CHILDREN AND FAMILIES
EDUCATION AND THE ARTS
The RAND Corporation is a nonprofit institution that helps improve policy and
decisionmaking through research and analysis.
ENERGY AND ENVIRONMENT
HEALTH AND HEALTH CARE
INFRASTRUCTURE AND
TRANSPORTATION
This electronic document was made available from www.rand.org as a public service
of the RAND Corporation.
INTERNATIONAL AFFAIRS
LAW AND BUSINESS
Skip all front matter: Jump to Page 16
NATIONAL SECURITY
POPULATION AND AGING
PUBLIC SAFETY
SCIENCE AND TECHNOLOGY
TERRORISM AND
HOMELAND SECURITY
Support RAND
Browse Reports & Bookstore
Make a charitable contribution
For More Information
Visit RAND at www.rand.org
Explore the Pardee RAND Graduate School
View document details
Limited Electronic Distribution Rights
This document and trademark(s) contained herein are protected by law as indicated in a notice appearing
later in this work. This electronic representation of RAND intellectual property is provided for noncommercial use only. Unauthorized posting of RAND electronic documents to a non-RAND website is
prohibited. RAND electronic documents are protected under copyright law. Permission is required from
RAND to reproduce, or reuse in another form, any of our research documents for commercial use. For
information on reprint and linking permissions, please see RAND Permissions.
This product is part of the Pardee RAND Graduate School (PRGS) dissertation series.
PRGS dissertations are produced by graduate fellows of the Pardee RAND Graduate
School, the world’s leading producer of Ph.D.’s in policy analysis. The dissertation has
been supervised, reviewed, and approved by the graduate fellow’s faculty committee.
Dissertation
China’s Health Insurance
Reform and Disparities
in Healthcare Utilization
and Costs
A Longitudinal Analysis
Henu Zhao
C O R P O R AT I O N
Dissertation
China’s Health Insurance
Reform and Disparities
in Healthcare Utilization
and Costs
A Longitudinal Analysis
Henu Zhao
This document was submitted as a dissertation in October 2014 in
partial fulfillment of the requirements of the doctoral degree in public
policy analysis at the Pardee RAND Graduate School. The faculty
committee that supervised and approved the dissertation consisted
of Hao Yu (Chair), Emmett Keeler, and Gema Zamarro.
PA R D E E R A N D GRADUATE SCHOOL
The Pardee RAND Graduate School dissertation series reproduces dissertations that
have been approved by the student’s dissertation committee.
The RAND Corporation is a nonprofit institution that helps improve policy and
decisionmaking through research and analysis. RAND’s publications do not necessarily
reflect the opinions of its research clients and sponsors.
R® is a registered trademark.
Permission is given to duplicate this document for personal use only, as long as it is unaltered
and complete. Copies may not be duplicated for commercial purposes. Unauthorized
posting of RAND documents to a non-RAND website is prohibited. RAND documents are
protected under copyright law. For information on reprint and linking permissions, please
visit the RAND permissions page (http://www.rand.org/publications/permissions.html).
Published 2015 by the RAND Corporation
1776 Main Street, P.O. Box 2138, Santa Monica, CA 90407-2138
1200 South Hayes Street, Arlington, VA 22202-5050
4570 Fifth Avenue, Suite 600, Pittsburgh, PA 15213-2665
RAND URL: http://www.rand.org/
To order RAND documents or to obtain additional information, contact
Distribution Services: Telephone: (310) 451-7002;
Fax: (310) 451-6915; Email: [email protected]
Table of Contents Tables.........................................................................................................................................................v
Figures.....................................................................................................................................................ix
Abstract...................................................................................................................................................xi
Acknowledgements...........................................................................................................................xiii
Chapter1Introduction........................................................................................................................1
Chapter2Background.........................................................................................................................3
2.1HealthinsurancereforminChina........................................................................................................3
2.1.1Collapseofhealthinsuranceschemesinthe1970sand1980s......................................................4
2.1.2Earlyeffortsinthe1980sandearly1990s..............................................................................................5
2.1.3Healthinsurancereformsincethelate1990s........................................................................................6
2.1.4Healthcarereformafter2009........................................................................................................................9
2.2ThreeMajorHealthInsuranceSchemes.........................................................................................10
2.2.1TheBasicMedicalInsuranceforUrbanEmployees...........................................................................10
2.2.2TheBasicMedicalInsuranceforUrbanResidents..............................................................................11
2.2.3TheNewRuralCooperativeMedicalInsurance...................................................................................13
2.3TrendsandCurrentStatusofHealthcareDisparities................................................................13
Chapter3LiteratureReviewandStudyObjectives................................................................19
3.1ExistingResearch....................................................................................................................................19
3.1.1LiteratureonRural–UrbanDisparitiesinHealthcareUtilization.................................................19
3.1.2LiteratureonDisparitiesinOut‐of‐PocketExpenditureandHealthcareCosts......................21
3.1.3LiteratureonDisparitiesinHealthInsuranceCoverage..................................................................22
3.1.4MethodologicalIssues.....................................................................................................................................22
3.2GapintheExistingLiterature.............................................................................................................26
3.3ObjectivesandResearchQuestions..................................................................................................27
Chapter4StudyDesign.....................................................................................................................28
4.1Data..............................................................................................................................................................28
4.2StudyPeriods............................................................................................................................................30
4.3ConceptualModelandVariableSelection......................................................................................30
4.3.1DependentVariables.......................................................................................................................................31
4.3.2IndependentVariables...................................................................................................................................33
4.4AnalyticApproach..................................................................................................................................38
4.4.1Difference‐in‐DifferencesAnalysiswithMultipleGroupsandMultipleTimePeriods........38
4.4.2MultivariateRegressionfortheVariablesthatdonotmeettheAssumptionofParallel
Trends...............................................................................................................................................................................44
4.5Sensitivityanalysis.................................................................................................................................46
4.5.1ControllingforInsuranceStatus.................................................................................................................46
4.5.2DroppingtheRichestProvinceorthePoorestProvince..................................................................46
iii
4.5.3IncludingInteractionTermswithHouseholdIncome......................................................................47
4.5.4DIDAnalysisResultsforVariablesinWhichParallelTrendsdidnotHold..............................47
Chapter5Results:DisparitiesinHealthcareUtilization......................................................48
5.1DescriptiveAnalysis...............................................................................................................................48
5.2DIDAnalysisforFormalCareUtilizationandOutpatientUtilization..................................51
5.3MultivariateAnalysisControllingforExistingTrendsforInpatientUtilization.............57
5.4SensitivityAnalysis................................................................................................................................64
5.4.1ControllingforInsuranceStatus.................................................................................................................64
5.4.2DroppingtheRichestProvinceorthePoorestProvince..................................................................71
5.4.3IncludingInteractionTermswithHouseholdIncome......................................................................80
5.4.4DIDAnalysisforInpatientCare..................................................................................................................84
5.5SummaryofFindings.............................................................................................................................85
Chapter6Results:Disparitiesinhealthcarecosts..................................................................88
6.1DescriptiveAnalysis...............................................................................................................................88
6.2MultivariateAnalysisControllingforExistingTrends..............................................................91
6.3SensitivityAnalysis..............................................................................................................................103
6.3.1controllingforhealthinsurancestatus.................................................................................................103
6.3.2droppingtherichestprovinceorthepoorestprovince.................................................................107
6.3.3.Includinginteractiontermswithhouseholdincome.....................................................................116
6.3.4DIDanalysisresultsforcostvariables..................................................................................................131
6.4SummaryofFindings...........................................................................................................................133
Chapter7Conclusion,Discussion,andPolicyImplications..............................................135
7.1Conclusion...............................................................................................................................................135
7.2Discussion................................................................................................................................................137
7.2.1ComparingWiththePublishedResearch............................................................................................137
7.2.2Strengths...........................................................................................................................................................138
7.2.3Limitations........................................................................................................................................................139
7.2.4FutureDirections...........................................................................................................................................140
7.3PolicyImplications...............................................................................................................................140
Appendix.............................................................................................................................................143
Reference............................................................................................................................................145
iv
Tables Table4.1SampleSizebyRuralandUrbanResidencesandRegistrations....................................................29
Table4.2DescriptiveStatisticsofIndependentVariablesbyRuralandUrbanResidencesand
Registrations..................................................................................................................................................................37
Table4.3ResultsofDIDAnalysisUsing1993and1997WavesforHealthcareUtilization...................42
Table4.4ResultsofDIDAnalysisUsing1993and1997WavesforHealthcareCosts.............................44
Table5.1DIDAnalysisResultsforFormalCareUtilizationandOutpatientUtilization..........................54
Table5.2TestResultsforDIDAnalysisofFormalCareUtilizationandOutpatientUtilization...........55
Table5.3MultivariateAnalysisResultsforInpatientCareUtilization..........................................................59
Table5.4TestResultsofDisparitiesforInpatientCareUtilization.................................................................60
Table5.5TestResultsofChangeinDisparitiesforInpatientCareUtilization............................................62
Table5.6DIDAnalysisResultsofFormalCareandOutpatientUtilization(ControllingforInsurance
Status)...............................................................................................................................................................................65
Table5.7TestResultsforDIDAnalysisofHealthcareUtilization(ControllingforInsuranceStatus)
.............................................................................................................................................................................................66
Table5.8MultivariateAnalysisResultsforInpatientCareUtilization(ControllingforInsurance
Status)...............................................................................................................................................................................67
Table5.9TestResultsofDisparitiesforInpatientCareUtilization(ControllingforInsuranceStatus)
.............................................................................................................................................................................................69
Table5.10TestResultsofChangeinDisparitiesforInpatientCareUtilization(Controllingfor
InsuranceStatus).........................................................................................................................................................70
Table5.11DIDAnalysisResultsforFormalCareandOutpatientUtilization(DroppingtheRichest
Province).........................................................................................................................................................................73
Table5.12TestResultsforFormalCareandOutpatientUtilization(DroppingtheRichestProvince)
.............................................................................................................................................................................................74
Table5.13DIDAnalysisResultsforFormalCareandOutpatientUtilization(DroppingthePoorest
Province).........................................................................................................................................................................75
Table5.14TestResultsforFormalCareandOutpatientUtilization(DroppingthePoorestProvince)
.............................................................................................................................................................................................76
v
Table5.15MultivariateAnalysisResultsforInpatientUtilization(DroppingtheRichest/Poorest
Province).........................................................................................................................................................................77
Table5.16TestResultsofDisparitiesinInpatientUtilization(DroppingtheRichest/poorest
Province).........................................................................................................................................................................78
Table5.17TestResultsofChangeinDisparitiesforInpatientCareUtilization(Droppingthe
Richest/poorestProvince).......................................................................................................................................79
Table5.18DIDAnalysisResultsforFormalCareandOutpatientUtilizations(IncludingInteraction
TermwithHouseholdIncome)..............................................................................................................................82
Table5.19TestResultsforFormalCareandOutpatientUtilizations(IncludingInteractionTerm
withHouseholdIncome)..........................................................................................................................................83
Table5.20DIDAnalysisResultsforInpatientCareUtilization..........................................................................84
Table5.21TestResultsforInpatientCareUtilization(DIDAnalysis)............................................................85
Table6.1MultivariateAnalysisResultsforOOPExceedingCertainPercentageofHouseholdIncome
.............................................................................................................................................................................................93
Table6.2MultivariateAnalysisResultsforTotalHealthcareCosts.................................................................95
Table6.3TestResultsofDisparitiesforOOPExceedingCertainPercentageofHouseholdIncome
..........................................................................................................................................................................................100
Table6.4TestResultsofChangesinDisparitiesforOOPExceedingCertainPercentageofHousehold
Income...........................................................................................................................................................................101
Table6.5BootstrapResultsforDisparitiesinTotalHealthCosts.................................................................103
Table6.6Multi‐variateAnalysisResultsforOOPExceedingCertainPercentageofHousehold
Income(ControllingforInsurance)..................................................................................................................104
Table6.7TestResultsofDisparitiesforOOPExceedingCertainPercentageofHouseholdIncome
(ControllingforInsurance)...................................................................................................................................105
Table6.8TestResultsofChangesinDisparitiesforOOPExceedingCertainPercentageofHousehold
Income(ControllingforInsurance)..................................................................................................................106
Table6.9BootstrapResultsforDisparitiesinTotalHealthCost(ControllingforInsurance)...........107
Table6.10Multi‐variateAnalysisResultsforOOPExceedingCertainPercentageofHousehold
Income(DroppingtheRichestProvince).......................................................................................................109
Table6.11TestResultsofDisparitiesforOOPExceedingCertainPercentageofHouseholdIncome
(DroppingtheRichestProvince)........................................................................................................................110
vi
Table6.12TestResultsofChangesinDisparitiesforOOPExceedingCertainPercentageof
HouseholdIncome(DroppingtheRichestProvince)................................................................................111
Table6.13Multi‐variateAnalysisResultsforOOPExceedingCertainPercentageofHousehold
Income(DroppingthePoorestProvince).......................................................................................................112
Table6.14TestResultsofDisparitiesforOOPExceedingCertainPercentageofHouseholdIncome
(DroppingthePoorestProvince).......................................................................................................................113
Table6.15TestResultsofChangesinDisparitiesforOOPExceedingCertainPercentageof
HouseholdIncome(DroppingthePoorestProvince)...............................................................................114
Table6.16BootstrapResultsforDisparitiesinTotalHealthCosts(DroppingtheRichestProvince)
..........................................................................................................................................................................................115
Table6.17BootstrapResultsforDisparitiesinTotalHealthCost(DroppingthePoorestProvince)
..........................................................................................................................................................................................116
Table6.18Multi‐variateAnalysisResultsforOOPExceedingCertainPercentageofHousehold
Income(Low‐incomeFamilies)..........................................................................................................................118
Table6.19TestResultsofDisparitiesforOOPExceedingCertainPercentageofHouseholdIncome
(Low‐incomeFamilies)...........................................................................................................................................119
Table6.20TestResultsofChangesinDisparitiesforOOPExceedingCertainPercentageof
HouseholdIncome(Low‐incomeFamilies)...................................................................................................120
Table6.21Multi‐variateAnalysisResultsforOOPExceedingCertainPercentageofHousehold
Income(Medium‐incomeFamilies)..................................................................................................................122
Table6.22TestResultsofDisparitiesforOOPExceedingCertainPercentageofHouseholdIncome
(Medium‐incomeFamilies)..................................................................................................................................123
Table6.23TestResultsofChangesinDisparitiesforOOPExceedingCertainPercentageof
HouseholdIncome(Medium‐incomeFamilies)...........................................................................................124
Table6.24Multi‐variateAnalysisResultsforOOPExceedingCertainPercentageofHousehold
Income(High‐incomeFamilies).........................................................................................................................126
Table6.25TestResultsofDisparitiesforOOPExceedingCertainPercentageofHouseholdIncome
(High‐incomeFamilies)..........................................................................................................................................127
Table6.26TestResultsofChangesinDisparitiesforOOPExceedingCertainPercentageof
HouseholdIncome(High‐incomeFamilies)..................................................................................................128
Table6.27BootstrapResultsforDisparitiesinTotalHealthCosts(Low‐incomeFamilies)..............129
Table6.28BootstrapResultsforDisparitiesinTotalHealthCosts(Medium‐incomeFamilies)......130
vii
Table6.29BootstrapResultsforDisparitiesinTotalHealthCosts(High‐incomeFamilies).............130
Table6.30DIDAnalysisResultsforOOPExceedingCertainPercentageofHouseholdIncome.......132
Table6.31TestResultsforOOPExceedingCertainPercentageofHouseholdIncome(DIDAnalysis)
..........................................................................................................................................................................................132
Table6.32BootstrapResultsforDisparitiesinTotalHealthCosts(DIDAnalysis)...............................133
viii
Figures Figure2.1HealthInsuranceCoverageinUrbanandRuralAreasinChina,SelectedYears1993‐2008
.............................................................................................................................................................................................15
Figure2.2HealthServiceUtilizationinUrbanandRuralAreasinChina(2003)......................................16
Figure2.3HealthcareSpendinginChina,bySourceandYear...........................................................................17
Figure2.4PerCapitaOut‐of‐PocketHealthExpensesasaPercentageofIncome.....................................18
Figure4.1UpdatedStructureofAndersonModel...................................................................................................31
Figure5.1ProbabilityofFormalCareUtilizationin4WeeksbyRuralandUrbanResidencesand
Registrations..................................................................................................................................................................48
Figure5.2ProbabilityofOutpatientCareUtilizationin4WeeksbyRuralandUrbanResidencesand
Registrations..................................................................................................................................................................49
Figure5.3ProbabilityofInpatientCareUtilizationin4WeeksbyRuralandUrbanResidencesand
Registrations..................................................................................................................................................................50
Figure5.4PredictedProbabilityofFormalCareUtilizationin4WeeksbyRuralandUrban
ResidencesandRegistrations.................................................................................................................................56
Figure5.5PredictedProbabilityofOutpatientCareUtilizationin4WeeksbyRuralandUrban
ResidencesandRegistrations.................................................................................................................................57
Figure5.6PredictedProbabilityofInpatientCareUtilizationin4WeeksbyRuralandUrban
ResidencesandRegistrations.................................................................................................................................63
Figure6.1ProbabilityofHavingOut‐of‐pocketMedicalExpenseExceeding20%ofHousehold
IncomebyRuralandUrbanResidencesandRegistrations.......................................................................89
Figure6.2ProbabilityofHavingOut‐of‐pocketMedicalExpenseExceeding40%HouseholdIncome
byRuralorUrbanResidencesandRegistrations...........................................................................................90
Figure6.3TotalHealthcareCostsbyRuralandUrbanResidencesandRegistrations............................91
Figure6.4PredictedProbabilityofHavingOOPExceeding20%ofHouseholdIncomebyRuraland
UrbanResidencesandRegistrations...................................................................................................................97
Figure6.5PredictedProbabilityofHavingOOPExceeding40%ofHouseholdIncomebyRuraland
UrbanResidencesandRegistrations...................................................................................................................98
Figure6.6PredictedTotalHealthcareCostsbyRuralandUrbanResidencesandRegistrations.......98
ix
Abstract China’seconomicsuccessduringthepast30yearswasnotmirroredinitshealth
caresystem.Asaresult,therural‐urbandisparitiesinhealthinsurancecoverageandthe
relatedhealthcareareasbecameprominent.Sincethelate1990s,Chinahasbeen
expandinginsurancecoverage,inordertoprovideaccessibleandaffordablehealthcareto
allresidents.Mystudyanalyzeswhethertheinsuranceexpansionreducesrural‐urban
disparitiesintermsofhealthcareutilizationandfinancialprotection.Tomyknowledge,
thisisthefirststudytoaddressthedisparityissuebyexaminingChina’shealthcarereform
policiesoveranextended18‐yearperiod(1993‐2011).Itisalsothefirststudytoaddress
thedynamicphenomenonofrural‐urbanmigrationduringthestudyperiodbyseparating
thestudygroupinto4subgroupsintermsofrespondentsinresidentialareasversus
householdregistrationtype.
DrawingonsevenwavesofdatafromtheChinaHealthandNutritionSurveyand
applyingmultivariateanalysistechniques,suchasdifference‐in‐differenceanalysisand
generalizedlinearmodel,Ifindthatrural‐urbandisparitiesinformalcareandoutpatient
utilizationweresignificantlyreducedbytheexpandedhealthinsurancecoverageinrural
areain2003.Therural‐urbandisparityintotalhealthcostsisalsosignificantlyreduced.
However,noevidenceshowsthatthepolicychangesinhealthinsurancecoveragehad
impactondisparitiesininpatientutilizationorhavinghighout‐of‐pocketpayments.By
conductingseveralsetsofsensitivityanalyses,mystudyalsofindsthattheexpanded
healthinsurancecoverageimpactedricherprovincemorethanpoorerprovinces,and
impacthigh‐incomefamiliesmorethanmedium‐andlow‐incomefamilies.
xi
ThestudyfindingshaveimportantpolicyimplicationsforChina’songoinghealth
carereform.First,China’spolicymakersshouldprovidebetterhealthcarecoverageand
morehealthcareresourcestoruralareastofurtherreducetherural‐urbandisparity.
Second,sincepriorpolicychangesaffectedrichprovincemorethanpoorprovince,new
policyshouldtargetspecificallypoorprovinces.Third,giventhefindingthatthepositive
impactonhealthcareutilizationofpolicychangein2003happeningmainlyinhigh‐income
groups,newpolicychangeshouldfocusmoreonmedium‐andlow‐incomegroup.
xii
Acknowledgements Iamgratefulforthesupportprovidedbymywonderfuldissertationcommittee:Dr.
HaoYu,Dr.GemaZamarro,andDr.EmmettKeeler.Thesuccessfulcompletionofthis
dissertationwasaconsequenceoftheirexcellentguidance.Iamespeciallythankfulfor
mentorshipofmyCommitteeChair,Hao.Histimelyfeedbacksonourweeklymeetings
werecrucialtokeepmeontherighttrack.IwouldalsoliketothankGemaandEmmettfor
theirinsightfulandconstructiveadvicesonthepolicycontextandmethodologicalissues.I
alsowanttothankmyoutsidereaderTeh‐weiHu,ProfessorEmeritusofHealthEconomics,
UniversityofCalifornia,Berkeley,forhishelpfulandresponsivecommentsonmy
dissertation.
IwouldalsoliketothankmyresearchmentorNelsonLim.Hetaughtmehowtodo
researchandhowtowrite,andprovidedmewithadvicesandencouragementduringmy
dissertationwork.IwouldalsoliketothankthePRGSfaculty,staffandstudentsfortheir
helpduringmydissertationwriting.
Thedissertationwouldnothavebeenpossiblewithoutthegenerousfinancial
supportprovidedbytheRosenfeldDissertationAward.
Lastly,Iwouldliketoextendspecialthankstomyparentsfortheirtrustand
encouragement,andtomyhusband,YongFu,forhisloveandsupport.
xiii
Chapter 1 Introduction Chinaexperiencedrapideconomicgrowthinthepasttwodecades,benefitingmany
sectorsoftheeconomy.However,theeconomicsuccesswasnotmirroredinthehealthcare
system.Instead,thetransitionfromacentrallyplannedeconomytoamarket‐oriented
economyhascausedproblemsinthepublichealtharena.Forexample,aftertheeconomic
reformsstartedin1978,theexistinghealthinsuranceprovidersfacedincreased
operationalchallenges,andasaresult,manyresidentslackedanyformofhealthinsurance.
Theconditionwasespeciallytroublesomeinruralareas,revealingsharprural‐urban
disparitiesinhealthinsurancecoverageandrelatedhealthcareservicesandcosts.Since
thelate1990s,therehavebeenattemptstoexpandpublichealthinsurancecoverageto
bothruralandurbanresidentsinordertoprovideaccessibleandaffordablehealthcareto
allresidents.Anothergoalofthehealthcarereformswastoprovidehealthcaretothepoor
anddisadvantagedpopulations.Asoftheendof2011,threehealthinsuranceprograms,
calledschemes,wereestablished,coveringmostoftheruralandurbanresidentswith
someformofhealthinsurance.However,theperformanceofthecurrenthealthinsurance
schemeshasnotbeenwellexamined.Mixedfindingshavebeenpresentedregardingthis
issue.Mydissertationfocusesontheroleofhealthinsuranceinreducingtherural‐urban
disparitiesintermsofhealthcareutilizationandfinancialprotection,inthecontextofthe
currenthealthinsuranceschemes.
Thedissertationisorganizedasfollows:Chapter2providesthebackgroundofthe
policychange.ThechapterbrieflyreviewsthehistoryoftheChinesehealthinsurance
systemreform,includingthecollapseandre‐establishmentofthesystems.Ialsoprovide
1
statisticsofthetrendsandcurrentstatusofrural‐urbanhealthcaredisparities.Chapter3
reviewsexistingliteratureonthetopicofrural‐urbanhealthcaredisparitiesand
summarizestheresearchquestions.Chapter4presentsthestudydesign,includingdata
used,conceptualframework,andanalyticalapproach.Chapters5and6presenttheresults
ofthestudy.InChapter7,Iconcludethestudyandpresentpolicyimplications.
2
Chapter 2 Background ThegreateconomicreforminChinabroughtchangestoallareasoftheeconomy,
includingthehealthcaresystem.Unfortunately,asaresult,manyresidentslosthealth
insurancecoverage.Theexistinghealthinsuranceschemesexperienceddifficultiesin
providingsufficienthealthcaretoinsuredresidents.Thecooperativemedicalscheme(CMS)
providingruralhealthinsuranceexperiencedthegreatestdamage.Inresponsetothe
emergingproblemsinitshealthcaresystem,Chinahasmadenumerousattemptstorebuild
universalcoveragesystemsincethelate1990s.Throughdecadesofeffort,theChinese
governmenthasdevelopedthreesystems,inbothurbanandruralareas,whichprovide
coverageformorethan90%ofthepopulation.Duringthelaunchofeachnewhealth
insurancescheme,thegovernmentalsoproposedothermeasurestoprovidemore
healthcareresourcestothetargetedpopulation.Thesemeasuresworktogetherwiththe
healthinsurancesystemstoprovidesufficientandaffordablehealthcaretoallresidents.
Althoughtherehasbeengreatprogress,thehealthinsurancesystemisfarfromperfect.
Thehealthinsurancereformisstillunderway,andtheeffectoftheexpandedinsurance
coverageinChinaisstillunderdebate.
2.1 Health insurance reform in China Inthissection,IreviewthehistoryofhealthinsurancereforminChina.Thehealth
insurancesystemcollapsedinthelate1970s,andagreatnumberofresidentsleft
uninsured.Startingfromthelate1990s,thegovernmentestablishedthreenewhealth
insurancesystemsinbothruralandurbanareas.In2009,thegovernmentstartedanew
roundofhealthcarereform.Inthenewroundofreform,themajorgoalwastoprovide
3
universalcoveragetoallresidents,andtotargetondisadvantagepopulationtoimprove
thehealthcareserviceforthemandreducedisparities.
2.1.1Collapseofhealthinsuranceschemesinthe1970sand1980s
Sincethelate1970s,theChineseeconomicreformshaveledtoaperiodof
prosperity.However,theeconomicsuccesswasnotmirroredinthehealthcaresystem.
Instead,theeconomictransitioncausedproblemsinthepublichealtharena.
Priortotheeconomicreforms,therewerethreebasicformsofinsurance,which
coveredalmostallChinesecitizens.TheGovernmentInsuranceScheme(GIS)covered
governmentemployees.TheLaborInsuranceScheme(LIS)coveredstate‐ownedenterprise
(SOE)workers.Finally,thecooperativemedicalscheme(CMS)coveredtherural
agriculturalworkers.Theeconomicreformsbroughtchangestothehealthcaresector,
weakeningallthreeformsofinsurancetosomeextent.First,thegovernment‐runhospitals
undertheGISexperiencedfinancialdifficultiesandthuswerehardpressedtoprovide
sufficienthealthcareservicetothoseinsuredunderGIS.Onereasonforthefinancialcrisis
wasthattheeconomicreformsledtorelaxationofpricecontrols,andasaresult,thecosts
incurredbythegovernment‐runhospitalsincreased.Anotherreasonisthatthe
governmentcontributedlesstopublichospitals:Governmentcontributionsshrankfrom50%
inthe1980stolessthan10%in2000(Wang2004).Second,duringthereform,financial
autonomywasgrantedtotheSOEs.Asaresult,alargenumberofSOEsclosed,andmany
employeeslosttheirjobs.Thus,thenumberofthoseinsuredbytheLISwasreduced.Even
thosewhokepttheirjobsfoundthattheirSOEemployersfaceddifficultiesinfinancing
healthinsuranceforworkers(Li2008).Finally,intheruralareas,thebasicproductionunit
4
becamehouseholdsasthecollectivefarmsweredismantled.TheCMSalsocollapsedwith
thischange.Inthe1990s,thevastmajorityoftheruralpopulationlackedanyformof
healthinsurancecoverage(Hsiao1984;Liu2004).
Asmentioned,allthreemajorhealthinsurancesystemsexperienceddamagesasa
resultofthechangesbroughtbytheeconomicreforms,andamongthem,theruralhealth
insuranceschemeCMSfacedthebiggestchallenge.By1998,thepercentageofrural
residentswithanyformofhealthinsurancecoveragehaddroppedto13%,comparedto56%
forresidentscoveredinurbanareas(ChinaMinistryofHealth,2004).Astheurban‐rural
gapwidened,theurban‐ruraldisparityinhealthinsurancestartedtodrawmoreattention.
2.1.2Earlyeffortsinthe1980sandearly1990s
Beforethemajorhealthreformsbeganinthelate1990s,therehadbeenattemptsto
improvetheexistinghealthinsurancesystems.Evensincethe1980s,actionshadbeen
takeninurbanareastorelievethefinancialburdenonthehealthinsurancesystems.By
introducingdemand‐andsupply‐sidecostsharing,theattemptsinthe1980sfocusedon
reducingcosts.Theseactionscurbedtherapidhealthcarecostgrowth,buttheywerenot
abletosolvethefundamentalfinancialproblems(Liu2002).Beginningintheearly1990s,
thegovernmentintroducedmoreactionstoincreasethelevelofriskpooling.In1995,the
governmentintroducedanewmodelcombiningindividualresponsibilityandsocial
protectionwithcity‐wideriskpooling.However,pilotprogramsofthenewsystemwere
launchedinonlytwocitiesandwerenotspreadnationwideuntilthelate1990s.
Inruralareas,debateandresearchhasfocusedonhowtomaintainthecollapsing
corporativeinsuranceschemefromthe1980sand1990s.Thecentralgovernment’seffort
5
mainlyfocusedonurbanarea;thelocalgovernmentswereadvisedtodevelopand
completethecurrentCMSsystemsbasedonlocaleconomicconditions.However,thelocal
actionsonlyslightlyincreasedthehealthinsurancecoverageinruralareas.Mostofthe
coverageconcentratedonlyondevelopedprovincesandcities,suchasShanghai,Jiangsu,
Guangdong,andShandong.Bytheendof1990s,mostoftheruralresidentswereleft
uninsured.
2.1.3Healthinsurancereformsincethelate1990s
Inresponsetotheemergingproblemsinitshealthcaresystem,Chinahasmade
numerousattemptstorebuilduniversalcoveragesincethelate1990s.Thegoalof
universalcoverageistoprovidesafe,effective,convenient,andaffordablebasicmedical
servicestoallurbanandruralresidents(StateCouncil,2009).Oneofthemostimportant
componentsofuniversalcoverageishealthinsurance.Beforethisgoalofuniversal
coveragewasofficiallyintroducedin2009withtheChinesegovernment’sannouncement
oftheblueprintforhealthsystemreform,healthinsurancereformsinbothurbanandrural
areashadresultedingreaterhealthinsurancecoverage.Threemajorhealthinsurance
schemeswereestablished.TheUrbanEmployeesBasicMedicalInsurancewaslaunchedin
urbanareasin1998,andtheUrbanResidentsBasicMedicalInsurancewaslaunchedin
2007.Inruralareas,theNewRuralCooperativeMedicalInsurance(NRCM)was
establishedin2003.In2008,thetwourbanhealthinsuranceschemescoveredabout65%
ofurbanresidents,andtheruralschemecoveredabout90%ofruralresidents(National
HealthServicesSurvey,2008).Thethreemajorhealthinsuranceschemesarediscussedin
detailinthenextsection.
6
Theexpandedhealthinsurancecoverageprovidedresidentswithmorefinancial
protectionandencouragedresidentstousehealthcarewhenneeded.However,the
utilizationofhealthcarewasalsosubjectedtomedicalresourcesavailable.Insteadofonly
providinghealthinsurancecoveragetoresidents,thehealthcarereformwasa
comprehensivesystemwithothermeasuresandactions.Thesemeasuresandactions
workedtogetherwithhealthinsuranceexpansion,providingresidentswithmore
healthcareresourcesandgrantingthemadequatehealthcareaccess.
First,themedicalservicesystemwithbasicfacilitieswasconstructedinruralareas.
In2003,togetherwiththelaunchofNRCM,theStateCouncilannouncedothermeasures
designedtorebuildtheruralmedicalsystem(StateCouncil,2002).Oneofthemeasures
wastoconstructthemedicalservicesystemwithbasicfacilities.Inordertoachievethis
goal,centralandlocalgovernmentsincreasedtheirfinancialsupporttothemedicalsystem
eachyear.From2003to2010,theincreasedfundingwaspartiallyusedontheconstruction
ofthemedicalsystem.Localgovernmentsatthecountylevelwereresponsibleforthe
operationalcostofthelocalmedicalfacilities.Thecentralgovernmentandlocal
governmentsattheprovincelevelprovidedundevelopedareaswithsubsidiesfor
infrastructureconstruction.
Second,amedicalassistanceprogramwasestablishedinbothruralandurbanareas.
Inruralareas,theprogramwaslaunchedin2003.Theprogramwastoprovidefinancial
assistancetolow‐incomehouseholds.Theassistancecouldeitherbeusedtotreat
catastrophicdiseaseorbeusedaspremiumstojointhelocalNRCM.Governmentsubsidies
fortheprogramhavebeenincreasingsincetheprogramwaslaunched.Inurbanareas,the
7
programwaslaunchedin2005.Thetargetedpopulationswere(a)urbanresidentsliving
belowthepovertylinewhodidnotjointheUrbanResidentsBasicMedicalInsurance;and
(b)urbanresidentswhojoinedtheURBMIbutwerestillcarryingheavyfinancialburdens.
Theprogramwasdesignedandfundedbylocalgovernments.Thecentralgovernmentalso
providedassistancethroughgovernmenttransfers.
Third,trainingofmedicalprofessionalswasenhancedinruralareas.Inits2002
documentNo.13,theStateCouncilannouncedmeasurestoimprovethequalityofmedical
professionalsinruralareas.Inpost‐secondarymedicalschools,theCouncilintroduceda5‐
yearprogramaftermiddleschoolanda3‐yearprogramafterhighschool,inaneffortto
producemoremedicalprofessionals,especiallyforruralareas.Medicalgraduatesand
retiredmedicalprofessionalsfromurbanareaswereencouragedtogobacktoworkin
ruralareas(StateCouncil,2002).Asareflectionofongoingprogress,measurestoimprove
educationandtrainingofmedicalprofessionalwereintroducedagaininanewroundof
healthreform(StateCouncil,2009).Healthcareworkerswereencouragedtoattendformal
educationprogramsandobtainofficiallicenses.Thetrainingofgeneralpractitionersfor
ruralareaswasincludedintheMinistryofEducation2010workplan.Thegovernment
providedthetrainingcosts(MengandTang2010).
Finally,thegovernmentundertookotheractionstorefinethewholemedicalsystem,
suchasregulationofdrugpolicy,allocationofmedicalfunding,andstrengtheningof
administrationandsupervisionsystem.Allthemeasuresworkedasawholetoimprovethe
medicalserviceforbothruralandurbanareas.
8
2.1.4Healthcarereformafter2009
Asmentionedintheprevioussection,thegoalofuniversalcoveragewasbroughtup
bytheStateCouncilin2009.ThegoalwaspublishedintheOpinionsonDeepeningthe
ReformoftheHealthcareSystem(StateCouncil,2009),whichmarkedaneweraofhealth
carereforminChina.Inthisroundofhealthcarereform,theStateCouncilsetupthegoalof
theuniversalcoverageforthefirsttime.ItwasalsothefirsttimefortheChinese
governmentstobreaktheurban‐ruraldichotomyandtoprovideequivalentpublic
healthcareservicetobothurbanandruralresidents.
Inordertoachievethegoalofuniversalcoverage,allthreeexistinghealthinsurance
programsweretobeimproved.Inadditiontoextendinginsurancecoveragetothe
uninsuredpopulation,thebenefitcoverageoftheinsuredwastobeincreasedand
expandedtocovercatastrophicillnessesandoutpatientvisits.Anothergoalofthenew
roundofhealthinsurancereformwastoprovidebetterhealthcarecoveragetovulnerable
population,suchasruralresidents,low‐incomefamilies,unemployedformerSOE
employees,seniorpopulation,theretired,thedisabledandchildren.Therural‐urbangapof
benefitcoveragewasexpectedtobeclosed,andthemedicalassistantprogramsweregoing
tobestrengthened.
Inadditiontoimprovingthehealthinsurancesystem,theStateCouncilalso
launchedotherinitiativestochangethehealthcaresystem(StateCouncil,2009).Thefirst
wastoprovideequivalentpublichealthcareservicetobothruralandurbanresidents.The
publichealthcareserviceincludedpreventativecare,healthcareeducation,aswellas
healthserviceforwomenandchildren.Thesecondwastoestablishbasicdrugsupply
system.Inordertoensurethesupplyofaffordablebasicdrugs,thecentralgovernment
9
establishedalistofessentialdrugs,andguaranteedthesupplyofthelisteddrugstoall
levelsofmedicalfacilities.Moreover,thehealthinsuranceprogramsprovidedmore
coverageforthesebasicdrugs.Thethirdwastostrengthenthegrassrootlevelmedical
servicesystem.Inruralareas,acomprehensivemedicalsystem,includingmedicalfacilities
incounty,townandvillagelevels,wastobeestablished,inordertoprovidemedical
serviceateachlocallevel.Inurbanareas,communitymedicalfacilitiesweretobe
strengthened.Trainingformedicalprofessionalswerealsoimprovedatlocallevels.Finally,
pilotprogramsforpublichospitalreformwerestartedbythecentralgovernmentafter
2009.
2.2 Three Major Health Insurance Schemes Asdiscussedinthelastsection,Chinaisnowimplementingambitiousreformsofthe
healthinsurancesystem,andthreetypesofhealthinsuranceschemeshavebeenlaunched.
Thesethreeschemeswerelaunchedindifferentyearstargetingdifferentpopulation
groups.Twoinsuranceschemescovertheurbanresidents,andthethirdonecoversthe
ruralresidents.
2.2.1TheBasicMedicalInsuranceforUrbanEmployees
In1998,theChineseStateCouncilissuedtheDecisionoftheStateCouncilon
EstablishingtheUrbanEmployees’BasicMedicalInsuranceSystem.Thiswasthefirststep
inre‐establishingthehealthinsurancesysteminurbanareas.TheUrbanEmployeesBasic
MedicalInsurance(UEBMI)iscompulsorybasedonemployment.Itprovidesbasicmedical
insurancecoverageforurbanemployeesinboththepublicandprivatesectors(State
Council,1998).Localgovernments,mainlyatthemunicipallevel,setthelevelof
deductibles,copayments,andreimbursementcapsaccordingtolocaleconomiclevels.
10
Thepolicywaslaunchedinearly1999,andinlate1999,itwasexpanded
nationwide.Bytheendof2002,about94millionpeopleparticipatedintheUEBMI.In
ordertofurtherexpandthecoverage,theMinistryofHumanResourcesandSocialSecurity
issuedNotificationofFurtherExpandingtheCoverageoftheUrbanEmployeesBasic
InsuranceCoveragein2003.Bytheendof2008,thenumberofinsuredtotaled200million.
TheUEBMIisfinancedbypremiumsfrombothemployersandemployees.Intheir
decision,theStateCouncilsuggestedthattheemployers’contributionbe6%ofthe
employee’ssalaryandtheemployees’percentagebe2%.Therevenuecollectedfrom
premiumsisdistributedevenlyintotwoindependentaccounts:theMedicalSavings
Account(MSA)andtheSocialPoolingAccount(SPA).Allemployees’contributionsand
about30%ofemployers’contributionsgointotheMSA,andtheremainderofthe
employers’contributionsgoestoSPA.Thetwoaccountsaremanagedseparatelyandpay
fordifferentservices:theMSAcoversoutpatientandemergencyservicesanddrug
expenses,andtheSPAcoversinpatientservices.
2.2.2TheBasicMedicalInsuranceforUrbanResidents
In2007,theStateCouncilissuedguidelinestolaunchtheUrbanResidentsBasic
MedicalInsurance(URBMI).Accordingtotheguidelines,theURBMIcoversprimaryand
secondaryschoolstudentswhoarenotcoveredbytheUEBMI(includingstudentsin
professionalseniorhighschools,vocationalmiddleschools,andtechnicalschools),young
children,andotherunemployedurbanresidentsonavoluntarybasis(StateCouncil,2007).
Themainpurposeoftheguidelinesistoprovidecoverageforurbanresidentswithout
11
formalemploymentwiththeintentionofeliminatingimpoverishmentresultingfrom
chronicorfataldiseases,whichcanleadtocatastrophicmedicalexpenditures.
TheURBMIwaspilotedin79cities,includingtwotothreecitiesineachofthe
provincesthatwereabletoparticipate,andexpandedtomorecitiesin2008and2009,
withtheobjectiveofcovering80%ofallcitiesintheparticipatingprovinces.In2010,this
insuranceschemewasexpandednationwideandgraduallyextendedtoallunemployed
urbanresidents.Thenumberofinsuredwasabout43millionbytheendof2007and
increasedto118millionbylate2008(ChinaMinistryofHealth,2010).
Thefinancingofthisinsuranceprogrammainlycomesfromparticipants’premiums.
Thegovernmentalsoprovidesasmalleramountofsubsidies,comparedtothepremium
contributions.Thepremiumofthepolicyisdeterminedbythelocalgovernment,according
tothelocaleconomiclevel,themedicalcareexpenselevel,andtheparticipants’household
incomelevel.Whenthepolicywaslaunched,thegovernmentcontributionwasatleast40
Yuanperparticipant.Fromthisamount,thecentralgovernmenttransfers20Yuanto
centralandwesternareasresidents.Thereareextragovernmentsubsidiesforlow‐income
families,disabledstudents,andyoungchildren(StateCouncil,2007).TheURMBImainly
targetspeoplewithchronicandfataldiseases;therefore,itcoversmoreexpensesfor
inpatientservices.In2008,theURMBIcovered45%ofexpensesfrominpatientservice
relatedtochronicandfataldiseases,whichequaled1436Yuanperinpatientstay(State
CouncilEvaluationGroupfortheURBMIPilotProgram,2008).
12
2.2.3TheNewRuralCooperativeMedicalInsurance
In2003,theStateCouncilissuedtheDecisiontoFurtherEnhancetheRuralHealth
CareSystem,aimedatre‐establishingtheRuralCooperativeMedicalInsurance(NRCM).
TheNRCMschemecoveredtheruralresidentsonavoluntarybasisinordertoavoid
impoverishmentcausedbycatastrophicexpensesfrominfectiousandendemicdiseases.
TheNRCMwaspilotedin2003inselectedcounties.In2006,coverageincreasedto40%of
allcounties,andabout60%in2007.In2010,theNRCMcoveredmorethan90%ofallrural
residents.
TheNRCMwasfundedbypremiumsfromboththeinsuredandbysubsidiesfrom
thelocalandcentralgovernments.In2003,thecentralgovernmentprovidedasubsidyof
10Yuanforeachinsuredresident.TheCouncil’s2003decisionalsorequiredlocal
governmentstoprovidenolessthan10Yuan.In2011,thesubsidizedamountwasraised
toatotalof200Yuan.TheNRCMprovidespartialcoverageforallkindsofmedical
expenses,excludingsomeoutpatientexpensesanddrugexpenses.Thereimbursement
capsvarybylocaleconomicdevelopmentlevels.
2.3 Trends and Current Status of Healthcare Disparities Chinaisavastcountrywithuneveneconomicdevelopment.Ruralandurban
residentsarecategorizedseparatelyaccordingtothehouseholdregistrationsystem.The
governmentfinancingsystemsforruralandurbansectorsarealsoseparate.Mostofthe
governmentrevenuecomesfromtheurbaneconomy,andmostisspentonurbaneconomy
aswell.Thisisespeciallytrueinpublicserviceareas,resultingintheurban‐ruraldisparity.
13
Asmentionedbefore,by1998,theurban‐ruraldisparityinhealthinsurance
coveragehadbecomeprominent.Thecoveragegappersistedinsubsequentyears.For
example,in2003,theurbanhealthinsurancecoverageratewasstillmorethan50%,while
onlyabout20%oftheruralresidentswerecoveredbysomeformofhealthinsurance
coverage,andabouthalfofthe20%wascoveredbypurecommercialhealthinsurance.
ThisisshowninFigure2.1,whichpresentsthepercentageofresidentscoveredbyhealth
insuranceinbothurbanandruralareasovertime.Duringtheselectedperiod,publichealth
insurancecoveragewasreducedyearbyyearinbothruralandurbanareasuntil2003.
However,thepercentageofcoveragehadalwaysbeenmuchlowerinruralareasthanin
urbanareas.
Then,in2008,therewasalargeincreaseininsurancecoverage,especiallyforrural
areas.Coverageincreasedtomorethan90%,andalargerportionofruralresidentswas
coveredbyhealthinsuranceatthistime,comparedtotheportionofurbanresidents.We
canalsoobservetheshiftintheurban‐ruralratio(thegreenline).Before2003,theurban‐
ruralratioofhealthinsurancecoveragewasextremelyhigh;however,in2008,theratio
decreasedtolessthan1,indicatingmorecoverageinruralareas.Betweenthetwotime
points,therewereseveralpolicychangesthataffectedhealthinsurancecoverage.Inthe
urbanareas,thebasicmedicalinsuranceforurbanemployeeswaslaunchedin1998,andin
2007,thebasicmedicalinsuranceforurbanresidentswasestablished.Intheruralareas,in
2003,thegovernmentstartedtorebuildthecooperativehealthinsurancesystem(NRCM),
whichinfluencedaverylargepopulation.Mostoftheruralcoveragein2008wasfrom
NRCM.Therefore,IbelievetheinitiationandexpansionoftheNRCMdiminishedthe
disparitiesinhealthinsurancecoverage;however,itisstillunknownwhetherthe
14
expansionhelpedreducedisparitiesinotherhealthcareareas,suchashealthcare
utilizationandcost.
Figure 2.1 Health Insurance Coverage in Urban and Rural Areas in China, Selected Years 1993-2008
Disparitywasalsoobservedinotherhealthcareissuesrelatedtohealthinsurance
coverage,suchasinhealthcareutilizationandout‐of‐pocketcost,especiallybeforetheyear
2003.Ononehand,theurban‐ruraldisparityonhealthcareutilizationdecreasedfrom
1993to2003.Forexample,in1993,thepercentagesofhospitaloutpatientserviceusein
thetwoweekspriortothesurveyforurbanandruralresidentswere19.9%and16.0%,
respectively;in2003,thepercentagesbecame11.8%and13.9%,respectively(China
MinistryofHealth,2004).Ontheotherhand,in2003,abouthalfoftheresidentsinrural
areaswhosoughtoutpatientserviceswenttoinformalhealthcareinstitutionsinsteadofto
formalhospitals,whilethepercentageinurbanareaswasonlyabout25%.Theshrinkage
15
oftheurban‐ruralgapofhealthcareutilizationwasduetothereductionininformal
healthcareinstitutionsinurbanareas(ChinaMinistryofHealth,2004).Moreover,the
percentageofunmetneedswashighestamongthelow‐incomepopulationinruralareas
(ChinaMinistryofHealth,2004).
Thehealthcareutilizationdisparitywasmostprominentinthehealthservicearea.
Figure2.2showsthepercentageofpregnancyhealthcareutilizationandthepercentageof
womenwhogavebirthinhospitalin2003.Wecanseethatruralwomenusedlessofthese
services,especiallylow‐incomewomen.By2008,thedisparityinhealthserviceutilization
hadbeenrelievedbutstillexisted.Thepercentageofpregnancyhealthcareutilizationhad
risento93.7%forruralwomen.Comparedtothe97.6%ratioforurbanwomen,therateof
healthcareutilizationwasstilllowerbutthegapbetweenurbanandruralhadbecome
narrower.
Health Service Utilization in Urban and Rural Areas in China, by Income (2003)
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
98%
85%
81%
45%
lowest
percentile
highest
percentile
give birth in
hospital
lowest
percentile
Urban
pregnancy
health care
highest
percentile
Rural
Source: China Ministry of Health, The Third National Health Services Survey Report (in Chinese), 2004, http://www.moh.gov.cn/publicfiles///business/cmsresources/mohwsbwstjxxzx/cmsrsdocument/doc9908.pdf (accessed Aug. 28, 2012)
Figure 2.2 Health Service Utilization in Urban and Rural Areas in China (2003)
16
Drivenbylimitedhealthinsurancecoverageandrapidlygrowinghealthcarecosts,
highout‐of‐pocketexpensescomprisedamajorchallengeforthoseseekinghealthcare.
ChinabecameoneoftheAsiancountrieswiththehighestratioofout‐of‐pocketcosttototal
healthcarecostsin2002(YipandHsiao2008).Atthattime,theout‐of‐pocketratiowas60%
(Hu,Tangetal.2008),andruralresidentsboreanevenhigherratio.Thetrendofhealth
spendingisshowninFigure2.3.Thepercentageofout‐of‐pocketpaymentsbyindividual
patientrosesteadilyfrom1980to2001.Thistrendindicatesthatthefinancialburdenof
healthcareshiftedmoreandmoretotheindividualpatientsduringthatperiod.However,
after2001,thegovernmentandsocialprogramsstartedtotakeonmoreofthecost,and
thisresultedinadownwardinfluenceonindividualout‐of‐pocketpayments.
70
Healthcare Spending in China, by Source and Year
60
Individual Patient, 38.2
Percentage 50
40
Social Programs, 34.6
30
Government, 27.2
20
10
0
1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008
Source: China Ministry of Health, China Health Statistics Yearbook(in Chinese), 2010, http://www.moh.gov.cn/publicfiles/business/htmlfiles/zwgkzt/ptjnj/year2010/index2010.html Figure 2.3 Healthcare Spending in China, by Source and Year
17
30.0%
Per Capita Out‐of‐pocket Health Expenses as a Percentage of Income
25.0%
20.0%
15.0%
10.0%
1993
5.0%
1998
2003
0.0%
lowest
percentile
middle
highest
percentile
lowest
percentile
Urban
middle
highest
percentile
Rural
Source: China Ministry of Health, The Third National Health Services Survey Report (in Chinese), 2004, http://www.moh.gov.cn/publicfiles///business/cmsresources/mohwsbwstjxxzx/cmsrsdocument/doc9908.pdf (accessed Aug. 28, 2012)
Figure 2.4 Per Capita Out-of-Pocket Health Expenses as a Percentage of Income
Figure2.4showsthepercapitaout‐of‐pockethealthexpenditureasapercentageof
incomebyurbanandruralareas.Ruralresidentspaidformedicalservicewithalarger
portionoftheirincomesthandidurbanresidents.Amongthepoorerruralresidents,out‐
of‐pocketpaymentsforhealthcareservicesconstituted26.7%oftheirtotalincomein2003,
alargeincreasefromthepercentagetenyearsearlier.
18
Chapter 3 Literature Review and Study Objectives 3.1 Existing Research Tworesearchareasinformmystudy.Thefirstareacomprisesresearchon
healthcaredisparities.Asdiscussed,urban–ruraldisparitiesinhealthandhealthcarehave
drawnattentioninChinainrecentyears.Manystudieshaveprovidedempiricalevidence
ontheconditions,trends,andassociatedfactorsofsuchdisparitiesinhealthstatus,
healthcareutilization,healthcarecosts,andrelatedissuessuchashealthinsurance
coverage.Otherresearchinthisareahasfocusedonexaminingthedeterminantsofthe
disparities.Thesecondareaofresearchincludesassessmentsoftheinsuranceschemesin
Chinaintermsofimpactonhealthcareutilization,out‐of‐pocketcost,andhealthoutcomes.
Althoughthesestudiesareusuallynotfocusedonhealthcaredisparities,Iviewedthemasa
goodfoundationformyresearch.Ialsofoundthesestudieshelpfulintermsofdataand
methodology.Inthenextsection,Ireviewsomeofthekeyresearch.
3.1.1LiteratureonRural–UrbanDisparitiesinHealthcareUtilization
Recentstudieshaveprovidedempiricalevidenceontheconditionsandtrendsof
rural–urbanhealthcaredisparities(Liu,Hsiaoetal.1999;Zhao2006;Tang,Mengetal.
2008;Meng,Zhangetal.2012).Liu,Hsiao,andEggleston(1999)examinedthechangesin
disparityinhealthstatusandhealthcareutilizationinChinafrom1985to1993andfound
thatthegapinhealthstatusandhealthcareutilizationbetweenurbanandruralresidents
widenedduringthetransitionalperiodwhentheChineseeconomywasshiftingfroma
commandeconomytoamarketeconomy.Theauthorsconcludedthatthetrendswere
correlatedwiththereductionofruralhealthinsurancecoverage.Zhao(2006)provided
evidenceforlateryears,showingthattherural–urbandisparitiesinmorbidityand
19
mortalitylevelswereassociatedwithdisparitiesinhealthcareaccess.Meng,Zhangetal.
(2012)providedsimilarevidenceondisparitiesinmaternalandunder‐fivemortalityrates.
Tang,Mengetal.(2008)pointedoutthattherewererural–urbandisparitiesinasetof
childhealthindicators,includinginfantmortalityrate,levelofmalnutrition,childstunting,
andunderweightstatus.However,theresearchersbelievedthatChinahastheabilityto
carryoutthenecessaryreformstoimprovehealthequity.
Severalresearchersspecificallyexamineddisparitiesinhealthcareaccessand
utilizationtoidentifythedeterminantsofhealthcareutilization.(Gao,Tangetal.2001;
Wang,Yipetal.2005;Gao,Ravenetal.2007;Liu,Zhangetal.2007;Fang,Chenetal.2009;
Jian,Chanetal.2010;Long,Zhangetal.2010;Feng,Guoetal.2011;XuandShort2011;Liu,
Tangetal.2012;Meng,Zhangetal.2012).Amongthesestudies,researcherspresented
mixedfindings.Generally,theauthorsagreedthatmosthealthcareresourceswerebeing
allocatedtourbanareasandthaturbanresidentsusemoreformalhealthcarethandorural
residents.However,Fang,Chenetal.(2009)examinedtheevolutionofrural–urban
disparitiesinhealthcareutilizationfrom1997to2006andconcludedthatruralresidents
actuallyvisitphysiciansmoreoftenthandourbanresidentswhentheyareill.Someofthe
researcherspointedoutthatbetterinsurancecoveragewasassociatedwithincreased
healthcareutilization.Liu,Zhangetal.(2007)notedthathospitalutilizationwaslower
amongtheuninsured.
Someofthestudiesfocusedoncertainsubpopulationsandreachedsimilar
conclusions.Gao,Ravenetal.(2007)examinedthetrendofinpatientutilizationamongthe
elderlyinurbanChina,andtheyfoundthatwithinthissubpopulation,theinsuredwere
20
morelikelytouseinpatientcare.Jian,Chanetal.(2010)analyzedchangesintherural–
urbangapforpatientswithchronicdisease,drawingondatacollectedbetween2003and
2008.Theyconcludedthatthegapbetweenurbanandruralresidentswasnarrowedin
termsofhospitaladmissionrates;however,therewasnochangeintermsofearlyself‐
dischargefromhospital.Liu,Tangetal.(2012)analyzedtheimpactofhealthinsuranceon
utilizationofoutpatientandinpatientservices.Theyconcludedthathavinghealth
insurancecoveragehadnosignificantimpactonoutpatientserviceutilization;however,
inpatientserviceutilizationincreased.
Someoftheresearchersfoundthatchangesindisparitiesandtheimpactsofhealth
insurancecoverageweredifferentamongdifferentincomegroups.Gao,Tangetal.(2001)
concludedthatfrom1993to1998,healthcareaccessforlow‐incomegroupsshrankmore
thandidhealthcareaccessforhigh‐incomegroup.Liu,Tangetal.(2012)pointedoutthat
theeffectofinsurancecoverageoninpatientserviceutilizationwasgreatestforhigh‐
incomegroups,whilelow‐incomegroupenjoyedfewerbenefits.
3.1.2LiteratureonDisparitiesinOut‐of‐PocketExpenditureandHealthcare
Costs
Severalstudiesfocusedonthedisparitiesanddeterminantsofout‐of‐pocket
expendituresandhealthcarecost(Pan,Dibetal.2009;Sun,Jacksonetal.2009;Long,
Zhangetal.2010).Theresearchersgenerallyagreedthatruralresidentstendedtobeat
increasedriskforhighandcatastrophicmedicalpayments;thecurrentinsuranceschemes
inruralareasofferlimitedfinancialprotection.Pan,Dibetal.(2009)concludedthat
hospitalizationcostswerehigheramonginsuredpatientsbecausetheinsuredgenerally
stayedlongerinhospitalthandidtheuninsured.Long,Zhangetal.(2010)foundthat
21
participatingintheNRCMreducedout‐of‐pocketexpendituresonaverage,buttherural
poorwerestillfacedwithhighpaymentproblems.Sun,Jacksonetal.(2008)pointedout
thatout‐of‐pocketpaymentsremainedaburdenforruralresidentsaftertheinitiationof
NRCM.
3.1.3LiteratureonDisparitiesinHealthInsuranceCoverage
Researchhasfocusedonthetrendsofdisparitiesinhealthinsurancecoverage(Akin,
Dowetal.2004;Xu,Wangetal.2007;XuandShort2011).Akin,Dow&Lance(2004)
examinedchangesinhealthinsurancecoveragefrom1989to1997andconcludedthatthe
overallcoveragedecreasedslightly,from26%in1989to23%in1997.Theyfurther
pointedoutthaturbanareas(citiesandtowns)experiencedreductionsinhealthinsurance
coverage,whileruralareacoverageincreased.However,thechangeswereverysmall,and
therural–urbandisparityinhealthinsurancecoveragepersists.Xu,Wangetal.(2007)
useddatafromtheNationalHealthServicesSurveysof1998and2003toexaminethe
impactofthereformonpopulationcoverage,andtheyconcludedthattheoverallhealth
insurancecoveragestayedalmostthesameamongurbanresidents.XuandShort(2011)
examinedthetrendsofhealthinsurancecoveragefrom1997to2006.Theypointedouta
sharpincreaseofcoveragein2006inruralresidents,whichresultedinasmallergapin
healthinsurancecoveragebetweenruralandurbanresidents.
3.1.4MethodologicalIssues
3.1.4.1 Definition of Rural and Urban TwodefinitionsareusedtodetermineruralandurbanstatusinChina.Thefirst
definitionclassifiesresidentsbygeographicalresidentialareas,whichareofficiallydivided
intourbanandruralareasbytheNationalBureauofStatisticsofChina,accordingto
22
China’sadministrativedivisions.Theseconddefinitionisbyhouseholdregistrationtype.
Chinaclassifiespeopleaseitheragricultural(rural)ornonagricultural(urban).These
categorizationdataarerecordedbythehouseholdregistration(Hukou,户口)system.
Thesetwodefinitionsofruralandurbanstatusarenotentirelyconsistent.
Differentdefinitionsofruralareascanleadtodifferentresultswhenstudyinghealth
policy,becausethedefinitionofruralareasaffectstheresourcestowhichpeoplehave
access(Hart,Larsonetal.2005).However,fewexistingstudiesaddressthedefinition
specifically.Formostofthestudies,Iidentifiedtheauthors’definitionsofrural/urban
areasonlybytheterminologyused.Forexample,iftheauthorsusedtermssuchas
residents,areas,orgeographicregions,Iviewedthesetermsasbeingconsistentwiththe
firstdefinition.Iftheauthorsmentionedhouseholdregistrationorusedthetermpopulation,
Iviewedthesetermsasconsistentwiththeseconddefinition.Inallofthecitedpapers,the
researchersadoptedthefirstdefinitionexceptforonestudyassessingNRCM.Lei&Lin
(2009)adoptedboththefirstandseconddefinitionswhentheyevaluatedNRCM.However,
theyrestrictedtheirsamplebyonlyincludingpeoplewholivedinruralareasandwere
withruralhouseholdregistration.
3.1.4.2 Modelling Intermsofmethodology,mostofthestudiesmentionedweredescriptive,andsome
ofthepapersusedcross‐sectionaldatatofitlogit/probitmodels.Theresearchers
emphasizedtheproblemofurban–ruraldisparitiesinhealthcareinChinaandclarifiedthe
trendsandcurrentconditions,aswellasprovideddirectionforfurtherstudyofthisissue.
However,noresearchhasprovidedacompletepictureofhowthedisparitiesinhealth
23
insurancecoverage,healthcareutilization,andhealthcarecostchangeovertime.Little
researchhasfocusedontheroleofhealthinsurancecoverageonclosingtherural–urban
gapinhealthcareutilizationandhealthcarecosts,whileconsideringallmajorinsurance
changes.
Asdiscussedbefore,someresearchershaveevaluatedNRCM,andthistypeof
researchprovidedmewithmethodologicalhelp.Wagstaff&Lindelow(2009)drewon
multipledatasourcestostudytheinsuranceandfinancialriskinChinabefore2003.They
appliedfixed‐effectmodelsfortwopaneldatasetsandaninstrumentalvariable(IV)
techniqueforacross‐sectionaldataset,andtheyconcludedthathavinghealthinsurancein
Chinadoesnotalwaysreducefinancialrisk.Theyexplainedthiscuriousphenomenonby
adverseselection,i.e.,peoplewithhigherriskofhighmedicalexpensetendtojointhe
insurancescheme.Theadvantageofthisresearchisthatitusedpaneldataandadvanced
analysistechniques.However,therewerestilldrawbacksinthisstudy’smethodology.
Theirlongestpanelhadonlyfourwaves,andthesewavescoveredatimeperiodbeforethe
NRCMwaslaunched.Asdiscussedbefore,allhealthinsurancesystemshadexperienced
changestosomeextentatthattime.Itwouldbemorecomprehensiveandconvincingto
extendtheresearchbyincorporatingthemostrecentdata.
Morerecently,threeotherpapersaddressedtheNRCMusingdifferentdataand
methodologies,reachingmixedconclusions(LeiandLin2009;Yu,Mengetal.2010;Lu,Liu
etal.2012).Inthefirststudy,Lei&Lin(2009)concentratedonevaluatingthehealthcare
serviceandhealthoutcomeaftertheinitiationofNRCM.Theyusedpaneldatafromthe
ChinaHealthandNutritiousSurveytoestimatefixed‐effectandIVmodels,andtheyalso
24
appliedadifference‐in‐differencesestimationwithpropensityscorematching.The
researchersfoundnoevidencethattheNRCMdecreasedout‐of‐pocketexpendituresor
increasedutilizationofhealthcareservice.Therefore,theyconcludedthattheimpactofthe
NRCMwaslimited.Intheirstudy,theyincludedonlythreewavesofdata,onebeforeNRCM
waslaunchedandtwowavesafter.Thispanelcouldstillbeexpandedtoincludericher
information.
Inthesecondstudy,Yu,Mengetal.(2010)useddatafromsixcountiesintwo
provincestoconductacross‐sectionalstudytoexaminewhetherthelaunchofNRCM
increasedhealthcareutilization.TheyfoundthatNRCMdidnotsignificantlyincrease
outpatientserviceutilizationinruralareas,whileinpatientserviceingeneralincreased.
Further,theypointedouttheassociationbetweenhouseholdincomeandhealthcare
utilization.Theauthorsconcludedthattheincreasehappenedonlyamongthemost
affluent.Forpeoplewithmiddleandlowerincomes,theincreasewasnotsignificant.
Inthethirdstudy,Lu,Liuetal.(2012)useddatafromthe2001ChinaHealth
SurveillanceBaselineSurveytoinvestigatewhetherthelaunchofNRCMledtoanincrease
inhealthcareutilizationandadecreaseinpossiblecatastrophicmedicalexpenseforrural
residents.SimilartothemethodusedbyLei&Li(2009),Lu,Liuetal.alsousedpropensity
scorematching,andappliedtheIVmethod.TheyfoundthatNRCMdidnotdecreaseout‐of‐
pocketexpenses.However,unlikeLei&Li(2009),theyfoundthatNRCMdidsignificantly
increasehealthcareutilization.
25
3.2 Gap in the Existing Literature Tosumup,currentresearchprovidesempiricalevidencesontherural–urban
disparitiesinhealthinsurancecoverage,healthcareutilization,andhealthcarecosts.
However,currentresearchcouldbeimprovedinseveralways.First,incurrentstudies,
researchershaveexaminedrural–urbandisparitiesindifferenttimeperiods,buthavenot
providedacompletepictureofthetrendsinrural–urbandisparities.Second,the
determinantsofrural–urbandisparitieshavenotbeenwellexamined.Theimpactofhealth
insurancestatus,whichcanbeaveryimportantpolicyinterventiontoreducedisparities,
hasnotbeenwellstudied.Third,inthepapersonhealthinsuranceorhealthcaredisparities,
theauthorshavenotdrawnconsistentconclusions;thestudiescouldbeimprovedinterms
ofdataqualityandmethodology.Fourth,thepapersontheimpactofhealthinsurance
usuallyfocusoncertainpopulationgroups.Forexample,whenstudyingtheeffectsof
NRCM,researchersusuallyfocusonlyonruralresidents.
Thefirstpossibleexpansiontoexistingliteratureistoincludemorewavesofdatato
showamorecompletepictureofthetrendsofchangeinrural–urbandisparityinhealth
insurancecoverage,healthcareutilization,andhealthcarecost.Thesecondpossible
expansiontothesestudiesistoincludemorewavesofdataandtouseadvancedtechniques
toexaminethedeterminantsofthedisparitiesandthusprovidepolicysuggestionsonways
tofurtherrelievethedisparities.Inaddition,amongthefactorsassociatedwiththe
disparities,healthinsuranceisanimportantissuetostudy.Thethirdareaofexpansionis
toincludeurbanareasasacontrolgroupwhenexaminingtheimpactofhealthinsurance
expansion.Toaddressthesegapsintheexistingliterature,Iexploredallpossibilitiesinmy
research.
26
3.3 Objectives and Research Questions Theobjectivesofmyresearchweretoexaminethestatusandtrendsofrural–urban
disparitiesinhealthcareutilizationandcosts,toanalyzetheroleofhealthinsurance
coverageinreducingthesedisparities,andtoprovideevidenceandsuggestionstopolicy
makersabouthowtofurtherreducerural–urbanhealthcaredisparities.
Myresearchquestionswere:
1. Whatdotherural–urbandisparitiesinhealthcareutilizationandcostslooklike?
Howdothedisparitieschangealongwithmajorhealthinsurancepolicychanges?
2. Doesmorehealthinsurancecoverageinruralareareducetherural–urban
disparitiesinhealthcareutilization?
3. Doesmorehealthinsurancecoverageinruralareareducethedisparitiesinhigh
out‐of‐pockethealthcareexpenditureandtotalhealthcarecosts?
4. Doestheimpactofhealthinsuranceondisparitiesdifferbyincomegroupandby
region?
27
Chapter 4 Study Design 4.1 Data Forthisstudy,Idrewonthedetailedindividual‐levellongitudinaldatafromthe
ChinaHealthandNutritionSurvey(CHNS),whichisacollaborativeprojectbetweenthe
CarolinaPopulationCenterattheUniversityofNorthCarolinaatChapelHillandthe
NationalInstituteofNutritionandFoodSafetyattheChineseCenterforDiseaseControl
andPrevention.Asapanelsurvey,CHNSstartedin1989andhasbeenconductedroughly
everyotheryear.Iusedthemostrecentsevenwavesofdata(1993,1997,2000,2004,
2006,2009,and2011)intheanalysis.The1989and1991datasetswerenotusedbecause
thesedatasetsdidnotcontainhealthinsuranceinformationorhouseholdregistration
information.
CHNSusedamultistage,randomcluster‐samplingapproach,andwasconductedin
nineprovinces,1whicharemostlyrepresentativeofCentralandEasternChinaandvary
substantiallyingeography,economicdevelopment,publicresources,andhealthindicators.
Countiesinthenineprovinceswerestratifiedintothreelayersbyincome,andaweighted
samplingschemewasusedtorandomlyselectfourcountiesineachprovince.Villagesand
townships(theCHNSdefinitionofcommunities)withinthecountiesandurbanand
suburbanneighborhoodswithinthecitieswerethenselectedrandomlyintoprimary
samplingunits(PSUs).Thesamehouseholdsweresurveyedovertimewheneverpossible
andnewlyformedhouseholdswereincludedbeginningin1993.Inthesample,rural
communitieshadpopulationsrangingfrom125to14,964people,andurbancommunities
1
The nine provinces are Guangxi, Guizhou, Heilongjiang, Henan, Hubei, Hunan, Jiangsu, Liaoning, and Shandong.
In the 2011 wave, three municipalities (Beijing, Shanghai and Chongqing) were added into the sample.
28
hadpopulationsrangingfrom167to86,733people.Inthisstudy,Iincludedall
respondentswhorespondedtothehealthinsurancesection.Thisfinalsampleincluded
morethan90,000respondents.ThesamplesizesareshowninTable4.1.
CHNSwasagooddatasourcefortheresearchbecauseitprovideddetailed
informationoninsurancecoverage,medicalproviders,healthservicesuse,andhealthcare
costs.Therefore,CHNSallowedmetolookathowinsurancecoverageaffectshealthservice
useandhealthfinancing.Questionsabouthealthcareaccessibility,timeandtravelcoststo
healthfacilities,andperceivedqualityofcarewerealsoasked.
Table 4.1 Sample Size by Rural and Urban Residences and Registrations
Rural Residents Wave Rural Registration Urban Residents Urban Registration Rural Registration Urban Registration Total 1993 7,663 2,253 1,433 2,470 13,819 1997 7,255 2,492 1,661 2,801 14,209 2000 7,956 2,601 1,563 3,015 15,135 2004 6,016 2,081 1,188 2,858 12,143 2006 5,774 2,059 1,228 2,679 11,740 2009 5,931 2,064 1,241 2,688 11,924 2011 6,489 2,874 1,420 4,717 15,500 Total 47,084 16,424 9,734 21,228 94,470 29
4.2 Study Periods Forthisanalysis,Iclassifiedthestudyperiodof1993–2011intofourperiods:
1. 1993–1997,aperiodbeforethemajorhealthinsuranceexpansioninChina
2. 2000,aperiodaftertheinitiationofUEBMIin1998
3. 2004–2006,aperiodaftertheinitiationofNRCMin2003
4. 2009–2011,aperiodaftertheinitiationofURBMIin2007
4.3 Conceptual Model and Variable Selection ThevariableselectionwasbasedontheAndersenmodel(Andersen1968).The
modelfocusedontheindividualastheunitofanalysisand,whenfirstdeveloped,wasused
toexplainwhypeopleusehealthcareservices.Afterseveralgenerations,themodelgrewto
includeotherendpointsofinterest,suchashealthcarequalityandhealthoutcomes
(Andersen1995).
Figure4.1showsthemostrecentAndersenmodel.Thisfiguredepictsthe
interactionbetweenenvironment,populationcharacteristics,healthbehavior,andhealth
outcomes.Specifically,thehealthcaresystemincludespolicy,resources,andorganizations;
predisposingcharacteristicsincludedemographiccharacteristics,healthbeliefs,andsocial
structure;enablingresourcesincludesincome,healthinsurance,andotherresourcesfor
healthcareservices.Allthesecharacteristicscanimpactthedecisiontousehealthservices
andfurtherinfluencehealthcareoutcomes.Healthbehaviorcaninfluenceenabling
resources;healthoutcomescanaffectenablingresourcesandhealthbehaviors(Andersen
1995).Therefore,byincludingpersonaldemographicinformation,familyandsocial
structure,income,insurancestatus,healthconditions,andpolicychangeinthemodel,I
wasabletoexaminehowthesefactorsaffectedpeoples’healthcare‐seekingbehaviorsand
30
healthcarecosts.Thevariablesofhealthinsurancecoverageandtypesofcoverageare
viewedasenablingfactorsinthemodel.Byincludinglocationinformationabouturban
versusruralareas,Ialsocontrolledtheimpactoftheexternalenvironment.
Figure 4.1 Updated Structure of Anderson Model
Moreover,Andersenassignedadegreeofmutabilitytothemodelcomponentswhen
hedevelopedthemodel.AccordingtoAndersen,themostmutablepopulation
characteristiccomponentwasenablingresources,whichincludedinsurancecoverage.In
myanalysis,statusofhealthinsurancewasaffectedbypolicychanges.Therefore,when
interpretingtheresults,Ifocusedontheimpactofhealthinsurancecoverageonhealthcare
utilizationandcosts,andtheresultingpolicyimplications.
4.3.1DependentVariables
Theanalysisfocusedonurban–ruraldisparitiesinhealthcareutilizationand
healthcarecosts.AllthehealthcareutilizationquestionsinCHNSfocusedonafour‐week
periodrightbeforetheinterview.Forhealthcareutilization,Iconstructedthreevariables:
31
formalcareutilization,outpatientcareutilization,andinpatientcareutilization.Formal
careutilizationisabinaryvariableindicatingwhethertherespondentsoughtformal
medicalcarefromahospitalorclinicinthefourweeksbeforetheinterview.Theformal
careutilizationvariablewasconstructedfromseveralrawvariables:(a)whetherthe
respondentwassickorinjuredorsufferedfromachronicoracutedisease,(b)whetherthe
respondentsoughtcarefromaformalmedicalprovider,and(c)whattherespondentdid
whenheorshewasillorinjured.Iftheanswertothefirstquestionwas“yes,”the
respondentwasaskedthesecondandthirdquestions.Iftheanswertothesecondquestion
was“yes”ortheanswertothethirdquestionwas“sawadoctor(clinic,hospital)”,I
consideredtherespondenttohavesoughtformalmedicalcareinthepastfourweeks.
Thereweresomeinconsistencesinthequestionsettingandwordingacrosswaves.In
waves1993to2000,CHNSonlyaskedthesecondquestion,andrepeatedthequestionfora
secondfacility.Inthelatterwaves,CHNSaskedbothquestions.2Theoutpatientand
inpatientutilizationwerealsobinaryvariables.Theywereconstructedfromtheraw
variableofwhetherthevisitwasaninpatientoroutpatientvisit.
Forhealthcareexpenses,Iconstructedtwotypesofvariables.Thefirsttypeof
variableinvolvedtheamountoftotalhealthcarecosts.Thesecondtypecontainedseveral
binaryvariablesindicatingwhethertheout‐of‐pockethealthcarecostsweremorethana
certainpercentageofthehouseholdincome.Iusedtwocut‐offpointsforthepercentage:
20%and40%.Theamountoftotalhealthcarecostswasderivedfromtherawvariables
underlyingthetreatmentcosts.Theamountofout‐of‐pocketcostswasconstructedfrom
2
There has been a jump of percentage of people who use formal medical care since the 2004 wave. However, the
change is not a result from the setting of the questions.
32
thetotaltreatmentcostsandpercentageoftreatmentcostspaidbyinsuranceandother
costoftreatingtheillnessorinjury.Thesevariableswerealsoconstrainedtothefour‐week
periodbeforetheinterview.Iinflatedtheamountsto2011valuesusingtheindexfrom
CHNSdata.Inthesurvey,thequestionabouthouseholdincomereferredtoatimeperiodof
oneyear.Therefore,Imultipliedtheout‐of‐pockethealthcareexpensesby12tomatchthe
twotimeframes.Thehealthcarecostsvariablesmeasuredthecostswithin4weeksbefore
theinterview,thusthecostscouldbefromacuteillnessandbeoverestimatedwhen
transportedtocostsinoneyear.Therefore,Ididnotpickalowercut‐offpointforhighout‐
of‐pocketcosts.
4.3.2IndependentVariables
4.3.2.1 Key Independent Variable: Dummies Indicating the Respondents’ Residence and Household Registration Type Mykeyindependentvariablewasasetofdummiesindicatingtherespondents’
residentareaandhouseholdregistrationtype.Therearetwodefinitionsofruralandurban
inChina.Thefirstconsistsofgeographicresidentialareas,whichareofficiallydividedinto
urbanandruralareas.TheNationalBureauofStatisticsofChinaofficiallyassignsthese
levels.ThisvariablewasdirectlycreatedfromtheprimarysamplingunitsofCHNS,which
drewsamplesfromcities,suburbs,towns,orvillages.Thefirsttwodesignations—cities
andsuburbs—areconsideredurbanareas;thelattertwoareclassifiedasruralareas.The
seconddefinitionofruralityisbytypeofhouseholdregistration.Chinaclassifiespeopleas
eitheragricultural(rural)ornonagricultural(urban)population,recordedbythe
householdregistration(Hukou,户口)system.Thesetwodefinitionsarenotcompletely
consistent,forthreepossiblereasons:(a)thereareareasinChinacalledurban–ruralmixed
33
areas(城乡结合部),buttheycanonlybeclassifiedaseitherurbanorruralarea;(b)
increasingnumbersofpeoplewithruralhouseholdregistrationmigratetourbanareasto
work,buttheirhouseholdregistrationsdonotchange;and(c)somepeoplewithurban
householdregistration,especiallyinrecentyears,havechosentoliveinruralareas.Mostof
theagriculturalpopulationresidesinruralareas.InmyCHNSsample,75%ofpeoplewith
agriculturalhouseholdregistrationlivedinruralareas,and67%ofpeoplewith
nonagriculturalhouseholdregistrationlivedinurbanareas.Thesepercentagesstayed
relativelyconsistentacrosswaves;therefore,myassumptionwasthatthesamplecovered
fewmigratingruralworkers.Ifthiswerenotthecase,thereshouldbegreaternumbersof
ruralworkersmigratingtourbanareasastheeconomydevelopsandthecontrolof
residencyrelaxes.
Asdiscussedintheliteraturereview,mostofthestudiesonthedisparityissueused
residentialareatodefinerurality,whilemoststudiesevaluatingNRCMusedthehousehold
registrationsystemtodefinerurality.Inmyresearch,Isoughttoexaminethechangesin
disparities,aswellastoestablishalinkbetweeninsuranceanddisparity.Therefore,Iused
bothofthetwoclassificationstodividepeopleintofourcategories:ruralresidentswith
ruralregistration(GroupRR),ruralresidentswithurbanregistration(GroupRU),urban
residentswithruralregistration(GroupUR)andurbanresidentswithurbanregistration
(GroupUU).IusedGroupUUasthereferencegroupandcomparedthethreeothergroups
withit.
Byadoptingthefourcategories,Iwasabletotrackallthreehealthinsurancepolicy
changesthatexpandedhealthinsurancecoveragetopeoplewithcertainhousehold
34
registrationtypesandtopeoplelivingincertainareas.Iwasalsoabletoexaminehowthe
disparitylevelschangedwiththeresidingenvironment.Asdiscussed,thepolicychanges
alsoincludedconstructionofhealthcarefacilities,trainingofmedicalserviceworkers,and
drugpolicychanges.Theseareallappliedtotheresidingenvironmentandcanaffectthe
residents’healthcareutilizationandcosts.
4.3.2.2 Descriptive Statistics of Independent Variables Otherindependentvariablesincludedbasicdemographiccharacteristics,familysize
andwealth,healthmeasures,andhealthinsurancestatus.Table4.2showsdescriptive
statisticsofalltheindependentvariables.Inordertoreflectthedifferencebetweenrural
andurbanresidents,Ireportthestatisticsseparatelyforruralandurbanresidents.From
thedescriptivestatistics,ruralandurbanresidentsweresubstantiallydifferent.Inmy
sample,ruralresidentscontainedaslightlylargerportionofmalesandminoritiesthan
urbanresidents.Ruralresidentswereyoungerthanurbanresidents,onaverage,althoughI
observedagingtrendsinbothgroups.Moreurbanresidentsweremarried,butrural
residentsusuallyhadlargerhouseholdsizes.Urbanresidentshadhighereducationlevels
andincomesthandidruralresidents.
4.3.2.3 Equivalence Scale for Adjusting Household Income Inordertoprovideamoreaccuratemeasureofhouseholdincome,Iusedthe
equivalencescaletoadjustthesizeofhouseholdandthencomputedtheper‐capita
householdincomeusingtheadjustedhouseholdsize.Ichosetoapplyoneofthemost
commonlyusedscales,thesquare‐rootscale,whichinvolvesdividinghouseholdincomeby
thesquarerootofhouseholdsize.ThisscalewasadoptedbysomerecentOECD
publicationsonincomeinequalityandpoverty(e.g.,OECD2011).
35
4.3.2.4 Missing Value Imputation for Independent Variables Iperformedbasicimputationformissingvalues.Formaritalstatus,Ireplacedthe
missingvalueswith“nevermarried”iftherespondentwasyoungerthan18.Accordingto
China’smarriagelaw,theyoungestagetogetmarriedis18.Forhouseholdsizeand
householdincome,Iimputedthemissingvaluesusingotherhouseholdmembers’answers.
Forhouseholdregistrationtype,ifthevaluewasmissinginonewave,butthepreviousand
postwaveshadthesamevalues,Iassignedthisvaluetothemissingwave.
Formissingvaluesineducationyears,Iassigned0tothevariableiftherespondent
wasyoungerthanseven.Ifthevaluesinthepreviousandpostwaveswereequal,Iassigned
thesamevaluetothemissingwave.Ifthevaluesinlasttwowavesdidnotchange,I
assignedthesamevaluetothemissingwave.Iftherespondentwasolderthan30,I
assignedthevaluefromthepreviouswavetothemissingwave.Iusedthevaluefromthe
variableindicatingyearsofformaleducationtoimputethemissingvaluesinhighestlevel
offormaleducation,whichwasusedintheanalysis.Formissingvaluesforthevariableof
whethertherespondentwasstillinschool,Ireplacedthevaluewith0iftherespondent
wasyoungerthansevenorolderthan30.
Formissingvaluesinthevariableofhavinganymedicalinsurancecoverage,I
assigned1tothevariableiftherespondentclaimedtohaveanytypeofmedicalinsurance.
Afterthebasicimputation,therewerestillafewmissingvalues.Thepercentageof
missingvalueswasgenerallylessthan1%.Inordertobetterusetheinformationinthe
dataset,Icreatedadditionalcategoriesineachvariableindicatingwhetherthevaluewas
missingandincludedthecategoriesinmyanalysis.
36
Table 4.2 Descriptive Statistics of Independent Variables by Rural and Urban Residences and Registrations
Group RR Group RU Group UR Group UU N=47,084 N=16,424 N=9,734 N=21,228 gender male 0.496 0.514 0.482 0.485 female 0.504 0.486 0.518 0.515 Ethnicity Han 0.843 0.873 0.826 0.945 Minority 0.156 0.121 0.173 0.049 unreported 0.001 0.006 0.002 0.006 age age equal or below 5 0.063 0.044 0.057 0.035 age between 6 and 17 0.179 0.141 0.177 0.114 age between 18 and 60 0.621 0.632 0.642 0.625 age equal or above 61 0.136 0.183 0.123 0.225 unreported 0.001 0.001 0.002 0.001 marital status married 0.602 0.640 0.602 0.665 never married 0.332 0.281 0.328 0.251 other(divorced, widowed, or separated) 0.059 0.071 0.063 0.076 unreported 0.007 0.008 0.007 0.008 education level primary school 0.623 0.381 0.560 0.329 middle school 0.295 0.293 0.307 0.251 high school 0.074 0.256 0.116 0.287 college and above 0.003 0.063 0.011 0.126 unreported education status 0.005 0.006 0.006 0.007 whether still in school whether still in school 0.141 0.136 0.157 0.118 not in school 0.839 0.847 0.831 0.873 unreported whether in school 0.020 0.017 0.012 0.009 income groups low income group 0.384 0.261 0.314 0.173 medium income group 0.345 0.351 0.325 0.305 high income group 0.270 0.387 0.359 0.517 unreported 0.001 0.001 0.002 0.006 Note:
1.Income was adjusted for inflation to 2011 value 2. Adjusted per‐capita household income was used
37
4.4 Analytic Approach Difference‐in‐differences(DID)analysiscomprisedmymainanalyzingtechnique.
DIDanalysisassumesparalleltrendsincontrolandtreatmentgroupsbeforethepolicy
intervention.Forthevariablesforwhichtheparalleltrendsdidnothold,Iperformed
multivariatemodels,controllingforexistingtrends.Ialsoperformedseveralsensitivity
analyses,eachofwhichhaddifferentfocuses,asdiscussedinthenextsection.
4.4.1Difference‐in‐DifferencesAnalysiswithMultipleGroupsandMultiple
TimePeriods
Usingthelongitudinaldatacollectedinsevenwavesbetween1993and2011
enabledmetotakeaDIDapproachinmyempiricalanalysis.Thisapproachhasbecome
increasinglypopularintheempiricalliteratureontheeffectsofpublicpolicyinterventions.
TheDIDestimationisbasedonthesimpleideaofcomparingthedifferenceinoutcomes
beforeandafteraninterventionforgroupsaffectedbyittothedifferenceforunaffected
groups.ThegreatappealofDIDestimationcomesfromitssimplicityaswellasfromits
potentialtomitigatebiasesinthecomparisonbetweenthetreatmentandcontrolgroup
thatcouldbetheresultofpermanentdifferencesbetweenthosegroups,aswellasto
mitigatebiasesfromthepre‐postcomparisonofthetreatmentgroupthatcouldbethe
resultofseculartrendsunrelatedtotheintervention(CardandKrueger2000;Atheyand
Imbens2002;Bertrand,Dufloetal.2004;Abadie2005;ConleyandTaber2005).My
researchfocusedonthechangeindisparities.Further,thesettingoftheresearchquestions
madeDIDthemostsuitableapproach.
TheDIDanalysiscanbeexpandedtoincludemorethantwotimeperiods(Bertrand,
Dufloetal.2004;Hansen2007).Asdiscussed,therehavebeenthreemajorpolicychanges
38
inhealthinsuranceinChina.Iincludedallthreemajorpolicyinterventionsonhealth
insuranceinmymodel.Mymainhypothesiswasthatthesecondpolicychange,which
expandedinsurancecoverageinruralareasin2003helpedreducerural–urbandisparities
inhealthcareutilizationandcosts.However,itwasimportanttotaketheothertwopolicy
changesinurbanareasintoconsiderationandseparatetheeffectsfromdifferentpolicy
changes.
AftertheDIDmodel,Iinterpretedtheresultsusingthewholesampletomake
predictionsfordifferentresidenceandregistrationgroupsineachperiod.Theresultsare
presentedinbargraphs.Usingtheadjustedoutcomevariables,Iwasabletoobservethe
trendsindisparities.
4.4.1.1 Econometric Models Inthissection,IelaborateonhowIbuilteconometricmodelstoperformthe
analysisbasedontheconceptualframework.Fordifferentoutcomes,Iapplieddifferent
techniques.
Consideringthedichotomousvariables,suchaswhetherapersonusedoutpatient
care,IappliedlogisticregressionmodelandageneralframeworkconsideredbyBertrand,
Dufloetal.(2004)andHansen(2007).Empirically:
1, 2, 3, 4,
,
,
,
,
where p̂ denotestheprobabilitythatthedependentvariableequals1,and1‐ p̂ isthe
probabilitythatthedependentvariableequals0, t istheeffectofruralorurban
39
residence/registration,  r istheeffectofeachdifferenttimeperiod, xrt istheinteraction
termofresidence/registrationandtimeperiods, zirt istheindividualspecificcovariates,
 rt istheunobservedtime/groupeffect,and irt istheindividualspecificerror.Thus,  wasthepolicyeffectthatIplannedtoestimate.
Forthecontinuousvariableoftheamountoftotalhealthcareexpenditures,I
estimatedatwo‐partmodel,whichwasdevelopedtoaddresstwoproblemstypicalof
expendituresdata—first,thatmanyindividualshavezeroexpenditureandthatthe
distributionofnonzeroexpendituresishighlyskewed(DuanManningetal.1983).The
firstpartofthemodelwasalogitmodelontheprobabilityofhavingnonzerototalhealth
expenditures,andthesecondpartfocusedontheamountofhealthexpenditures
conditionalonnonzerohealthexpenditures.Forthesecondpartofthemodel,Iuseda
generalizedlinearmodel(GLM;ManningandMullahy2001).IperformedBox‐Coxtestto
selecttheappropriatelinkfunctionandaGLMfamilytest(Parktest)toselectGLMfamily.
Basedonthetestresults,gammafamilyandloglinkwereselected.Empirically:
Part1:
1, 2, 3, 4,
,
,
,
,
,
,
Part2:
yirt  t   r  xrt   z irt  rt   rt   irt ,
1, 2, 3, 4,
,
40
Totesttheresultsofthetwo‐partmodel,Iperformedabootstrapapproachwhen
producingthepredictionafterthemodelfitting.Iprovidethe95%confidenceintervalof
theadjustedresults.
4.4.1.2 Test of Trends Before the Policy Intervention TheDIDanalysesassumedsimilartrendsinthestudyoutcome,suchashealthcare
costs,amongthestudypopulationsbeforetheexpansionofhealthinsurancecoverage.To
testthisassumption,Iexaminedtrendsinthestudyoutcomesamongthestudy
populationsbyanalyzingthe1993–1997data,whichreflectedthesituationbeforethe
dramaticexpansionofhealthinsuranceinthelate1990s.
Table4.3showsthetestresultsusingthe1993and1997data.Column1shows
resultsofwhethertherespondentusedanyformalcareinthepreviousfourweeks.Asseen
intheresults,theinitialrural–urbandisparityestimatorsrangefrom0.560to0.789,
indicatingsignificantdisparitiesintheyear1993.TheDIDestimatorshowschangein
disparityin1997.Iobservednosignificantresultsinthechangeofdisparityforformalcare
utilization,indicatingsimilardisparitiesfrom1993to1997.Theseresultsruleoutthe
possibilityofchangesindisparitiesbeforethepolicyinterventions,suggestingthatthe
paralleltrendholdsforthevariableofformalcare.Therefore,Iconcludedthatstandard
DIDanalysiswassuitableforformalcareutilization.
Similarresultswereobservedforoutpatientutilization,whichareshownincolumn
2.Again,allgroupsusedlessformalmedicalcarethanGroupUUin1993.Thechangesin
disparityin1997werenotsignificantforanyofthethreegroups.Therefore,DIDanalysis
wasalsosuitableforoutpatientutilization.Resultsforinpatientutilizationareshownin
41
column3.Forthisvariable,however,Ididnotobserveanysignificantresultsintheinitial
disparityin1993,althoughthechangesindisparityin1997weresignificantforGroupRR
andGroupRU.Therefore,theparalleltrendassumptiondidnotholdforinpatientcare
utilization,preventingmefromusingstandardDIDanalysisforthisvariable.
Table 4.3 Results of DID Analysis Using 1993 and 1997 Waves for Healthcare Utilization
Formal care Odds Ratio Robust Std. Err. Outpatient Inpatient Odds Ratio Robust Std. Err. Odds Ratio Robust Std. Err. Disparity with Group UU in 1993 Group RR 0.674*** 0.080 0.695** 0.095 0.885 0.244 Group RU 0.789 0.113 0.666* 0.117 1.526 0.453 Group UR 0.560*** 0.100 0.591* 0.120 0.525 0.252 Change in disparity in 1997 Group RR 0.866 0.130 0.915 0.155 0.307** 0.116 Group RU 0.740 0.144 0.874 0.199 0.396* 0.165 Group UR 1.376 0.303 1.440 0.353 0.834 0.506 Note:
1.Significancelevel:***0.001,**0.01,*0.05.
2.Resultsforotherindependentvariablesareomitted.
Table4.4showstestresultsforhealthcarecosts.Theresultsaresimilartothose
observedforinpatientcareutilization.Columns1and2showresultsforwhetherhaving
OOPexceeding20%/40%ofhouseholdincome.Nosignificantdisparitieswereobservedin
1993,whilethedisparitiessignificantlydecreasedin1997forGroupRRandGroupRU.
Columns3and4showresultsforthetwo‐partmodelfortotalhealthcarecost.Fromthe
firstpart,nosignificantresultswereobservedforinitialdisparitiesin1993,whilethere
wasasignificantincreaseindisparityforGroupRRin1997.Forthesecondpart,therewas
significantdecreaseindisparitiesforallthreegroups.Theresultsindicatethattheparallel
42
trendsdidnotholdforthesevariables.Therefore,DIDanalysiswasnotsuitableforanyof
thevariables.
AsdiscussedinChapter2,in1990s,policychangeshavebeenimplementedinurban
areastoalleviatefinancialproblems,andthesemeasuresmayhaveincreasedcosts.In
ruralareas,however,thesituationwasnotimprovedduringthesameperiod.Therefore,
forsomeoftheoutcomevariables,Iobservedsignificantchangesinrural–urbandisparity
during1990s,evenbeforethefirstmajorhealthinsuranceexpansionin1998.Assuming
thetrendscontinuedinthefollowingyears,Iestimatedthefollowingmodels,which
includedvariablestocontrolforthetrendsbefore1998,andthenIexaminedthedeviation
fromtheexistingtrendsineachofthesubsequentwaves.
43
Table 4.4 Results of DID Analysis Using 1993 and 1997 Waves for Healthcare Costs
OOP>20% Household Income OOP>40% Household Income Odds Ratio Odds Ratio Robust Std. Err. Disparity with Group UU in 1993 Total Healthcare Cost Odds Ratio Robust Std. Err. Coef. Robust Std. Err. Group RR 1.096 0.217 1.261 0.318 0.880 0.107 95.617 266.246 Group RU 1.110 0.260 1.448 0.410 0.752 0.117 511.350 334.908 Group UR 1.283 0.339 1.370 0.452 0.846 0.144 344.013 371.668 Change in disparity in 1997 Robust Std. Err. Having any Healthcare Cost Group RR 0.499** 0.120 0.457** 0.135 0.627** 0.094 Group RU 0.505* 0.151 0.411* 0.146 0.753 Group UR 0.589 0.192 0.517 0.209 0.977 Note:
‐662.942* 302.294 0.148 ‐1212.681*** 375.975 0.201 ‐856.702* 424.727 1.Significancelevel:***0.001,**0.01,*0.05.
2.Resultsforotherindependentvariablesareomitted.
4.4.2MultivariateRegressionfortheVariablesthatdonotmeetthe
AssumptionofParallelTrends
Forthedependentvariablesinwhichtheparalleltrendsdidnothold,Iapplied
anothertechniquetoaccountforthepre‐existingtrendsin1990s.
Consideringthedichotomousvariables,suchasuseofinpatientcare,Iapplied
logisticregressionmodel.Empirically:
1993, 1997, … , 2011,
1993 ,
,
,
,
where p̂ denotestheprobabilitythatthedependentvariableequals1,and1‐ p̂ isthe
probabilitythatthedependentvariableequals0, istheeffectofruralorurban
residence/registration,
1993 isthetrendin1990sfordifferentgroups,
44
isthe
interactionbetweengroupsandyeardummyvariables,
istheinteractiontermof
residence/registrationandtimeperiods, zirt istheindividualspecificcovariate, rt isthe
unobservedtime/groupeffect,and irt istheindividualspecificerror.Thus,  wasthe
policyeffectthatIplannedtoestimate.
Afterthemultivariatemodelwascompleted,IcarriedoutaWaldtesttoexamine
whetherthedisparitiesweresignificantineachwaveandtoexaminewhetherthechange
indisparitiesbetweendifferentwaveswassignificant.
Forthecontinuousvariableoftheamountoftotalhealthcareexpenditures,I
estimatedatwo‐partmodel,discussedindetailinsection4.4.1.1.Empirically:
Part1:
1993 ,
1993, 1997, … , 2011,
,
,
,
Part2:
1993 1993, 1997, … , 2011,
,
,
,
,
Totesttheresultsofthetwo‐partmodel,Iperformedabootstrapapproachwhen
producingthepredictionafterthemodelfitting.Iprovidedthe95%confidenceintervalof
theadjustedresults.
45
4.5 Sensitivity analysis Iperformedseveralsensitivityanalysesinadditiontothebaselineresults,whichI
discussinthefollowingsection.
4.5.1ControllingforInsuranceStatus
Inthebasecase,Ididnotcontrolforinsurancestatus.Insurancecoverageisoneof
theaspectsthattheChinesehealthcarereformhasbeendesignedtochange.Iplannedto
examinehowinsurancecoveragechangesthedisparities.However,therewerepolicy
changesotherthaninsurancecoverageoccurringinthesameperiod.Asdiscussedin
Chapter2,therewereusuallyothermeasuresimplementedwhileChinaprovidedmore
healthinsurancecoveragetoresidents.Forexample,whenprovidingmorehealth
insurancecoverageforruralresidentsin2003,thegovernmentalsoprovidedfundingfor
medicalfacilityconstructionandtrainingofmedicalworkers.Medicalassistanceprograms
werealsoestablishedinbothruralandurbanareasindifferentyears.Thesemeasures
couldalsobeimportantinpromotinghealthcareutilization,aswellasreducingout‐of‐
pocketcosts.Therefore,IperformedtheDIDmodelswhilecontrollingforinsurancestatus
asasensitivityanalysistoexaminetheimpactofotherpolicychanges.Ithencompared
howmuchdisparitychangedwithandwithoutcontrollingforinsurance.
4.5.2DroppingtheRichestProvinceorthePoorestProvince
MyCHNSsamplecontainednineprovincesandthreemunicipalities,andthese
provincesvariedintermsofeconomicdevelopment.Inordertoexaminedifferenteffectsof
thepolicychangesindifferentprovinceswithunevendevelopment,Iperformedanalysis
withouttherichestandpoorestprovinces(intermsofGDPin2012,seeAppendixfor
details)andcomparedtheresultswithresultsfrommodelsusingthewholesample.
46
4.5.3IncludingInteractionTermswithHouseholdIncome
WhenstudyingtheimpactofNRCM,severalresearchersfounddifferenteffects
amongresidentswithdifferentincomelevels.Inordertoexaminewhetherthepolicyeffect
differedamongdifferentincomegroups,Iincludedaninteractiontermofhousehold
incomewithrural/urbanresidencesandregistrations.Inthisanalysis,Iclassifiedresidents
intothreecategoriesbyadjustedper‐capitahouseholdincome.Thethreegroupsarehigh‐,
medium‐,andlow‐incomegroups,representingthethreedifferentquintilesofadjusted
per‐capitahouseholdincome.Byincludingthisterm,Iwasabletostudythedifferentpolicy
effectsamongdifferentincomegroups.
4.5.4DIDAnalysisResultsforVariablesinWhichParallelTrendsdidnotHold
Asdiscussedpreviously,theparalleltrendsdidnotholdforinpatientcare
utilization,OOPexceeding20%/40%ofhouseholdincome,andtotalhealthcarecosts.
Therefore,Iusedamodelcontrollingforexistingtrendsbeforepolicyinterventionasthe
basemodelforthesevariables.Inthesemodels,Iassumedtheexistingtrendscontinuedin
thefollowingyears.IalsoperformedDIDanalysistodeterminewhethertheresultswere
differentwhennotcontrollingforexistingtrends.
47
Chapter 5 Results: Disparities in Healthcare Utilization 1.40
0.20
0.18
1.20
0.16
1.00
0.14
0.12
0.80
0.10
0.60
0.08
0.06
0.40
0.04
0.20
0.02
0.00
Ratio of other study groups to Group UU
Proportion of redients using medical care during the past 4 weeks
5.1 Descriptive Analysis Group RR
Group RU
Group UR
Group UU
Ratio: Group
RR/Group UU
Ratio: Group
Ru/Group UU
Ratio: Group
UR/Group UU
0.00
1
2
3
4
Period
Figure 5.1 Probability of Formal Care Utilization in 4 Weeks by Rural and Urban Residences and
Registrations
ColumnsinFigure5.1showthetrendsofproportionofresidentsseekingformal
medicalcarebyrural/urbanresidencesandregistrationtypes.Therearealsolines
showingtheratios,usingurbanresidentswithurbanregistrationasthebasegroup.Among
thefourgroups,residentsinGroupUUhadalwaysbeenusingthemostformalmedicalcare,
andGroupRRresidentshadalwaysbeenusingtheleast.GroupRUandGroupURremained
inthemiddle.However,theratiosbetweengroupschangedovertime.Inperiod1,before
thefirstpolicychangein1998,GroupRRusedabout60%asmuchformalmedicalcareas
didGroupUU.GroupURusedmoreformalcarethandidGroupRU.Inperiod2,afterthe
policychangein1998andbeforethe2003policychange,GroupUUusedagreateramount
ofmedicalcarethaninperiod1,andutilizationwithinGroupRRandRUalsoincreased
48
slightly.However,GroupURusedlessformalcarethaninperiod1.Asaresult,alltheratios
decreasedinthisperiod,andtheratiobetweenGroupURandUUdroppedthemost.In
period3,afterthe2003ruralpolicychange,utilizationwithinallgroupsincreased
dramatically.UtilizationwithinGroupRRRUandURincreasedmorethanGroupUU
utilization,resultinginhigherratios.Inperiod4,afterthepolicychangein2007,GroupUU
utilizationincreasedsteadilyagainwhileutilizationwithintheothergroupsonlyincreased
1.40
0.20
0.18
1.20
0.16
1.00
0.14
0.12
0.80
0.10
0.60
0.08
0.06
0.40
0.04
0.20
0.02
0.00
0.00
1
2
3
Ratio of other study groups to group UU
Proportion of residents using outpatient care during the past 4 weeks
slightlyinthisperiod.Therefore,theratiosdroppedinthisperiod.
Group RR
Group RU
Group UR
Group UU
Ratio: Group
RR/Group UU
Ratio: Group
Ru/Group UU
Ratio: Group
UR/Group UU
4
Period
Figure 5.2 Probability of Outpatient Care Utilization in 4 Weeks by Rural and Urban Residences and
Registrations
Similartrendswereobservedinoutpatientcareutilization.Figure5.2showsthe
trendsofproportionofresidentsusingoutpatientcarebyrural/urbanresidencesand
registrationtypes.Again,GroupRRhadalwaysbeenusinglessoutpatientservicesthan
othergroups,andGroupUUhadbeenusingthemost.Utilizationwithinallgroups
49
increasedalongtheperiods,exceptthattheutilizationofGroupURdecreasedinperiod2.
Theratiosdecreasedafterthe1998policychange,anditdecreasedmostforGroupUR.
Thentheratiosincreasedafterthe2003policychange,andfinallydroppedfollowingthe
0.2
1.4
UEBMI,
1998
0.18
0.16
NRCM,
2003
URBMI,
2007
1.2
1
0.14
0.12
0.8
0.1
0.6
0.08
0.06
0.4
0.04
0.2
0.02
0
0
1993
1997
2000
2004
Wave
2006
2009
2011
Ratio of other study groups to group UU
Proportion of residents using inpatient care duing the past 4 weeks
2007policychange.
Group RR
Group RU
Group UR
Group UU
Ratio: Group
RR/Group UU
Ratio: Group
RU/Group UU
Ratio: Group
UR/Group UU
Figure 5.3 Probability of Inpatient Care Utilization in 4 Weeks by Rural and Urban Residences and
Registrations
Similartotheresultsfoundforoverallformalcareandoutpatientutilization,Group
RRalmostalwaysusedlessinpatientcarethandidalltheothergroups.However,inpatient
careutilizationshowsdifferenttrends.Theratiosalsodroppedinthe1990sandincreased
inthe2000s,withdifferentslopesfordifferentgroups.Notethattheamountofinpatient
carewasverysmallinthesample.Fewerthan1%oftherespondentsusedinpatientcare
50
withinthefour‐weekperiodbeforetheinterview.Therefore,datathataremore
informativemightbeneededtodiscerntherealpattern.
5.2 DID Analysis for Formal Care Utilization and Outpatient Utilization TheDIDanalysisresultsforrural–urbandisparitiesinformalhealthcareutilization
andoutpatientutilizationarepresentedinTable5.1.TheseDIDmodelsincludedfour
categoriesofdifferentruralandurbansettings,usingGroupUUasthereferencegroup.
Table5.1column1reportsresultsforformalcare.UsingGroupUUasthereference
group,theinitialrural–urbandisparityestimatorsrangedfrom0.586to0.688,indicating
thatthereweregreatrural–urbandisparitiesgoingbacktotheearly1990s.Amongthe
threegroups,GroupRRusedtheleastformalcare;GroupURusedthemost.Changein
disparitiescanbeindicatedfromtheDIDestimators.Thedisparitiesincreasedforallthree
groupsinperiod2sincethepolicychangein1998.Subsequently,inperiods3and4,the
disparitiesdecreasedcomparedwiththeinitialperiod.However,mostofthechangeswere
notsignificantexceptforGroupRRinperiod4andGroupRUinperiods3and4.
Inordertotestthechangeofdisparitiesbetweentwoadjacentperiods,Iperformed
WaldtestsaftertheDIDanalysis.Ifthetestresultwassignificant,Irejectedthenull
hypothesisthatchangeinperiod2equaledchangeinperiod3.Thetestresultsareshown
inTable5.2.Forformalcareutilization,testresultscomparingthechangeinperiod2with
changeinperiod3weresignificantforallthreegroups.Therefore,Irejectedthenull
hypothesisthatthechangeinperiod2equaledchangeinperiod3.Theseresultsshowthat
GroupsRR,RU,andURallimprovedafterthepolicychangein2003,comparedwiththeir
counterpartsfromGroupUU.
51
Ialsoobservedsignificanteffectsinotherindependentvariables.Malerespondents
usedlessformalmedicalcarethandidfemales.Minoritiesusedlessformalmedicalcare
thandidHanChinese.Childrenundertheageofsixandseniorsovertheageof60used
moreformalmedicalcarethanmiddle‐agedgroups.Peoplewhowerenevermarriedused
lessformalmedicalcarethandidthoseinthemarriedgroup.Peoplewhosehighest
educationlevelwaslowerthanprimaryschoolusedmoremedicalcare,butthismay
becausethesampleincludedchildrenwhowerestillinschool.Finally,therewere
differencesacrossdifferentprovinces.UsingtheprovincewiththehighestGDPlevelasthe
referencegroup,theotherprovincesgenerallyusedlessformalmedicalcare,exceptfor
GuangxiandHenan.3Thisdifferencemayhavebeenduetodifferenthealthcarepoliciesin
differentprovinces.
Similarresultswereobservedforoutpatientcareutilization.Inthefirstperiod,
GroupsRR,RU,andURusedabout60%to78%ofoutpatientservicescomparedtothe
amountusedbyurbanresidents.Inperiod2,however,thedisparitiesincreasedforall
threegroups,asdeterminedfromDIDestimatorssmallerthan1.Inperiod3,the
disparitiesshrankcomparedwiththefirsttwoperiods.Finally,inperiod4,thedisparities
diminished,comparedwithperiod1.However,comparedwiththeadjacentperiod3,the
disparitiesincreasedslightlyforGroupRU.Thefluctuationofdisparitiesoverthefour
periodsindicatesthatrural–urbandisparitiesinoutpatientcareutilizationincreasedafter
thepolicychangein1998,diminishedafterthepolicychangein2003,andslightly
decreasedafterthepolicychangein2007(exceptforGroupRU).TheWaldtestresultsfor
outpatientcareweresignificantforallgroupsinperiods2and3,showingthatallthree
3
Jiangsu province, which had the biggest GDP value in 2012, was used as the reference group.
52
groupshadimprovedoutcomesafterthe2003policychange,comparedwiththeir
counterpartsfromGroupUU.Theotherindependentvariablesshowthesameeffectsfor
outpatientcareutilizationasforoverallformalcareutilization.
53
Table 5.1 DID Analysis Results for Formal Care Utilization and Outpatient Utilization
Formal care
Independent Variable Odds Ratio Robust Std. Err.
disparities with Group UU in period 1 Group RR 0.586***
0.046
Group RU 0.652***
0.064
Group UR 0.688***
0.077
periods period 1 1
n/a
period 2 1.368***
0.126
period 3 2.359***
0.175
period 4 2.055***
0.153
change in disparities Group RR in period 2 0.807
0.095
Group RR in period 3 1.145
0.106
0.115
1.248*
Group RR in period 4 Group RU in period 2 0.872
0.133
Group RU in period 3 1.383**
0.160
Group RU in period 4 1.263*
0.143
Group UR in period 2 0.740
0.138
Group UR in period 3 1.093
0.151
Group UR in period 4 1.219
0.166
gender 0.886***
0.025
male female 1
n/a
ethnicity minority 0.788***
0.044
Han 1
n/a
age age equal or below 5 2.458***
0.203
age between 6 and 17 1.036
0.108
age between 18 and 60
1
n/a
age equal or above 61 2.282***
0.080
marital status married 1
n/a
0.037
0.558***
never married other (divorced, widowed or separated) 0.997
0.047
education level primary school 1
n/a
middile school 0.730***
0.027
high school 0.712***
0.033
college or higher 0.727***
0.050
whether still in school in school 0.948
0.093
not in school 1
n/a
adjusted per capita household income low household income 1.039
0.034
medium household income 1
n/a
high household income
1.026
0.033
province Jiangsu 1
n/a
Liaoning 0.682***
0.046
Heilongjiang 0.444***
0.033
Shandong 0.641***
0.041
Henan 1.233***
0.069
Hubei 0.851**
0.052
Hunan 0.856**
0.050
Guangxi 1.300***
0.072
0.051
Guizhou 0.759***
Beijing 2.583***
0.213
Shanghai 2.749***
0.212
Chongqing 1.216*
0.111
Note:
1.Significancelevel:***0.001,**0.01,*0.05.
54
Outpatient
Odds Ratio Robust Std. Err. 0.628***
0.609***
0.783*
0.055 0.068 0.094 1
1.495***
2.561***
2.159***
n/a 0.151 0.210 0.178 0.801
1.143
1.234*
0.985
1.513***
1.386**
0.659*
0.974
1.108
0.103 0.115 0.124 0.167 0.196 0.176 0.132 0.142 0.161 0.862***
1
0.025 n/a 0.752***
1
0.044 n/a 2.548***
1.087
1.000
2.150***
0.226 0.123 n/a 0.081 1
0.557***
0.997
n/a 0.039 0.050 1
0.739***
0.726***
0.715***
n/a 0.029 0.035 0.052 0.983
1
0.105 n/a 1.057
1
1.039
0.037 n/a 0.035 1
0.671***
0.454***
0.610***
1.214***
0.814**
0.831**
1.349***
0.774***
2.914***
3.188***
1.324**
n/a 0.047 0.036 0.042 0.072 0.053 0.052 0.079 0.055 0.248 0.253 0.125 Table 5.2 Test Results for DID Analysis of Formal Care Utilization and Outpatient Utilization
Formal Care chi2 Group RR Change in disparity in period 2 = Change in disparity in period 3 8.19** 0.0042 7.85** 0.0051 Change in disparity in period 3 = Change in disparity in period 4 0.07 0.7977 0.09 0.7636 Group RU Change in disparity in period 2 = Change in disparity in period 3 10.83*** Prob>chi Outpatient chi2 Prob>chi 0.0010 8.25** 0.0041 0.1478 1.56 0.2114 Change in disparity in period 3 = Change in disparity in period 4 2.09 Group UR Change in disparity in period 2 = Change in disparity in period 3 4.87* 0.0274 4.25* 0.0393 Change in disparity in period 3 = Change in disparity in period 4 0.12 0.7331 0.36 0.5460 Note:
1.Significancelevel:***0.001,**0.01,*0.05.
BasedontheresultsfromDIDanalysis,Ipredictedtheprobabilitiesofformalcare
andoutpatientinfourweeksbyruralandurbanresidencesandregistrationsforfourtime
periods.TheresultsareshowninFigure5.4andFigure5.5.
Figure5.4showspredictedprobabilityofformalmedicalcareutilizationinfour
weeks.AlltheratiostoGroupUUhadalwaysbeenlessthan1,butchangedovertime.The
ratiosdecreasedbetweenperiods1and2andincreasedbetweenperiods2and3.
Subsequently,theratioforGroupRUdecreasedslightlybetweenthelasttwoperiodsand
increasedslightlyforGroupsRRandUR.ThesetrendswereconsistentwithwhatI
observedindescriptivefiguresandshowthatthepolicychangesresultedinfirstmore,
55
thenlessrural–urbandisparityinformalcareutilization.Asdiscussedpreviously,the
changebetweenperiods2and3wassignificant.Amongthethreegroups,GroupRRhad
alwaysbeenthelowest.Figure5.5showsthepredictedprobabilityofoutpatientcare
1.40
0.20
0.18
1.20
0.16
1.00
0.14
0.12
0.80
0.10
0.60
0.08
0.06
0.40
0.04
0.20
0.02
0.00
0.00
1
2
3
Ratio of other study groups to group UU
Predicted probability of redients using medical care during the past 4 weeks
utilization.Asimilarpatternwasobservedinthisfigure.
Group RR
Group RU
Group UR
Group UU
Ratio: Group
RR/Group UU
Ratio: Group
RU/Group UU
Ratio: Group
UR/Group UU
4
Period
Figure 5.4 Predicted Probability of Formal Care Utilization in 4 Weeks by Rural and Urban Residences and
Registrations
56
Predicted probability of residents using outpatient care during the past 4 weeks
0.18
1.20
0.16
1.00
0.14
0.12
0.80
0.10
0.60
0.08
0.06
0.40
0.04
0.20
0.02
0.00
0.00
1
2
3
Ratio of other study groups to group UU
1.40
0.20
Group RR
Group RU
Group UR
Group UU
Ratio: Group
RR/Group UU
Ratio: Group
RU/Group UU
Ratio: Group
UR/Group UU
4
Period
Figure 5.5 Predicted Probability of Outpatient Care Utilization in 4 Weeks by Rural and Urban Residences
and Registrations
5.3 Multivariate Analysis Controlling for Existing Trends for Inpatient Utilization Forinpatientcare,Iappliedmultivariateregression,controllingforexistingtrends,
andtheresultsareshowninTable5.3.Theinitialcoefficientsofdisparitiesweresmaller
than0forGroupRRandUR,meaningthatthetwogroupsusedlessinpatientcarethandid
GroupUU.GroupRUusedmoreinpatientcarecomparedwithGroupUU.However,noneof
thedisparitieswassignificant.ForGroupRR,thetrendinthe1990swasnegative,andthe
resultwassignificant.Ifthetrendpersisted,GroupRRwoulduselessandlessinpatient
careinthefollowingyears.However,thisgroupexperiencedapositivedeviationfromthe
trendinallofthesubsequentyears.Thisdeviationcouldbecausethepolicychangein2003
providedmorehealthinsurancecoverageforGroupRR.Thedeviationsinallyearsafter
2004weresignificant.Thisindicatesthepolicyimpactpersistedinthesubsequentyears.
GroupRUfollowedthesamepatternasGroupRR.However,noneoftheresultsforGroup
57
RUwassignificant.ForGroupUR,thetrendwaspositive;deviationin2000wasnegative,
andthenallthedeviationsinthesubsequentyearswerepositive.ForGroupUU,thetrend
waspositivebutnotsignificant.Inallthefollowingyears,thedeviationfromtrendwas
negative,andthedeviationin2000wassignificant.
Table5.4showstestresultsofdisparitiesbetweenGroupUUandothergroups.As
discussed,thedisparityisthedifferencebetweentheprobabilityofhavinganyinpatient
carevisitforGroupUU,comparedtotheothergroups.Column1showsdisparities,andthe
testresultsareincolumns2and3.ForGroupRR,disparitywithGroupUUin1997is0.012,
indicatingthattheprobabilityofhavinginpatientcarevisitwasgreaterinGroupUUthan
inGroupRR.Thedifferenceinprobabilitieswas0.012.Thetestresultshowsthatthe
disparitywasnotsignificant.ForGroupRR,disparitieswithGroupUUwereallpositive,
meaningthatGroupRRhadalwaysbeenusinglessinpatientcarecomparedwithGroupUU.
In2000,2004,2006,and2011,thedisparitiesweresignificant.ForGroupRU,similarly,the
disparitieswereallpositiveexceptforthedisparityin2009.ForGroupUR,disparities
wereallpositive.However,noneoftheresultswassignificantforGroupRUandonly
significantin2000forGroupUR.
58
Table 5.3 Multivariate Analysis Results for Inpatient Care Utilization
Independent Variables Coef. disparity with Group UU in 1993 Group RR ‐0.235 0.261 Group RU 0.319 0.291 Group UR ‐0.688 0.465 Group RR ‐0.204** 0.066 deviation from trend in 2000 0.566 0.464 deviation from trend in 2004 1.812** 0.687 deviation from trend in 2006 2.370** 0.809 deviation from trend in 2009 3.328*** 0.996 deviation from trend in 2011 4.159*** 1.121 Group RU trend in 1990s trend in 1990s and change in later waves trend in 1990s Robust Std. Err. ‐0.131 0.077 deviation from trend in 2000 0.171 0.524 deviation from trend in 2004 0.882 0.774 deviation from trend in 2006 1.554 0.916 deviation from trend in 2009 2.106 1.129 deviation from trend in 2011 2.390 1.284 Group UR trend in 1990s 0.047 0.135 deviation from trend in 2000 ‐1.018 0.905 deviation from trend in 2004 0.143 1.253 deviation from trend in 2006 0.063 1.509 deviation from trend in 2009 0.229 1.895 deviation from trend in 2011 0.180 2.160 Group UU trend in 1990s 0.105 0.066 deviation from trend in 2000 ‐0.756* 0.380 deviation from trend in 2004 ‐0.820 0.599 deviation from trend in 2006 ‐1.125 0.728 deviation from trend in 2009 ‐1.481 0.918 deviation from trend in 2011 ‐1.163 1.041 constant ‐4.586*** 0.274 Note:
1.Significancelevel:***0.001,**0.01,*0.05.
59
Table 5.4 Test Results of Disparities for Inpatient Care Utilization
Disparity Chi2 Group RR disparity (Group UU probability‐Group RR probability) 1997 0.0117 0.13 2000 0.0065 13.24*** 2004 0.0088 15.96*** 2006 0.0067 9.50** 2009 0.0037 2.71 2011 0.0076 7.96** Group RU disparity (Group UU probability‐Group RU probability) 1997 0.0070 0.44 2000 0.0033 1.48 2004 0.0061 3.71 2006 0.0008 0.07 2009 ‐0.0017 0.28 2011 0.0063 3.37 Group UR disparity (Group UU probability‐Group UR probability) 1997 0.0091 2.39 2000 0.0073 4.89* 2004 0.0042 0.99 2006 0.0028 0.50 2009 0.0094 0.10 2011 0.0169 2.05 Note:
1.Significancelevel:***0.001,**0.01,*0.05.
Table5.5showstestresultsforthechangeindisparities.Themajorhealth
insurancepolicychangesoccurredin1998,2003,and2007.Therefore,Icomparedthe
disparitiesintheyearsbeforetheinitiationofeachpolicyintervention(1997,2000,and
2006)withallthewavesthatoccurredafterwardandthentestedforthesignificanceofthe
changeindisparities.ForGroupRR,thedisparitydecreasedin2000by0.5%,compared
withthedisparityin1997.However,thechangewasnotsignificant,asshownbythetest
resultsincolumns2and3.Inallthesubsequentwaves,thedisparitiesweresmallerthanin
60
1997.Thechangesweresignificantfor2009and2011.Thedisparitywasreducedby0.8%
in2009andby0.4%in2011.ComparedtothedisparitywithGroupUUin2000,the
disparitywaslargerin2004and2011andsmallerin2009.However,noneofthechanges
wassignificant.Comparedwithdisparityin2006,thedisparitywassmallerin2009and
largerin2011.Again,thechangeswerenotsignificant.ForGroupRU,thechangewasnot
significantforanyofthefollowingyearscomparedwithdisparitiesin1997,2000,or2006.
ForGroupUR,thedisparityincreasedin2009comparedwith1997and2000,andthe
changewassignificant.Insum,therewasnosignificantchangeindisparitiesintheyears
immediatelyafterthemajorpolicyinterventions.ThedisparitybetweenGroupRRandUU
decreasedfrom1997in2009and2011.However,noevidenceshowsthatitwasduetothe
policychangein2000.
61
Table 5.5 Test Results of Change in Disparities for Inpatient Care Utilization
Change In Disparity
Group RR compare with disparities with Group UU in 1997
2000 ‐0.0052
2004 ‐0.0029
2006 ‐0.0050
2009 ‐0.0080
2011 ‐0.0041
compare with disparities with Group UU in 2000
2004 0.0023
2006 0.0002
2009 ‐0.0027
2011 0.0012
compare with disparities with Group UU in 2006
2009 ‐0.0030
2011 0.0009
Group RU compare with disparities with Group UU in 1997
2000 ‐0.0038
2004 ‐0.0010
2006 ‐0.0062
2009 ‐0.0087
2011 ‐0.0007
compare with disparities with Group UU in 2000
2004 0.0028
2006 ‐0.0024
2009 ‐0.0050
2011 0.0031
compare with disparities with Group UU in 2006
2009 ‐0.0025
2011 0.0055
Group UR compare with disparities with Group UU in 1997
2000 ‐0.0018
2004 ‐0.0049
2006 ‐0.0063
2009 0.0003
2011 0.0077
compare with disparities with Group UU in 2000
2004 ‐0.0031
2006 ‐0.0045
2009 0.0021
2011 0.0095
compare with disparities with Group UU in 2006
2009 0.0066
2011 0.0140
Note:
1.Significancelevel:***0.001,**0.01,*0.05.
62
Chi2 0.96 1.58 3.75 9.61** 9.64** 0.04 0.71 3.74 3.24 1.45 1.01 0.24 0.01 2.02 3.92 0.49 0.14 0.64 1.65 0.01 0.30 0.86 0.36 1.10 1.60 4.14* 1.26 1.97 2.46 4.51* 2.12 0.59 0.10 Figure5.6showsthepredictedprobabilityofinpatientcareutilization.ForGroup
RR,theratiodecreasedinthe1990s,increasedinthe2000s,andfinallydecreasedin2011.
ForGroupRUandUR,theratiodoesnotshowanypattern.Asdiscussedbefore,the
variableonlymeasuredinpatientvisitsinafour‐weekperiod,andtheproportionof
residentsusinginpatientcarewasverysmall.Thedatamaynotbesufficienttoshowthe
realpattern,andmoredetaileddataisneeded.
1.4
UEBMI Launch, 1998
NRCM Launch, 2003
0.02
URBMI Launch, 2007
1.2
1
0.015
0.8
0.6
0.01
0.4
0.005
0.2
0
0
1993
1997
2000
2004 2006
Wave
2009
2011
RatioofotherstudygroupstogroupUU
Predicetedprobabilityofresidentsusing
inpatientcareduringthepast4weeks
0.025
GroupRR
GroupRU
GroupUR
GroupUU
Ratio:Group
RR/GroupUU
Ratio:Group
RU/GroupUU
Ratio:Group
UR/GroupUU
Figure 5.6 Predicted Probability of Inpatient Care Utilization in 4 Weeks by Rural and Urban Residences
and Registrations
63
5.4 Sensitivity Analysis 5.4.1ControllingforInsuranceStatus
Myfirstsensitivityanalysisinvolvedcontrollingforinsurancestatus.Afterthe
analysis,Ialsoperformedteststoexaminewhetherthereweresignificantchanges
betweenadjacentperiods/waves.Theregressionresultsforformalcareandoutpatient
utilizationareshowninTable5.6,andthetestresultsareshowninTable5.7.
Fromthesemodels,Iobservedsimilareffectsaswereobservedinthebasemodels.
Column1showsresultsforformalcareutilization.Havinghealthinsurancecoveragehada
positiveeffectonformalcareutilization.Whencontrollingforinsurancestatus,therewere
rural–urbandisparitiesinperiod1,astheoddsratioforallgroupswerelessthan1.Group
RRwasstilltheworstperformingintermsofusingformalmedicalcare.Comparedwith
modelsnotcontrollingforinsurance,theoddsratioswerelarger.Theresultsindicatethat
havinginsurancecoveragecouldexplainpartofthedisparitiesinformalcareutilization.
Themagnitudeofchangesindisparitieswassmallercomparedwiththebasemodels.
However,thedisparitiesinthelastthreewavesweregenerallynotsignificantfromperiod
1.Thetrendsofchangesindisparitiesweresimilarwiththebasemodels.ForGroupRR
andUR,thedisparitiesincreasedinperiod2anddecreasedinperiods3and4.ForGroup
RU,thedisparitiesincreasedinperiod2,decreasedinperiod3,andfinallyincreasedagain
inperiod4.TheWaldtestresultsindicatedthatthechangesindisparitiesforallgroups
fromperiods2to3weresignificant.Thiswasalsoconsistentwiththebasemodels.The
oddsratioforchangeindisparitiesdecreasedcomparedwithbasemodels.After
controllingforinsurancestatus,thechangesindisparitieswerestillsignificant,butsmaller.
64
Theresultsindicatethatthedisparitieswerereducednotonlybecauseofmorehealth
insurancecoveragebutalsobecauseofotherpolicyinterventions.Iobservedthesame
resultsforoutpatientcareutilization.
Table 5.6 DID Analysis Results of Formal Care and Outpatient Utilization (Controlling for Insurance Status)
Formal Care Independent Variable Odds Ratio Outpatient Robust Std. Err. disparities with Group UU in period 1 Robust Std. Err. Odds Ratio Group RR 0.645*** 0.051 0.680*** 0.060 Group RU 0.688*** 0.068 0.635*** 0.071 Group UR 0.741** 0.083 0.829 0.100 periods period 1 1 period 2 1.401*** 0.129 1.524*** 0.154 period 3 2.411*** 0.179 2.605*** 0.213 period 4 1.940*** 0.146 2.059*** 0.171 change in disparities Group RR in period 2 0.792* 0.093 0.789 0.101 Group RR in period 3 1.071 0.100 1.083 0.110 Group RR in period 4 1.091 0.102 1.108 0.113 Group RU in period 2 0.864 0.132 0.978 0.165 Group RU in period 3 1.342* 0.155 1.476** 0.191 Group RU in period 4 1.185 0.135 1.317* 0.167 Group UR in period 2 0.730 0.136 0.652* 0.130 Group UR in period 3 1.063 0.147 0.953 0.139 Group UR in period 4 1.103 0.150 1.023 0.149 whether having insurance insurance 1.264*** not having insurance 1 Note:
n/a 1 n/a 0.043 n/a 1.Significancelevel:***0.001,**0.01,*0.05.
2.Resultsforotherindependentvariablesareomitted.
65
1.208*** 1 0.044 n/a Table 5.7 Test Results for DID Analysis of Healthcare Utilization (Controlling for Insurance Status)
Formal Care chi2 Prob>chi GroupRR
chi2 Prob>chi Changeindisparityinperiod2=Changein
disparityinperiod3
8.19** 0.0042 7.85** 0.0051 Changeindisparityinperiod3=Changein
disparityinperiod4
0.07 0.7977 0.09 0.7636 GroupRU
Changeindisparityinperiod2=Changein
disparityinperiod3
10.83*** Changeindisparityinperiod3=Changein
disparityinperiod4
2.09 0.0010 8.25** 0.0041 0.1478 1.56 0.2114 GroupUR
Changeindisparityinperiod2=Changein
disparityinperiod3
4.87* 0.0274 4.25* 0.0393 Changeindisparityinperiod3=Changein
disparityinperiod4
0.12 0.7331 0.36 0.5460 Note:
1.Significancelevel:***0.001,**0.01,*0.05.
Outpatient 66
Table 5.8 Multivariate Analysis Results for Inpatient Care Utilization (Controlling for Insurance Status)
Independent Variables Coef.
Robust Std. Err. disparity with Group UU in 1993 Group RR 0.138
0.272 Group RU 0.472
0.292 Group UR ‐0.418
0.468 trend in 1990s and change in later waves
Group RR trend in 1990s ‐0.219***
0.066 deviation from trend in 2000 0.657
0.464 deviation from trend in 2004 1.921**
0.687 deviation from trend in 2006 2.269**
0.808 deviation from trend in 2009 3.000**
0.994 3.850***
1.119 deviation from trend in 2011 Group RU ‐0.112
0.077 trend in 1990s deviation from trend in 2000 0.116
0.524 deviation from trend in 2004 0.713
0.774 deviation from trend in 2006 1.281
0.918 deviation from trend in 2009 1.539
1.133 deviation from trend in 2011 1.750
1.289 Group UR trend in 1990s 0.054
0.135 deviation from trend in 2000 ‐1.014
0.905 deviation from trend in 2004 0.106
1.254 deviation from trend in 2006 ‐0.130
1.510 deviation from trend in 2009 ‐0.307
1.898 deviation from trend in 2011 ‐0.381
2.161 Group UU trend in 1990s 0.125
0.066 deviation from trend in 2000 ‐0.792*
0.381 ‐0.922
0.600 deviation from trend in 2004 deviation from trend in 2006 ‐1.316
0.730 deviation from trend in 2009 ‐1.879*
0.920 deviation from trend in 2011 ‐1.634
1.044 whether having insurance insurance 0.655***
0.107 not having insurance 0
n/a constant ‐5.121***
0.288 Note:
1.Significancelevel:***0.001,**0.01,*0.05.
2.Resultsforotherindependentvariablesareomitted.
MultivariateanalysisresultsforinpatientcareutilizationareshowninTable5.8,
andthecorrespondingtestresultsareshowninTables5.9and5.10.Havinginsurance
coveragehadapositiveeffectonusinginpatientcare.Similartotheresultsseenforthe
basemodel,disparitiesininpatientcareutilizationforallothergroupswithGroupUUin
67
1993werenotsignificant.Lookingatthetrends,therewasasignificanttrendinthe1990s
onlyforGroupRR.ThetrendforGroupRRwasnegative,andtherewasasignificant
deviationfromthetrendinlateryears.Forothergroups,similarresultswereobservedas
thoseobservedinthebasemodel,andtheresultsweregenerallynotsignificant.After
controllingforinsurancestatus,themagnitudesofothercoefficientsweregenerally
smaller.Theresultsindicatethatthechangeindisparitiescouldpartlybeexplainedby
insurancestatus.
Testresults,showninTable5.9,wereconsistentwiththebasemodel.The
disparitiesbetweenGroupUUandGroupRRwerepositiveinallyears,indicatingthat
GroupRRwaslesslikelytouseinpatientcarecomparedwithGroupUU.Thedisparities
weresignificantfromyears2000to2011.ForGroupRUandUR,thedisparitieswerealso
positive,butonlythedisparitybetweenGroupRUandUUin2011wassignificant.
68
Table 5.9 Test Results of Disparities for Inpatient Care Utilization (Controlling for Insurance Status)
Disparity Chi2 Group RR disparity (Group UU probability‐Group RR probability) 1997 0.0069 0.26 2000 0.0240 7.37** 2004 0.0064 9.99** 2006 0.0081 9.16** 2009 0.0075 4.57* 2011 0.0023 11.29** Group RU disparity (Group UU probability‐Group RU probability) 1997 0.0037 0.91 2000 0.0220 0.67 2004 0.0049 2.76 2006 0.0039 0.00 2009 0.0043 0.17 2011 0.0012 3.93* Group UR disparity (Group UU probability‐Group UR probability) 1997 0.0049 1.46 2000 0.0248 3.44 2004 0.0027 0.23 2006 0.0050 0.20 2009 0.0049 0.00 2011 0.0016 2.77 Note:1.Significancelevel:***0.001,**0.01,*0.05.
69
Table 5.10 Test Results of Change in Disparities for Inpatient Care Utilization (Controlling for Insurance
Status)
Change In Disparity
Chi2 Group RR compare with disparities with Group UU in 1997
2000 0.0171
1.07 2004 ‐0.0005
1.37 2006 0.0012
1.75 2009 0.0006
4.36* 2011 ‐0.0046
4.25* compare with disparities with Group UU in 2000
2004 ‐0.0176
0 2006 ‐0.0159
0.04 2009 ‐0.0165
0.81 2011 ‐0.0217
0.55 compare with disparities with Group UU in 2006
2009 ‐0.0005
0.67 2011 ‐0.0057
0.41 Group RU compare with disparities with Group UU in 1997
0.20 2000 0.0183
2004 0.0012
0.00 2006 0.0003
1.42 2009 0.0006
2.26 2011 ‐0.0024
0.04 compare with disparities with Group UU in 2000
2004 ‐0.0171
0.24 2006 ‐0.0181
0.40 2009 ‐0.0177
0.81 2011 ‐0.0207
0.11 compare with disparities with Group UU in 2006
2009 0.0004
0.09 2011 ‐0.0027
1.45 Group UR compare with disparities with Group UU in 1997
0.36 2000 0.0199
2004 ‐0.0021
0.98 2006 0.0001
1.05 2009 0.0000
2.00 2011 ‐0.0033
0.28 compare with disparities with Group UU in 2000
2004 ‐0.0220
1.86 2006 ‐0.0198
1.93 2009 ‐0.0199
2.80 2011 ‐0.0232
1.06 compare with disparities with Group UU in 2006
2009 ‐0.0001
0.14 0.47 ‐0.0035
2011 Note:
1.Significancelevel:***0.001,**0.01,*0.05.
70
Table5.10showstestresultsofchangeindisparities.Aftercontrollingforinsurance,
thechangewasstillsignificantforGroupRRin2009and2011.However,thedirectionof
changewasdifferentin2009.Thiswasalsotrueforchangesindisparitiesforothergroups.
However,theresultswerenotsignificantforanyofthechangesindisparitiesforother
groups.Aftercontrollingforinsurance,someofthechangesindisparitieswerenot
significant,asseeninthebasemodel.Thismaybebecausethechangeindisparitiescan
partlybeexplainedbyinsurancestatus.However,themagnitudeofchangeindisparities
wasverysmall.Asdiscussedbefore,theproportionofresidentsusinginpatientcarewas
verysmall.Furtherdatacollectionisneededtorevealthepatternofinpatientcare
utilization.
5.4.2DroppingtheRichestProvinceorthePoorestProvince
Thesecondsetofsensitivityanalysistechniquesinvolveddroppingoneofthe
provincesfromtheanalysistocheckwhethertheresultsstillheld.Idroppedtherichest
province,Jiangsu,inthefirstsetofmodels,andthendroppedthepoorestprovince,
Guizhou,inthesecondsetofmodels.Theresultsforformalcareandoutpatientutilization
areshowninTable5.11andTable5.13.Aftertheregression,IalsoperformedWaldteststo
examinethechangebetweentwoperiods,andtheresultsareshowninTables5.12and
5.14.
AsshowninTable5.11,column1,forformalcareutilization,theresultswerevery
similartothebasemodelafterdroppingJiangsu,therichestprovince.Theoddsratiosfor
allthreegroupsweresmallerthan1,indicatingthattherewasrural–urbandisparityin
termsofformalcareutilizationinitiallyinperiod1.Thechangeindisparityinperiod2was
smallerthan1,andinperiods3and4weregreaterthan1.Thisindicatesthatthe
71
disparitieswerelargerinperiod2comparedwithperiod1,andinperiods3and4,the
disparitiesweresmaller.ThechangesindisparityforGroupsRRandURkeptincreasing
fromperiods2to4.Thistrendindicatesthatthedisparitiesshrankthroughoutthelast
threeperiods.AsshowninTable5.12,column1,therewassignificantchangeindisparities
betweenperiods2and3forGroupsRRandRU.Thechangewasassociatedwiththe2003
policychangeinruralarea.Nosignificantchangeindisparitywasobservedbetweenother
periods.
Thedifferencewiththebasemodelwasthatnosignificantchangeindisparitywas
observedbetweenperiods2and3forGroupUR.AlthoughGroupURwasalsounderrural
householdregistrationandprovidedmorehealthinsurancecoveragebetweenperiods2
and3,nosignificantpolicyeffectwasobservedafterdroppingtherichestprovince.The
observationindicatesthatthepolicywasmoreeffectiveinreducingdisparitiesinformal
careutilizationinrichprovinces.Whendroppingtherichestprovince,theeffect
disappeared.ThereasonIstillobservedpositiveeffectsinGroupsRRandRUmaycome
fromtheothermeasuresaffectingruralresidents,suchastheconstructionofbasic
facilitiesinruralareas.Thesameresultswereobservedforoutpatientcareutilization.
WhendroppingGuizhou,thepoorestprovince,exactlythesameresultsandtrends
wereobservedasinthebasemodels.
72
Table 5.11 DID Analysis Results for Formal Care and Outpatient Utilization (Dropping the Richest Province)
Formal care Independent Variables Odds Ratio disparities in period 1 Outpatient Robust Std. Err. Odds Ratio Robust Std. Err. Group UU 1 Group RR 0.624*** 0.053 0.614*** 0.057 Group RU 0.655*** 0.071 0.556*** 0.068 Group UR 0.706** 0.086 0.764* 0.097 periods period1 1 period2 1.481*** 0.150 1.510*** 0.166 period3 2.533*** 0.206 2.491*** 0.220 period4 2.186*** 0.179 2.078*** 0.184 changes in disparities Group RR in period 2 0.735* 0.094 0.778 0.107 Group RR in period 3 1.060 0.106 1.166 0.126 Group RR in period 4 1.136 0.113 1.242* 0.133 Group RU in period 2 0.777 0.132 0.934 0.175 Group RU in period 3 1.329* 0.169 1.606*** 0.227 Group RU in period 4 1.191 0.147 1.467** 0.201 Group UR in period 2 0.737 0.145 0.676 0.141 Group UR in period 3 1.017 0.150 0.963 0.149 Group UR in period 4 1.160 0.169 1.130 0.173 Note:
n/a 1 n/a n/a 1 n/a 1.Significancelevel:***0.001,**0.01,*0.05.
2.Resultsforotherindependentvariablesareomitted.
73
Table 5.12 Test Results for Formal Care and Outpatient Utilization (Dropping the Richest Province)
Formal care Outpatient chi2 Prob>chi chi2 Prob>chi Change in disparity in period 2 = Change in disparity in period 3 10.35** 0.0013 10.97*** 0.0009 Change in disparity in period 3 = Change in disparity in period 4 0.81 0.3676 Group RR 0.60 0.4392 Group RU Change in disparity in period 2 = Change in disparity in period 3 12.97*** 11.39*** 0.0007 Change in disparity in period 3 = Change in disparity in period 4 1.40 Group UR Change in disparity in period 2 = Change in disparity in period 3 3.31 0.0690 3.39 0.0657 Change in disparity in period 3 = Change in disparity in period 4 1.34 0.2475 1.69 0.1938 Note:1.Significancelevel:***0.001,**0.01,*0.05.
0.0003 74
0.2363 0.83 0.3614 Table 5.13 DID Analysis Results for Formal Care and Outpatient Utilization (Dropping the Poorest Province)
Formal care Independent Variables Odds Ratio Robust Std. Err. disparities in period 1 Odds Ratio n/a 1 Robust Std. Err. Group UU 1 Group RR 0.649*** 0.054 0.734*** 0.069 Group RU 0.727** 0.076 0.718** 0.085 Group UR 0.633*** 0.081 0.723* 0.100 n/a 1 periods n/a period1 1 period2 1.417*** 0.137 1.584*** 0.171 period3 2.586*** 0.202 2.908*** 0.255 period4 2.273*** 0.179 2.486*** 0.219 changes in disparities n/a Group RR in period 2 0.796 0.098 0.770 0.104 Group RR in period 3 1.043 0.102 0.991 0.106 Group RR in period 4 1.112 0.108 1.050 0.112 Group RU in period 2 0.810 0.130 0.900 0.160 Group RU in period 3 1.273* 0.155 1.334* 0.181 Group RU in period 4 1.090 0.130 1.143 0.153 Group UR in period 2 0.708 0.150 0.596* 0.139 Group UR in period 3 1.130 0.174 0.987 0.162 Group UR in period 4 1.316 0.200 1.191 0.195 Note:
1.Significancelevel:***0.001,**0.01,*0.05.
2.Resultsforotherindependentvariablesareomitted.
Outpatient 75
Table 5.14 Test Results for Formal Care and Outpatient Utilization (Dropping the Poorest Province)
Formal care Prob>chi Outpatient chi2 chi2 Prob>chi Group RR Change in disparity in period 2 = Change in disparity in period 3 6.04* 0.0140 4.55* 0.0330 Change in disparity in period 3 = Change in disparity in period 4 0.76 0.3830 0.54 0.4616 Group RU Change in disparity in period 2 = Change in disparity in period 3 10.32** 0.0013 6.85** 0.0088 Change in disparity in period 3 = Change in disparity in period 4 3.05 0.0809 2.67 0.1021 Group UR Change in disparity in period 2 = Change in disparity in period 3 5.91* 0.0151 5.57* 0.0183 Change in disparity in period 3 = Change in disparity in period 4 1.67 0.1959 2.19 0.1387 Note:
1.Significancelevel:***0.001,**0.01,*0.05.
TheanalysisresultsforinpatientcareutilizationareshowninTable5.15,andthe
correspondingtestresultsareshowninTables5.16andTable5.17.
76
Table 5.15 Multivariate Analysis Results for Inpatient Utilization (Dropping the Richest/Poorest Province)
Dropping the Richest Province Dropping the Poorest Province Robust Robust Independent Variables Coef. Std. Err. Coef. Std. Err. disparity with Group UU in 1993 Group RR 0.452 0.339 ‐0.384 0.276 Group RU 0.998** 0.364 0.288 0.303 Group UR ‐0.634 0.654 ‐0.711 0.503 trend in 1990s and change in later waves Group RR trend in 1990s ‐0.230*** 0.070 ‐0.168* 0.071 deviation from trend in 2000 0.736 0.496 0.397 0.492 deviation from trend in 2004 2.016** 0.739 1.615* 0.727 0.869 1.979* 0.860 deviation from trend in 2006 2.643** deviation from trend in 2009 3.673*** 1.070 2.833** 1.060 deviation from trend in 2011 4.528*** 1.204 3.566** 1.197 Group RU trend in 1990s ‐0.131 0.079 ‐0.136 0.082 deviation from trend in 2000 0.185 0.542 ‐0.037 0.575 deviation from trend in 2004 0.757 0.801 0.899 0.823 deviation from trend in 2006 1.503 0.957 1.634 0.974 deviation from trend in 2009 1.974 1.165 2.129 1.201 deviation from trend in 2011 2.261 1.326 2.374 1.367 Group UR trend in 1990s 0.206 0.170 0.064 0.147 deviation from trend in 2000 ‐1.500 0.957 ‐1.325 1.030 deviation from trend in 2004 1.356 ‐1.120 1.449 ‐0.027 deviation from trend in 2006 ‐1.475 1.773 ‐0.137 1.632 deviation from trend in 2009 ‐1.740 2.261 ‐0.129 2.052 deviation from trend in 2011 ‐2.146 2.596 ‐0.079 2.336 Group UU trend in 1990s 0.224* 0.088 0.111 0.069 deviation from trend in 2000 ‐1.055* 0.461 ‐0.830* 0.394 deviation from trend in 2004 ‐1.506* 0.758 ‐0.852 0.621 deviation from trend in 2006 ‐1.994* 0.928 ‐1.258 0.757 deviation from trend in 2009 ‐2.608* 1.183 ‐1.606 0.954 deviation from trend in 2011 ‐2.654* 1.350 ‐1.292 1.083 constant ‐5.360*** 0.333 ‐4.552*** 0.281 Note: 1.Significancelevel:***0.001,**0.01,*0.05.
2.Resultsforotherindependentvariablesareomitted.
Afterdroppingtherichestprovince,theresultsweresimilartothebasemodel.The
onlydifferencewasthatthetrendinthe1990sbecamesignificantforGroupUU.Thetrend
77
waspositive,andthedeviationfromtrendinlateryearswasnegativeandalsosignificant.
Theresultsindicatethatinpoorerprovinces,effortsin1990saffectedinpatientutilization.
However,theimpactwasnotmaintainedinlateryears.Afterdroppingthepoorest
province,theresultswerethesameasthoseseeninthebasemodel.AsshowninTables
5.16and5.17,thelevelsofdisparitiesandchangesindisparitieswerethesameasthosein
thebasemodelafterdroppingtherichest/poorestprovinces.
Table 5.16 Test Results of Disparities in Inpatient Utilization (Dropping the Richest/poorest Province)
Dropping the Richest Province
Dropping the Poorest Province
Disparity Chi2
Disparity
Chi2
Group RR disparity (Group UU probability‐Group RR probability)
1997 0.0085 0.82
0.0125
0.38
2000 0.0047 8.62**
0.0066
12.41***
2004 0.0077 14.34***
0.0093
14.97***
2006 0.0065 10.02**
0.0066
8.86**
2009 0.0051 5.23*
0.0040
2.99
2011 0.0076 9.20**
0.0086
9.54**
Group RU disparity (Group UU probability‐Group RU probability)
1997 0.0039 2.61
0.0081
0.04
2000 0.0017 0.49
0.0047
2.92
2004 0.0057 3.86*
0.0073
4.66*
2006 0.0014 0.19
0.0004
0.01
2009 0.0012 0.15
‐0.0009
0.07
2011 0.0069 4.80*
0.0080
5.07*
Group UR disparity (Group UU probability‐Group UR probability)
0.0095
2.45
0.0057 1.83
1997 2000 0.0054 3.72
0.0077
4.30*
2004 0.0043 1.32
0.0049
1.02
2006 0.0034 0.85
0.0022
0.26
2009 0.0006 0.02
‐0.0002
0.00
2011 0.0065 2.51
0.0059
1.42
Note: 1.Significancelevel:***0.001,**0.01,*0.05.
78
Table 5.17 Test Results of Change in Disparities for Inpatient Care Utilization (Dropping the Richest/poorest
Province)
Dropping the Richest Province Change in disparity Chi2 Group RR compare with disparities with Group UU in 1997 2000 ‐0.0037 0.93 2004 ‐0.0007 0.70 2006 ‐0.0020 1.86 2009 ‐0.0034 4.61* 2011 ‐0.0009 5.40* compare with disparities with Group UU in 2000 2004 0.0030 0.04 2006 0.0017 0.09 2009 0.0003 1.07 2011 0.0029 1.21 compare with disparities with Group UU in 2006 2009 ‐0.0014 0.69 2011 0.0011 0.85 Group RU compare with disparities with Group UU in 1997 2000 ‐0.0022 0.13 2004 0.0018 0.25 2006 ‐0.0025 0.49 2009 ‐0.0027 0.60 2011 0.0030 0.02 compare with disparities with Group UU in 2000 2004 0.0040 0.68 2006 ‐0.0004 0.08 2009 ‐0.0005 0.10 2011 0.0052 0.31 compare with disparities with Group UU in 2006 2009 ‐0.0001 0.00 2011 0.0056 1.03 Group UR compare with disparities with Group UU in 1997 2000 ‐0.0003 0.45 2004 ‐0.0014 0.22 2006 ‐0.0024 0.43 2009 ‐0.0052 1.65 2011 0.0008 0.28 compare with disparities with Group UU in 2000 2004 ‐0.0011 1.10 2006 ‐0.0021 1.42 2009 ‐0.0049 2.76 2011 0.0010 1.25 compare with disparities with Group UU in 2006 0.40 2009 ‐0.0028 2011 0.0031 0.05 Note: 1.Significancelevel:***0.001,**0.01,*0.05.
79
Dropping the Poorest Province Change in disparity Chi2 ‐0.0059 ‐0.0032 ‐0.0058 ‐0.0084 ‐0.0039 0.85 1.81 3.52 8.71** 8.03** 0.0027 0.0001 ‐0.0025 0.0020 0.10 0.70 3.33 2.53 ‐0.0026 0.0019 1.18 0.59 ‐0.0034 ‐0.0008 ‐0.0077 ‐0.0090 ‐0.0001 0.01 0.00 2.67 3.69 0.30 0.0026 ‐0.0043 ‐0.0056 0.0033 0.00 1.72 2.41 0.13 ‐0.0013 0.0076 0.07 1.77 ‐0.0018 ‐0.0046 ‐0.0073 ‐0.0097 ‐0.0036 0.56 0.75 1.53 2.87 1.24 ‐0.0029 ‐0.0055 ‐0.0080 ‐0.0019 1.84 2.59 3.65 2.30 ‐0.0024 0.0037 0.19 0.09 5.4.3IncludingInteractionTermswithHouseholdIncome Thethirdsetofsensitivityanalysisinvolvedincludinganinteractiontermwith
householdincometoexaminedifferenteffectswithindifferentincomegroups.Thethree
differentincomecategorieswerebasedontheadjustedhouseholdper‐capitaincome.The
resultsforformalcareandoutpatientutilizationareshowninTable5.18.Afterthe
regression,IalsoperformedasetofWaldteststocheckthechangesindisparitiesin
adjacentperiods,andtheresultsforformalcareandoutpatientutilizationareshownin
Table5.19.
AsshowninthefirstcolumnofTable5.18,allgroupsexperienceddisparities
comparedwithGroupUUinthefirstperiod,exceptthatthedisparitywasreversedfor
mediumincomeinGroupUR.Thereverseddisparitywasnotsignificant.Thechangesin
disparitiesgenerallyfollowedthesametrendsasinthebasemodels,althoughtherewere
severalexceptions.Disparitiesincreasedforallgroupsinperiod2,exceptforlow‐income
familiesinGroupUR.Inperiod3,thedisparitiesdroppedforallgroups.Inthefourth
period,someofthegroupsexperiencedanincreaseindisparities,andsomeexperienceda
decrease,butthedisparitiesinthisperiodweresmallercomparedwithperiod1forall
groups.Fromthetestresults,Iobservedsignificantchangesfromperiods2to3only
withinthehigh‐incomefamiliesinGroupsRRandUR.ForGroupRU,thechangeswere
significantforthehigh‐andlow‐incomefamilies.
Foroutpatientcareutilization,Iobservedsimilarresultsasforformalcare
utilization.Forinpatientcare,similartrendsasthoseseeninthebasemodelswere
observed,butnoneofthetestresultswassignificant.
80
Insum,byincludinginteractiontermwithhouseholdincome,Ifoundsignificant
evidencetosupporttheconclusionthattherural–urbandisparityshrankafterthe2003
policychange.However,thisreductionindisparityonlybenefitedhigh‐incomefamiliesin
termsofformalcareutilizationandoutpatientcareutilization.OnlyinGroupRUdidlow‐
incomefamiliesalsoreceivethebenefit.
Thissensitivityanalysiswasnotconductedforinpatientcarebecausetherewas
onlysmallnumberofresidentsusinginpatientcareduringafour‐weektimeperiod,and
therewerenotsufficientobservationsineachsubgroup.
81
Table 5.18 DID Analysis Results for Formal Care and Outpatient Utilizations (Including Interaction Term
with Household Income)
Formal care Robust Independent Variables Odds Ratio Std. Err. disparities in period 1 Group UU medium income 1 n/a Group RR low income 0.569*** 0.059 Group RU low income 0.812 0.114 Group UR low income 0.468*** 0.100 Group UU low income 1.102 0.078 Group RR medium income 0.674*** 0.072 Group RU medium income 0.695* 0.103 Group UR medium income 1.088 0.174 Group RR high income 0.645*** 0.074 Group RU high income 0.566*** 0.096 Group UR high income 0.700* 0.115 Group UU high income 1.145* 0.064 periods period1 1 n/a period2 1.340*** 0.123 period3 2.313*** 0.170 period4 2.004*** 0.149 changes in disparities Group RR low income in period 2 0.988 0.144 Group RR low income in period 3 1.224 0.142 Group RR low income in period 4 1.456*** 0.169 Group RU low income in period 2 0.904 0.197 Group RU low income in period 3 1.527** 0.246 Group RU low income in period 4 1.251 0.209 Group UR low income in period 2 1.203 0.374 Group UR low income in period 3 1.687* 0.417 Group UR low income in period 4 2.099** 0.501 Group RR medium income in period 2 0.773 0.123 Group RR medium income in period 3 1.020 0.123 Group RR medium income in period 4 1.177 0.140 Group RU medium income in period 2 0.903 0.212 Group RU medium income in period 3 1.311 0.225 Group RU medium income in period 4 1.318 0.218 Group UR medium income in period 2 0.644 0.172 Group UR medium income in period 3 0.878 0.178 Group UR medium income in period 4 0.784 0.153 Group RR high income in period 2 0.693* 0.123 Group RR high income in period 3 1.348* 0.177 Group RR high income in period 4 1.208 0.158 Group RU high income in period 2 0.928 0.227 Group RU high income in period 3 1.534* 0.289 Group RU high income in period 4 1.473* 0.270 Group UR high income in period 2 0.561 0.188 Group UR high income in period 3 1.056 0.216 Group UR high income in period 4 1.279 0.262 Note: 1.Significancelevel:***0.001,**0.01,*0.05.
2.Resultsforotherindependentvariablesareomitted.
82
Outpatient Robust Odds Ratio Std. Err. 1 n/a 0.627*** 0.073 0.822 0.131 0.578* 0.131 1.168* 0.090 0.763* 0.089 0.637** 0.112 1.335 0.224 0.731* 0.092 0.571** 0.108 0.777 0.138 1.195** 0.071 n/a 1 1.476*** 0.149 2.525*** 0.206 2.120*** 0.174 0.960 0.154 1.243 0.158 1.440** 0.184 0.246 1.030 1.594** 0.285 1.372 0.255 0.891 0.310 1.475 0.382 1.886* 0.477 0.773 0.131 0.987 0.129 1.144 0.147 1.099 0.283 1.510* 0.300 1.521* 0.292 0.610 0.169 0.675 0.146 0.681 0.141 0.685* 0.130 1.318 0.186 1.177 0.167 0.919 0.250 1.618* 0.337 1.502* 0.306 0.514 0.186 1.086 0.234 1.161 0.257 Table 5.19 Test Results for Formal Care and Outpatient Utilizations (Including Interaction Term with
Household Income)
Formal care chi2 Prob>chi Group RR high income group change in disparity in period 2 = Change in disparity in period 3 16.99*** 0.0000 change in disparity in period 3 = Change in disparity in period 4 1.13 0.2876 medium income group change in disparity in period 2 = Change in disparity in period 3 3.65 0.0560 change in disparity in period 3 = Change in disparity in period 4 2.39 0.1224 low income group change in disparity in period 2 = Change in disparity in period 3 2.67 0.1022 change in disparity in period 3 = Change in disparity in period 4 3.84* 0.0500 Group RU high income group change in disparity in period 2 = Change in disparity in period 3 5.96* 0.0146 change in disparity in period 3 = Change in disparity in period 4 0.12 0.7345 medium income group change in disparity in period 2 = Change in disparity in period 3 3.16 0.0755 change in disparity in period 3 = Change in disparity in period 4 0.00 0.9662 low income group change in disparity in period 2 = Change in disparity in period 3 7.06* 0.0079 change in disparity in period 3 = Change in disparity in period 4 2.04 0.1532 Group UR high income group change in disparity in period 2 = Change in disparity in period 3 3.91* 0.0479 change in disparity in period 3 = Change in disparity in period 4 1.17 0.2786 medium income group change in disparity in period 2 = Change in disparity in period 3 1.52 0.2174 change in disparity in period 3 = Change in disparity in period 4 0.45 0.5043 low income group change in disparity in period 2 = Change in 1.60 0.2062 disparity in period 3 change in disparity in period 3 = Change in disparity in period 4 1.66 0.1980 Note:1.Significancelevel:***0.001,**0.01,*0.05.
83
chi2 Outpatient Prob>chi 14.78*** 0.0001 1.07 0.3012 2.57 0.1091 2.30 0.1290 3.36 0.0668 2.42 0.1197 6.24* 0.0125 0.33 0.5630 2.08 0.1488 0.00 0.9589 4.43* 0.0353 1.03 0.3103 4.61* 0.0317 0.13 0.7235 0.14 0.7036 0.00 0.9645 2.88 0.0897 1.84 0.1754 5.4.4DIDAnalysisforInpatientCare
ThelastsetofsensitivityanalysisinvolvedDIDanalysisforinpatientcareutilization.
TheresultsareshowninTable5.20,andthecorrespondingtestresultsareshowninTable
5.21.
AsshowninTable5.20,thereweredisparitiesforGroupsRR,RU,andURwith
GroupUU.ForGroupsRRandUR,thedisparitiesweresignificant,andbothofthetwo
groupsonlyusedlessthanhalfofinpatientcarecomparedwiththeusageofGroupUUin
period1.Thedisparitydidnotchangesignificantlyinanyofthefollowingperiodsforany
ofthegroups.
Table 5.20 DID Analysis Results for Inpatient Care Utilization
Independent Variable disparities in period 1 Group UU Group RR Group RU Group UR periods period 1 period 2 period 3 period 4 change in disparities Group RR in period 2 Group RR in period 3 Group RR in period 4 Group RU in period 2 Group RU in period 3 Group RU in period 4 Group UR in period 2 Group UR in period 3 Group UR in period 4 Note:
Odds Ratio
1
0.452***
0.851
0.438**
Robust Std. Err.
n/a
0.082
0.173
0.133
1
0.764
1.040
1.290
n/a
0.180
0.191
0.224
0.757
0.924
1.427
0.784
0.869
0.980
0.592
1.689
1.883
0.257
0.234
0.324
0.305
0.242
0.254
0.403
0.684
0.683
1.Significancelevel:***0.001,**0.01,*0.05.
2.Resultsforotherindependentvariablesareomitted.
FromTable5.21,thechangeindisparityinperiod4wassignificantlydifferentfrom
thechangeindisparityinperiod3.Thisindicatesthatthedisparitywasreducedbetween
84
periods3and4forGroupRR.However,thedisparitywasnotsignificantlydifferentfrom
theoriginaldisparityinperiod1.Thereisnoevidencetoshowthatmorehealthinsurance
coveragereduceddisparityininpatientcareutilization.
Table 5.21 Test Results for Inpatient Care Utilization (DID Analysis)
Group RR Change in disparity in period 2 = Change in disparity in period 3
Change in disparity in period 3 = Change in disparity in period 4
Group RU Change in disparity in period 2 = Change in disparity in period 3
Change in disparity in period 3 = Change in disparity in period 4
Group UR Change in disparity in period 2 = Change in disparity in period 3
Change in disparity in period 3 = Change in disparity in period 4
chi2 Prob>chi
0.35 4.33* 0.07 0.23 2.47 0.12 0.5566
0.0375
0.7869
0.6321
0.1162
0.7296
Note:1.Significancelevel:***0.001,**0.01,*0.05.
5.5 Summary of Findings 1. Rural–urbandisparityinformalcareutilizationandoutpatientvisitwas
associatedwithpolicychangeinhealthinsurancecoverage,aswellasother
relatedmeasures.Whenthegovernmentprovidedmorehealthinsurance
coverageforresidentswithruralregistration,thedisparitiesinformalcareand
outpatientutilizationdecreasedforGroupsURandRR.
2. OnlyforGroupRR,thenegativetrendofusinginpatientcarewasalleviated
duringlateryears.However,noevidenceshowsthatdisparityininpatientcare
utilizationwasalsocorrelatedtohealthinsurancecoverage.
3. The2003policychangeinruralareasamongresidentswithruralhousehold
registrationreducedrural–urbandisparities.Byprovidingmorehealth
insurancecoveragetoresidentswithruralhouseholdregistration,thepolicy
changereducedthedisparitybetweenGroupsRRandUR,motivatingresidents
withruralhouseholdregistrationtousemoreformalhealthcareandoutpatient
85
visits,comparedtoGroupUU.Throughothermeasuresenablingresourcesin
ruralareas,thepolicychangealsoreduceddisparitiesbetweenGroupsRUand
GroupUU.AlthoughGroupRUhadurbanhouseholdregistration,thesemembers
residedinruralareasandbenefitedfromtheimprovedenvironment.
4. The2003policychangeinruralareasnotonlyreducedthedisparityfromthe
levelofthe1990s,butalsofromtheoriginallevel.Thischangehappenedfor
ruralresidentswitheitherruralorurbanhouseholdregistration.
5. Aftercontrollingforinsurancestatus,thepositiveeffectscouldstillbeobserved
inallgroups.Thisindicatesthatthepositiveeffectscamenotonlyfrommore
healthinsurancecoveragebutalsofromotherrelatedmeasures.Comparedwith
thebasemodel,thechangeindisparitiesshrankmostforruralresidentswith
ruralhouseholdregistration.Thisindicatesthattheruralresidentswithrural
householdregistrationbenefitedmostfromtheexpandedhealthinsurance
coverage.
6. The2003policychangeaffectedbothpoorprovincesandrichprovinces.
However,theexpandedhealthinsurancecoveragewasmoreeffectiveinricher
provincesinreducingdisparitiesinhealthcareutilization.Thepolicyeffecton
poorerprovinceswasassociatedmorecloselywiththeothermeasureson
changingtheenvironmentinruralareas,suchasconstructionofbasicmedical
facilities.
7. Thepositiveimpactonformalcareandoutpatientutilizationofpolicychangein
2003occurredmainlyinhigh‐incomefamilies.Inthemedium‐incomegroup,I
86
observednosignificantimpact.Inthelow‐incomegroup,thepositiveimpact
wasobservedonlyinruralresidentswithurbanhouseholdregistration.
87
Chapter 6 Results: Disparities in healthcare costs 6.1 Descriptive Analysis Figure6.1showsthetrendsofproportionofrespondentswhoseout‐of‐pocket(OOP)
healthcarecostwasmorethan20%ofthehouseholdgrossincomebyruralandurban
residencesandregistrations.Fromthefigure,itcanbeseenthatthepercentageofOOP
exceeding20%householdincomehadalwaysbeenbelow5%.Bothofthetwogroupsof
ruralresidentshadalwaysbeenlesslikelytohaveOOPexceeding20%ofhousehold
incomecomparedwithGroupUU.Itseemsthatruralresidentsexperiencedlessfinancial
riskthantheirurbancounterparts.However,giventhefactthatruralresidentsusedless
formalcare,thelowpossibilityofhavinghighOOPmaybeduetoalackofformalcareor
foregonecare.Initially,theratiobetweenGroupUUandallothergroupswaslessthan1,
indicatingthatalowerproportionofthethreegroupshadOOPexceeding20%of
householdincome,comparedwithGroupUU.TheratioforGroupRRdroppedslightlyin
period2,whenmorehealthinsurancecoveragewasprovidedtourbanworkersin1998.In
periods3and4,theratioincreased,andfinallygrewtomorethan1.TheratioforGroup
RUstayednearlyconsistentinperiod2,andthenincreasedinperiod3.Inthisperiod,
healthinsurancedidnotchangeforeitherGroupRUorUU.However,morehealthcare
resourceswereallocatedtoruralareas.Inperiod4,theratiodroppedslightly.Inthis
period,morehealthinsuranceandhealthcareresourceswereallocatedtourbanresidents.
ForGroupUR,theratiodroppedinperiod2,whenmorehealthinsurancecoveragewas
providedtoGroupUU.Subsequently,theratioincreasedinperiods3and4,whenmore
healthinsuranceormorehealthcareresourceswereallocatedtoruralareas.The
88
observationwascontrarytomyhypothesisthatmorehealthinsurancecoveragereduces
Proportion of redients whose OOP medical expense exceeds 20% of the household income
0.20
1.40
0.18
1.20
0.16
0.14
1.00
0.12
0.80
0.10
0.08
0.60
0.06
0.40
0.04
0.20
0.02
0.00
0.00
1
2
3
Ratio of other study groups to Group UU
financialrisk.
Group RR
Group RU
Group UR
Group UU
Ratio: Group
RR/Group UU
Ratio: Group
RU/Group UU
Ratio: Group
UR/Group UU
4
Period
Figure 6.1 Probability of Having Out-of-pocket Medical Expense Exceeding 20% of Household Income by
Rural and Urban Residences and Registrations
SimilarresultscanbeobservedinFigure6.2,whichshowsthetrendsofthe
proportionofrespondentswhoseout‐of‐pockethealthcarecostwasmorethan40%ofthe
householdgrossincomebyruralandurbanresidents.Again,thetwogroupsofrural
residentshadalwayshadalowerpossibilityofhavingveryhighOOP(morethan40%of
householdincome)untilthelastperiod.Thetrendsofratiochangeareconsistentwiththe
resultsshowninFigure6.1.Again,thisresultwascontrarytomyhypothesisthatmore
healthinsurancecoveragereducesfinancialrisk.
89
0.18
1.20
0.16
1.00
0.14
0.12
0.80
0.10
0.60
0.08
0.06
0.40
0.04
0.20
0.02
0.00
Ratio of other study groups to Group UU
Proportion of redients whose OOP medical expense exceeds 40% of the household income
1.40
0.20
Group RR
Group RU
Group UR
Group UU
Ratio: Group
RR/Group UU
Ratio: Group
RU/Group UU
Ratio: Group
UR/Group UU
0.00
1
2
3
4
Period
Figure 6.2 Probability of Having Out-of-pocket Medical Expense Exceeding 40% Household Income by Rural
or Urban Residences and Registrations
Figure 6.3 shows the trends of average healthcare cost. All three groups had always spent less on healthcare than Group UU. For Groups RR and RU, the ratio to Group UU decreased in period 2, and then increased in periods 3 and 4. For group UR, the ratio to Group UU decreased in period 2, increased in period 3, and then decreased again in period 4. This indicates that rural residents started to have more medical expenses after the rural policy change in 2003. Urban residents with rural registration also began to pay more compared to Group UU after the health insurance expansion in 2003. However, their total healthcare cost shrank compared to Group UU when health insurance covered more urban residents in 2007. 90
350.00
1.40
300.00
1.20
250.00
1.00
200.00
0.80
150.00
0.60
100.00
0.40
50.00
0.20
0.00
0.00
1
2
3
Ratio of other study groups to Group UU
Average medical expense
4
Group RR
Group RU
Group UR
Group UU
Ratio: Group
RR/Group UU
Ratio: Group
RU/Group UU
Ratio: Group
UR/Group UU
period
Figure 6.3 Total Healthcare Costs by Rural and Urban Residences and Registrations
6.2 Multivariate Analysis Controlling for Existing Trends The results ofananalysisassumingpersistenttrendsforrural–urbandisparitiesin
healthcarecostsarepresentedinTables6.1and6.2.Inthesemodels,IusedGroupUUas
thereferencegroup,calculatingtheinitialdisparitiesbetweenGroupUUandothergroups.
Themodelalsocontrolsforthepre‐existingtrendsinthe1990s,andanalyzeschangesin
theyearsafter.Usingthemodel,Iwasabletocalculatetheoddsratiosoftrendsandactual
valuesineachperiodforeachgroup.Afterproducingtheresults,IperformedWaldteststo
determineifthedisparitiesandchangesindisparitiesweresignificant.
Table6.1showsresultsforthemultivariateanalysisforOOPexceeding20%/40%
ofhouseholdincome.Column1showsresultsfortheindicatorofOOPexceeding20%
householdincome.Itseemsthatthedisparitiesarereversed,sincealltheruralresidents
(GroupsRRandRU)werelesslikelytohavehighOOPexceeding20%oftheirhousehold
incomeinitiallyin1993.GroupURwasmorelikelytohavehighOOPcomparedwithGroup
91
UU.However,theresultswerenotsignificantforanyofthegroups.GroupRRshowed
negativetrendsinthe1990s;thus,peopleinthisgroupshouldbehavingadecreasing
chanceofhavinghighOOPifthetrendpersists.Incontrast,trendsfortheotherthree
groupswerepositive,meaninganincreasinglikelihoodofhavinghighOOPifthetrend
persisted.Again,thetrendinthe1990swasnotsignificantforGroupsRR,RU,andUR.The
trendwassignificantforGroupUU.InGroupsRRandRU,Iobservedsignificantpositive
deviationfromthetrendsinyear2004,whichwasrightaftertheNRCMwasinitiated.The
positivedeviationcontinuedtobesignificantforGroupRRinthefollowingyears.InGroup
UU,Iobservedanegativedeviationfromthetrendin2009,whichwasrightafterthe
initiationofURBMI.Column2inTable6.3showsresultsfortheindicatorofOOPexceeding
40%householdincome.SimilarresultsareshowninColumn1.Allthreegroupsweremore
likelytohaveOOPexceeding40%oftheirhouseholdincomecomparedwithGroupUU.
Thedisparitieswerenotsignificantforanyofthegroups.Iobservedsignificantpositive
trendsinthe1990swithinGroupUU,andtrendsforothergroupswerenotsignificant.
SignificantpositivedeviationswereobservedforGroupsRRandRUin2004,reflectingthe
initiationofNRCMtheyearbefore.Asignificantnegativetrendwasobservedin2009and
2011withinGroupUU,occurringimmediatelyafterURBMIwasinitiatedandcontinuingin
thelaterwave.
92
Table 6.1 Multivariate Analysis Results for OOP Exceeding Certain Percentage of Household Income
OOP>20% Household Income
OOP>40% Household Income
Robust Robust Coef. Std. Err. Coef. Std. Err. Disparity with Group UU in 1993 Group RR ‐0.054
0.194
0.081 0.244
Group RU ‐0.016
0.234
0.228 0.282
Group UR 0.253
0.261
0.341 0.324
trend in 1990s and change in later waves
Group RR trend in 1990s ‐0.001
0.032
0.013 0.038
deviation from trend in 2000 0.304
0.190
0.256 0.219
deviation from trend in 2004 0.959**
0.304
0.795* 0.356
deviation from trend in 2006 0.926*
0.369
0.614 0.429
1.061*
0.460
0.747 0.540
deviation from trend in 2009 deviation from trend in 2011 0.814
0.527
0.568 0.616
Group RU trend in 1990s 0.008
0.054
‐0.010 0.061
deviation from trend in 2000 0.414
0.319
0.631 0.361
deviation from trend in 2004 1.107*
0.505
1.220* 0.573
deviation from trend in 2006 1.178
0.607
1.282 0.686
0.870
1.185 deviation from trend in 2009 0.997
0.769
deviation from trend in 2011 0.747
0.874
1.045 0.986
Group UR trend in 1990s 0.027
0.064
0.025 0.078
deviation from trend in 2000 ‐0.171
0.369
0.065 0.437
deviation from trend in 2004 0.805
0.580
0.875 0.708
deviation from trend in 2006 0.564
0.707
0.463 0.864
deviation from trend in 2009 0.672
0.894
0.651 1.092
0.453 1.247
0.316
1.015
deviation from trend in 2011 Group UU trend in 1990s 0.171***
0.051
0.199** 0.064
deviation from trend in 2000 ‐0.390
0.253
‐0.345 0.306
deviation from trend in 2004 ‐0.356
0.437
‐0.494 0.541
deviation from trend in 2006 ‐0.945
0.538
‐1.181 0.668
deviation from trend in 2009 ‐1.572*
0.692
‐1.865* 0.859
deviation from trend in 2011 ‐1.907*
0.785
‐2.185* 0.978
0.184
‐4.236*** 0.232
constant ‐3.709***
Note:
1.Significancelevel:***0.001,**0.01,*0.05.
2.Resultsforotherindependentvariablesareomitted.
Table6.2showsresultsfortotalhealthcarecosts.Column1showsresultsfromthe
firstpartexaminingwhethertherespondenthadanyhealthcarecost,andColumn2shows
resultsfromtheGLMmodelexaminingthetotalhealthcarecostforusers.Fromtheresults,
in1993,itcanbeseenthatallthreegroupswerelesslikelytohavehadanyhealthcare
costs,comparedtoGroupUU.ThiswasconsistentwithwhatIfoundinChapter5:thethree
93
groupsuselessmedicalcarethanGroupUU.However,theresultswerenotsignificant.In
laterwaves,thetrendforGroupRRwaspositive,andthedeviationfromtrendwasstill
positivestartingfrom2004.ThissuggeststhatGroupRRwasmorelikelytohavehad
healthcarecostsafterthesecondpolicychangein2003.TrendanddeviationsforGroupRU
followthesamepattern,butthedeviationsfromtrendwerenotsignificant.ForGroupUR,
thetrendwaspositive,anddeviationsfromtrendwerenegative.However,onlythe
deviationin2000wassignificant.GroupUUfollowedthesamepatternasGroupUR,but
thedeviationsweresignificantforthisgroup.Lookingatthetotalhealthcarecost,usersin
GroupsRR,RU,andURpaidmorehealthcarecostthanusersinGroupUUin1993.Groups
RR,RU,andURfollowednegativetrendsinthe1990s,andthedeviationsfromtrendin
lateryearswerepositive.ForGroupRU,thedeviationswereallsignificant.GroupUUhada
positivetrend,andthedeviationswerenegativebutnotsignificant.ForGroupRR,I
observedasignificantlyincreasedprobabilityofhavinghealthcarecostimmediateafterthe
2003policychange,andtheeffectcontinuedinthefollowingwaves.
94
Table 6.2 Multivariate Analysis Results for Total Healthcare Costs
Having Any Healthcare Cost Robust Std. Coef. Disparity with Group UU in 1993 Group RR Group RU Group UR trend in 1990s and change in later waves Group RR trend in 1990s deviation from trend in 2000 deviation from trend in 2004 deviation from trend in 2006 deviation from trend in 2009 deviation from trend in 2011 Group RU trend in 1990s deviation from trend in 2000 deviation from trend in 2004 deviation from trend in 2006 deviation from trend in 2009 deviation from trend in 2011 Group UR trend in 1990s deviation from trend in 2000 deviation from trend in 2004 deviation from trend in 2006 deviation from trend in 2009 deviation from trend in 2011 Group UU trend in 1990s deviation from trend in 2000 deviation from trend in 2004 deviation from trend in 2006 deviation from trend in 2009 deviation from trend in 2011 constant Note:
‐0.121 ‐0.280 ‐0.037 0.038 ‐0.006 0.904*** 0.653** 0.798** 0.495 0.086* ‐0.066 0.639 0.414 0.162 ‐0.148 0.147*** ‐0.616** ‐0.174 ‐0.543 ‐0.867 ‐1.344 0.166*** ‐0.533** ‐0.121 ‐0.716* ‐1.317** ‐1.742*** ‐2.618*** 0.118 0.155 0.167 0.021 0.121 0.192 0.232 0.292 0.333 0.038 0.208 0.341 0.414 0.525 0.600 0.040 0.222 0.359 0.437 0.556 0.631 0.031 0.160 0.270 0.330 0.420 0.479 0.111 Total Healthcare Cost Robust Std. Coef. 0.236 0.664 0.305 ‐0.051 0.671* 0.744 0.787 1.020 1.743 ‐0.218** 1.443** 2.107** 2.114* 3.688*** 4.486*** ‐0.077 1.012 1.651 0.844 1.307 1.810 0.193* 0.512 ‐1.269 ‐1.743 ‐2.028 ‐2.354 6.342*** 0.359 0.399 0.449 0.057 0.342 0.525 0.635 0.797 0.910 0.077 0.456 0.692 0.845 1.067 1.213 0.110 0.655 1.045 1.226 1.535 1.761 0.087 0.396 0.715 0.892 1.143 1.307 0.336 1.Significancelevel:***0.001,**0.01,*0.05.
2.Resultsforotherindependentvariablesareomitted.
Usingthecoefficientsfromthemodels,Iwasabletocalculatethepredicted
probabilityofOOPexceeding20%/40%foreachgroupineachyear.Ithenusedthe
95
differencesintheprobabilitiesbetweenGroupUUandothergroupsasthemeasureof
disparity.ThepredictedprobabilitiesareshowninFigures6.4to6.6.Ialsoincluderatios
betweenothergroupswithGroupUUtoshowthetrendofdisparities.
Figure6.4showsthepredictedprobabilitiesofOOPexceeding20%ofhousehold
income.Again,Iobservedthereverseddisparity.GroupUUwasalmostalwaysmorelikely
tohaveahighchanceofOOPexceeding20%ofhouseholdincome,exceptforwaves1993
and2009,whileGroupRRalwaysenjoyedthelowestchanceofhavinghighOOP.ForGroup
RR,inthe1990s,theratiowithGroupUUdecreased.In2000,afterthegovernment
providedmorehealthinsurancecoveragetourbanworkers,theratioforGroupRRstarted
toincrease.In2004,aftertheinitiationofNRCM,theratiodecreasedagain.Thenin2006,
theratioonceagainincreased.Afterthegovernmentofferedmorehealthinsurance
coverageforurbanresidents,theratiofinallydecreasedin2011.Fromthetrend,itseems
that,comparedwithGroupUU,GroupRRbenefitedwhenhealthinsurancecoverage
expandedforpeoplewithruralregistration,butwasharmedwhenmorehealthinsurance
coveragewasprovidedforurbanresidents.However,theratioforGroupsRRandRU
followedsimilartrends,althoughthesetwogroupsdidnothavethesametypeof
householdregistration.ForGroupUR,theratiodecreasedinthe1990s,startedtoincrease
in2004,anddecreasedin2011.Thisindicatesthatprovidingmorehealthinsurance
coveragedidnotalwaysreducefinancialrisk,sincetheratioincreasedin2004afterthe
initiationofNRCM.Forallthreegroups,theratiowasalmostalwayshigherthanin1997,
suggestingincreaseddisparitiesinlateryears.Insteadofgainingfinancialprotection,the
threegroupswerelosingtheinitialadvantage.
96
Figure6.5showsthepredictedprobabilitiesofhavingOOPexceeding40%of
householdincome.Again,GroupUUalmostalwaysshowedahigherpossibilityofhaving
extremelyhighOOPexceeding40%oftheirhouseholdincome,andGroupRRalways
enjoyedthelowestpossibility.ThetrendsofratiochangeweresimilartowhatIobserved
0.12
1.4
0.1
1.2
1
0.08
0.8
0.06
0.6
0.04
0.4
0.02
0.2
0
0
Ratio of other study groups to Group UU
Predicted probability of having OOP exceeding 20% of household income
inthepreviousvariable,buttheslopeswereflatter.
Group RR
Group RU
Group UR
Group UU
Ratio: Group
RR/Group UU
Ratio: Group
RU/Group UU
Ratio: Group
UR/Group UU
1993 1997 2000 2004 2006 2009 2011
Wave
Figure 6.4 Predicted Probability of Having OOP Exceeding 20% of Household Income by Rural and Urban
Residences and Registrations
97
1.4
0.1
1.2
1
0.08
0.8
0.06
0.6
0.04
0.4
0.02
0.2
0
Ratio of study groups to Group UU
Predicted probability of having OOP exceeding 40% of household income
0.12
0
1993 1997 2000 2004 2006 2009 2011
Group RR
Group RU
Group UR
Group UU
Ratio: Group
RR/Group UU
Ratio: Group
RU/Group UU
Ratio: Group
UR/Group UU
Wave
Figure 6.5 Predicted Probability of Having OOP Exceeding 40% of Household Income by Rural and Urban
Residences and Registrations
Predicted total health cost
1.4
250
1.2
200
1
150
0.8
0.6
100
0.4
50
0.2
0
0
1993
1997
2000
2004
2006
2009
2011
Ratio of study groups to Group UU
1.6
300
Group RR
Group RU
Group UR
Group UU
Ratio: Group
RR/Group UU
Ratio: Group
RU/Group UU
Ratio: Group
UR/Group UU
Wave
Figure 6.6 Predicted Total Healthcare Costs by Rural and Urban Residences and Registrations
Figure6.6showspredictedtotalhealthcarecosts.Similartotheprevioustwo
variables,GroupUUalmostalwayshadhigherhealthcarecost.TheratioofGroupRRto
GroupUUdecreaseduntil2000.In2004,theratiostartedtoincrease,andcontinuedto
98
increaseintheyearsafter.Theratiosfortheothertwogroupsfollowedsimilartrends.
Althoughtheslopesdiffered,Iobservedaclearincreaseforallgroupsinthe2004wave,
afterthe2003policychangeinruralareas.The2003policychangeinruralareasseemed
toincreasetotalhealthcarecostforallaffectedgroups.
IthenusedthedifferencebetweenprobabilitiesforGroupUUandothergroupsas
anestimatefordisparity.Afterthedisparitieswerecalculated,IperformedaWaldtestto
determinewhetherthedisparitiesweresignificant.TheresultsforOOPexceeding20%/40%
ofhouseholdincomeareshowninTable6.3.Disparitiesinbothofthetwooutcomeswere
greaterthan0,indicatingthatrespondentsinGroupUUweremorelikelytohaveOOP
exceedingcertainpercentageofhouseholdincome.Thedisparitywasreversedinthiscase.
DisparitiesbetweenGroupsRRandUUweresignificantin2000,2004,2006,and2011.For
GroupUR,thedisparitiesweresignificantin2000,2004,and2011.ForGroupUR,the
disparitywassignificantin2000.ForOOPexceeding40%ofhouseholdincome,disparities
werealsosignificantinyears2006and2011.
99
Table 6.3 Test Results of Disparities for OOP Exceeding Certain Percentage of Household Income
OOP>20% Household Income
OOP>40% Household Income
Disparity
Chi2
Disparity Chi2
Group RR disparity (Group UU probability‐Group RR probability)
1997 0.0237
0.04
0.0257 0.05
2000 0.0217
17.66***
0.0324 16.57***
2004 0.0445
37.25***
0.0600 32.01***
2006 0.0259
13.63***
0.0445 16.22***
2009 0.0106
2.32
0.0380 4.50*
2011 0.0241
17.85***
0.0438 19.60***
Group RU disparity (Group UU probability‐Group RU probability)
1997 0.0220
0.24
0.0214 1.31
2000 0.0148
4.45*
0.0221 2.03
2004 0.0272
7.49**
0.0430 6.16*
2006 0.0006
0.00
0.0245 0.03
2009 0.0034
0.14
0.0237 0.17
2011 0.0175
5.94*
0.0316 4.88*
Group UR disparity (Group UU probability‐Group UR probability)
0.02
0.0123
0.1
0.0222 1997 2000 0.0208
6.15*
0.0295 3.90
2004 0.0142
1.26
0.0482 1.42
2006 0.0075
0.41
0.0383 1.91
2009 ‐0.0145
1.67
0.0288 0.16
2011 0.0072
0.61
0.0357 0.92
Note:
1.Significancelevel:***0.001,**0.01,*0.05.
IalsoperformedWaldteststoexaminewhetherthechangesindisparitieswere
significant,andtheresultsforOOPexceeding20%/40%ofhouseholdincomeareshownin
Table6.4.Years1997,2000,and2006werethewavesbeforeeachpolicyintervention.
Therefore,Icompareddisparitiesinthesethreeyearswithdisparitiesintheyearsafter.
Column1showstestresultsforOOPexceeding20%householdincome.ForGroupRRand
UR,thedisparityin2009wassignificantlysmallerthandisparitiesin1997and2000.For
GroupRU,thedisparitiesin2006and2009weresignificantlysmallerthanthedisparityin
1997.Column2showstestresultsforOOPexceeding40%householdincome.The
disparitywassignificantlydifferentonlybetweenyear1997and2006forGroupRU.For
GroupsRRandRU,disparitiesin2011weresignificantlyreducedfromdisparitiesin2000.
100
Fromtheresults,noimmediatereductionofdisparitieswasobservedaftereachpolicy
intervention,althoughthedisparitieswerefinallyreducedinlateryears.
Table 6.4 Test Results of Changes in Disparities for OOP Exceeding Certain Percentage of Household Income
OOP>20% Household Income OOP>40% Household Income Group RR compare with disparity with Group UU in 1997 2000 0.88 2004 0.41 2006 3.17 2009 9.75** 2011 3.22 compare with disparity with Group UU in 2000 2004 0.16 2006 0.74 2009 5.35* 2011 0.72 compare with disparity with Group UU in 2006 2009 2.84 2011 0.00 Group RU compare with disparity with Group UU in 1997 1.60 2000 2004 2.02 2006 8.19** 2009 6.74** 2011 2.78 compare with disparity with Group UU in 2000 2004 0.00 2006 2.66 2009 1.92 2011 0.09 compare with disparity with Group UU in 2006 2009 0.06 2011 2.60 Group UR compare with disparity with Group UU in 1997 2000 0.53 2004 0.39 2006 0.69 2009 4.06* 2011 0.73 compare with disparity with Group UU in 2000 2004 2.05 2006 2.54 2009 7.72** 2011 2.73 compare with disparity with Group UU in 2006 2009 2.09 2011 0.00 Note: 1.Significancelevel:***0.001,**0.01,*0.05.
0.3482 0.5198 0.0752 0.0018 0.0728 0.04 0.00 0.34 3.23 0.57 0.8417 0.9648 0.5598 0.0724 0.4493 0.6862 0.3890 0.0207 0.3950 0.08 0.17 3.10 0.38 0.7739 0.6816 0.0784 0.5391 0.0921 0.9480 0.1323 0.8523 0.2058 0.1556 0.0042 0.0094 0.0956 2.27 0.03 1.52 0.89 4.75* 3.82 1.34 0.9687 0.1032 0.1660 0.7689 0.18 0.97 0.64 0.04 0.6707 0.3248 0.4249 0.8430 0.8137 0.1068 0.8543 0.1725 0.4659 0.5326 0.4058 0.0440 0.3936 0.03 1.86 0.11 0.25 0.06 1.90 0.45 0.1524 0.1112 0.0055 0.0985 0.81 0.36 3.25 1.20 0.3669 0.5496 0.0716 0.2741 0.1484 0.9797 1.92 0.24 0.1662 0.6210 0.2183 0.3462 0.0293 0.0505 0.2475 0.7350 0.6138 0.8095 0.1676 0.5018 Table6.5showsestimateofdisparitiesintotalhealthcarecost.Theresultsare
basedon500iterationsofbootstrap.ForGroupRR,theconfidenceintervalsforthefour
101
periodswerenotoverlapped.Icanconcludethatthechangesindisparitiesbetween
adjacentperiodsaresignificant.SimilarresultswereobservedforGroupRUandUR.
Lookingatthetrendsofdisparities,forGroupRR,thedisparitiesincreasedduringthe
1990s,andstartedtodecreasebetween2000and2004,whichwasafterthepolicychange
inruralareas.Theresultsindicatethattheruralrelatedgroupspaidmoretotalhealthcare
costscomparedwithurbancounterpartsafterthepolicychange.Between2006and2009,
whichwasafterthepolicychangeinurbanareas,thedisparitybetweenGroupRRand
GroupUUincreased,indicatingthatGroupRRexperiencedmorehealthcarecosts
comparedwithGroupUU.TheresultisconsistentwithChapter5,whereIfoundthatthe
ruralgroupsusemoreformalcareandoutpatientserviceafterthepolicychange.The
increasedvisitthenledtoincreasedtotalhealthcarecosts.ForGroupRR,thedisparityin
totalcostsincreasedwhenmorehealthinsurancecoveragewasprovidedtopeoplewith
ruralhouseholdregistration,anddecreasedwhenmorehealthinsurancecoveragewas
providedtourbanresidents.GroupUR,whichhadthesamehouseholdregistrationtype
withGroupRR,borethesametrendasGroupRR.ForGroupRU,thetrendwasalsosimilar,
exceptthatthedisparitycontinuedtodecreaseafter2006.Thisgrouphadurban
householdregistration,thusnosignificantchangeindisparityintotalcostswithGroupUU
afterurbangroupsreceivemorehealthinsurancecoverage.
102
Table 6.5 Bootstrap Results for Disparities in Total Health Costs
Variable Mean Std. Err. Group RR disparity with Group UU in 1993 ‐3.417 0.463 disparity with Group UU in 1997 76.630 0.923 disparity with Group UU in 2000 240.683 3.019 disparity with Group UU in 2004 147.003 1.904 disparity with Group UU in 2006 99.688 1.836 disparity with Group UU in 2009 115.866 2.056 disparity with Group UU in 2011 50.281 1.836 Group RU disparity with Group UU in 1993 ‐13.889 0.624 disparity with Group UU in 1997 80.683 0.942 disparity with Group UU in 2000 233.155 3.097 disparity with Group UU in 2004 137.568 2.005 disparity with Group UU in 2006 119.026 2.058 disparity with Group UU in 2009 52.818 2.759 disparity with Group UU in 2011 6.739 2.557 Group UR disparity with Group UU in 1993 ‐8.478 0.678 disparity with Group UU in 1997 59.943 1.076 disparity with Group UU in 2000 210.383 3.302 disparity with Group UU in 2004 ‐10.805 4.429 disparity with Group UU in 2006 95.359 2.006 disparity with Group UU in 2009 104.133 2.270 disparity with Group UU in 2011 72.882 2.046 [95% Conf. ‐4.326 74.816 234.751 143.262 96.081 111.827 46.673 ‐15.116 78.832 227.070 133.629 114.982 47.397 1.716 ‐9.811 57.829 203.896 ‐19.507 91.417 99.674 68.862 Interval] ‐2.508 78.444 246.615 150.744 103.295 119.905 53.888 ‐12.663 82.535 239.240 141.506 123.070 58.240 11.762 ‐7.146 62.057 216.870 ‐2.102 99.301 108.592 76.902 6.3 Sensitivity Analysis 6.3.1controllingforhealthinsurancestatus
Thefirstsetofsensitivityanalysisiscontrolforhealthinsurancestatus.Fromthe
results,havinginsurancehaspositiveeffectonhavingOOPexceeding20%/40%of
householdincome.ForOOPexceeding20%householdincome,aftercontrollingfor
insurancestatus,thedisparitybetweenGroupRRandGroupUUbecamepositive,
indicatingthatGroupRRwasmorelikelytohaveOOPexceeding20%ofhouseholdincome.
ThesamehappenedforGroupRU.Thisindicatesthatinsurancestatuscanexplainsomeof
103
thedisparities.However,theresultswerestillnotsignificant.Thetrendsandchangesin
laterwavesfollowedthesamepatternasthebasemodels.
Table 6.6 Multi-variate Analysis Results for OOP Exceeding Certain Percentage of Household Income
(Controlling for Insurance)
OOP>20% household income OOP>40% household income independent variables Coef. Robust Std. Err. Coef.
Robust Std. Err. disparity with Group UU in 1993 Group RR 0.059 0.198
0.163
0.247
Group RU 0.041 0.236
0.269
0.283
Group UR 0.340 0.262
0.405
0.326
trend in 1990s and change in later waves Group RR trend in 1990s ‐0.004 0.032
0.011
0.038
deviation from trend in 2000 0.327 0.190
0.273
0.219
deviation from trend in 2004 0.984*** 0.304
0.814*
0.356
deviation from trend in 2006 0.885** 0.368
0.585
0.429
0.656
0.540
deviation from trend in 2009 0.935* 0.460
deviation from trend in 2011 0.690 0.527
0.479
0.616
Group RU trend in 1990s 0.014 0.054
‐0.006
0.061
deviation from trend in 2000 0.404 0.319
0.624
0.361
deviation from trend in 2004 1.058* 0.504
1.184*
0.573
deviation from trend in 2006 1.095 0.606
1.221
0.686
0.768
1.045
0.870
deviation from trend in 2009 0.808 deviation from trend in 2011 0.531 0.874
0.886
0.986
Group UR trend in 1990s 0.028 0.064
0.025
0.078
deviation from trend in 2000 ‐0.162 0.369
0.074
0.437
deviation from trend in 2004 0.806 0.580
0.879
0.708
deviation from trend in 2006 0.526 0.707
0.438
0.864
deviation from trend in 2009 0.511 0.894
0.538
1.093
1.016
0.338
1.248
deviation from trend in 2011 0.153 Group UU trend in 1990s 0.179*** 0.051
0.204***
0.064
deviation from trend in 2000 ‐0.402 0.253
‐0.352
0.306
deviation from trend in 2004 ‐0.395 0.437
‐0.521
0.541
deviation from trend in 2006 ‐1.014 0.539
‐1.229
0.669
deviation from trend in 2009 ‐1.720* 0.695
‐1.973*
0.861
‐2.317*
0.981
deviation from trend in 2011 ‐2.090** 0.789
whether having insurance insurance 0.210*** 0.051
0.153*
0.060
not having insurance 0 n/a
0
n/a
constant ‐3.879*** 0.191
‐4.357*** 0.237
Note:1.Significancelevel:***0.001,**0.01,*0.05.
2.Resultsforotherindependentvariablesareomitted.
104
Table 6.7 Test Results of Disparities for OOP Exceeding Certain Percentage of Household Income
(Controlling for Insurance)
OOP>20% household income
OOP>40% household income
Disparity Chi2
Disparity
Chi2 Group RR disparity (Group UU probability‐Group RR probability)
1997 0.0194 0.06
0.0127
0.00 2000 0.0174 13.41***
0.0154
13.81***
2004 0.0375 30.63***
0.0303
28.24***
2006 0.0232 14.00***
0.0205
16.53***
2009 0.0111 3.72
0.0114
5.63*
2011 0.0211 20.33***
0.0189
21.21***
Group RU disparity (Group UU probability‐Group RU probability)
1997 0.0183 0.46
0.0121
1.58 2000 0.0118 3.38
0.0072
1.56 2004 0.0230 6.44*
0.0184
5.57*
2006 ‐0.0006 0.00
0.0007
0.01 2009 0.0036 0.23
0.0032
0.24 5.15*
6.40*
0.0123
2011 0.0150 Group UR disparity (Group UU probability‐Group UR probability)
1997 0.0091 0.01
0.0074
0.09 2000 0.0169 4.86*
0.0123
3.22 2004 0.0091 0.61
0.0099
0.97 2006 0.0050 0.22
0.0112
1.66 2009 ‐0.0098 1.11
‐0.0020
0.06 2011 0.0069 0.83
0.0071
1.09 Note:1.Significancelevel:***0.001,**0.01,*0.05.
Table6.7showspredicteddisparitiesandtestresultsforthedisparities.Theresults
wereverysimilartobasemodels.Thedifferencewasthatthemagnitudesofdisparities
weregenerallysmalleraftercontrollingforinsurance.Theresultssuggestthathaving
insurancecanexplainpartofthedisparities.However,significantdisparitieswerestill
observed,whichindicatesthatinsurancewasnotthesourcefordisparities.
Table6.8showstheresultsforchangesindisparitiesandthetestresults.Fromthe
results,themagnitudesofchangesindisparitiesweregenerallysmallerthanthebase
model.Someofthechangeswerenotsignificantanymore,suchasdisparityforGroupRU
in2009comparedwithdisparitiesin2007.Theresultsindicatethatthechangein
105
disparitiescanbepartlyexplainedbyinsurancecoverage.Insomewaves,moreinsurance
coverageiscrucialforchangingthedisparitiesinOOP.
Table 6.8 Test Results of Changes in Disparities for OOP Exceeding Certain Percentage of Household Income
(Controlling for Insurance)
OOP>20% household income change in disparity Chi2 Group RR compare with disparities with Group UU in 1997 2000 ‐0.0020 0.88 2004 0.0180 0.31 2006 0.0038 1.88 2009 ‐0.0083 6.20* 2011 0.0016 1.5 compare with disparities with Group UU in 2000 2004 0.0200 0.25 2006 0.0058 0.18 2009 ‐0.0064 2.65 2011 0.0036 0.06 compare with disparities with Group UU in 2006 2 2009 ‐0.0122 2011 ‐0.0022 0.06 Group RU compare with disparities with Group UU in 1997 2000 ‐0.0065 1.55 2004 0.0047 1.72 2006 ‐0.0189 7.42** 2009 ‐0.0146 5.43 2011 ‐0.0033 1.94 compare with disparities with Group UU in 2000 0 2004 0.0112 2006 ‐0.0124 2.25 2009 ‐0.0081 1.24 2011 0.0032 0 compare with disparities with Group UU in 2006 2009 0.0043 0.17 2011 0.0156 3.15 Group UR compare with disparities with Group UU in 1997 2000 0.0079 0.53 2004 0.0001 0.35 2006 ‐0.0041 0.5 2009 ‐0.0189 2.7 2011 ‐0.0022 0.29 compare with disparities with Group UU in 2000 2004 ‐0.0078 1.96 2006 ‐0.0120 2.18 2009 ‐0.0268 5.83* 2011 ‐0.0100 1.81 compare with disparities with Group UU in 2006 2009 ‐0.0148 1.29 2011 0.0019 0.06 Note:1.Significancelevel:***0.001,**0.01,*0.05.
106
OOP>40% household income change in disparity 0.0027 0.0176 0.0078 ‐0.0013 0.0062 0.0149 0.0051 ‐0.0040 0.0035 ‐0.0091 ‐0.0016 ‐0.0049 0.0063 ‐0.0114 ‐0.0089 0.0003 0.0112 ‐0.0065 ‐0.0040 0.0052 0.0025 0.0116 0.0049 0.0025 0.0038 ‐0.0094 ‐0.0003 ‐0.0024 ‐0.0011 ‐0.0142 ‐0.0052 ‐0.0131 ‐0.0041 Chi2 0.04 0.01 0.11 1.97 0.16 0.12 0.02 1.75 0.05 1.79 0.01 1.48 0.76 4.36* 3.19 0.97 0.24 0.81 0.38 0.14 0.08 2.14 0.12 0.23 0.03 1.33 0.23 0.77 0.27 2.46 0.8 1.45 0.13 Table 6.9 Bootstrap Results for Disparities in Total Health Cost (Controlling for Insurance)
Variable Mean
Std. Err.
[95% Conf.
Interval] Group RR disparity with Group UU in 1993 ‐10.911
0.479
‐11.851
‐9.971 disparity with Group UU in 1997 71.176
0.926
69.358
72.995 disparity with Group UU in 2000 242.659
3.167
236.437
248.882 disparity with Group UU in 2004 132.532
1.981
128.641
136.423 disparity with Group UU in 2006 99.663
1.831
96.066
103.261 disparity with Group UU in 2009 109.833
1.811
106.275
113.391 1.591
50.390
56.640 disparity with Group UU in 2011 53.515
Group RU disparity with Group UU in 1993 ‐18.136
0.640
‐19.394
‐16.878 disparity with Group UU in 1997 77.388
0.944
75.533
79.243 disparity with Group UU in 2000 240.529
3.255
234.134
246.924 disparity with Group UU in 2004 131.123
2.067
127.062
135.184 113.268
121.286 117.277
2.041
disparity with Group UU in 2006 disparity with Group UU in 2009 49.269
2.431
44.493
54.044 disparity with Group UU in 2011 8.129
2.212
3.783
12.476 Group UR disparity with Group UU in 1993 ‐15.588
0.758
‐17.077
‐14.098 disparity with Group UU in 1997 54.204
1.103
52.037
56.372 disparity with Group UU in 2000 207.627
3.540
200.672
214.582 ‐58.002
‐37.977 ‐47.990
5.096
disparity with Group UU in 2004 disparity with Group UU in 2006 92.671
2.005
88.731
96.611 disparity with Group UU in 2009 99.117
1.985
95.216
103.018 disparity with Group UU in 2011 68.205
1.779
64.710
71.700 Table6.9showsestimateofdisparitiesintotalhealthcarecostcontrollingfor
insurancecoverage.Theresultsarebasedon500iterationsofbootstrap.Consistentwith
basemodel,theconfidenceintervalswerenotoverlappedbetweenanyofthetwoadjacent
periods,soIcanconcludethatthechangesindisparitiesbetweenadjacentperiodswere
significant.Thetrendofchangesindisparitiesisalsosimilartobasemodel.
6.3.2droppingtherichestprovinceorthepoorestprovince
Thesecondsetofsensitivityanalysisisdroppingtherichestprovinceorthepoorest
province.TheresultsafterdroppingtherichestprovinceareshowninTable6.10.Column
1showsresultsforOOPexceeding20%ofhouseholdincome.Afterdroppingtherichest
provinceJiangsu,theresultsweresimilarasbasemodel.Thedifferencewasthatsomeof
107
thedeviationfromtrendswasnotsignificantanymorecomparedwithbasemodels,suchas
deviationforGroupRUin2004,anddeviationforGroupUUin2009and2011.IntheOOP
exceeding40%ofhouseholdincome,thedifferencewasmoreprominent.Noneofthe
deviationswassignificantafterdroppingtherichestprovince.Theresultssuggestthatthe
deviationsfromexistingtrendsweremoresignificantinrichprovinces.Table6.11shows
predicteddisparitiesandtestresults.Themagnitudeofdisparitieswasgenerallysmaller
thaninbasemodel,butthedisparitieswerestillsignificantasobservedinthebasemodel.
Theresultsindicatethatthedisparitiesweremoresignificantwithinrichprovinces.Table
6.12showsresultsforthechangesindisparities.Afterdroppingtherichestprovince,none
ofthechangesindisparitieswassignificantanymore.Theresultsindicatethatthechanges
indisparitiesarealsohappenedmainlyinricherprovince.
Table6.13to6.15showresultsafterdroppingthepoorestprovince.Iobservethat
theresultswereverysimilartobasemodels.Droppingthepoorestprovincedidnothave
significantimpactoneitherthemagnitudeorsignificanceofresults.
108
Table 6.10 Multi-variate Analysis Results for OOP Exceeding Certain Percentage of Household Income
(Dropping the Richest Province)
OOP>20% household income OOP>40% household income independent variables Coef. Robust Std. Err. Coef.
Robust Std. Err. disparity with Group UU in 1993 Group RR 0.015 0.212
0.133
0.262
Group RU ‐0.030 0.261
0.184
0.310
Group UR 0.319 0.283
0.274
0.358
trend in 1990s and change in later waves Group RR trend in 1990s 0.009 0.035
0.021
0.041
deviation from trend in 2000 0.272 0.201
0.253
0.231
deviation from trend in 2004 0.896** 0.325
0.719
0.379
deviation from trend in 2006 0.818* 0.394
0.503
0.458
0.493
0.639
0.577
deviation from trend in 2009 0.927 deviation from trend in 2011 0.670 0.565
0.431
0.659
Group RU trend in 1990s 0.054 0.060
0.036
0.068
deviation from trend in 2000 0.178 0.345
0.379
0.391
deviation from trend in 2004 0.740 0.550
0.885
0.625
deviation from trend in 2006 0.738 0.664
0.900
0.752
0.845
0.442
0.959
deviation from trend in 2009 0.321 deviation from trend in 2011 0.034 0.962
0.346
1.087
Group UR trend in 1990s 0.028 0.068
0.045
0.086
deviation from trend in 2000 ‐0.070 0.385
0.126
0.462
deviation from trend in 2004 0.709 0.619
0.695
0.769
deviation from trend in 2006 0.619 0.752
0.354
0.941
deviation from trend in 2009 0.724 0.949
0.481
1.187
0.205
1.358
1.079
deviation from trend in 2011 0.325 Group UU trend in 1990s 0.155** 0.057
0.172*
0.070
deviation from trend in 2000 ‐0.191 0.288
‐0.083
0.347
deviation from trend in 2004 ‐0.091 0.493
‐0.159
0.605
deviation from trend in 2006 ‐0.553 0.606
‐0.683
0.745
deviation from trend in 2009 ‐1.080 0.775
‐1.247
0.954
1.085
deviation from trend in 2011 ‐1.475 0.879
‐1.605
‐4.707*** 0.244
constant ‐4.313*** 0.200
Note:1.Significancelevel:***0.001,**0.01,*0.05.
109
Table 6.11 Test Results of Disparities for OOP Exceeding Certain Percentage of Household Income
(Dropping the Richest Province)
OOP>20% household income OOP>40% household income
Disparity
Chi2
Disparity Chi2
Group RR disparity (Group UU probability‐Group RR probability)
1997
0.0104
0.18
0.0066 0.22
2000
0.0130
13.92***
0.0118 13.20***
2004
0.0273
28.93***
0.0227 25.45***
2006
0.0209
18.08***
0.0191 20.98***
2009
0.0130
6.84**
0.0125 8.55**
2011
0.0169
18.77***
0.0160 20.75***
Group RU disparity (Group UU probability‐Group RU probability)
1997
0.0077
0.04
0.0053 0.08
2000
0.0085
4.09*
0.0071 2.49
2004
0.0138
5.19*
0.0114 3.25
2006
0.0002
0.13
0.0005 0.01
2009
0.0087
2.66
0.0103 3.33
2011
0.0115
7.09**
0.0105 5.17*
Group UR disparity (Group UR probability‐Group 0.0041
0.04
0.0036 0.00
1997
2000
0.0112
4.08*
0.0089 0.00
2004
0.0144
2.95
0.0111 2.14
2006
0.0075
0.83
0.0115 2.65
2009
‐0.0050
0.39
0.0011 0.03
2011
0.0068
1.21
0.0069 1.45
Note:1.Significancelevel:***0.001,**0.01,*0.05.
110
Table 6.12 Test Results of Changes in Disparities for OOP Exceeding Certain Percentage of Household
Income (Dropping the Richest Province)
OOP>20% household income change in disparity Group RR compare with disparities with Group UU in 1997 2000 0.0026 2004 0.0170 2006 0.0105 2009 0.0026 2011 0.0066 compare with disparities with Group UU in 2000 2004 0.0144 2006 0.0079 2009 0.0000 2011 0.0040 compare with disparities with Group UU in 2006 2009 ‐0.0079 2011 ‐0.0040 Group RU compare with disparities with Group UU in 1997 2000 0.0008 2004 0.0062 2006 ‐0.0075 2009 0.0010 2011 0.0038 compare with disparities with Group UU in 2000 2004 0.0053 2006 ‐0.0083 2009 0.0002 2011 0.0030 compare with disparities with Group UU in 2006 2009 0.0085 2011 0.0113 Group UR compare with disparities with Group UU in 1997 2000 0.0071 2004 0.0103 2006 0.0034 2009 ‐0.0090 2011 0.0027 compare with disparities with Group UU in 2000 2004 0.0032 2006 ‐0.0037 2009 ‐0.0161 2011 ‐0.0044 compare with disparities with Group UU in 2006 2009 ‐0.0124 2011 ‐0.0007 Note:1.Significancelevel:***0.001,**0.01,*0.05.
OOP>40% household income change in disparity Chi2 111
Chi2 0.02 0.03 0.09 1.68 0.30 0.0053 0.0161 0.0125 0.0060 0.0094 0.11 0.04 1.66 0.22 0.0109 0.0073 0.0007 0.0042 1.69 0.09 0.06 0.25 2.39 0.56 0.16 ‐0.0066 ‐0.0031 0.0018 0.0060 ‐0.0048 0.0050 0.0052 0.07 1.97 0.29 0.02 0.0043 ‐0.0066 0.0033 0.0034 0.96 2.38 0.71 0.10 0.01 1.07 0.01 0.0099 0.0101 0.0052 0.0075 0.0079 ‐0.0025 0.0032 0.42 1.11 4.07 1.18 0.0022 0.0027 ‐0.0077 ‐0.0020 1.37 0.00 ‐0.0104 ‐0.0046 0.23 0.61 0.54 0.10 0.11 0.10 0.07 0.85 0.04 1.82 0.29 0.03 0.09 1.50 0.02 0.02 0.02 1.41 0.00 0.00 1.74 2.20 0.24 0.01 0.10 0.38 0.01 0.20 0.04 1.50 0.49 1.42 0.29 Table 6.13 Multi-variate Analysis Results for OOP Exceeding Certain Percentage of Household Income
(Dropping the Poorest Province)
OOP>20% household income OOP>40% household income independent variables Coef. Robust Std. Err. Coef.
Robust Std. Err. disparity with Group UU in 1993 Group RR ‐0.102 0.205
0.049
0.257
Group RU ‐0.032 0.248
0.250
0.296
Group UR 0.303 0.275
0.455
0.338
trend in 1990s and change in later waves Group RR trend in 1990s 0.028 0.034
0.039
0.040
deviation from trend in 2000 0.166 0.198
0.108
0.228
deviation from trend in 2004 0.727* 0.319
0.604
0.372
deviation from trend in 2006 0.630 0.385
0.341
0.449
0.482
0.353
0.565
deviation from trend in 2009 0.643 deviation from trend in 2011 0.323 0.552
0.123
0.645
Group RU trend in 1990s 0.022 0.057
‐0.013
0.063
deviation from trend in 2000 0.256 0.334
0.564
0.379
deviation from trend in 2004 0.988 0.529
1.262*
0.604
deviation from trend in 2006 1.098 0.634
1.357
0.719
0.807
1.163
0.913
deviation from trend in 2009 0.723 deviation from trend in 2011 0.523 0.913
1.060
1.030
Group UR trend in 1990s 0.022 0.068
0.008
0.082
deviation from trend in 2000 ‐0.180 0.399
0.038
0.467
deviation from trend in 2004 0.791 0.623
0.927
0.751
deviation from trend in 2006 0.644 0.760
0.716
0.914
deviation from trend in 2009 0.753 0.960
0.822
1.156
0.705
1.317
1.089
deviation from trend in 2011 0.412 Group UU trend in 1990s 0.177*** 0.053
0.202
0.067
deviation from trend in 2000 ‐0.449 0.262
‐0.367
0.319
deviation from trend in 2004 ‐0.386 0.454
‐0.471
0.564
deviation from trend in 2006 ‐1.035 0.559
‐1.223
0.698
deviation from trend in 2009 ‐1.657* 0.720
‐1.858*
0.897
‐2.205*
1.022
deviation from trend in 2011 ‐1.995* 0.816
‐3.725*** 0.192
‐4.251*** 0.242
constant Note:1.Significancelevel:***0.001,**0.01,*0.05.
112
Table 6.14 Test Results of Disparities for OOP Exceeding Certain Percentage of Household Income
(Dropping the Poorest Province)
OOP>20% household income
OOP>40% household income
Disparity Chi2 Disparity
Chi2
Group RR disparity (Group UU probability‐Group RR probability)
1997 0.0230 0.14 0.0138
0.21
2000 0.0203 14.56***
0.0175
14.91***
2004 0.0458 35.88***
0.0353
30.27***
2006 0.0238 10.73** 0.0208
13.20***
2009 0.0122 2.95 0.0149
6.73**
2011 0.0268 21.22***
0.0230
22.16***
Group RU disparity (Group UU probability‐Group RU probability)
1997 0.0235 0.03 0.0138
0.93
2000 0.0189 5.51* 0.0103
2.67
2004 0.0343 8.15** 0.0222
6.29*
2006 ‐0.0005 0.32 ‐0.0025
0.09
2009 0.0132 0.98 0.0072
0.86
0.0225 6.66** 0.0154
5.66*
2011 Group UR disparity (Group UU probability‐Group UR probability)
1997 0.0123 0.02 0.0083
0.05
2000 0.0197 4.88* 0.0148
3.59
2004 0.0193 1.97 0.0171
2.12
2006 0.0026 0.04 0.0059
0.33
2009 ‐0.0171 2.03 ‐0.0020
0.04
2011 0.0056 0.33 0.0058
0.49
Note:1.Significancelevel:***0.001,**0.01,*0.05.
113
Table 6.15 Test Results of Changes in Disparities for OOP Exceeding Certain Percentage of Household
Income (Dropping the Poorest Province)
OOP>20% household income
OOP>40% household income
change in disparity Chi2
change in disparity
Chi2
Group RR compare with disparities with Group UU in 1997
2000 ‐0.0026 0.69
0.0037
0.00
2004 0.0228 0.13
0.0216
0.09
2006 0.0008 2.79
0.0071
0.20
2009 ‐0.0108 7.02**
0.0011
1.33
2011 0.0038 1.48
0.0092
0.05
compare with disparities with Group UU in 2000
2004 0.0254 0.35
0.0178
0.07
2006 0.0034 0.72
0.0033
0.29
2009 ‐0.0082 3.66
‐0.0026 1.77
2011 0.0064 0.11
0.0055
0.10
compare with disparities with Group UU in 2006
2009 ‐0.0116 1.50
‐0.0060 0.77
2011 0.0030 0.41
0.0021
0.08
Group RU compare with disparities with Group UU in 1997
2000 ‐0.0046 0.82
‐0.0036 0.94
2004 0.0108 1.42
0.0084
0.69
2006 ‐0.0241 9.38**
‐0.0163 5.45*
2009 ‐0.0104 4.19*
‐0.0066 2.39
2011 ‐0.0010 2.10
0.0016
0.95
compare with disparities with Group UU in 2000
0.05
0.0120
0.05
2004 0.0154 2006 ‐0.0194 4.96*
‐0.0127 2.14
2009 ‐0.0057 1.41
‐0.0030 0.40
2011 0.0036 0.22
0.0051
0.01
compare with disparities with Group UU in 2006
2009 0.0137 1.32
0.0097
0.80
2011 0.0230 4.76*
0.0179
3.39
Group UR compare with disparities with Group UU in 1997
2000 0.0074 0.40
0.0065
0.24
2004 0.0070 0.11
0.0088
0.02
2006 ‐0.0097 1.03
‐0.0024 0.35
2009 ‐0.0294 3.95*
‐0.0103 1.18
2011 ‐0.0067 0.79
‐0.0024 0.41
compare with disparities with Group UU in 2000
2004 ‐0.0004 1.07
0.0023
0.46
2006 ‐0.0171 2.73
‐0.0089 1.21
2009 ‐0.0368 6.94**
‐0.0168 2.69
2011 ‐0.0142 2.47
‐0.0089 1.49
compare with disparities with Group UU in 2006
2009 ‐0.0197 1.43
‐0.0079 0.37
2011 0.0030 0.05
0.0000
0.00
Note:1.Significancelevel:***0.001,**0.01,*0.05.
114
Table6.16showstheestimateddisparitiesintotalhealthcarecostsafterdropping
therichestprovince.Theestimationisbasedon500iterationsofbootstrapping.Underthis
scenario,the95%confidenceintervalswerenotoverlappedasthebasemodel.This
indicatesthatthechangesindisparitiesbetweenadjacentperiodsweresignificant,andthis
wasconsistentwiththebasemodel.Table6.17showsbootstrapestimateddisparitiesin
totalhealthcarecostsafterdroppingtherichestprovince.Thetrendofchangesin
disparitieswasconsistentwithbasemodel.However,themagnitudeofdisparitieswas
largerthanthebasemodelingeneral.Thisindicatesthatthedisparitiesintotalhealth
costsweremoreprominentinrichprovinces.
Table 6.16 Bootstrap Results for Disparities in Total Health Costs (Dropping the Richest Province)
Variable Mean
Std. Err.
[95% Conf.
Interval] Group RR disparity with Group UU in 1993 ‐16.903
0.406
‐17.700
‐16.106 disparity with Group UU in 1997 41.546
0.708
40.156
42.936 disparity with Group UU in 2000 256.342
4.029
248.427
264.258 disparity with Group UU in 2004 89.262
2.162
85.014
93.509 disparity with Group UU in 2006 94.805
1.721
91.423
98.186 disparity with Group UU in 2009 133.149
2.207
128.813
137.486 1.516
24.348
30.305 disparity with Group UU in 2011 27.326
Group RU disparity with Group UU in 1993 ‐17.351
0.511
‐18.354
‐16.347 disparity with Group UU in 1997 42.071
0.734
40.629
43.513 disparity with Group UU in 2000 258.348
4.028
250.435
266.262 disparity with Group UU in 2004 103.216
2.355
98.589
107.844 111.462
118.576 115.019
1.810
disparity with Group UU in 2006 disparity with Group UU in 2009 87.561
2.670
82.316
92.806 disparity with Group UU in 2011 ‐5.687
2.058
‐9.731
‐1.644 Group UR disparity with Group UU in 1993 ‐25.492
0.798
‐27.059
‐23.925 disparity with Group UU in 1997 19.287
1.003
17.316
21.259 disparity with Group UU in 2000 212.697
4.294
204.261
221.132 ‐51.705
5.136
‐61.795
‐41.614 disparity with Group UU in 2004 disparity with Group UU in 2006 83.536
1.860
79.881
87.190 disparity with Group UU in 2009 120.392
2.249
115.974
124.811 disparity with Group UU in 2011 42.984
1.688
39.668
46.300 115
Table 6.17 Bootstrap Results for Disparities in Total Health Cost (Dropping the Poorest Province)
Variable Mean
Std. Err.
[95% Conf.
Interval] Group RR disparity with Group UU in 1993 ‐4.675
0.512
‐5.681
‐3.670 disparity with Group UU in 1997 83.959
1.049
81.898
86.019 disparity with Group UU in 2000 280.757
4.073
272.756
288.759 disparity with Group UU in 2004 148.829
2.331
144.249
153.408 disparity with Group UU in 2006 91.114
1.845
87.490
94.739 disparity with Group UU in 2009 125.656
2.192
121.349
129.963 1.787
60.812
67.833 disparity with Group UU in 2011 64.322
Group RU disparity with Group UU in 1993 ‐15.870
0.683
‐17.212
‐14.528 disparity with Group UU in 1997 93.248
1.027
91.230
95.265 disparity with Group UU in 2000 298.455
3.954
290.688
306.223 disparity with Group UU in 2004 151.651
2.438
146.862
156.440 100.108
107.873 103.990
1.976
disparity with Group UU in 2006 disparity with Group UU in 2009 81.023
2.561
75.992
86.055 disparity with Group UU in 2011 17.365
2.395
12.659
22.071 Group UR disparity with Group UU in 1993 ‐14.102
0.851
‐15.774
‐12.430 disparity with Group UU in 1997 66.376
1.365
63.694
69.058 disparity with Group UU in 2000 233.949
4.566
224.978
242.919 ‐82.643
‐56.928 ‐69.786
6.544
disparity with Group UU in 2004 disparity with Group UU in 2006 66.581
2.177
62.304
70.857 disparity with Group UU in 2009 107.848
2.257
103.413
112.283 disparity with Group UU in 2011 69.363
2.013
65.407
73.318 6.3.3.Includinginteractiontermswithhouseholdincome
Thenextsetofsensitivityanalysisistotakehouseholdincomelevelinto
consideration.Table6.18to6.26showresultsformulti‐variateanalysisforOOPexceeding
certainpercentageofhouseholdincome.Themodelwasestimatedbyasingleregression
includinginteractiontermbetweenfourgroupsandhouseholdincomegroupsand
presentedseparatelyforlow,mediumandhigh‐incomefamilies.Medium‐incomefamilies
inGroupUUwereusedasreferencegroupintheanalysis.
Table6.18showsregressionresultsformulti‐variateanalysisforlow‐income
families.PresentedinTable6.18Column1,in1993,thelow‐incomefamilieswithinGroup
RR,RUandURallhadgreaterprobabilitytohaveOOPexceeding20%ofhouseholdincome
thantheircounterpartsinGroupUU.Thiswasdifferentfromthebasemodel.Table6.19
116
showstheestimateddisparitiesandtheresultsfromWaldtestindicatingstatistical
significanceofthedisparities.Generally,thedisparitieswerenotsignificantanymorewhen
using20%asthecut‐offpoint,indicatingthatrurallow‐incomefamiliesdidnothave
significantlylowerpossibilitytohavehighOOPcoststhanurbanlow‐incomefamilies.
Table6.20showsresultsforchangeindisparitiesandtestresults.Differentfromthebase
model,thedisparitiesgenerallyincreasedinlateryearscomparedwithdisparitiesin1997,
2000and2006.However,theresultswerenotsignificantexceptforchangesindisparities
betweenGroupRUandUUafter2006. 117
Table 6.18 Multi-variate Analysis Results for OOP Exceeding Certain Percentage of Household Income
(Low-income Families)
OOP>20% household income OOP>40% household income independent variables Coef. Robust Std. Err.
Coef.
Robust Std. Err. disparity with Group UU medium income in 1993 Group RR low income 0.798* 0.382
1.401*
0.603
Group RU low income 0.798 0.435
1.512*
0.645
Group UR low income 1.298** 0.466
1.882** 0.685
Group UU low income 0.771 0.506
1.751** 0.681
trend in 1990s and change in later waves Group RR low income trend in 1990s 0.027 0.047
0.061
0.055
deviation from trend in 2000 0.264 0.269
0.144
0.306
0.506
0.439
0.591
deviation from trend in 2004 0.906* deviation from trend in 2006 0.765 0.533
0.311
0.611
deviation from trend in 2009 0.918 0.668
0.418
0.773
deviation from trend in 2011 0.640 0.765
0.106
0.885
Group RU low income trend in 1990s 0.114 0.077
0.116
0.086
deviation from trend in 2000 ‐0.232 0.428
‐0.122
0.475
0.247
0.774
0.692
deviation from trend in 2004 0.269 deviation from trend in 2006 0.387 0.831
0.383
0.929
deviation from trend in 2009 ‐0.383 1.066
‐0.526
1.201
deviation from trend in 2011 ‐0.630 1.218
‐0.543
1.358
Group UR low income trend in 1990s ‐0.104 0.111
‐0.133
0.138
deviation from trend in 2000 0.969 0.638
1.467
0.788
1.055
2.804*
1.320
deviation from trend in 2004 2.303* deviation from trend in 2006 2.040 1.263
2.563
1.592
deviation from trend in 2009 2.725 1.593
3.295
1.991
deviation from trend in 2011 2.902 1.808
3.491
2.268
Group UU low income trend in 1990s 0.147 0.105
0.046
0.111
deviation from trend in 2000 ‐0.365 0.514
0.140
0.572
deviation from trend in 2004 0.281 0.880
1.103
0.962
1.175
1.085
0.865
deviation from trend in 2006 ‐0.278 deviation from trend in 2009 ‐0.553 1.404
0.887
1.510
deviation from trend in 2011 ‐0.858 1.601
0.816
1.718
Note:1.Significancelevel:***0.001,**0.01,*0.05.
118
Table 6.19 Test Results of Disparities for OOP Exceeding Certain Percentage of Household Income (Lowincome Families)
OOP>20% household income
OOP>40% household income
Disparity
Chi2
Disparity
Chi2 Group RR disparity (Group UU probability‐Group RR probability)
1997 0.0186
3.81
0.0087
4.53* 2000 0.0091
1.12
0.0096
4.29* 2004 0.0754
0.04
0.0618
2.23 2006 0.0482
0.27
0.0556
2.13 2009 0.0482
0.44
0.0542
2.81 2011 0.0666
0.07
0.0696
1.88 Group RU disparity (Group UU probability‐Group RU probability)
1997 0.0049
2.17
‐0.0015
5.51* 2000 0.0034
1.48
0.0003
5.45* 2004 0.0431
0.63
0.0326
3.99* 2006 ‐0.0355
3.61
‐0.0215
7.31**
2009 0.0379
0.68
0.0499
2.92 2011 0.0383
0.68
0.0342
4.02* Group UR disparity (Group UU probability‐Group UR probability)
0.0153
9.51**
7.24**
1997 0.0194
2000 ‐0.0059
2.14
‐0.0099
6.56* 2004 0.0278
1.03
0.0149
4.97* 2006 0.0421
0.37
0.0422
2.77 2009 0.0258
1.05
0.0332
3.92* 2011 0.0278
1.01
0.0419
3.50 Note:1.Significancelevel:***0.001,**0.01,*0.05.
119
Table 6.20 Test Results of Changes in Disparities for OOP Exceeding Certain Percentage of Household
Income (Low-income Families)
OOP>20% household income
OOP>40% household income
change in disparity
Chi2
change in disparity Chi2
Group RR compare with disparities with Group UU in 1997
2000 ‐0.0096
0.62
0.0009
0.02
2004 0.0568
0.53
0.0531
1.40
2006 0.0296
0.01
0.0469
1.42
2009 0.0295
0.01
0.0455
0.72
2011 0.0480
0.40
0.0609
2.30
compare with disparities with Group UU in 2000
2004 0.0664
2.75
0.0522
2.00
2006 0.0391
0.99
0.0460
1.93
2009 0.0391
0.64
0.0446
1.09
2011 0.0575
2.56
0.0600
3.17
compare with disparities with Group UU in 2006
‐0.0014
0.25
2009 0.0000
0.08
0.40
0.0140
0.11
2011 0.0184
Group RU compare with disparities with Group UU in 1997
2000 ‐0.0015
0.01
0.0018
0.01
2004 0.0382
0.46
0.0341
0.80
2006 ‐0.0404
1.14
‐0.0200
0.15
2009 0.0330
0.36
0.0514
1.83
2011 0.0334
0.43
0.0357
0.90
compare with disparities with Group UU in 2000
2004 0.0396
0.6
0.0323
0.65
2006 ‐0.0389
0.89
‐0.0218
0.27
2009 0.0345
0.47
0.0495
1.63
2011 0.0348
0.56
0.0339
0.73
compare with disparities with Group UU in 2006
3.89* 0.0714
4.67*
2009 0.0734
2011 0.0738
4.48* 0.0557
2.99
Group UR compare with disparities with Group UU in 1997
2000 ‐0.0253
1.40
‐0.0253
1.90
2004 0.0084
0.35
‐0.0004
0.67
2006 0.0227
0.02
0.0268
0.02
2009 0.0064
0.35
0.0179
0.23
2011 0.0084
0.34
0.0265
0.10
compare with disparities with Group UU in 2000
2004 0.0337
0.63
0.0249
0.63
0.0521
2.19
2006 0.0480
1.43
2009 0.0317
0.60
0.0431
1.46
2011 0.0337
0.75
0.0518
2.23
compare with disparities with Group UU in 2006
2009 ‐0.0164
0.34
‐0.0090
0.20
2011 ‐0.0143
0.32
‐0.0003
0.05
Note:1.Significancelevel:***0.001,**0.01,*0.05.
120
Table6.21showsmulti‐variateregressionresultsformedium‐incomefamilies.
Similartolow‐incomefamilies,in1993,medium‐incomefamiliesfromGroupRR,RUand
URweremorelikelytohavehighOOPthanmedium‐incomefamiliesfromGroupUUwhen
using20%cut‐offpoint.Thiswasdifferentfromthebasemodel.Table6.22showsthe
estimateddisparitiesformediumincomefamilies.Similartolow‐incomefamilies,mostof
thedisparitieswerenotsignificant,whichwasdifferentfromthebasemodel.Insomeof
theyears,medium‐incomefamiliesinGroupRR,URandRUhadgreaterprobabilitythan
medium‐incomefamiliesinGroupUUtohavehighOOPexceeding20%/40%oftheir
householdincome.Forexample,in1997,2006and2009,GroupURhadhigherprobability
tohaveOOPexceeding20%ofhouseholdincomethanGroupUU.Table6.23showsthe
estimatedchangesindisparities.Thedirectionofchangewasverysimilartothebase
model.
121
Table 6.21 Multi-variate Analysis Results for OOP Exceeding Certain Percentage of Household Income
(Medium-income Families)
OOP>20% household income OOP>40% household income independent variables Coef. Robust Std. Err.
Coef.
Robust Std. Err. disparity with Group UU medium income in 1993 Group RR medium income
0.581 0.395
1.267*
0.613
Group RU medium income 0.917* 0.431
1.795** 0.634
Group UR medium income 1.064* 0.495
1.824** 0.697
trend in 1990s and change in later waves Group RR medium income
trend in 1990s ‐0.027 0.061
‐0.025
0.071
deviation from trend in 2000 0.374 0.363
0.477
0.413
deviation from trend in 2004 1.084 0.587
1.025
0.680
0.703
1.019
0.817
deviation from trend in 2006 1.204 deviation from trend in 2009 1.452 0.883
1.205
1.022
deviation from trend in 2011 1.001 1.008
0.940
1.168
Group RU medium income trend in 1990s ‐0.144 0.100
‐0.241*
0.121
deviation from trend in 2000 1.136 0.635
1.827*
0.786
deviation from trend in 2004 2.656** 0.986
3.536** 1.228
1.473
1.184
3.426*
deviation from trend in 2006 2.621* deviation from trend in 2009 3.172* 1.476
4.549*
1.824
deviation from trend in 2011 2.883 1.675
4.205*
2.068
Group UR medium income trend in 1990s 0.086 0.105
0.046
0.121
deviation from trend in 2000 ‐1.335* 0.648
‐1.208
0.778
deviation from trend in 2004 ‐0.210 0.917
‐0.086
1.077
1.318
1.117
0.052
deviation from trend in 2006 0.041 deviation from trend in 2009 ‐0.329 1.425
‐0.016
1.667
deviation from trend in 2011 ‐1.450 1.636
‐0.675
1.908
Group UU medium income trend in 1990s 0.294** 0.099
0.442** 0.157
deviation from trend in 2000 ‐0.839 0.454
‐1.001
0.626
deviation from trend in 2004 ‐1.200 0.811
‐2.297
1.212
1.525
deviation from trend in 2006 ‐2.079* 1.009
‐3.523*
1.302
‐4.926*
1.989
deviation from trend in 2009 ‐3.146* deviation from trend in 2011 ‐3.755* 1.494
‐5.818*
2.294
Note:1.Significancelevel:***0.001,**0.01,*0.05.
122
Table 6.22 Test Results of Disparities for OOP Exceeding Certain Percentage of Household Income (Mediumincome Families)
OOP>20% household income
OOP>40% household income
Disparity
Chi2
Disparity
Chi2 Group RR disparity (Group UU probability‐Group RR probability)
1997 0.0220
1.49
0.0135
2.33 2000 0.0163
3.44
0.0166
4.14* 2004 0.0454
13.50***
0.0267
6.79**
2006 0.0187
2.40
0.0287
1.30 2009 ‐0.0031
0.06
0.0036
0.13 2011 0.0210
0.06*
0.0138
3.79 Group RU disparity (Group UU probability‐Group RU probability)
1997 0.0247
4.84*
0.0181
3.96* 2000 0.0071
0.35
0.0058
0.26 2004 0.0039
0.05
‐0.0073
0.23 2006 0.0052
0.10
0.0057
0.20 2009 ‐0.0151
0.95
‐0.0167
1.60 2011 0.0167
2.38
0.0160
3.12 Group UR disparity (Group UU probability‐Group UR probability)
‐0.0079
0.00 5.00*
1997 ‐0.0111
2000 0.0267
3.81
0.0279
4.56* 2004 0.0185
0.74
0.0170
0.95 2006 ‐0.0421
3.40
‐0.0126
0.52 2009 ‐0.0431
4.44*
‐0.0200
1.47 2011 0.0183
1.70
0.0062
0.27 Note:1.Significancelevel:***0.001,**0.01,*0.05.
123
Table 6.23 Test Results of Changes in Disparities for OOP Exceeding Certain Percentage of Household
Income (Medium-income Families)
OOP>20% household income
OOP>40% household income
change in disparity
Chi2
change in disparity Chi2
Group RR compare with disparities with Group UU in 1997
2000 ‐0.0056
0.52
0.0031
0.04
2004 0.0234
0.02
0.0132
0.01
2006 ‐0.0033
1.47
0.0152
0.70
2009 ‐0.0251
5.08*
‐0.0099
1.69
2011 ‐0.0009
0.71
0.0004
0.30
compare with disparities with Group UU in 2000
2004 0.0290
0.50
0.0101
0.01
2006 0.0024
0.23
0.0121
0.53
2009 ‐0.0195
2.52
‐0.0130
1.54
2011 0.0047
0.00
‐0.0028
0.16
compare with disparities with Group UU in 2006
1.69
‐0.0251
0.31
2009 ‐0.0218
2011 0.0023
0.25
‐0.0149
0.15
Group RU compare with disparities with Group UU in 1997
2000 ‐0.0176
1.95
‐0.0319
1.92
2004 ‐0.0208
3.47
‐0.0213
4.01*
2006 ‐0.0195
2.92
‐0.0271
2.06
2009 ‐0.0398
5.81*
‐0.0398
5.54*
2011 ‐0.0080
1.39
‐0.0326
0.72
compare with disparities with Group UU in 2000
2004 ‐0.0032
0.13
0.0106
0.49
2006 ‐0.0019
0.07
0.0048
0.00
2009 ‐0.0222
1.13
‐0.0079
1.47
2011 0.0096
0.18
‐0.0007
0.60
compare with disparities with Group UU in 2006
2009 ‐0.0203
0.87
‐0.0224
1.46
2011 0.0115
0.63
0.0103
0.73
Group UR compare with disparities with Group UU in 1997
2000 0.0379
4.15*
0.0358
4.48*
2004 0.0297
1.32
0.0248
1.33
2006 ‐0.0310
0.39
‐0.0047
0.00
2009 ‐0.0319
0.64
‐0.0121
0.10
2011 0.0294
2.04
0.0141
0.67
compare with disparities with Group UU in 2000
‐0.0373
1.74
2004 ‐0.0082
1.58
2006 ‐0.0688
6.73** ‐0.0065
4.74*
2009 ‐0.0698
7.54** ‐0.0327
5.99*
2011 ‐0.0085
0.96
‐0.0245
2.54
compare with disparities with Group UU in 2006
2009 ‐0.0010
0.04
‐0.0074
0.10
2011 0.0604
4.93*
‐0.0180
0.77
Note:1.Significancelevel:***0.001,**0.01,*0.05.
124
Table6.24showsresultsforhigh‐incomefamilies.Table6.24showsregression
resultsformulti‐variateanalysisforhigh‐incomefamilies.Theresultsweresimilartothe
basemodel.Table6.25showstheestimateddisparitiesandtheresultsfromWaldtest
indicatingstatisticalsignificanceofthedisparities.Generally,thedisparitieswerenot
significantanymorewhenusing20%asthecut‐offpoint,indicatingthatruralhigh‐income
familiesdidnothavesignificantlylowerpossibilitytohavehighOOPcoststhanurbanhigh‐
incomefamilies.Table6.26showsresultsforchangeindisparitiesandtestresults.The
trendofchangesindisparitieswasconsistentwithbasemodel.
Insum,inthesensitivityanalysisincludinginteractiontermsbetweenfourgroups
andincomegroups,thedisparitiesinhighOOPwerenotsignificantinlow‐andhigh‐
incomefamilies.Ialsofindthatthechangesindisparitieswereindifferentdirectionin
low‐incomefamilies,althoughthechangesindisparitieswerenotsignificant.
125
Table 6.24 Multi-variate Analysis Results for OOP Exceeding Certain Percentage of Household Income
(High-income Families)
OOP>20% household income OOP>40% household income independent variables Coef. Robust Std. Err.
Coef.
Robust Std. Err. disparity with Group UU medium income in 1993 Group RR high income 0.550 0.406
1.316*
0.620
Group RU high income ‐0.137 0.573
0.349
0.821
Group UR high income 0.053 0.546
0.378
0.821
Group UU high income 0.556 0.423
0.945
0.654
trend in 1990s and change in later waves Group RR high income trend in 1990s ‐0.037 0.069
‐0.052
0.078
deviation from trend in 2000 0.317 0.414
0.227
0.483
0.669
0.910
0.765
deviation from trend in 2004 0.882 deviation from trend in 2006 0.904 0.803
0.609
0.925
deviation from trend in 2009 0.621 1.001
0.493
1.150
deviation from trend in 2011 0.937 1.137
1.054
1.293
Group RU high income trend in 1990s 0.002 0.152
0.084
0.184
deviation from trend in 2000 0.866 0.835
0.600
0.942
1.410
‐0.731
1.666
deviation from trend in 2004 0.553 deviation from trend in 2006 0.928 1.694
‐0.241
1.988
deviation from trend in 2009 0.890 2.142
‐0.438
2.526
deviation from trend in 2011 0.806 2.437
‐0.474
2.881
Group UR high income trend in 1990s 0.095 0.137
0.197
0.178
deviation from trend in 2000 ‐0.253 0.815
‐0.459
0.959
1.562
1.258
‐0.397
deviation from trend in 2004 0.442 deviation from trend in 2006 ‐0.337 1.525
‐1.705
1.921
deviation from trend in 2009 ‐0.307 1.918
‐1.973
2.433
deviation from trend in 2011 ‐0.571 2.188
‐2.065
2.773
Group UU high income trend in 1990s 0.159* 0.071
0.229*
0.090
deviation from trend in 2000 ‐0.420 0.373
‐0.710
0.451
0.632
‐1.252
0.778
deviation from trend in 2004 ‐0.798 deviation from trend in 2006 ‐1.319 0.776
‐2.104
0.959
deviation from trend in 2009 ‐2.111* 0.983
‐3.211** 1.222
deviation from trend in 2011 ‐2.393* 1.120
‐3.446*
1.391
Note:1.Significancelevel:***0.001,**0.01,*0.05.
126
Table 6.25 Test Results of Disparities for OOP Exceeding Certain Percentage of Household Income (Highincome Families)
OOP>20% household income
OOP>40% household income
Disparity
Chi2
Disparity
Chi2
Group RR disparity (Group UU probability‐Group RR probability)
1997 0.0242
0.22
0.0172
2.55
2000 0.0045
0.03
‐0.0005
0.16
2004 ‐0.0052
0.02
‐0.0080
0.31
2006 0.0006
0.21
0.0021
0.28
2009 ‐0.0014
0.08
‐0.0032
0.52
2011 ‐0.0031
0.54
‐0.0051
1.18
Group RU disparity (Group UU probability‐Group RU probability)
1997 0.0328
0.62
0.0226
1.50
2000 0.0009
0.01
‐0.0076
0.77
2004 0.0115
0.95
0.0100
0.81
2006 0.0057
0.02
0.0029
0.19
2009 ‐0.0070
0.45
‐0.0076
1.14
2011 0.0002
0.22
‐0.0041
0.99
Group UR disparity (Group UU probability‐Group UR probability)
0.00
0.0163
0.28
1997 0.0236
2000 0.0079
0.14
‐0.0026
0.24
2004 ‐0.0294
1.23
‐0.0244
1.37
2006 0.0014
0.14
0.0024
0.21
2009 ‐0.0248
2.39
‐0.0138
1.97
2011 ‐0.0166
1.91
‐0.0161
2.50
Note:1.Significancelevel:***0.001,**0.01,*0.05.
127
Table 6.26 Test Results of Changes in Disparities for OOP Exceeding Certain Percentage of Household
Income (High-income Families)
OOP>20% household income
OOP>40% household income
change in disparity
Chi2
change in disparity Chi2
Group RR compare with disparities with Group UU in 1997
2000 ‐0.0197
1.04
‐0.0176
0.05
2004 ‐0.0294
0.03
‐0.0252
0.24
2006 ‐0.0237
0.02
‐0.0151
0.18
2009 ‐0.0257
0.21
‐0.0204
0.55
2011 ‐0.0274
0.08
‐0.0223
2.00
compare with disparities with Group UU in 2000
2004 ‐0.0097
0.26
‐0.0075
0.07
2006 ‐0.0039
0.91
0.0025
0.05
2009 ‐0.0060
0.47
‐0.0028
0.31
2011 ‐0.0076
2.00
‐0.0046
1.66
compare with disparities with Group UU in 2006
2009 ‐0.0020
0.06
‐0.0053
0.12
2011 ‐0.0037
0.18
‐0.0072
1.03
Group RU compare with disparities with Group UU in 1997
2000 ‐0.0319
2.34
‐0.0301
2.18
2004 ‐0.0213
0.20
‐0.0126
0.52
2006 ‐0.0271
2.47
‐0.0196
0.82
2009 ‐0.0398
4.32* ‐0.0301
2.78
2011 ‐0.0326
3.86* ‐0.0267
2.73
compare with disparities with Group UU in 2000
2004 0.0106
1.74
0.0176
5.71*
2006 0.0048
0.00
0.0105
0.42
0.53
0.0000
0.10
2009 ‐0.0079
2011 ‐0.0007
0.25
0.0035
0.02
compare with disparities with Group UU in 2006
2009 ‐0.0127
0.50
‐0.0105
0.85
2011 ‐0.0055
0.20
‐0.0070
0.69
Group UR compare with disparities with Group UU in 1997
2000 ‐0.0157
0.00
‐0.0188
0.05
2004 ‐0.0530
2.43
‐0.0406
1.13
2006 ‐0.0222
0.60
‐0.0139
0.03
2009 ‐0.0484
4.07* ‐0.0301
1.79
2011 ‐0.0402
3.38
‐0.0324
2.53
compare with disparities with Group UU in 2000
‐0.0218
0.60
2004 ‐0.0373
2.25
2006 ‐0.0065
0.61
0.0050
0.00
2009 ‐0.0327
3.81
‐0.0113
1.13
2011 ‐0.0245
3.16
‐0.0136
1.65
compare with disparities with Group UU in 2006
2009 ‐0.0262
2.04
‐0.0162
1.63
2011 ‐0.0180
1.37
‐0.0185
2.07
Note:1.Significancelevel:***0.001,**0.01,*0.05.
128
Table6.27to6.29showestimateddisparitiesintotalhealthcostsfordifferent
incomegroups.Theestimateisbasedononesingletwo‐partmodel,andpresented
separatelyfordifferentincomegroups.Table6.29showsresultsforlow‐incomefamilies.
Themagnitudeofdisparitieswasgenerallysmallerwithinlow‐incomefamilies.Mostofthe
disparitieswerestillsignificant,exceptforGroupRRin2011andGroupRUin2009and
2011.Differentfromthebasemodel,intheselateryears,thedisparitiesinlow‐income
familiesinGroupRRandGroupRUwerenotsignificantanymore.Thetrendofchangesin
disparitiesintotalhealthcostswasthesameasthebasemodel.Table6.28and6.29show
bootstrapresultsformedium‐andhigh‐incomefamilies.Theresultswereconsistentwith
thebasemodels.
Table 6.27 Bootstrap Results for Disparities in Total Health Costs (Low-income Families)
Variable Mean
Std. Err.
[95% Conf.
Interval] Group RR disparity with Group UU in 1993 ‐14.94171
0.5385447
‐16.00619
‐13.87724 disparity with Group UU in 1997 56.83302
3.057011
50.7906
62.87543 disparity with Group UU in 2000 177.9387
12.62493
152.9845
202.8928 disparity with Group UU in 2004 134.318
7.863278
118.7757
149.8604 disparity with Group UU in 2006 167.8766
10.2749
147.5675
188.1857 disparity with Group UU in 2009 155.0509
9.737621
135.8038
174.298 7.159147
‐23.07021
5.230974 disparity with Group UU in 2011 ‐8.919616
Group RU disparity with Group UU in 1993 ‐51.11267
2.525351
‐56.10422
‐46.12112 disparity with Group UU in 1997 59.15513
3.02742
53.17121
65.13905 disparity with Group UU in 2000 212.4249
12.75617
187.2114
237.6384 disparity with Group UU in 2004 79.0237
8.962753
61.30815
96.73926 10.31639
120.9764
161.7587 141.3675
disparity with Group UU in 2006 disparity with Group UU in 2009 7.849936
15.6367
‐23.05717
38.75704 disparity with Group UU in 2011 9.953452
6.343274
‐2.584507
22.49141 Group UR disparity with Group UU in 1993 ‐14.98737
0.7037759
‐16.37844
‐13.59631 disparity with Group UU in 1997 35.24916
4.008824
27.32542
43.1729 disparity with Group UU in 2000 149.1536
12.82396
123.806
174.5011 disparity with Group UU in 2004 ‐128.8861
19.50158
‐167.4325
‐90.3398 disparity with Group UU in 2006 164.5178
10.36703
144.0266
185.009 disparity with Group UU in 2009 157.6196
9.344997
139.1485
176.0907 disparity with Group UU in 2011 24.96881
7.429766
10.28331
39.6543 129
Table 6.28 Bootstrap Results for Disparities in Total Health Costs (Medium-income Families)
Variable Group RR disparity with Group UU in 1993 disparity with Group UU in 1997 disparity with Group UU in 2000 disparity with Group UU in 2004 disparity with Group UU in 2006 disparity with Group UU in 2009 disparity with Group UU in 2011 Group RU disparity with Group UU in 1993 disparity with Group UU in 1997 disparity with Group UU in 2000 disparity with Group UU in 2004 disparity with Group UU in 2006 disparity with Group UU in 2009 disparity with Group UU in 2011 Group UR disparity with Group UU in 1993 disparity with Group UU in 1997 disparity with Group UU in 2000 disparity with Group UU in 2004 disparity with Group UU in 2006 disparity with Group UU in 2009 disparity with Group UU in 2011 Mean ‐29.1525 63.15983 268.4552 105.42 15.80023 69.28482 79.61505 ‐38.86723 71.6066 218.0608 48.78306 69.7419 44.86233 ‐190.2435 ‐51.46965 48.45871 286.6617 106.3634 ‐15.27882 33.43513 93.39053 Std. Err. 1.301184 3.104018 13.2618 5.152178 6.096454 8.371618 5.349504 1.308961 3.059992 14.02427 7.31754 6.951842 8.349629 14.18549 2.644083 3.596754 13.02051 6.631208 7.554651 9.072895 6.207767 [95% Conf. ‐31.72439 57.02451 242.2423 95.23633 3.750134 52.73769 69.04135 ‐41.45449 65.5583 190.3408 34.31939 56.00106 28.35866 ‐218.2822 ‐56.69588 41.34946 260.9256 93.25627 ‐30.21115 15.50187 81.12042 Interval] ‐26.58061 69.29516 294.6681 115.6037 27.85033 85.83195 90.18874 ‐36.27997 77.6549 245.7808 63.24672 83.48273 61.366 ‐162.2048 ‐46.24342 55.56796 312.3977 119.4704 ‐0.3464823 51.36839 105.6607 Table 6.29 Bootstrap Results for Disparities in Total Health Costs (High-income Families)
Variable Group RR disparity with Group UU in 1993 disparity with Group UU in 1997 disparity with Group UU in 2000 disparity with Group UU in 2004 disparity with Group UU in 2006 disparity with Group UU in 2009 disparity with Group UU in 2011 Group RU disparity with Group UU in 1993 disparity with Group UU in 1997 disparity with Group UU in 2000 disparity with Group UU in 2004 disparity with Group UU in 2006 disparity with Group UU in 2009 disparity with Group UU in 2011 Group UR disparity with Group UU in 1993 disparity with Group UU in 1997 disparity with Group UU in 2000 disparity with Group UU in 2004 disparity with Group UU in 2006 disparity with Group UU in 2009 disparity with Group UU in 2011 Mean 6.369809 93.67514 236.2302 123.896 103.3722 127.0916 44.33455 29.87013 94.11589 223.0315 207.175 136.4541 70.0443 97.0841 11.8881 83.18063 140.3317 ‐34.97923 108.8807 116.541 62.56974 Std. Err. 1.974789 2.733506 8.237553 6.393577 4.66994 5.445521 5.518837 1.589092 2.687684 8.303527 5.230706 4.550388 6.180751 4.45719 2.298468 3.040064 9.95022 14.60023 5.119126 5.593961 6.686264 130
[95% Conf. 2.46649 88.27216 219.9481 111.2586 94.14169 116.3281 33.42615 26.72917 88.80348 206.619 196.8362 127.4599 57.82758 88.27413 7.345005 77.17171 120.6644 ‐63.83769 98.7624 105.4841 49.35383 Interval] 10.27313 99.07812 252.5123 136.5334 112.6027 137.8551 55.24294 33.01109 99.4283 239.4441 217.5139 145.4483 82.26102 105.8941 16.43119 89.18954 159.9991 ‐6.120781 118.9991 127.5979 75.78564 6.3.4DIDanalysisresultsforcostvariables
ThelastsetofsensitivityanalysisisDIDanalysisforthecostrelatedvariables.For
OOPexceeding20%/40%ofhouseholdincome,theresultsareshowninTable6.30.
Column1showsresultsforOOPexceeding20%ofhouseholdincome.Thedisparitiesin
period1weresmallerthan1,indicatingthatallthreegroupswerelesslikelytohavehigh
OOPcomparedwithGroupUU.However,thechangesindisparitieswerenotsignificantin
thefollowingperiods.ThiswasconsistentwithwhatIfoundfromthemulti‐variatemodel.
ThesamepatternwasobservedforOOPexceeding40%ofhouseholdincome.Thetest
resultsfordisparitychangesbetweenadjacentperiodsareshowninTable6.31.Theonly
significantresultwasbetweenperiods3and4forGroupRR.FromTable6.31,thedisparity
reducedbetweenperiods3and4forGroupRR.Thedisparitywasreversed,sothe
reductionindisparitymeansthatGroupRRwasmoreandmorelikelytohighOOP
comparedwithGroupUU.ThiswasconsistentwithwhatIfoundfromthebasemodel.The
otherchangesindisparitiesbetweenadjacentperiodswerenotsignificant.
131
Table 6.30 DID Analysis Results for OOP Exceeding Certain Percentage of Household Income
OOP>20% household income OOP>40% household income
independent variable Odds Ratio Robust Std. Err. Odds Ratio Robust Std. Err.
disparities in period 1 Group UU 1 n/a
1
n/a
Group RR 0.616*** 0.072
0.666**
0.093
Group RU 0.651** 0.097
0.737
0.129
Group UR 0.890 0.145
0.889
0.175
periods period 1 1 n/a
1
n/a
period 2 1.464** 0.202
1.712***
0.274
period 3 2.680*** 0.300
2.865***
0.378
period 4 2.147*** 0.241
2.373***
0.316
change in disparities Group RR in period 2 0.923 0.157
0.807
0.158
0.131
0.812
0.131
Group RR in period 3 0.954 Group RR in period 4 1.182 0.161
0.995
0.161
Group RU in period 2 1.079 0.235
1.045
0.256
Group RU in period 3 1.274 0.221
1.105
0.224
Group RU in period 4 1.263 0.220
1.101
0.225
Group UR in period 2 0.656 0.174
0.701
0.210
Group UR in period 3 0.971 0.190
0.881
0.210
1.053
0.244
0.221
Group UR in period 4 1.145 Note:1.Significancelevel:***0.001,**0.01,*0.05.
Table 6.31 Test Results for OOP Exceeding Certain Percentage of Household Income (DID Analysis)
Group RR change in disparity in period 2 = Change in disparity in period 3 change in disparity in period 3 = Change in disparity in period 4 Group RU change in disparity in period 2 = Change in disparity in period 3 change in disparity in period 3 = Change in disparity in period 4 Group UR change in disparity in period 2 = Change in disparity in period 3 change in disparity in period 3 = Change in disparity in period 4 Note:1.Significancelevel:***0.001,**0.01,*0.05.
OOP>20% household income OOP>40% household income chi2 Prob>chi chi2 Prob>chi 0.05 4.51* 0.8286 0.0337 0.00 2.98 0.9751 0.0844 0.79 0.00 0.3753 0.9452 0.07 0.00 0.7893 0.9824 2.64 1.24 0.1043 0.2649 0.68 1.01 0.4099 0.3157 Estimateddisparitiesintotalhealthcostsareshownintable6.32.Theestimatesare
fromtheDIDanalysisandbasedon500iterationsofbootstrap.Theresultsshowedsimilar
trendofchangeindisparitiesasthetwo‐partmodel.Thechangesindisparitiesbetween
twoadjacentperiodswerealsosignificant.
132
Table 6.32 Bootstrap Results for Disparities in Total Health Costs (DID Analysis)
disparity Mean Std. Err. [95% Conf. Interval] Group RR period 1 37.136 0.485 36.183 38.089 Group RR period 2 242.434 3.155 236.236 248.632 Group RR period 3 120.964 1.393 118.226 123.701 Group RR period 4 68.133 1.385 65.413 70.853 Group RU period 1 34.784 0.550 33.703 35.865 Group RU period 2 233.906 3.186 227.646 240.165 Group RU period 3 129.381 1.479 126.476 132.287 Group RU period 4 22.255 1.753 18.811 25.699 Group UR period 1 27.080 0.600 25.902 28.259 Group UR period 2 211.009 3.431 204.267 217.751 Group UR period 3 42.139 2.338 37.545 46.733 Group UR period 4 75.695 1.442 72.861 78.529 6.4 Summary of Findings 1.ThedisparityinhavinghighOOPexceeding20%/40%ofhouseholdincomewas
reversed.Ruralresidentsandpeoplewithruralregistrationswerealllesslikelytohave
highOOPexceedingacertainpercentageoftheirhouseholdincomecomparedwithGroup
UU.Thesamewastruewithtotalhealthcarecosts.Ruralresidentsexperiencedlower
healthcarecoststhandidurbanresidents.
2.DisparitiesinhighOOPcostwithGroupUUweremoresignificantinGroupRR
thantheothertwogroups.
3.ThedisparitiesinhighOOPweresignificantlyreducedin2009comparedwith
disparitiesin1997.
4.Thereisnoevidenceshowingthatmorehealthinsurancecoveragehadan
immediateimpactonhighlevelofOOP.
133
5.Disparitiesintotalhealthcostswereassociatedwithinsurancecoverage.
Providingmorehealthinsurancewouldincreasethechanceofhavinganyhealthcost,as
wellastheaverageamountoftotalhealthcosts.
6.Havinghealthinsurancecoveragecouldpartlyexplainthedisparitiesandchanges
indisparities.Providingmoreinsurancecoverageactuallymadepeopleworseoffinterms
ofbeingmorelikelytohavehighOOPexpenditures.
7.Thedisparitiesandchangesindisparitiesweremoresignificantinrichprovinces
thaninpoorprovinces.
8.ThedisparitiesinhighOOPwerenotsignificantinlow‐andhigh‐incomefamilies.
Thechangesindisparitieswereindifferentdirectioninlow‐incomefamilies,althoughthe
changesindisparitieswerenotsignificant.Intermsoftotalhealthcosts,themagnitudeof
disparitieswasgenerallysmallerwithinlow‐incomefamilies.Inlateryears,thedisparities
inlow‐incomefamiliesinGroupRRandGroupRUwerenotsignificant.
134
Chapter 7 Conclusion, Discussion, and Policy Implications 7.1 Conclusion UsingDIDandmultivariateanalysisanddrawingonsevenwavesoflongitudinal
datafromCHNS,Iwasabletoillustratethetrendsofrural–urbandisparitiesinhealthcare
utilizationandcost,inconjunctionwiththemajorhealthinsurancepolicychanges.Iwas
alsoabletoexaminewhetherthegovernment’shealthinsurancepolicychangesaffected
changesindisparities.
Frommyresults,itseemsclearthattherehavealwaysbeenrural–urbandisparities
informalcareutilizationandoutpatientvisits.Urbanresidentsusedformalcareand
outpatientvisitsmorethandidruralresidents.ResultsfromDIDanalysisindicatethatthe
rural–urbandisparitiesinformalcareutilizationandoutpatientvisitweresignificantly
affectedbythepolicychangesinhealthinsurancecoverage.Whenthegovernment
providedmorehealthinsurancecoverageforresidentswithruralregistration,the
disparitiesinformalcareandoutpatientutilizationdecreasedforGroupsURandRR.Only
forGroupRR,thenegativetrendofusinginpatientcarewasalleviatedduringlateryears.
However,therewasnoevidenceshowingthatdisparityininpatientcareutilizationwas
alsocorrelatedwithhealthinsurancecoverage.
The2003policychangeinruralareasamongresidentswithruralhousehold
registrationreducedrural–urbandisparities.Providingmorehealthinsurancecoverageto
residentswithruralhouseholdregistrationreducedthedisparitybetweenGroupsRRand
UR,allowingresidentswithruralhouseholdregistrationtousemoreformalhealthcareand
outpatientvisitscomparedwithGroupUU.Thereformalsoreduceddisparitiesbetween
GroupsRUandGroupUU,suggestingthatpeopleinGroupRUwhohadurbanhousehold
135
registrationbutresidedinruralareas,benefitedfromtheimprovedhealthcare
environment.The2003policychangeinruralareasbroughtthedisparitydowntothe
originallevelin1990s.ThischangeoccurredforbothGroupRRandUR.Aftercontrolling
forinsurancestatus,thepositiveeffectscouldstillbeobservedinthetwogroups.This
findingindicatesthatthepositiveeffectsnotonlycamefrommorehealthinsurance
coveragebutalsofromotherrelatedmeasuresthatimprovedthehealthcareenvironment.
Comparedwiththebasemodel,thechangeindisparitieswasthelargestforGroupRR.This
indicatesthattheGroupRRbenefitedmostfromtheexpandedhealthinsurancecoverage.
Thepolicychangein2003affectedbothpoorandrichprovinces.However,the
expandedhealthinsurancecoveragewasmoreeffectiveinricherprovinces.Thepolicy
effectonpoorerprovincewasassociatedmorecloselywithothermeasuresaimedat
changingtheenvironmentinruralareas,suchasconstructionofbasicmedicalfacilities.
Thepositiveimpactonformalcareandoutpatientutilizationofthe2003policychange
occurredmainlyamonghigh‐incomefamilies.Inthemedium‐incomegroup,therewasno
significantimpact.Inthelow‐incomegroup,thepositiveimpactwasobservedonlyin
GroupUR.
Thedisparityinfinancialriskwasreversed.In2009,thedisparitiesinhighOOP
weresignificantlyreducedfromthelevelin1997.However,therewasnoevidence
showingthatthe2003policychangeinruralareasaffectedrural–urbandisparitiesin
financialrisk.
Therural‐urbandisparityintotalhealthcarecostswasalsoreduced.Whenthe
governmentprovidedmorehealthinsurancecoverageinurbanarea,therural‐urban
136
disparityinhealthcarecostsincreased,andviceversa.Thiswasconsistentwiththefinding
forhealthcareutilization.Morehealthinsurancecoverageinruralareasledtoasmaller
rural–urbandisparityinhealthcareutilization.
Inordertotestthesensitivityofresults,Ialsoperformedsensitivityanalysisby
droppingtherichestandthepoorestprovincesfromthesample.ForbothhighOOPand
totalhealthcosts,sensitivityanalysisshowedthatthedisparitiesandchangesindisparities
weremoresignificantintherichprovinces.
Ifurtherexaminedthedifferentimpactsfordifferentincomegroups.Thedisparities
inhighOOPwerenotsignificantforthelow‐andhigh‐incomefamilies.Intermsoftotal
healthcosts,themagnitudeofthedisparitieswasgenerallysmallerwithinthelow‐income
families.Inlateryears,thedisparitiesinlow‐incomefamiliesbetweenGroupRRandUUor
betweenGroupRUandUUwerenolongersignificant.Thisindicatesthatthedisparitiesin
totalhealthcostsfinallydiminishedinlow‐incomefamilies.Low‐incomefamiliesinGroups
RRandRUhadsimilarlevelsoftotalhealthcoststothecostsoflow‐incomefamiliesin
GroupUU.
7.2 Discussion 7.2.1ComparingWiththePublishedResearch
Myfindingsagreewithpreviousresearcherswhoclaimedtherearerural–urban
disparitiesinhealthcareutilization.Myresearchfurthershowsthatthedisparitieswere
themostsignificantwithinruralresidentswithruralregistration,andthedisparitywas
alleviatedafterasetofhealthpolicychanges.Regardinghealthcarecosts,myresearch
conclusionsagreewiththoseofWagstaff&Lindelow(2009),whoclaimedthatproviding
137
morehealthinsurancecoveragedoesnotnecessarilymeanmorefinancialprotection.
Instead,althoughnotstatisticallysignificant,IfoundthedisparityinhighOOPwas
reversed.RuralresidentswerelesslikelytohavehighOOPcomparedwithurban
counterparts.Wagstaff&Lindelow(2009)explainedthiscasebynotingthebalance
betweenbetterhealthandhighercosts.Thiscouldrepresentapossibleexplanationofthe
Chinesecase.Theinsuredtendtousemoreformalhealthcare,andtheirtotalhealthcosts
arealsohigh.However,thebenefitcoveragefromNRCMislimitedforoutpatientvisits,and
thereimbursementcapisrelativelylow.Therefore,thebenefitcoveragemaybeenoughto
encouragetheinsuredtousemoreformalcarebutnotsufficienttoprovideenough
financialprotection.Thisexplanationisalsosupportedbythefindingsfromtheanalyses
forhealthcareutilizationandtotalhealthcarecosts.
7.2.2Strengths
1. Myresearchusedanewclassificationofruralandurban.Byclassifyingthe
respondentsintofourcategories,Iwasabletoobtainamoreaccurateestimate
oftheeffectfrominsurancecoverageexpansion,aswellastoexaminethe
impactoftheresidingenvironment.
2. Myresearchprovidedaholisticpictureoftrendsofrural–urbandisparitiesin
healthinsurancecoverage,healthcareutilization,andhealthcarecostinChina
over20yearsoftherapid‐reformera,whichencompassedthreemajorhealth
insurancepolicychanges.
3. Inmyresearch,Iexaminedthecorrelationbetweenexpansionofinsurance
coverageandhealthcareutilizationandhealthcarecost,contributingnew
knowledgetoatopicnotwellstudied.
138
4. MyDIDmodelincludedthreemajorpolicychangesinChina,providingmore
thoroughevidenceontheimpactofpolicychangeinhealthinsurancecoverage
onrural–urbandisparitiesinChina.
5. Iexploredthepolicyeffectsindifferentsubgroupsofthepopulation,providing
newevidencetoanswertheresearchquestionsandenablingpolicymakersto
examinepolicyeffectsatadeeperandmoredetailedlevel.
7.2.3Limitations
Fivelimitationsshouldbementioned.First,theremightbeanunderestimationof
thepolicyeffect,sincethedefinitionofrural/urbanresidentsandthedefinitionof
rural/urbanhouseholdregistrationwerenotconsistent.Someoftheurbanresidentsheld
ruralhouseholdregistration,andthesamewastrueforruralresidents.Therefore,no
matterthedefinitionused,Iwasnotabletoprovideapreciseestimateofthepolicyeffect
onrural–urbandisparities.
Second,thethreemajorpolicychangesfocusedonpublichealthinsurancecoverage,
andinvolvedprovidingmorecoveragetocertaingroupsofpeopleeachtime.However,
duringthesametimeperiods,therewereotherpolicychanges,whichalsoaffectedrural
andurbanresidentsdifferently,suchasconstructionofhealthfacilities,trainingofhealth
workers,andchangesindrugpolicy.Duetothemethodology,Icouldnotseparatethe
effectofpolicyexpansionofhealthinsurancecoverage.
Third,mystudydidnotdistinguishtheeffectsbetweenthe2007insurance
expansionforurbanresidentsandthe2009nationalhealthcarereformduetoalackof
datain2008.
139
Fourth,Ididnotuseanationallyrepresentativesample.
Fifth,inpatientcareutilizationconstitutedaverylowpercentageinmysample;thus,
Iwasnotabletofullyexaminethechangeofdisparityininpatientcareutilization.
Finally,Istudiedonlyhealthcareutilizationandcosts;otherrelatedareassuchas
healthoutcomeandmortalitywereoutsidethescopeofthisproject.
7.2.4FutureDirections
Futureresearchshouldinvolvethefollowing:
1. Examinetheeffectofdifferentpolicychangesotherthaninsuranceusingmore
detaileddata.
2. Futurestudiesneedtodifferentiatetheeffectsofthe2007insuranceexpansion
andthe2009nationalhealthcarereform.
3. Useanationallyrepresentativesampletoestimatetheaveragepolicyeffectin
China.
4. Conductmoreresearchondisparitiesininpatientcareutilization.
5. Studydisparitiesinotherhealthcare‐relatedareas,suchashealthstatusand
mortality.
7.3 Policy Implications Threeimportantpolicyimplicationscanbedrawnfromtheresultsofthisstudy.
First,morehealthinsuranceandbetterbenefitcoverageisneeded.AsIfoundfromthe
analysis,thepolicychangesthatprovidedincreasedhealthinsurancecoveragetorural
groupsreducedrural–urbandisparitiesinhealthcareutilizationandtotalhealthcarecosts.
However,currentpolicyhasnotbeenabletoreducetherural–urbandisparityin
140
healthcaretotheoriginal1980slevel.Disparitiesstillexistinthestudiedareas.Therefore,
policymakersshouldprovidemorehealthcarecoverageandhealthcareresourcestorural
areastofurtherreducethedisparity.Ialsofoundthatruralgroupswereinitiallylesslikely
tohavehighOOP,comparedtotheurbangroups.Ruralgroupsalsohadlowertotalhealth
coststhandidurbangroups.Whenthegovernmentprovidedmorehealthinsuranceto
ruralgroups,thedisparitiesdecreasedinhighOOPaswellasintotalhealthcarecosts.
Insurancefailedtoprovidefinancialprotectioninthiscase.Thisresultmayindicatethat
thebenefitcoveragewasnotsufficient.Therefore,betterbenefitcoverageshouldbe
providedtoruralgroups.
Second,inordertoreducerural–urbandisparities,policymakersshouldalso
considerpolicydirectionsotherthanofferingincreasedhealthinsurancecoverage,suchas
constructionofhealthcarefacilities,healtheducation,andsoon.Inmyanalysis,Ifound
thattheenvironmentwasalsoimportantbecausetheenvironmentdeterminedthe
resourcesapersonreceived.Thepolicyactionschangedtheenvironmentandprovided
morehealthcareresourcestoruralresidents.Theseactionsareimportantpolicy
alternativesinreducingrural–urbandisparities.
Third,disadvantagedgroupsshouldreceivemoreattention.Intermsofhealthcare
utilizationaswellasintotalhealthcosts,currentpolicyaffectsrichprovincesmorethanit
affectspoorprovinces.Therefore,newpolicycouldprovidemorebenefitcoveragetorural
residentsinpoorprovinces.Thepositiveimpactonhealthcareutilizationofthe2003
policychangeoccurredmainlyinhigh‐incomeandmedium‐incomegroups.Therefore,new
policychangesshouldfocusmoreonlow‐incomegroupsinruralarea.Intermsoffinancial
141
protection,high‐incomegroupsalsobenefitedmorethandidlow‐incomegroups.When
designingnewhealthinsurancepolicy,policymakersshouldprovidedifferentbenefit
coveragetodifferentincomegroups,andlow‐incomegroupsshouldreceivemorecoverage.
AsdiscussedinChapter2,thenewroundofhealthcarereformisintendedto
provideuniversalcoveragetoallresidents;thefocusofthenewreformisthe
disadvantagedpopulation.Theseactionsareallconsistentwithmyresearchfindings.
142
Appendix TableA1Majorhealthinsuranceschemes
Urban Employee Basic Medical Insurance Launching Time 1998 2007 2003 Urban Employee Urban Resident who are not covered by UEBMI Rural Resident Risk Pools County level City level City level Premium Paid By Employer and Employee Government and insured individual Government and insured individual Annual Premium Level (2012) Employer pays 6% of employee's wage, employee pays 2% of the wage At least 300 CNY, in which At least 300 CNY, in government pays 240 CHY/ which government pays insured 240 CHY/ insured Reimbursement Cap (2012) 6 times of local average salary (at least 60000 CNY) 6 times of local per capita income (at least 60000 CNY) 8 times of local per capita income (at least 60000 CNY) Covered Services Inpatient Services Covered Covered Covered Outpatient Services for Catastrophic Illnesses Covered Covered Covered General outpatient services Covered Limited and vary by location Limited and vary by location 237 195 New Rural Cooperative Medical Insurance Insured Population Number of Insured at 2010 Year‐end (Million) Urban Resident Basic Medical Insurance 143
836 TableA2GDPin2012ofthesampledprovinces
Province GDP in 2012 (Unit: billion Chinese Yuan) Jiangsu 5405.8 Shandong 5001.3 Henan 3000.0 Liaoning 2480.1 Hubei 2225.0 Hunan 2215.4 Shanghai 2010.1 Beijing 1780.1 Heilongjiang 1369.2 Guangxi 1303.1 Chongqing 1145.9 Guizhou 680.2 144
Reference Abadie, A. (2005). "Semiparametric Difference‐in‐Differences Estimators." The Review of Economic Studies 72(1): 19. Akin, J. S., W. H. Dow, et al. (2004). "Did the distribution of health insurance in China continue to growless equitable in the nineties? Results from a longitudinal survey." Social Science & Medicine 58: 12. Andersen, R. M. (1995). "Revisiting the Behavioral Model and Access to Medical Care: Does It Matter?" Journal of health and social behavior 36(1): 10. Athey, S. and G. W. Imbens (2002). Identification and Inference in Nonlinear Difference‐In‐Differences Models. SIEPR Discussion Paper Stanford,CA, Stanford Institute for Economic Policy Research: 61. Bertrand, M., E. Duflo, et al. (2004). "How Much Should We Trust Differences‐in‐Differences Estimates?" The Quarterly Journal of Economics: 27. Card, D. and A. B. Krueger (2000). "Minimum Wages and Employment: A Case Study of the Fast‐Food Industry in New Jerseyand Pennsylvania: Reply." The American Economic Review 90(5): 5. Conley, T. and C. Taber (2005). Inference with “Difference in Differences” with a Small Number of Policy Changes. NBER Technical Working Paper. China Ministry of Health (2004). "The Third National Health Services Survey Report" (In Chinese). China Ministry of Health (2005). "China Statistical Yearbook 2005" (In Chinese). China Ministry of Health (2010). "China Statistical Yearbook 2010" (In Chinese). China State Council (1998). "State Council Policy Document No. 44, 1998" (In Chinese). China State Council (2002). "State Council Policy Document No. 13, 2002" (In Chinese). China State Council (2007). "State Council Policy Document No. 20, 2007" (In Chinese). China State Council (2009). "State Council Policy Document No. 6, 2009" (In Chinese). Fang, H., J. Chen, et al. (2009). "Explaining Urban‐Rural Health Disparities in China." Medical Care 47(12): 1209‐1216. Feng, X. L., S. Guo, et al. (2011). "Regional disparities in child mortality within China 1996‐2004: epidemiological profile and health care coverage." Environmental health and preventive medicine 16(4): 209‐216. Gao, J., J. H. Raven, et al. (2007). "Hospitalisation among the elderly in urban China." Health policy 84(2‐
3): 210‐219. Gao, J., S. Tang, et al. (2001). "Changing access to health services in urban China: implications for equity." Health policy and planning 16(3): 11. Hansen, C. B. (2007). "GENERALIZED LEAST SQUARES INFERENCE IN PANEL AND MULTILEVEL MODELS WITH SERIAL CORRELATION AND FIXED EFFECTS." Journal of Econometrics 140(2): 40. Hart, L. G., E. H. Larson, et al. (2005). "Rural Definitions for Health Policy and Research." American Journal of Public Health 95(7): 7. Hsiao, W. C. (1984). "Transformation of health care in China." The New England journal of medicine 310(14): 932‐936. Hu, S., S. Tang, et al. (2008). "Reform of how health care is paid for in China: challenges and opportunities." The Lancet 372: 8. Jian, W., K. Y. Chan, et al. (2010). "China's rural‐urban care gap shrank for chronic disease patients, but inequities persist." Health affairs 29(12): 2189‐2196. Lei, X. and W. Lin (2009). "The New Cooperative Medical Scheme in rural China: does more coverage mean more service and better health?" Health economics 18 Suppl 2: S25‐46. 145
Li, D. (2008). "China's great economic transformation." Choice: Current Reviews for Academic Libraries 46(1): 157‐157. Liu, M., Q. Zhang, et al. (2007). "Rural and Urban Disparity in Health Services Utilization in China." Medical Care 45(8): 8. Liu, X., S. Tang, et al. (2012). "Can rural health insurance improve equity in health care utilization? A comparison between China and Vietnam." International journal for equity in health 11: 10. Liu, Y. (2002). "Reforming China’s urban health insurance system." Health Policy 60: 18. Liu, Y. (2004). "Development of the rural health insurance system in China." Health policy and planning 19(3): 159‐165. Liu, Y., W. C. Hsiao, et al. (1999). "Equity in health and health care: the Chinese experience." Social Science & Medicine 49: 8. Long, Q., T. Zhang, et al. (2010). "Utilisation of maternal health care in western rural China under a new rural health insurance system (New Co‐operative Medical System)." Tropical medicine & international health : TM & IH 15(10): 1210‐1217. Lu, C., Y. Liu, et al. (2012). "Does China's Rural Cooperative Medical System Achieve Its Goals? Evidence from the China Health Surveillance Baseline Survey in 2001." Contemporary Economic Policy 30(1): 93‐112. Manning, W. G. and J. Mullahy (2001). "Estimating log models: to transform or not to transform?" Journal of health Economics 20: 34. Meng, Q. and S. Tang (2010). "Universal Coverage of Health Care in China: Challenges and Opportunities." World Health Report (2010) Background Paper: 23. Meng, Q., J. Zhang, et al. (2012). "One country, two worlds ‐ the health disparity in China." Global public health 7(2): 124‐136. OECD (2011). "Divided We Stand – Why Inequality Keeps Rising"
(http://www.oecd.org/social/inequality.htm) Pan, X., H. H. Dib, et al. (2009). "Absence of appropriate hospitalization cost control for patients with medical insurance: a comparative analysis study." Health economics 18(10): 1146‐1162. State Council Evaluation Group for the URBMI Pilot Program (2008). "Report on URBMI Pilot Programs". (In Chinese). Sun, X., S. Jackson, et al. (2009). "Catastrophic medical payment and financial protection in rural China: evidence from the New Cooperative Medical Scheme in Shandong Province." Health economics 18(1): 103‐119. Tang, S., Q. Meng, et al. (2008). "Tackling the challenges to health equity in China." The Lancet: 9. Wagstaff, A., M. Lindelow, et al. (2009). "Extending health insurance to the rural population: an impact evaluation of China's new cooperative medical scheme." Journal of health economics 28(1): 1‐19. Wang, H., W. Yip, et al. (2005). "Community‐based health insurance in poor rural China: the distribution of net benefits." Health policy and planning 20(6): 366‐374. Wang, S. (2004). "China’s Health System: From Crisis to Opportunity." Yale‐China Health Journal 3: 45. Xu, H. and S. E. Short (2011). "Health insurance coverage rates in 9 provinces in China doubled from 1997 to 2006, with a dramatic rural upswing." Health affairs 30(12): 2419‐2426. Xu, L., Y. Wang, et al. (2007). "Urban health insurance reform and coverage in China using data from National Health Services Surveys in 1998 and 2003." BMC health services research 7: 37. Yip, W. and W. C. Hsiao (2008). "The Chinese health system at a crossroads." Health affairs 27(2): 460‐
468. Yu, B., Q. Meng, et al. (2010). "How does the New Cooperative Medical Scheme influence health service utilization? A study in two provinces in rural China." BMC health services research 10: 116. Zhao, Z. (2006). "Income Inequality, Unequal Health Care Access, and Mortality in China." Population and Development Review 32(3): 23.
146
This product is part of the Pardee RAND Graduate School (PRGS) dissertation series.
PRGS dissertations are produced by graduate fellows of the Pardee RAND Graduate
School, the world’s leading producer of Ph.D.s in policy analysis. The dissertation has
been supervised; reviewed; and approved by the faculty committee composed of
Hao Yu (Chair), Emmett Keeler, and Gema Zamarro.
PA R D E E R A ND G R A D UATE S C H OOL
www.rand.org
RGSD-345
Download

The RAND Corporation is a nonprofit institution that helps improve... decisionmaking through research and analysis.