http:// www.ncbi.nlm.nih.gov/pubmed/30213495.[Cited15November 2015].
7.HirlekarG,KarlssonT,AuneS,etal.Survivalandneurological outcomeintheelderlyafterin-hospitalcardiacarrest.Resuscitation 2017;118:101–6Availablefrom:https://www.sciencedirect.com/ science/article/pii/S0300957217302940?via%3Dihub.[Cited15 November2018].
8.SkrifvarsMB,CastrenM,NurmiJ,ThorenAB,AuneS,HerlitzJ.Do patientcharacteristicsorfactorsatresuscitationinfluencelong-term outcomeinpatientssurvivingtobedischargedfollowingin-hospital cardiacarrest?JInternMed2007;262:488–95Availablefrom:https:// doi.org/10.1111/j.1365-2796.2007.01846.x.[Cited22March2018]. 9.HirlekarG,JonssonM,KarlssonT,HollenbergJ,AlbertssonP,Herlitz
J.Comorbidityandsurvivalinout-of-hospitalcardiacarrest. Resuscitation2018;133:118–23Availablefrom:https://www. sciencedirect.com/science/article/pii/S0300957218309882?via% 3Dihub.[Cited15November2018].
10.AndrewE,NehmeZ,BernardS,SmithK.Theinfluenceofcomorbidity onsurvivalandlong-termoutcomesafterout-of-hospitalcardiac arrest.Resuscitation2017;110:42–7..Availablefrom:http://www. embase.com/search/results?subaction=viewrecord&from= export&id=L613130514.
11.PiscatorE,HedbergP,GöranssonK,DjärvT.Survivalafterin-hospital cardiacarresthighlyassociatedtoage-adjustedCharlsonco-morbidity indexacohortstudyfromatwo-sitedSwedishUniversityhospital. Resuscitation2015;96:109–10..Availablefrom:http://www.embase. com/search/results?subaction=viewrecord&from=export&id= L72184025.
12.SchluepM,RijkenbergS,StolkerRJ,HoeksS,EndemanH.One-year mortalityofpatientsadmittedtotheintensivecareunitafterin-hospital cardiacarrest:aretrospectivestudy.JCritCare2018;48:345–51.. Availablefrom:https://doi.org/10.1016/j.jcrc.2018.09.029.
13.TaglieriN,SaiaF,BacchiReggianiML,etal.Prognosticsignificanceof shockableandnon-shockablecardiacarrestinST-segmentelevation myocardialinfarctionpatientsundergoingprimaryangioplasty. Resuscitation[Internet]Elsevier2018;123:8–14Availablefrom:
https://www.sciencedirect.com/science/article/pii/
S0300957217307669?via%3Dihub.[Cited15November2018]. 14.KaragiannidisC,BrodieD,StrassmannS,etal.Extracorporeal membraneoxygenation:evolvingepidemiologyandmortality. IntensiveCare2016;42:889–96Availablefrom:http://link.springer. com/10.1007/s00134-016-4273-z.[Cited15November2018]. 15.ExtracorporealLifeSupportOrganization.ExtracorporealLife SupportOrganizationECLSregistryreportoveralloutcomes. Availablefrom:..https://www.elso.org/Registry/Statistics/ InternationalSummary.aspx.
16.MassettiM,TasleM,LePageO,etal.Backfromirreversibility: extracorporeallifesupportforprolongedcardiacarrest.AnnThorac Surg2005;79:178–83..Availablefrom:http://www.embase.com/ search/results?subaction=viewrecord&from=export&id=L40037462. 17.TrammR,IlicD,DaviesAR,PellegrinoVA,RomeroL,HodgsonC.
Extracorporealmembraneoxygenationforcriticallyilladults. CochraneDatabaseSystRev20151:..CD010381.Availablefrom:
https://doi.org/10.1002/14651858.CD010381.pub2.
18.ChenYS,ChouNK,WangCH,etal.Comparisonofoutcomeof extracorporealcardiopulmonaryresuscitationforout-of-hospitaland in-hospitalcardiacarrest.Circulation[Internet]2013128:..Available from:http://www.embase.com/search/results?
subaction=viewrecord&from=export&id=L71340882.
19.OuweneelDM,SchotborghJV,LimpensJ,etal.Extracorporeallife supportduringcardiacarrestandcardiogenicshock:asystematic reviewandmeta-analysis.IntensiveCareMed2016;42:1922–34.. Availablefrom:http://www.embase.com/search/results?
subaction=viewrecord&from=export&id=L612271576.
20.HolmbergMJ,GeriG,WibergS,etal.Extracorporealcardiopulmonary resuscitationforcardiacarrest:asystematicreview.Resuscitation 2018;131:91–100Availablefrom:http://www.ncbi.nlm.nih.gov/ pubmed/30063963.[Cited5December2018].
21.LuceJM,RubenfeldGD.Canhealthcarecostsbereducedbylimiting intensivecareattheendoflife?AmJRespirCritCareMed 2002;165:750–4Availablefrom:https://doi.org/10.1164/ajrccm.165.6. 2109045.[Cited15November2018].
22.BharmalMI,VenturiniJM,ChuaRFM,etal.Cost-utilityof
extracorporealcardiopulmonaryresuscitationinpatientswithcardiac arrest.Resuscitation2019;136:126–30Availablefrom:https://doi.org/ 10.1016/j.resuscitation.2019.01.027.[Cited15November2018]. 23.DennisM,ZmudzkiF,BurnsB,etal.Costeffectivenessandqualityof
lifeanalysisofextracorporealcardiopulmonaryresuscitation(ECPR) forrefractorycardiacarrest.Resuscitation2019;139:49–56..Available from:https://doi.org/10.1016/j.resuscitation.2019.03.021.
24.SonnenbergFA,BeckJR.Markovmodelsinmedicaldecisionmaking. MedDecisMak1993;13:322–38Availablefrom:https://doi.org/10. 1177/0272989X9301300409.[Cited15November2018].
25.GoldM,SiegelJ,RussellL,WeinsteinM.Cost-effectivenessinhealth andmedicine.NewYork:OxfordUniversityPress;1996Available from:http://doi.wiley.com/10.1002/%28SICI%291099-176X% 28199906%292%3A2%3C91%3A%3AAID-MHP46%3E3.0.CO% 3B2-I.[Cited15November2018].
26.HusereauD,DrummondM,PetrouS,etal.ConsolidatedHealth EconomicEvaluationReportingStandards(CHEERS)statement.Eur JHealthEcon2013;14(3):367–72.
27.CallawayCW,SoarJ,AibikiM,etal.Part4:advancedlifesupport. Circulation2015;132:S84–S145Availablefrom:https://doi.org/10. 1161/CIR.0000000000000273.[Cited31July2018].
28.CookI,KirkupAL,LanghamLJ,MalikMA,MarlowG,SammyI.Endof lifecareanddonotresuscitateorders:howmuchdoesageinfluence decisionmaking?Asystematicreviewandmeta-analysis.Gerontol GeriatrMed20173:2333721417713422.Availablefrom:http://www. ncbi.nlm.nih.gov/pubmed/28638855.[Cited15November2018]. 29.KhanAM,KirkpatrickJN,YangL,etal.Age,sex,andhospitalfactors
areassociatedwiththedurationofcardiopulmonaryresuscitationin hospitalizedpatientswhodonotexperiencesustainedreturnof spontaneouscirculation.JAmHeartAssoc2014;3:e001044.. Availablefrom:https://doi.org/10.1161/JAHA.114.001044. 30.SheuJ-J,TsaiT-H,LeeF-Y,etal.Earlyextracorporealmembrane
oxygenator-assistedprimarypercutaneouscoronaryintervention improved30-dayclinicaloutcomesinpatientswithST-segment elevationmyocardialinfarctioncomplicatedwithprofoundcardiogenic shock.CritCareMed2010;38:1810–7Availablefrom:https://insights. ovid.com/crossref?an=00003246-201009000-00007.[Cited28 August2018].
31.MullerG,FlecherE,LebretonG,etal.TheENCOURAGEmortalityrisk scoreandanalysisoflong-termoutcomesafterVA-ECMOforacute myocardialinfarctionwithcardiogenicshock.IntensiveCareMed 2016;42:370–8Availablefrom:http://link.springer.com/10.1007/ s00134-016-4223-9.[Cited28August2018].
32.SakamotoS,TaniguchiN,NakajimaS,TakahashiA.Extracorporeal lifesupportforcardiogenicshockorcardiacarrestduetoacute coronarysyndrome.AnnThoracSurg2012;94:1–7Availablefrom:
http://linkinghub.elsevier.com/retrieve/pii/S0003497512001439. [Cited28August2018].
33.Extracorporeallifesupportorganisation.ECLSregistryreport. Availablefrom:..https://www.elso.org/Registry/Statistics/ InternationalSummary.aspx.
34.C.B.voordeStatistiek.Levensverwachting;geslacht,leeftijd(perjaar enperiodevanvijfjaren). Availablefrom:..http://statline.cbs.nl/ Statweb/publication/?DM=SLNL&PA=37360ned&D1=a&D2=a&D3= 0-1,11,21,31,41,51,66&D4=0,10,20,30,40,50,56-l&HDR=G1, T&STB=G2,G3&VW=T.
35.FeingoldPL,MinaMJ,BurkeRM,etal.Long-termsurvivalfollowing in-hospitalcardiacarrest:amatchedcohortstudy.Resuscitation 2016;99:72–8Availablefrom:http://www.ncbi.nlm.nih.gov/pubmed/ 26703463.[Cited21February2018].
36.IsraelssonJ,BremerA,HerlitzJ,etal.Healthstatusandpsychological distressamongin-hospitalcardiacarrestsurvivorsinrelationto gender.Resuscitation2017;114:27–33Availablefrom:https://www. sciencedirect.com/science/article/pii/S030095721730059X?via% 3Dihub.[Cited25September2018].
37.Lansink-HartgringAO,VanDenHengelB,VanDerBijW,etal. Hospitalcostsofextracorporeallifesupporttherapy.CritCareMed 2016;44(4):717–23.
38.ZorginstituutNederland.Richtlijnvoorhetuitvoerenvaneconomische evaluatiesindegezondheidszorg.2016.
39.GirotraS,NallamothuBK,SpertusJA,LiY,KrumholzHM,ChanPS. Trendsinsurvivalafterin-hospitalcardiacarrest.Circulation 2011124:..Availablefrom:http://www.embase.com/search/results? subaction=viewrecord&from=export&id=L70619607.
40.Alarid-EscuderoF.dampack:anRpackagewithusefulfunctionsto developandanalyzedecision-analyticmodels.2018Availablefrom:
https://experts.umn.edu/en/publications/dampack-an-r-package-with-useful-functions-to-develop-and-analyze.[Cited27September 2018].
41.JalalH,PechlivanoglouP,KrijkampE,Alarid-EscuderoF,EnnsE, HuninkMGM.AnoverviewofRinhealthdecisionsciences.MedDecis Mak2017;37:735–46Availablefrom:https://doi.org/10.1177/ 0272989X16686559.[Cited5December2018].
42.SoarJ,NolanJP,BöttigerBW,etal.EuropeanResuscitationCouncil GuidelinesforResuscitation2015.Resuscitation2015;95:100–47 Availablefrom:https://linkinghub.elsevier.com/retrieve/pii/ S0300957215003287.[Cited20November2018].
43.GarberAM,SkinnerJ.IsAmericanhealthcareuniquelyinefficient?J EconPerspect2008;22(4):27–50.
44.AubinH,PetrovG,DalyanogluH,etal.Asuprainstitutionalnetworkfor remoteextracorporeallifesupportaretrospectivecohortstudy. Availablefrom:..https://ac.els-cdn.com/S2213177916301238/1-s2. 0-S2213177916301238-main.pdf?_tid=59de6c39-d87c-4f12-828d-d46baaeb8721&acdnat=1547104917_
bab83470f538b254f9ea8086f7a509b2.
45.HarveyMJ,GaiesMG,ProsserLA.USandinternationalin-hospital costsofextracorporealmembraneoxygenation:asystematicreview. ApplHealthEconHealthPolicy2015;13:341–57Availablefrom:
https://link.springer.com/content/pdf/10.1007%2Fs40258-015-0170-9.pdf.[Cited15November2018].
46.HirlekarG,JonssonM,KarlssonT,HollenbergJ,AlbertssonP,Herlitz J.Comorbidityandsurvivalinout-of-hospitalcardiacarrest. Resuscitation2018;133:118–23Availablefrom:https://linkinghub. elsevier.com/retrieve/pii/S0300957218309882.[Cited15November 2018].
47.KaragiannisC,GeorgiouM,KouskouniE,IacovidouN,XanthosT. Associationoflactatelevelswithoutcomeafterin-hospitalcardiac arrest.Resuscitation2012;83:e175–6..Availablefrom:https://doi.org/ 10.1016/j.resuscitation.2012.04.016.
48.HastingsSL,PellegrinoVA,PreovolosA,SalamonsenRF.Surveyof adultextracorporealmembraneoxygenation(ECMO)practiceand attitudesamongAustralianandNewZealandintensivists.CritCare Resusc2008;10:46Availablefrom:http://www.ncbi.nlm.nih.gov/ pubmed/18304017.[Cited15November2018].