cata.netcms

RESEARCHOpenAccessEquityofthepremiumoftheGhanaiannationalhealthinsuranceschemeandtheimplicationsforachievinguniversalcoverageEugeniaAmporfuAbstractTheGhanaianNationalHealthInsuranceScheme(NHIS)wasintroducedtoprovideaccesstoadequatehealthcareregardlessofabilitytopay.
BylawtheNHISismandatorybutbecausetheinformalsectorhastomakepremiumpaymentbeforetheyareenrolled,theauthoritiesareunabletoenforcemandatorynatureofthescheme.
TheultimategoaloftheSchemethenistoprovideallresidentswithaccesstoadequatehealthcareataffordablecost.
Inotherwords,theSchemeintendstoachieveuniversalcoverage.
Animportantfactorfortheachievementofuniversalcoverageisthatrevenuecollectionbeequitable.
ThepurposeofthisstudyistoexaminetheverticalandhorizontalequityofthepremiumcollectionoftheScheme.
TheKakwaniindexmethodaswellasgraphicalanalysiswasusedtostudytheverticalequity.
Horizontalinequitywasmeasuredthroughtheeffectofthepremiumonredistributionofabilitytopayofmembers.
Theextenttowhichthepremiumcouldcausecatastrophicexpenditurewasalsoexamined.
Theresultsshowedthatrevenuecollectionwasbothverticallyandhorizontallyinequitable.
Thehorizontalinequityhadagreatereffectonredistributionofabilitytopaythanverticalinequity.
Thecomputationofcatastrophicexpenditureshowedthatasmallminorityofthepoorwerelikelytoincurcatastrophicexpenditurefrompayingthepremiumasituationthatcouldimpedetheachievementofuniversalcoverage.
Thestudyprovidesrecommendationstoimprovetheinequitablesystemofpremiumpaymenttohelpachieveuniversalcoverage.
IntroductionUniversalcoverageisachievedinahealthsystemwhenallresidentsofacountryareabletohaveaccesstoadequatehealthcareataffordableprices[1].
Theachieve-mentofsuchagoalrequirestheprovisionofadequatehealthcareaswellastheavailabilityofahealthcarefinancingsystemthatensuresaccesstoadequatecarere-gardlessofabilitytopay.
Inotherwords,thehealthcarefinancingsystemrequiredforauniversalcoveragehastobeequitable.
Ahealthcarefinancingsystemisequitableifhouseholdpaymentforhealthcareisinaccordancewithabilitytopayandutilizationisinaccordancewithneed[2].
Equitythenreferstotheabilityofthehealthsystemtoprovidefinancialprotectiontoall(potential)usersofhealthcare.
Theethicaljustificationforhealthcareequitycomesfromthestronglinkbetweenhealthcareandhealth.
Peopledonotdesirehealthcarebuthealth.
Howeverhealthisnotsoldinthemarketsotheyneedtopurchasehealthcareinordertoproducehealth.
Goodhealthisneededforpeopletothriveashumanbeings[3].
Thuswhatdrivespolicymakerstofinditfairtolinkpaymentofhealthcarewithitsutilizationorabilitytopayisthefactthathealthcarepaymentsarenotvoluntaryitemofexpenditurebutitiscausedbyunwantedhealthshockandthatsocietyasawholeiswillingtoshareinabsorb-ingtheburden[4].
Alternatively,othersarguethattheneedforequityisdrivenbythebeliefthathealthcareisarightandsoitistheobligationofsocietytoensurethateveryonehasaccesstothegoodregardlessofabilitytopay[5].
Equityinhealthcarefinancingcanbevertical(vari-ationinfinancialcontributionisproportionaltoabilitytopay)orhorizontal(peopleofthesameabilitytopaymakingthesamecontribution).
Studiesonverticalequityhavefocusedontheprogressivityofhealthcarefinancingwhichreferstothetendencyofhealthcarepay-ment,asaproportionofincome,toriseasincomerisesCorrespondence:eamporfu@gmail.
comKwameNkrumahUniversityofScienceandTechnology,Kumasi,Ghana2013Amporfu;licenseeBioMedCentralLtd.
ThisisanOpenAccessarticledistributedunderthetermsoftheCreativeCommonsAttributionLicense(http://creativecommons.
org/licenses/by/2.
0),whichpermitsunrestricteduse,distribution,andreproductioninanymedium,providedtheoriginalworkisproperlycited.
AmporfuInternationalJournalforEquityinHealth2013,12:4http://www.
equityhealthj.
com/content/12/1/4[3].
Thisimpliesthatinaprogressivehealthcarefinan-cingsystemthepoorspendalowerproportionoftheirincomeonhealthcarethantherich.
Aprogressivehealthfinancingreducesfinancialburdenonthepooringen-eralandhenceislikelytoprovidefinancialprotectionandaccesstohealthcare.
Horizontalequityhasreceivedlessattentionintheliteraturebutisimportantinensur-ingthatfinancialprotectionisnotprovidedonlytosomepoorresidentswhileotherpoorresidentsareleftwithoutprotection.
Thereasonisthathorizontalin-equitycouldimplythatsomepoorpeoplemakehigherhealthcarepaymentthanotherpoorpeople.
Inequityinhealthcarefinancingcancausefinancialburdenthroughtheeffectonincomeredistribution.
Incomeredistributionthatresultsfromhealthcarefinan-cingcanbeduetoverticalorhorizontalequity/inequity,and/orincomere-ranking.
Incomere-rankingoccurswhenincomeafterhealthcarepaymentmovesthepayerintoadifferentincomegroup.
Progressiveandhorizon-tallyequitablehealthcarepaymentcanhaveapositiveeffectonincomedistributionbyreducinganypreexistinginequityinincomedistribution.
Thusbothtypesofequityareimportantforuniversalhealthcoverage.
Thepurposeofthispaperistousenewsurveydatatomeasurethever-ticalandhorizontalequityofthepremiumcontributionoftheGhanaianNationalHealthInsuranceScheme(NHIS)andtheimplicationforuniversalcoverage.
Eventhoughtheinterestofthisstudyisinthetwotypesofequity,incomere-rankingisincludedintheanalysisbecausethemodelusedfortheanalysisisunabletoseparatehorizon-talequityfromincomere-ranking.
TheGhanaiannationalhealthinsurance(NHIS)Inordertoimproveequityinhealthcarefinancingandutilization,theNHISwasintroducedin2003toreplaceanexistinguserfeesystemlocallyknownastheCashandCarrySystem.
UndertheCashandCarrysystem,patientshadtopayforhealthcareservicesandsoutilizationwasskewedinfavoroftherich.
Certainexemptionsmeasureswereputinplacetocushionthenegativeeffectofuserfeesonthepoorbutthesemeasuresdidnotworkduetopoorimplementationandunclearguidelines.
Thustherewasnofinancialprotec-tionavailableforthepoor,andhealthcarefinancingwashighlyinequitable.
Sinceeverybodypaidthesamefeeforagivenservice,andagivenfeewaslikelytoformagreaterproportionoftheincomeforthepoorthantherich,thecashandcarrysystemwasregressive.
Eventhoughtheflatfeeforagivenservicewasbornebyallandsoonewouldexpectthatpeopleofthesamein-comewouldpaythesamefeeforthegivenservice,thevariationintheprobabilityofbeingsickwithinagivenincomegroupimpliesthatthehighriskinthegroupwerelikelytospendmoreonhealthcarethanthelowrisk.
Hence,theCashandCarrysystemwasbothverti-callyandhorizontallyinequitable.
TheNHISisasocialinsuranceschemewiththevisionofprovidingfinancialriskprotectionforbasicqualityhealthcareforitsmembers.
MembershiptotheschemeiscurrentlyvoluntaryandaccordingtotheNHISreportfor2009,enrolmenttotheschemeis62%oftheGhanaianpopulation.
TheNHISraisesrevenuefromsixsources:thehealthinsurancelevy(67%),insurancepremiums(5.
0%),SocialSecurityofNationalInsuranceTrust(SSNIT)contributions(15.
6%)investmentincome(17%),sectorbudgetsupport(2.
3%),andothersources(0.
2%)[6].
Studieshaveexaminedtheprogressivityofthehealthinsurancelevywhichis2.
5percentageofanexistingValueAddedTaxandhaveconcludedthatitisprogressive[7,8].
However[7],showedthatwhiletheoverallfinancingoftheNHISwasprogressive,thepre-miumrevenuewasregressive.
Inarecentstudy[9],foundthatthehealthinsurancelevyismildlyprogres-sivebutthepremiumrevenuestillregressive.
EquityofthepremiumisimportantinensuringmembershipbecauseresidentsmayhavetopaypremiuminordertoregisterwiththeNHIS.
Ifthepremiumimposesfinan-cialburdenonexistingorpotentialmembers,member-shipcouldfallandhenceimpedetheabilityoftheNHIStoachieveuniversalcoverage.
Tominimizethepotentialfinancialburdenthatcouldbeimposedbythepremiumseveralcategoriesofmem-bersareexemptfrompremiumpayment.
Forexample,toensurecoverageforchildrenandtoreducematernalandchildmorbidityandmortality,childrenundereight-eenyearsofage(whoseparentsareregisteredwiththeNHIS)andmaternalpatientsareexemptfrompayingpre-mium.
Thesecategoriesofmembersformabout54.
9%oftheregisteredmembers.
ThematernalexemptioncategorywasmadepossiblebyagrantfromthegovernmentofUK.
Indigentsandmembersaboveseventyyearsofagearealsoexemptfrompremiumpayment.
Adultsintheformalsec-torwhomakepensioncontributionstotheSocialSecurityandNationalInsuranceTrust(SSNIT)andSSNITpen-sionersdonotpaypremium.
Onlyadultsintheinformalsectorwhoformaboutthirtypercentofthemembershippaypremium.
Thusthirtypercentofthememberspaythepremiumwhichcontributesonly5.
0%ofthetotalrevenue.
Theschemethenreliesontheothersourcesoffundsforoperation.
EventhoughthepremiumrevenueformsonlyasmallpercentageofthetotalrevenuethatgoestotheNHIS,thepremiumisaveryimportantdeterminingfactorfortheachievementofuniversalcoverage.
Accordingtoarecentstudy[10],thepoorarelesslikelytoenrollwiththeNHISthantherich.
Thiscouldaffecttheabilityofthepoorthatarecurrentlyenrolledontheschemetoremainonit.
GiventhatthepercentageofindigentsinAmporfuInternationalJournalforEquityinHealth2013,12:4Page2of9http://www.
equityhealthj.
com/content/12/1/4Ghanais18[10],theformalsectoremploysabout10%oftheworkforce[11](implyingthat90%isemployedbytheinformalsector),about50.
5%ofthepopulationformtheworkforce[12],andassumingthattheindigentsarenotpartoftheworkforceitfollowsthatabout45.
5(0.
9*0.
505)percentofthepopulationwouldpaypremiumifthewholeGhanaianpopulationwereenrolledwiththescheme.
ThusgiventhecurrentGhanaianpopulationofabout24.
9million,thereareabout3.
6millionadultsplustheirchildrenwhoarenotenrolledinthescheme.
Highpremiumcouldbeoneofthereasonsfortheirlackofparticipationinthescheme.
Focusingmainlyontherevenuecontributionofthepremiumcouldmakeonemisstheimportantroleplayedbythepremiumindeter-miningmembershipandhencethesustainabilityofthescheme.
Theimportanceofequityofthepremiumthencannotbeoveremphasized.
TheNHISismadeupofDistrictMutualHealthSchemes(DMHS)regulatedbytheNationalHealthInsuranceAuthority.
Registeredpatientspaynofeeatthepointofpurchaseofhealthcare.
Membersregister/renewtheirmembershipandpaytheirpremiumattheDMHS.
PremiumsrangebetweenGHc7.
2andGHc48(US$4.
8andUS$32)accordingtoabilitytopay.
Thereisnoclearguidelineregardinghowmuchistobepaidaccordingtoagivenlevelofincome.
Inaddition,thelargeinformalsectormakesitevendifficultfortheDMHStohaveaccesstoinformationontheincomesofregisteredmembers.
Often,aflatpremiumischargedtoallmembersofagivendistrictmutualhealthschemebutadjustmentscouldbemadewheninformationonareaofresidence,householdsize,etc.
arerevealed.
Withaginiindex,of0.
40[13],incomedistributioninGhanaishighlyinequitable.
EventhoughtheextentoftheinequalityislessthanthatinSouthAfrica,itisworsethanthatofTanzania[14].
Thesignificantdisparityofin-comeimpliesthatpremiumpaymentshouldalsovarysig-nificantlytoensureequityinpremiumcontribution.
SincetheDMHSthatdeterminethepremiumcontributionsofmembersdonothaveinformationonmembers'income,itispossibleformembersofequalincomeandhouseholdsizetopayunequalpremiums.
Withhighasymmetricinformationonincomeintheinformalsectorwherethepoorarelikelytobefound,itispossibleforthepremiumcontributionoftheNHIStobeverticallyand/orhorizon-tallyinequitable.
Thequestionis:towhatextentdoesanyexistinginequityimposefinancialburdenonthepoorandhenceimpedetheachievementofuniversalcoverage.
UniversalhealthcoverageandequityAsindicatedinthedefinition,universalcoverageisachievedwhenallresidentshavefinancialprotectionfromhealthexpenditureandhaveaccesstoadequatehealthcare.
Thisimpliesthatuniversalcoverageleadstoequity.
Universalcoveragehasthusbeenadvocatedasameansofachievinghealthequity.
Existingliteraturehavethusshownhowtheachievementofuniversalcoveragehasledtoequityinthequalityofcare.
Forexample[15],showedthataftertheintroductionoftheuniversalcover-ageprogram,healthcarefinancinginequityinThailandreducedasaresultofacontinualreductionincata-strophicexpenditure.
Othershavearguedthatiftheap-proachofuniversalcoveragedoesnotstartwithcoverageforthepoor,butfocusesonthosethatareeasiesttoreach,thereisthepossibilityofinitialinequityastherichgetbet-teraccesstohealthcarewhilethepoorwaittogetthetricklingdownbenefits[16].
CountrieslikeBrazilandMexicothattargetedthepoortoensurethatthepoorre-ceiveatleastthesamebenefitastherich,i.
e.
,progressiveuniversalism,arelikelytoavoidtheinitialinequitybetweentherichandpoor[16].
TheGhanaianNHISisac-cessibletothepooraslongasthepremiumcontributiondoesnotimposefinancialburdenonthepoor.
ForuniversalcoveragetobeachievedthroughtheNHISallGhanaianresidents,regardlessofeconomicstatus,aretoregisterwiththescheme.
Thiscanhappenifnewmembersjoinand/ortheexistingmembersdonotleave.
Whentherevenuecollectionisverticallyin-equitableitmightimposefinancialburdenonthepooringeneralanddiscouragethemfromremainingwiththescheme.
Hence,verticalequityisimportantforuniversalcoverage.
EventhoughhorizontalequityhasnotreceivedanyattentionintheliteratureregardingtheequityoftheNHIS,horizontalequitymustalsobeimportantinen-suringuniversalcoverage.
Thereasonisthatevenifthepaymentisprogressivebuthorizontallyinequitable,itispossibleforthepaymenttoimposefinancialburdenonsomeofthepoorandhencediscouragethemfromremainingorjoiningthescheme.
HorizontalinequitycanalsotaintthepublicimageoftheNHISandreducemembershipandhencepreventorslowdowntheachievementofuniversalcoverage.
Thisisbecausewhileitisdifficultforalowincomeearnertodeterminethathigherincomeearnersarepayingdisproportionatelysmallerpercentageoftheirincome(i.
e.
,contributionsareregressive),itismucheasiertoobserveinequityinthepremiumpaymentamongpeopleofthesameincomegroup.
Thusinequityinpremiumcontributionsbypeopleinthesameincomegroupcoulddispleasemembersandhencereducemembershipunlessthecriterionforthevariationissociallyacceptable.
Universalcoveragecanstillbeachievedeveninthepresenceofverticalandhorizontalinequity,ifthepremiumpaymentdoesnotleadtocatastrophicexpend-iture.
Ifthepremiumisverysmallrelativetopeoples'abilitytopaythenitwillnotcausefinancialburdenevenonthepoorandallcanreceivefinancialprotection.
Thusinadditiontofindingtheequityofthepremium,AmporfuInternationalJournalforEquityinHealth2013,12:4Page3of9http://www.
equityhealthj.
com/content/12/1/4thestudyalsoexaminedthetendencyofthepremiumtocausecatastrophicexpenditure.
Totheauthor'sknow-ledgenosuchcomprehensivestudyhasbeendoneonthepremiumoftheNHIS.
AbilitytopayvariableIntheformalsector,incomelevelisusedasindicatorofabilitytopayorwelfare.
Incomeisregularizedandveri-fiableintheformalsectorandsoitisagoodindicatorofwelfare.
Intheinformalsectorhoweverincomeisir-regularandsostudiesoftenusehouseholdconsumptionexpenditureasameasureofabilitytopayorwelfare.
Thisisinaccordancewiththepermanentincomehy-pothesisthatstatesthatpeoplesmooththeirconsump-tionovertimeevenifincomeisnotregular.
Consumptionexpenditureasameasureofabilitytopayhoweverisnotwithoutproblem.
Themainweaknessofthemeasureistheimpliedassumptionthathealthcareexpendituredoesnotaffectsavingdecision.
Theuseofconsumptionexpenditureasameasureofwelfareorabilitytopayassumesthathouseholdswithhigherabil-itytopayhavehigherconsumptionexpenditurethanthosewithlowerabilitytopay.
Themeasurethendoesnottakeintoaccountthefactthathouseholdsmayhavetoborrowinordertoincreaseconsumptionexpenditure[17].
Besidestheabilitytopayofhouseholdsthatareabletoproducetheirownfoodstuffwouldbeunderesti-mated.
Anincreaseinconsumptionexpenditureforahouseholdcouldalsobeduetodebtrepaymentandhencemaynotimplyanincreaseingoodsandservicescon-sumedbutafallornochangeinwelfare.
Tominimizethenegativeeffectsconsumptionexpenditureasameasureofwelfareiscomputedasgrosshouseholdexpendituresonfoodandnon-fooditemsincludingtaxes,socialsecurity,aswellasalloutofpocketexpendituresonhealthcare[17].
Sincetheinterestinthecurrentstudyisinthedistri-butionalimpactofthepremium,thegrossexpenditureofhouseholdsonfoodandallotherhouseholdneedsinclud-inghealthcareexpenditurewasusedasanindicatorofabilitytopay.
MethodologyTheequityanalysiswasthreefold.
ThestudyfirstassessedtheprogressivityoftheNHISpremiumandmeasuredthedegreeofprogressivity.
Second,thestudycomputedtheredistributiveeffectofthepremium,ameasurethatinvolvesthecomputationofthehorizontalinequity.
Lastlythestudycomputedtheincidenceandintensityofcatastrophicexpenditureofthepremium.
Toassesstheprogressivityofthepremium,house-holdsinthedatawerecategorizedintoabilitytopayquintiles.
ConcentrationcurveandLorenzcurvesweredrawnonthesamegraphfortheabilitytopayquintilesandcompared.
Theconcentrationcurveplottedthecumulativepercentageofthepremiumcontributionagainstthecumulativepercentageofthesampleaccord-ingtoabilitytopayinincreasingorder.
Ifthepremiumcontributionwasthesameregardlessofabilitytopaytheconcentrationcurvewouldbeequaltothe45°line.
Ifthepremiumcontributionsofthepoorexceededthatoftherichthenthecurvewouldlieabovethe45degreeline,otherwiseitwouldliebelowtheline.
TheLorenzcurveistherepresentationofthecumulativedistribu-tionofabilitytopayamongthegroup.
IfabilitytopaywasevenlydistributedtheLorenzcurvewouldbeequaltothe45degreelineotherwiseitwouldbeconvexlyslopedunderthe45degreeline.
IftheconcentrationcurvewaseverywherebelowtheLorenzcurvethenthepremiumcollectionwouldbeprogressive.
ThepremiumcollectionwouldberegressiveiftheconcentrationcurvewaseverywhereabovetheLorenzcurve[17].
Themethodintroducedby[18]wasusedtomeasuretheabilitytopayredistributiveeffectofthepremium.
Theredistributionismeasuredas:REVHRg1gKpXαxGxpGxpCxpwheregistheaverageshareofabilitytopaytakenupbythepremium,KpistheKakwaniindexofpremiumpro-gressivity,αx,weight,isequaltotheproductofthesquareofpopulationshareofthosewithpre-premiumabilitytopayxandpostpremiumabilitytopayshareofthepre-premiumabilitytopayofthegroup,Gx-pisthepostpremiumGinicoefficientsofthosewithprepay-mentabilitytopayx,Cx-pisthepostpremiumconcen-trationindexwhichisobtainedbyfirstrankingthehouseholdsbytheirpre-premiumabilitytopayandthenwithineachequalabilitytopaygrouprankthembytheirpostpremiumabilitytopay.
Thefirsttermontherighthandside(V)measurestheverticalredistributionortheinequalityreductionthatwouldoccurintheab-senceofhorizontalinequity.
Thesecondterm(H)mea-sureshorizontalinequityandisequaltotheweightedsumofthepost-premiumabilitytopayGinicoefficientsoftheabilitytopaygroups.
Thethirdterm,(R)mea-suresre-rankingthatoccursfromthemovefromanabilitytopaygrouptoanotherasaresultofthepre-miumcontribution.
IfRiszerothereisnore-ranking.
AccordingtoAronsonetal.
,HincreaseswhileRfallsastherangeofincomeusedtodefine'equals'widens.
ThusadistinctionbetweenHandRisnotinteresting.
ThefocusoftheanalysiswasonthecomponentofVversusH+Rinincomeredistribution.
HoweverrerankingistypicallycausedbyhorizontalinequityandsoH+Rwasconsideredascapturinghorizontalinequity[4].
AmporfuInternationalJournalforEquityinHealth2013,12:4Page4of9http://www.
equityhealthj.
com/content/12/1/4IngeneralKakwaniindexequalstwicetheareabe-tweenthepremiumconcentrationcurveandtheLorenzcurveandiscomputedasKp=C-GwhereCisthepre-premiumconcentrationindexandGisthepre-premiumGinicoefficient.
The-2zandzeroother-wise.
Thismeasurementofincidencedoesnottakeintoaccountthedistributionofcatastrophicexpenditureandsogivesthesameweighttohouseholdswhoincurredcatastrophicexpenditureregardlessofabilitytopay.
Aweightedincidence,HWwasthusused:HW=H(1CE)where;CEistheconcentrationindexforE.
HWistherankweightedincidenceorheadcountandittakesintoaccountthedistributionofthecatastrophicexpenditure.
Therankweightedheadcountputsagreaterweightonthepoorhouseholdthatincurcatastrophicpaymentthantherich.
Thus,theHAnegativeCEmeansthatthepooraremorelikelytoexceedthethresholdthantherich.
Theintensityofanycatastrophicexpenditure(i.
e.
,theamountbywhichcatastrophicexpenditureexceedsthethreshold)thatmightexistwasmeasuredas:OXNi1OiNwhereOiEiPiyizhi.
Again,toadjustforthedistributionofcatastrophicexpendituretherankweightedovershootwasused:OW=O(1CO),whereCOistheconcentrationindexforO.
Themeanpositiveovershoot(MPO)wasalsocomputedtoprovideinformationontheaverageover-shootamongthosewhoexceededthethreshold.
MPOOH,thusz+MPOrepresentstheaverageexpenditureonpremiumasashareofabilitytopaybythosewhosepre-miumexceededthethreshold.
Tofindthecharacteristicsofthosewhoarelikelytoincurcatastrophicexpendituremultivariatelogisticregres-sionswererun:Ei=α1+α2X2+α3X3+α4X4+α5X5+eiwhereX2isavectorofdemographiccharacteristics:ageandgender(female);X3isavectorofvariablesformaritalstatus,X4istheabilitytopayvariable,andfi-nally,X5isthelocationdummyforKumasi.
Threeregressionswererun;oneforeachofthethresholds:5,10,and15.
ResultsanddiscussionDatadescriptionDatausedforthestudywerecollectedfromtheadmin-istrationofquestionnairesonNHISmembersrandomlyselectedfromthetwomaincitiesinthecountry:Accra(thecapitalcity)andKumasi(thecommercialcity).
Thereasonforthechoiceofurbanlocationswasduetothepotentialadvantages.
First,itiseasiertogetalargesam-pleinanurbanareathanaruralareaduetothedensepopulationintheurbanarea.
Second,theinformationonabilitytopaywhichisimportantforthestudyiscloselylinkedtotheextentofsubsistencelevelinlife-style.
Highsubsistenceintheruralareacaneasilycauseunderestimationofabilitytopayofruralhouseholds.
Forexample,afarmerinaruralareamaynotspendasmuchonfoodasanurbandwellerbecausethefarmermaygetsomefoodsuppliesfromthefarm.
Thefoodexpenditureisthuslikelytobemuchhigherfortheurbandwellerthantheruraldweller.
TherespondentswerehouseholdheadswhopaidpremiumforNHIS.
AmporfuInternationalJournalforEquityinHealth2013,12:4Page5of9http://www.
equityhealthj.
com/content/12/1/4ThepopulationofAccraandKumasiismorethanfourmillionmakingitimpossibletousecensusdataforthestudy.
Thechoiceofthesamplesizeusedforthestudyfollowedthestepsin[19].
Toensureasamplesizethatpredictsthepopulationwell,asampleerrorof±3,aconfidenceintervalof95%,andaconservativevariabilityof50%-toincorporatetheheterogeneityofthepopula-tioninincomedistribution-wereused.
Hencethecorre-spondingsamplesizeaccordingtothesamplesizeTablein[19]cannotbelessthan1,111.
Aftertheremovalofobservationswithmissinginformation,thesamplesizeusedthenwas1,529.
Table1showsthedescriptivestatisticsofthedatausedforthestudy.
Theaverageageofhouseholdheadswasabout40yearsandmostlyeducated.
Thisisnotasur-prisebecauseeducatedpeoplearemorelikelytohaveabetterunderstandingofthevalueofhealthinsurancethanthelesseducatedortheuneducated.
Thehouse-holdheadsweremostlymarriedandabout40%ofthemwerefemales.
DataonAccraformed58%ofthedata.
AccordingtotheNHIS2009report,theGreaterAccraregionhasthelowestenrolmentinthecountry.
Therea-sonforthelowerenrolmentismostlybecausemostoftheworkersintheformalsectorhaveprivatehealthin-suranceprovidedbyemployers.
ThusalargepercentageoftheenrolleesintheGreaterAccraregionarelikelytobefromtheinformalsector.
TheaverageannualpremiumperhouseholdwasGHc21.
00(US$13.
13)andtheaverageannualexpend-itureonfoodandotherswasGHc5,915.
07(US$3,696.
91).
Thus,onaverage,thepremiumformedabout0.
35%ofthetotalabilitytopay.
However,thehighvari-abilityintheabilitytopay(standarddeviationis5119.
5)andthelowervariabilityinthepremium(standarddevi-ationis15.
67)impliedthattheproportionofpremiumsonabilitytopaycouldbeveryhighforsomehouse-holds.
Thefiveincomegroupsshowtheextentofthevariationinabilitytopay.
Thepoorestgroup(firstquin-tal)hadanaverageannualexpenditureofGHc1,818.
50(US$1,136.
56)whichimpliesthathouseholdsinthisgrouplivedonUS$3.
11aday.
ThisamountwasjustabovethepovertylineofUS$2.
00aday,indicatingthatindeedonaveragetheindigentarenotmadetopaypre-mium.
However,withastandarddeviationof710.
32,itispossibleforthepremiumtobeaburdentosomeofthepoororthatsomeindigentswerepayingpremium.
EquityanalysisLorenzandconcentrationcurvesweredrawntofindtheprogressivityofthepremiums.
ThegraphinFigure1showsthattheconcentrationcurvewaseverywheregreaterthantheLorenzcurve,anindicationoftheregres-sivenessofthepremiumcontribution.
TheestimationoftheKakwaniindexproducedanindexof-0.
32,asreportedinTable2.
Thenegativesignindicatestheregressiveness,henceconfirmingtheresultsfromthecomparisonoftheconcentrationandLorenzcurves.
Tofindouthowregressivityvariedacrossvariouscat-egoriesofmembersanotherconvenienceregressionfortheKakwaniindexwasrunwiththeinclusionofdummyvariablesforKumasi,maritalstatus,education,andgen-der.
TheresultsarereportedinTable2andtheyshowthattheslopedummieswereallnegativeandwiththeexceptionofgenderallwerealsostatisticallysignificant.
SuchresultsimplythatpremiumsweremoreregressiveinKumasithanAccra.
TheresultsalsoshowedthatregressivitywashigheramongthetertiaryandsecondaryTable1DatasummaryVariablePercentageAge(average)40.
27yearsPremiumGHc21.
00(US$13.
13)Females40.
5AnnualExpenditureonfoodandothersGHc5,915.
07(US$3,696.
91)FirstQuintalGHc1,818.
50(US$1,136.
56)SecondQuintalGHc3,574.
81(US$2,234.
26)ThirdQuintalGHc5,444.
97(US$3,403.
11)FourthQuintalGHc7,530.
99(US$4,706.
87)FifthQuintalGHc12,276.
19(US$7,672.
62)Accra58.
0Education:Tertiary31.
5Secondary34.
1Basic30.
2Illiterate4.
2MaritalStatusMarried68.
7Divorced10.
5Widow/er1.
5Single16.
5Separated2.
60.
0020.
0040.
0060.
0080.
00100.
00120.
00PoorestsecondthirdfourthRichestLorenzConcentration45degreeFigure1Measuringverticalequityforthepremium.
AmporfuInternationalJournalforEquityinHealth2013,12:4Page6of9http://www.
equityhealthj.
com/content/12/1/4educationthanthebasicanduneducated,andhigheramongthemarriedthantheunmarried.
Genderhoweverdidnotaffecttheregressivity.
Theresultsonincomeredistributioneffectofthepre-miumhadVas-0.
00113whichformedlessthanonepercent(0.
28%)ofthetotalredistributiveeffectofthepremium.
Thiscouldbeduetothesmallshareofthepremiuminabilitytopay.
TheH+Rwas0.
3933form-ing99.
6%oftheRE.
Giventhatrerankingistypicallycausedbyhorizontalinequitythesumwastreatedinthestudyashorizontalinequity.
Thustheunequaltreatmentofequalswasfarmoreimportantintheredistributiveef-fectofthepremiumsthantheunequaltreatmentofunequalswhichhasreceivedattentionintheliterature.
Itisnotasurpriseforthehorizontalinequitytobeimportantbecausethelackofinformationonabilitytopayintheinformalsectorimpliesthattheamountofpremiumpaidcoulddependonfactorsotherthanabilitytopay.
Tofindouthowthepremiumpaidvariedaccordingtothecharacteristicsofmembers,alinearre-gressionwasrunwiththenaturallogofpremiumasthedependentvariableandthemembers'characteristics:age,abilitytopayvariable,location,maritalstatus,edu-cation,andgender.
Thenaturallogofpremiumwasusedbecausethefunctionalformofalinearregressiondoesnotrestrictthepredictedvaluestopositiveandsoitispossibletogetnegativepredictedpremiumshenceweakeningtheresults[20].
Theresults,reportedinTable3,showthatpremiumincreasedslightlywithabilitytopaybutfellformembersinKumasiaswellasthosewithtertiaryeducation.
Thisimpliesthataftercontrollingforabilitytopay,peoplewithtertiaryeducationinagivenincomegroupwerelikelytopayalowerpremiumthanthosewithlowerlevelofeducation.
ThelowerpremiuminKumasiisnotsurprisingbecauseKumasihasalowercostoflivingthaninAccra.
Moreimportantly,thelowpremiuminKumasirelativetoAccracouldexplainwhytheper-centageofNHISenrolmentinKumasiishigherthanthatinAccra.
TheresultsoncatastrophicexpenditurearereportedintheTable4.
Asexpected,boththeincidenceandtheintensitydecreasedwiththethreshold.
Theresultsshowthattheconcentrationindiceswereallnegativeimplyingthatforanygiventhreshold,thepoorweremorelikelytoincurcatastrophicexpenditurethantherich.
Thisisconsistentwiththeresultsontheregressivenessofthepremium.
Theadditionalinformationhereisthatthere-gressivenessofthepremiumcouldbeimposingfinancialburdenonthepoor.
However,theresultsalsoshowthatthefractionofthosethatwerelikelytoincurcata-strophicpaymentformedlessthan1.
5%ofthesampleregardlessofthethresholdused.
Thefractionevenreducedtolessthan1%whenthethresholdincreasedfrom5to10%andhigher.
Thisisconsistentwiththehorizontalinequitymeaningthatsomeofthepooraremadeworseoffbythepremium.
Thisimpliesthatonlyaverysmallpercentageofthesamplemakecatastrophicpayment.
Andthissmallpercentageofthesampleislikelytobepoor.
Table2RegressionresultsofKakwaniindicesIndependentvariablesCoefficientP-valueIndependentvariablesCoefficientP-valueConstant0.
100.
00Constant0.
1050.
000FractionalrankofFractionalrank0.
0830.
056Abilitytopay0.
320.
00Tertiary0.
0130.
580Secondary0.
0030.
898Female0.
0010.
938Kumasi0.
0840.
001married0.
0510.
005Tertiary*slope0.
1000.
014Secondary*slope0.
0850.
029Female*slope0.
0050.
868Kumasi*slope0.
2490.
000Married*slope0.
0920.
004Table3RegressionresultsonpremiumIndependentvariablesCoefficientP-valueConstant2.
9340.
000Age0.
0030.
158Abilitytopay0.
00010.
001Numberofadults0.
0050.
068Tertiary0.
1820.
000Secondary0.
0420.
330Female0.
0410.
225Kumasi0.
1200.
006Married0.
0420.
487AmporfuInternationalJournalforEquityinHealth2013,12:4Page7of9http://www.
equityhealthj.
com/content/12/1/4Theresultsontheintensityofthecatastrophicex-penditureshowedthattheaveragedegreebywhichthepremiumpaymentasashareofexpenditureexceededthethresholdwaslessthanonepercentregardlessofthethresholdused.
Withtheexceptionoftheovershootforthe5%thresholdwhichwasabout0.
1%,therestwereallclosetozero.
Theimplicationisthatwhenspreadoverthewholesample,thepaymentswereweaklycata-strophicrelativetothethresholds.
Howeverthemeanovershootamongthosewhosepaymentexceededthethreshold(MPO)exceeded7%regardlessofthethresh-old.
Thisimpliesthatonaverage,thosespendingmorethan5%oftheirabilitytopayonpremiumonaveragespent14.
67%whilethosespendingmorethan20%onaveragespent27.
14%.
Thepremiumthenimposesfinan-cialburdenonasmallminorityofthepoor.
ThelogisticregressionsresultsarereportedinTable5.
Theresultsforthe20%thresholdwerenotreportedbe-causetheywereverysimilartothoseof15%threshold.
Theresultsshowthatwiththeexceptionofabilitytopay,noneofthecharacteristicsofthehouseholdheadsignificantlyaffectedtheprobabilityofincurringcata-strophicexpenditure.
Theseresultsareconsistentwiththeresultsthatonlyasmallminorityofthepeoplein-curredcatastrophicexpenditure.
Becauseonlytheabilitytopayvariablewasstatisticallysignificant,alinearre-gressionwasrunwiththenaturallogoftheabilitytopayvariableasthedependentvariableandthecharacter-isticsofhouseholdheadastheindependentvariables.
TheresultsarereportedinthelastcolumnofTable5andtheyshowedthatabilitytopayincreasedwiththemarriedhouseholdheads,age,numberofadultsinthehousehold,andthoseinKumasi.
Theresultsimplythathouseholdswiththesecharacteristicsarenotlikelytoincurcatastrophicexpendituresfromthepremiumpay-ment.
TheseresultsconfirmtheconjecturethattheNHISenroleesinAccraarelikelytobepoor.
Theresultsalsoshowedthatpeoplewithsecondaryandbasiceduca-tionwerelikelytobepoorerthantheuneducatedandthosewithtertiaryeducation.
Thereasoncouldbethatpeoplewithlittleeducation(basicand/orsecondary)maynotbewillingtotakestrenuousjobsthatmaybebetterpayingthanthe'suitable'jobs(suchasreception-ists,storekeeper,etc.
)forthenotsoeducated.
ConclusionThestudyhasexaminedequityofthepremiumcontri-butionoftheGhanaianNationalHealthInsuranceScheme.
Theanalysisfocusedonverticalandhorizontalequityaswellasthepossibilityofthepremiumimposingcatastrophicexpenditureonmembers.
Theresultsshowthathorizontalinequitydominatesverticalinequity.
Eventhoughthepremiumisregersssive,theregressive-nessdoesnotaffectredistributionofabilitytopayasthehorizontalinequity.
Aftercontrollingforabilitytopay,thestudyhasshownthatpremiumsarehigherinAccrathanKumasiandloweramongmemberswithtertiaryeducationthanthosewithouttertiaryeducation.
Inadditionthestudyhasshownthatthepremiumislikelytoimposecatastrophicexpenditureonasmallminorityofthepoor.
Sinceuniversalcoveragerequiresallresi-dentshavingfinancialprotection,havingsomeresidents,nomatterhowsmallthepercentage,facingcatastrophicexpenditurecanimpedetheachievementofuniversalcoverage.
Thestudymakesthreepolicyrecommenda-tionstoimproveequity.
Giventhattheindigentaretobeexemptedfromthepaymentofpremium,theresultinthestudycouldimplythatitispossibleforsomeindigenttobepayingpremium.
Thusifextraeffortsaremadetoidentifythepoorintheurbanareaswhoqualifyforexemptionthepercentageofthoseincurringcatastrophicexpenditurecouldbefurtherreduced.
ThereasonforthelargehorizontalinequitycouldbeduetothelackofuniformguidelinestoDistrictsMutualHealthSchemesastohowmuchtochargeaccordingtocharacteristicsthatarerelatedtotheeconomicstatusofTable4Incidenceandintensityofcatastrophicexpenditure5%10%15%20%Incidence(H)0.
62%0.
41%0.
21%0.
14%Standarddeviation0.
0780.
0640.
0450.
037Overshoot(O)0.
06%0.
03%0.
02%0.
01%Standarddeviation0.
0100.
0070.
0050.
003Meanpositiveovershoot(MPO)9.
67%7.
31%9.
52%7.
14%Averageexpenditure14.
67%17.
31%24.
52%27.
14%ConcentrationindexforEi(CE)0.
9600.
9550.
9640.
954Rankweightedincidence(HW)1.
22%0.
80%0.
41%0.
27%ConcentrationindexforO(CO)0.
9251.
1060.
8330.
932Rankweightedovershoot(OW)0.
12%0.
06%0.
04%0.
02%Table5Logisticregressionresultsonfactorsaffectingcatastrophicexpenditure5%10%15%AbilitytopayConstant7.
01323.
45028.
3547.
604*Age0.
0060.
0590.
1040.
011*Female0.
1720.
62026.
4770.
012Abilitytopay*0.
0030.
0020.
022Numberofadults0.
5600.
8147.
3990.
062*Tertiary5.
3587.
192110.
6080.
013Secondary3.
9836.
51578.
9810.
279*Basic4.
8706.
75072.
999-.
251*Kumasi0.
2180.
5445.
6620.
704*Married0.
31115.
5968.
7300.
183**Statisticallysignificanceat5%significancelevel.
AmporfuInternationalJournalforEquityinHealth2013,12:4Page8of9http://www.
equityhealthj.
com/content/12/1/4members.
Suchcharacteristicscouldincludeage,occu-pation,areaofresidence,housing,etc.
Forexample,smallrestaurantowners,seamstresses/tailors,andhairsalonownerscouldpaythesamepremiumwhilegar-deners,pettytraders,andporterspayonepremium.
Peopleinasimilaroccupationlivinginthesameorsimilarresidentialareaarealsolikelytobehomogeneousineconomicstatus.
FollowingtheGhanaLivelihoodEmpowermentprogram,thecommunitycouldbeusedtoconfirmtheeconomicstatusclaimedbyindividuals.
Certainly,suchaprogramwouldnotbewithoutpro-blemsbutwouldhelpimprovesignificantlythecurrentinequityproblem.
Giventhatthepremiumpaymentimposescatastrophicexpenditureononlyasmallminorityofthemembers,itmaynotbenecessarytocancelthepremiumasasourceofrevenue.
Minimizingthecatastrophicexpendituremaybemorebeneficialthanscrapingthepremium.
CompetinginterestTheauthordeclaresthattheyhavenocompetinginterests.
Author'scontributionEAconceivedtheidea,rantheregressionsandwrotethewholemanuscript.
Received:14March2012Accepted:3January2013Published:7January2013References1.
CarrinG,JamesC:Reachinguniversalcoverageviasocialhealthinsurance:keydesignfeaturesinthetransitionperiod,DiscussionPaperNumber2.
Geneva:WorldHealthOrganization;2004.
2.
McIntyreD:LearningfromExperience:Healthcarefinancinginlow-andmiddle-incomecountries.
Geneva:GlobalForumforHealthResearch;2007.
3.
WagstaffA,DoorslaerEV:Equityinhealthcarefinanceanddelivery.
InHandbookofHealthEconomics.
1Bthedition.
EditedbyCulyerAJ,NewhouseJP.
SouthHolland:Elsevier;2000:18.
3–1862.
4.
WagstaffA,MeasuringEquityinHealthCareFinancing:ReflectionsonandAlternativestotheWorldHealthOrganization'sFairnessofFinancingIndex.
;1999.
http://www-wds.
worldbank.
org/external/default/WDSContentServer/IW3P/IB/2001/03/20/000094946_01030305572311/additional/146546323_20041118130716.
pdf.
5.
HendersonJW:HealthEconomicsandPolicy.
3rdedition.
SouthWestern:Thomson;2005.
6.
2010Annualreport-NationalHealthInsurance.
2011.
http://www.
nhis.
gov.
gh/_Uploads/dbsAttachedFiles/8%281%29.
pdf.
7.
AkaziliJ,GyapongJ,McIntyreD:WhopaysforhealthcareinGhanaInternationalJournalforEquityinHealth2011,10:26.
http://www.
equityhealthj.
com/content/pdf/1475-9276-10-26.
pdf.
8.
PrichardW,BentumI:TaxationandDevelopmentinGhana:Finance,EquityandAccountability.
2009.
http://www.
taxjustice.
net/cms/upload/pdf/Ghana_0906_Report_printer_friendly.
pdf.
9.
AkaziliJ,GarshongJ,AikinsB,GyapongJ,McIntyreD:ProgressivityofhealthcarefinancingandincidenceofservicebenefitsinGhana.
HealthPolicyPlan2012,27(suppl.
1):i13–i22.
10.
LeppertG,SchützeichelP:IdentificationandIntegrationofIndigentsintheNHIS.
http://www.
microhealthinsurance-africa.
org/fileadmin/user_upload/Ghana-Workshop/Leppert_Schuetzeichel_2010_Identification_and_Integration_of_idigents_in_the_NHIS.
pdf.
11.
OtooKN,Osei-BoatengC,Asafo-AdjayeP:TheLabourMarketinGhana:AdescriptiveanalysisofthelabourmarketcomponentoftheGhanaLivingStandardSurvey(V).
http://www.
ghanatuc.
org/wp-content/uploads/2012/01/The-Labour-Market-in-Ghana.
pdf.
12.
GhanaStatisticalServices:PopulationandHousingCensus,SummaryofFinalResults.
2010.
https://www.
cia.
gov/library/publications/the-world-factbook/fields/2172.
html.
13.
CentralIntelligenceAgency:TheWorldFactbook.
https://www.
cia.
gov/library/publications/the-world-factbook/fields/2172.
html.
14.
USAID:Ghana:EconomicPerformanceAssessment.
http://egateg.
usaid.
gov/sites/default/files/Ghana_Economic_Performance_Assessment.
pdf.
15.
PrakongsaiP,LimwattananonS,TangcharoensathienV:\coverage:lessonsfromThailand.
AdvHealthEconHealthServRes2009,21:57–81.
16.
GwatkinDR,ErgoA:Universalhealthcoverage:friendorfoeofhealthequityLancet2011,377(9784):2160–2161.
17.
O'DonnellO,DoorslaerE,WagstaffA,LindelowM:AnalyzingHealthEquityUsingHouseholdSurveyData.
WashingtonDC:TheWorldBank;www.
worldbank.
org/analyzinghealthequity.
18.
AronsonJR,JohnsonP,LambertPJ:Redistributiveeffectandunequalincometaxtreatment.
EconJ1994,104:262–270.
19.
IsraelGD:DeterminingSampleSize.
https://edis.
ifas.
ufl.
edu/document%20pd005.
20.
KennedyP:AGuidetoEconometrics.
Massachusetts:MITPress;2003.
doi:10.
1186/1475-9276-12-4Citethisarticleas:Amporfu:EquityofthepremiumoftheGhanaiannationalhealthinsuranceschemeandtheimplicationsforachievinguniversalcoverage.
InternationalJournalforEquityinHealth201312:4.
SubmityournextmanuscripttoBioMedCentralandtakefulladvantageof:ConvenientonlinesubmissionThoroughpeerreviewNospaceconstraintsorcolorgurechargesImmediatepublicationonacceptanceInclusioninPubMed,CAS,ScopusandGoogleScholarResearchwhichisfreelyavailableforredistributionSubmityourmanuscriptatwww.
biomedcentral.
com/submitAmporfuInternationalJournalforEquityinHealth2013,12:4Page9of9http://www.
equityhealthj.
com/content/12/1/4