(cid:2)BUILDING NEW KNOWLEDGE SUPPLEMENT Moving towards universal coverage in South Africa? Lessons from a voluntary government insurance scheme Veloshnee Govender1*, Matthew F. Chersich2,3, Bronwyn Harris2, Olufunke Alaba1, John E. Ataguba1, Nonhlanhla Nxumalo2 and Jane Goudge2 1Health Economics Unit, SchoolofPublic Health and FamilyMedicine, Faculty ofHealth Sciences, UniversityofCapeTown,SouthAfrica;2CentreforHealthPolicy,SchoolofPublicHealth,Facultyof Health Sciences,University oftheWitwatersrand, Johannesburg, South Africa; 3Department of Obstetrics andGynaecology, International Centre forReproductive Health, GhentUniversity, Ghent,Belgium Background: In 2005, the South African government introduced a voluntary, subsidised health insurance schemeforcivilservants.Inlightoftheglobalemphasisonuniversalcoverage,empiricalevidenceisneededto understandtherelationshipbetweennewhealthfinancingstrategiesandhealthcareaccesstherebyimproving globalunderstandingof these issues. Objectives: This study analysed coverage of the South African government health insurance scheme, the populationgroupswithlowuptake,andtheindividual-levelfactors,aswellascharacteristicsofthescheme, thatinfluenced enrolment. Methods:Multi-stagerandomsamplingwasusedtoselect1,329civilservantsfromthehealthandeducation sectorsinfourofSouthAfrica’snineprovinces.Theywereinterviewedtodeterminefactorsassociatedwith enrolmentinthescheme.Theanalysisincludedbothdescriptivestatisticsandmultivariatelogisticregression. Results:Notwithstandingtheavailabilityofanon-contributoryoptionwithintheinsuranceschemeandaccess toprivately-providedprimarycare,aconsiderableportionofsocio-economicallyvulnerablegroupsremained uninsured (57.7% of the lowest salary category). Non-insurance was highest among men, black African or coloured ethnic groups, less educated and lower-income employees, and those living in informal-housing. Therelativelypooruptakeofthecontributoryandnon-contributoryinsuranceoptionswasmostlyattributed to insufficient information, perceived administrative challenges of taking up membership, and payment costs. Conclusion:Barrierstoenrolmentincludeinsufficientinformation,unaffordabilityofpaymentsandperceived administrativecomplexity.Achievinguniversalcoveragerequiresgoodphysicalaccesstoserviceprovidersand appropriatebenefitoptionswithinpre-paymenthealthfinancingmechanisms. Keywords: healthinsurance;civilservants;health-financereforms;universalcoverage;SouthAfrica Received:31 July 2012; Revised:11 November 2012; Accepted:12 November 2012; Published:24 January 2013 In 2005, member states of the World Health Organi- introduction of pre-payment schemes with tax-based zation (WHO) committed themselves to develop- funding or compulsory or voluntary health insurance ing health financing systems that would enable contributions (2). universal coverage (UC) by ensuring access to adequate Given the limits of, and competing demands on, tax- health care at an affordable cost for all citizens (1). based funding (3), the focus in many low- and middle- Although there is no one clear path to UC, the World income countries, has been on contributory health Health Report 2010 describes several strategies for insurance schemes (where employees contribute toward expanding access to care. These include the removal the premium). Nonetheless, there is the recognition that of direct payments, particularly user fees, and the for some groups, these contributions will need to be GlobHealthAction2013.#2013VeloshneeGovenderetal.ThisisanOpenAccessarticledistributedunderthetermsoftheCreativeCommonsAttribution-109 Noncommercial3.0UnportedLicense(http://creativecommons.org/licenses/by-nc/3.0/),permittingallnon-commercialuse,distribution,andreproductionin anymedium,providedtheoriginalworkisproperlycited. Citation:GlobHealthAction2013,6:19253-http://dx.doi.org/10.3402/gha.v6i0.19253 (pagenumbernotforcitationpurpose) VeloshneeGovenderetal. partially- or fully-subsidised by government (2). Some Box1.Timelineofhealth-financingpolicyinitiatives countries such as Iran are expanding coverage through andproposals since1994 voluntary, contributory insurance schemes (4), while Ghanaisoptingformandatory insurance(5).Countries 1994: African National Congress (ANC) National with existing insurance schemes, have attempted to Health Plan recommended that a Commission of expand coverage through the provision of lower cost Inquiry be appointed to investigate the feasibility of alternatives with similar benefit packages, but possibly a National Health Insurance (NHI) Fund (17). with limited choice of providers. The Seguro Popular 1994: National Department of Health’s Health Care Programme in Mexico and the UC scheme in Thailand, Finance Committee put forward three possibleman- are examples where the contributions of low-income datory insurance options, including NHI (18). individuals and families are subsided by government 1995: Hospital Strategy Project, initiated by the (6,7). National Ministry, tasked with setting out a frame- In South Africa the tax-funded public health system, work for the development and restructuring of the with free primary health care and minimal charges for public hospital sector (19). inpatient care, provides some form of UC. However, 1995:CommitteeofInquiryintoaNHISystem(20). despite substantial transformation of the public health 1997:NationalDepartmentofHealthreleasespolicy systempost-apartheid(8,9),perceptionsandexperiences document on Social Health Insurance Scheme for of poor quality of public health care persist (10). These formal sector employees (21). arise from a range of factors, including the quadruple 1997: White Paper on the transformation of the diseaseburden(11),poorstewardship,andinefficientuse health system in South Africabuilt upon the ANC’s of resources (8). This has lead to increased utilisation of 1994 Health Plan (22). private providers for primary health care. However, only 2002: The Committee of Inquiry into a Comprehen- the wealthiest 16% of the population can afford private sive System of Social Security for South Africa health insurance to cover the costs of private-sector recommends that South Africa move toward a NHI services (12). For the uninsured, direct payments are system (23). often catastrophic in nature (above 10% of household expenditure) (13), contributing to household poverty 2004: Ministerial Task Team on SHI recommended (14, 15). Therefore, despite a tax-funded public health implementation of SHI for the formally employed, caresystemavailabletoall,markedinequalitiesinhealth since it did not consider NHI feasible in the short care access persist (8, 16). term. It is against this backdrop that the goal of UC has 2005: Ministerial Task Team commissioned an in- taken centre stage in several health-financing reform vestigation into low-income medical schemes (24). policyproposalsandinitiativessincedemocracyin1994. 2005: Introduction of the Government Employee Box1showsatimelineofpolicyinitiativesandproposals Medical Scheme (referred to as the government (17(cid:2)26). scheme in the article), restricted to public-sector As indicated in these timelines, the earlier debates employees. considered the option of a NHI scheme which, by 2007: A policy resolution committed the ANC to definition ‘covers the entire population irrespective of introduce NHI (25). whether they have personally contributed to the scheme 2011: NHI Green Paper released by government or not’ (12, p. 73). Around 2005 strategies for insurance detailing a 14-year plan towards NHI (26). coverage of low-income households were considered in the country (24). In 2005, the government (as an employer) implemented a health insurance scheme (27), The scheme is heavily subsidised, particularly for low- restricted to government employees, that aimed to income members, to encourage enrolment and so extend achieve greater pooling of funds across this segment coverage.1 Employees appointed from 1 July 2006 on- of the employed population. Post 2009 the debate has wards were only eligible for the government subsidy if shiftedtotheimplementationofaNHIsystem,thataims they joined the government scheme and not another to strengthen the public health care system and ensure health insurance scheme.2 adequate provision offunding (26). South Africa’s government employees’scheme intends topoolresourcesfromabroadrangeofcivilservantsand 1The lowest cost benefit option is fully subsidised for those in the aims to attract members from all income groups. The twolowestsalarycategories;fortheotherfourbenefitoptions,the intention of a designated network of private general governmentpays75%oftheemployee’stotalmonthlycontribution, subjecttoanupper-limit. practitioners and private hospitals is to expand access 2Employees appointed prior to 2006 still received a subsidy for to benefits for low-income government employees. membershipofanyinsurancescheme. 110 Citation:GlobHealthAction2013,6:19253-http://dx.doi.org/10.3402/gha.v6i0.19253 (pagenumbernotforcitationpurpose) HealthinsurancecoverageofcivilservantsinSouthAfrica This study analysed coverage of the government thosewhowereuninsuredpriortojoiningthegovernment health insurance scheme, the population groups with scheme. The survey questionnaire included questions to low uptake, and the individual-level factors, as well as assesspossibleconsumerinertia,arisingfromthetransac- characteristics of the scheme that influenced enrolment. tioncostsofeitherswitchingfromoneschemetoanother The study findings are used to highlight lessons for (28)orfromjoiningaschemehavingnotbeenpreviously othercontributoryschemeswhichaimtoencompassboth insured. The questionnaire also included reasons under- high- and low-income population groups. lying inertia, specifically the lackof aperceived need for insuranceandadministrativecomplexityofthescheme. Adverseselection,arisingfromthetendencyforpeople Methods with perceived low risk (younger, healthier, low-income) to avoid insurance coverage is another challenge for Samplinganddatacollection voluntaryschemes(29).Inrecognitionofthis,age,health In 2008(cid:2)09, 1,329 currently employed civil servantswere interviewedacrossfourofSouthAfrica’snineprovinces.3 status and income (indicated by salary) were assessed as key potential determinants of the decision to take Health and education, two of the largest public sectors, insurance. The choice of socio-demographic (age, sex, were selected for the survey. Provinces were chosen on race, location, education, marital status and housing), thebasisofbeingurban,havingagreaterdistributionof economic(income)andhealthstatusvariableswasguided privateprovidersaswellasrelativelywell-resourcedpublic by previous research examining determinants of health health care facilities (Gauteng and Western Cape), and insuranceownershipinvoluntaryschemesinSouthAfrica being predominantly rural with few private facilities (30) and internationally (31(cid:2)34).4 Civil servants were and less-resourced public facilities (KwaZulu-Natal and classified by skill level into five categories (lower skilled, North West) in order to assess variation in enrolment skilled, highly-skilled, supervisory and senior manage- relatedtogeographicalaccess.Theminimumsamplesize ment);thesecategoriesdeterminesalarylevelswithinthe per province was 245 and this was increased to 309 to civilservice. allowforpossibleincompletequestionnaires. Data were double-entered by an independent survey Multi-stage random sampling was used. First, the company, cross-checked by the research team and then numberofhealthandeducationemployeestobesampled analysed using Stata† 10 (Stata Corporation, College in each salary category was determined by their relative Station, TX, United States). In addition to descriptive proportion in each province. Second, districts in each statistics of the uninsured and insured populations, the provincewereselectedwithaprobabilityproportionateto respondent’s decision to enrol in the government scheme number of employees, following which 15 schools and was modelled using multivariate logistic regression. four hospitals within each of the selected districts were The dichotomous dependent variable was enrolment in randomly selected. Finally, within the selected schools the government scheme (combining the ‘previously- and hospitals, a sampling frame was constructed of all insured’ and ‘newly-insured’) and the explanatory varia- employees, stratified bysalarycategory, to allow specific bles were categorical and included socio-demographics, quotasofinterviewstobeconductedacrossthedifferent salary level and health status. Variables associated with salary categories. These employees were then invited for government scheme membership in univariate analysis an interview until the required number in each salary (pB0.1) were included in the initial multivariate model category was reached. Study procedures received ethics in addition to important potential confounders such as clearance from the Universities of Cape Town and the gender, and retained if their removal markedly altered Witwatersrand, as well as relevant Provincial Depart- themodelfit.Educationlevelwasexcludedasoneofthe ments of Health. All respondents provided informed independent variables since it correlated closely with signed consent. salary level. Studyvariablesanddatamanagement Information was collected on health insurance uptake, Results including membership of the government and other schemes, factors influencing membership of the govern- Descriptionofstudypopulation mentscheme,choiceofbenefitoptionandthereasonsfor Two-thirds of respondents worked in the education such choice. Those who transferred to the government sector, and one third in health. More than half (58.6%) scheme from another medical scheme were classified werefemale.Athirdofrespondentswere30(cid:2)39yearsand as ‘previously-insured’, while ‘newly-insured’referred to 4Asbrokerscouldnotmakeanyfinancialgainsfromenrollingcivil 3Retiredcivilservantswerenotincludedinthestudysample.This servantsinthegovernmentscheme,weconsideredthemunlikelyto population was initially excluded from participation in the haveanyroleorinfluenceonthedecisiontoenrol.Hencetheywere governmentscheme,butlaterincluded. notconsideredinthisstudy. 111 Citation:GlobHealthAction2013,6:19253-http://dx.doi.org/10.3402/gha.v6i0.19253 (pagenumbernotforcitationpurpose) VeloshneeGovenderetal. a similar proportion was 40(cid:2)49 years (35.2%). Approxi- lowest salary category were more likely to have enrolled mately two-thirds had tertiary-level education, while in the government scheme (Table 2). Enrolment in the 12.5% had only primary or no education. Median total government scheme was 72% lower among those (cid:2)60 monthlyhouseholdexpenditurewasUS$533.3.5Halfthe years and 43% lower in those 40(cid:2)49 years, compared respondents were classified as highly skilled employees with those aged 20(cid:2)29 years. Similarly, those living (53.3%);almostathird(31.1%)asbeinglowskilled;and in the relatively rural provinces (KwaZulu-Natal and only 3.1% were senior managers. Only 2.6% reported North West) were less likely to be insured under the theirhealthasbeingpoororverypoor,butalmostathird government scheme than the urban provinces. was taking chronic medication. Choiceofbenefitoptionunderthegovernment Insurancestatusofpublic-sectoremployees scheme Three-quarters of respondents (74.3%) were insured The government scheme has five benefit packages rang- (witheitherthegovernmentoranotherinsurancescheme ing from low-cost options, which are fully subsidised Table1)and25.7%areuninsured.Lessthanhalf(41.9%) for those in the lower income categories, to high-cost of the insured are members of the government scheme; packagesthatareincreasinglycomprehensiveintherange morethanhalfoftheinsured(58.1%)belongedtoanother of services covered. The two lower-cost options (options scheme.Ofthemembersofthegovernmentscheme,29.5% 1 and 2) offer members outpatient benefits through a had insurance prior to joining the government scheme, limited network of private healthcare providers (general while 12.4% were newly-insured. Of the 9.2% of the practitioners, dentists or optometrists). These two op- respondents who joined the civil service after 2006, only tionsdifferwith respect tohospital benefits;inoption1, 18.9%hadenrolledinthegovernmentscheme.Afurther members have access to a networkof state hospitals and 16.4% were members of another scheme, considerably option 2 to a limited private hospital network.6 Options fewer than amongst civil servants employed before 3, 4 and 5 allow access to any private hospital. 2006 where almost half were enrolled in another scheme Option4wasthemostpopularbenefitoption,withthe (46.0%);theremainder(64.7%)wereuninsured. proportionselectingthisoptionrisingassalaryincreased The insured (those belonging to the government or (Table 3). However, a substantive proportion (28.3%) of other schemes) were more likely to be above 40 years, those in the highest salary category selected comprehen- women,educatedattertiarylevel,livinginformalhousing, sive option 5. Of the two low-cost options, the fully Indian/Asian or white, in the higher salary categories subsidisedoption1wasmorepopularamongstthelowest (highly skilled to senior management) and living in a salary employees (19.4%). household with an individual on chronic medication. Self-assessed health status was not a predictor of health Factorsaffectinguptakeofthegovernmentscheme insurance. Inunivariate analysis, all socio-economic and For the insured, the most important reasons for joining demographic variables, besides gender, were associated the government scheme across all salary categories were with uptake of the government scheme (either newly or the affordability of member contributions (67.4%), per- previouslyinsured). ceptions that it had better benefits and covered more Taking up insurance for the first time (newly insured) dependents (37.9%) (Table 4). was highest amongst those aged 20(cid:2)29, females, single Amongst the uninsured, 40.2% of those in the lower- people, black Africans, and those living in informal salary categories (lower skilled and skilled) cited lackof housing or with a lower-income (salary categories lower affordability as a reason for not joining and almost a skilled and skilled) (Table 1). In contrast, factors asso- third of all the uninsured across the three lower-salary ciated with switching from a previous scheme to the categories noted that they would join if the scheme was government scheme were having a skilled job, age 50(cid:2)59 made more affordable. Among respondents, 28.9% of years, being divorced, separated or widowed, secondary thoseinthelowerskilledand21.7intheskilledcategories educationlevel,livinginformalhousingandinanurban statedperceivedadministrativecomplexitieshaddeterred province (i.e. Gauteng or the Western Cape). them from joining the scheme, while 23.7% of lower Multivariate analysis allowed for the simultaneous skilled and skilled) stated lackof information about the examination of the effect of several demographic, socio- scheme asimportant obstacles to enrolment. Among the economic and health status factors on the uptake of the uninsured,26.7%ofthoseinthehigher-salarygrades,did governmentinsuranceschemebyboththepreviouslyand not join because they believed they did not need health newly insured (Table 2). Multivariate analysis showed insurance.However,morethanathirdofthemsaidthey that employees who were female, no longer married or would join if they had a health need. cohabiting (i.e. divorced, separated, widowed), or in the 6Networks refer to designated health-care providers contracted 51US$(cid:3)7.5SouthAfricanRand. throughthegovernmentschemetoprovideservicestomembers. 112 Citation:GlobHealthAction2013,6:19253-http://dx.doi.org/10.3402/gha.v6i0.19253 (pagenumbernotforcitationpurpose) HealthinsurancecoverageofcivilservantsinSouthAfrica Table1. Associationsbetweenhealthinsurance,demographiccharacteristicsandincomelevelamongpublic-sectoremployees in SouthAfrica Insurancescheme(%) Governmentscheme Variable(n) Insured%(n) Newlyinsured Previouslyinsured Otherschemes p Age(years) 20(cid:2)29(141) 53.9(76) 39.5 19.7 40.8 B0.001 30(cid:2)39(402) 74.4(299) 18.7 26.4 54.9 40(cid:2)49(468) 78.6(368) 6.3 31.3 62.5 50(cid:2)59(268) 76.5(205) 5.9 35.1 59.0 ]60(46) 80.5(37) 2.7 27.0 70.3 Sex Female(778) 78.8(613) 13.4 28.2 58.4 0.317 Male(548) 67.7(371) 10.8 31.8 57.4 Maritalstatus Married/cohabiting(806) 77.8(627) 7.3 30.6 62.1 B0.001 Div./sep./widow(149) 77.2(115) 8.7 38.3 53.0 Single(375) 65.6(246) 26.8 22.8 50.4 Educationlevel None/prim.comp(168) 48.8(82) 18.3 30.5 51.2 B0.001 Incomp.secondary(102) 66.7(68) 14.7 33.8 51.5 Comp.secondary(184) 63.0(116) 23.3 39.7 37.0 Diploma(360) 79.2(285) 14.7 27.7 57.6 Degree(516) 84.7(437) 6.4 27.2 66.4 Housing Formal(1272) 75.9(966) 11.8 29.8 58.4 0.003 Informal(50) 40.0(20) 40.0 20.0 40.0 Race BlackAfrican(858) 71.1(610) 16.1 27.9 56.0 B0.001 Coloured(253) 70.7(179) 10.0 33.0 57.0 Indian/Asian(77) 87.0(67) 6.0 34.3 59.7 White(132) 93.9(124) 0.8 31.5 67.7 Salarycategory Lowerskilled(168) 42.3(71) 31.0 22.5 46.5 B0.001 Skilled(246) 60.6(149) 27.5 37.6 34.9 Highly-skilled(709) 81.2(576) 9.6 28.3 62.1 SupervisoryandSeniorManagement(206) 92.7(191) 2.1 29.8 68.1 Province Gauteng(344) 70.4(242) 17.7 32.3 50.0 B0.001 KwaZulu-Natal(310) 72.6(225) 17.3 24.9 57.8 NorthWest(329) 82.4(271) 5.9 29.2 64.9 WesternCape(343) 72.6(249) 9.6 31.8 58.6 Self-assessedhealthstatus Excellent(320) 71.3(228) 17.6 29.8 52.6 0.021 Good(633) 75.8(480) 10.2 31.0 58.8 Average(342) 74.5(255) 11.0 25.5 63.5 Poor(34) 73.5(25) 20.0 40.0 40.0 Individualonchronicmedication Yes(385) 86.7(334) 8.1 32.3 59.6 0.013 No(923) 69.5(641) 14.4 27.9 57.7 Total 74.3(988) 12.4 29.5 58.1 113 Citation:GlobHealthAction2013,6:19253-http://dx.doi.org/10.3402/gha.v6i0.19253 (pagenumbernotforcitationpurpose) VeloshneeGovenderetal. Table2. Multivariate logistic regressionanalysisoffactors In order to assess affordability, we examined employ- associatedwithuptakeofthegovernmentinsurancescheme ees’ contribution as a percentage of the lowest monthly (newlyinsuredand previously insuredbyotherschemes) income for each income category. As noted earlier, the governmentschemecontributionsareincome-based,with Adjusted the monthly contributions varying according to the oddsratio (95%CI) employee’s salary, choice of benefit option and number of dependents. Despite the subsidy for low-income Age employees,onaveragethoseinthelower-salarycategories 20(cid:2)29 1.0 still paid a higher percentage of their salaries for health 30(cid:2)39 0.83 0.47(cid:2)1.48 insurance than those with greater income (with the 40(cid:2)49 0.57* 0.32(cid:2)1.03 exception of the fully subsidised option) (Table 5). For 50(cid:2)59 0.62 0.33(cid:2)1.17 example, in the low-cost option 2, the health insurance ]60 0.28* 0.11(cid:2)0.74 payment constitutes 7.6% of income for someone in the Sex lowestsalarycategory1whoearns$475permonth,while Female 1.0 thisoption is only 1.7% of monthlyincome forsomeone Male 0.69* 0.51(cid:2)0.93 in category 5 who earns $6,000 per month. Maritalstatus Married/cohabiting 1.0 Discussion Divorced/separated/widowed 1.60* 1.03(cid:2)2.48 In 2003, prior to the government scheme, insurance Single 1.35 0.97(cid:2)1.90 coverage among South African civil servants was Housing 56% (35). Our analysis shows that two years after the Formal 1.0 government scheme was initiated in 2006, 74.3% of civil Informal 1.73 0.63(cid:2)4.72 servants were insured, and 41.9% of these belonged to Salarycategory thegovernmentscheme.Althoughevidencesuggeststhat Lowerskilled 1.0 membership has increased, with 53.8% of civil servants Skilled 1.51 0.82(cid:2)2.75 enrolled in the government scheme in 2012 (27), other Highly-skilled 0.48** 0.28(cid:2)0.81 studies on enrolment in health insurance schemes in SupervisoryandSenior 0.39** 0.22(cid:2)0.71 Ecuador,Ghana,Mali,SenegalandUganda,havefound Management similar low levels of enrolment (36(cid:2)40). The newly-insured group included those from pop- Province ulation groups who commonly experience financial and Gauteng 1.0 otheraccessbarriers,suchasyoungeremployees,women, KwaZulu-Natal 0.69* 0.46(cid:2)1.01 unmarriedsinglepeople,blackAfricans,andthoseliving NorthWest 0.58** 0.40(cid:2)0.85 in informal housing or with lower-incomes. Therefore, WesternCape 0.76 0.52(cid:2)1.11 in contrast to private health insurance in the general Individualonchronicmedication population, where 71% of members are located in the Yes 1.0 richest 20% of the population (41), the government No 0.94 0.69(cid:2)1.27 scheme is comparatively pro-poor. Nevertheless, a con- siderableportionofsocio-economicallyvulnerablegroups CI,confidenceinterval. *pB0.1;**pB0.05. remained uninsured (more than halfof the lowest salary category for example), including men, black African or Table3. Choice ofgovernment scheme’sbenefit options Salarycategory(%) Supervisoryand Benefitoptionchosen Lowerskilled Skilled Highly-skilled SeniorManagement Total Option1:Lowcost 19.4 9.4 0.5 1.7 4.4 Option2:Lowcost 13.9 7.3 0.5 0.0 3.2 Option3:Mid-rangesavings 11.1 5.2 11.1 5.0 8.8 Option4:Comprehensive 55.6 75.0 80.6 65.0 74.8 Option5:Comprehensive 0.0 3.1 7.4 28.3 8.8 Total 100(36) 100(96) 100(216) 100(60) 100(408) 114 Citation:GlobHealthAction2013,6:19253-http://dx.doi.org/10.3402/gha.v6i0.19253 (pagenumbernotforcitationpurpose) HealthinsurancecoverageofcivilservantsinSouthAfrica Table4. Factorsinfluencing the decisionto join the subsidisedgovernment scheme Salarycategory(%) Factors Supervisoryand Affordability Lowerskilled Skilled Highly-skilled SeniorManagement Overall(%) Insuredjoinedbecauseschemeaffordable 57.9 65.0 70.2 67.2 67.4 Uninsureddidnotjoinbecauseschemeexpensive 40.2 40.2 27.1 33.3 34.8 Uninsuredwouldjoinifschememoreaffordable 34.0 32.0 34.6 13.3 32.8 Benefitoptionsandcoverage Insuredjoinedasschemeofferedbetterbenefitsand 36.8 41.2 33.5 49.2 37.9 coveredmoredependents Uninsuredwouldjoinifmoredependentswere 2.1 2.1 0.0 0.0 1.2 covered Administrativecomplexityofscheme Insuredjoinedasadministrativeprocedureseasy 2.6 10.3 6.0 11.5 7.5 Uninsureddidnotjoinasadministrativeprocedures 28.9 21.7 17.3 20.0 21.9 complex Informationaboutscheme Uninsureddidnotjoinschemeaslackedinformation 23.7 23.7 15.8 6.7 19.9 Uninsuredwouldjoinifmoreinformationprovided 2.1 2.1 0.0 0.0 1.2 Needforhealthinsurance Uninsureddidnotjoinasinsurancenotneeded 10.3 15.5 17.3 26.7 15.2 Uninsuredwouldjoiniftherewasahealthneed 41.2 44.3 31.6 33.3 38.0 Multiple-responsequestions. coloured race groups,7 less educated and lower-income network of primary healthcare service providers and employees, and those living in informal housing. This is geo-mapped members’ homes and workplaces against despite membership for the lowest salary tier being fully theproviderinordertoimproveavailability(43).In2010, subsidised. the scheme reported reaching a target of having at least Factors discouraging or deterring enrolment included 90%ofmemberswithin 10kmfromthenearest network affordability, the perceived administrative complexity of provider (43). At a provincial level, this target was joiningthescheme,anddifficultiesinobtaininginforma- achieved in four of the country’s nine provinces (Free tion about the benefit options. Moreover, the compara- State, Gauteng, KwaZulu-Natal and the Western Cape); tively poorer uptake of the government scheme in the intheruralNorthWestprovince84.8%ofmemberswere more rural North West and KwaZulu-Natal provinces, within10kmofaregisteredprovider.Itwillbeimportant may reflect underlying variations in geographical access to document whether these changes have diminished the toservices(Table2).Proximityofaprimarycareprovider differentials between membership across provinces. contractedwith the scheme is likely an important factor Affordability(orlackthereof)ofmembercontributions influencing a potential member’s decision to join the was an important factor encouraging (or discouraging) scheme. Transport costs have been shown to be an enrolment in the government scheme. As Carrin et al. important barrier to accessing care in the South African observe (44, p. 803), ‘Affordability of premiums or con- setting (42), and the distance to a scheme-contracted tributions is often mentioned as one of the main deter- providermayincreaseproblemsofaffordability.Thiswas minants of membership.’ The South African Ministerial likely an important issue in some provinces at the time Task Team commissioned investigation of low-income of the survey. As argued in a recent review of UC in medicalschemesfoundthat‘...thefundamentalobstacle Thailand, ‘Financing reform must go hand in handwith to expanding coverage to low-income households in ensuring physical access to services.’ (7, p. 17). In 2009, South Africa remains affordability’ (24, p. 124). Several other studies have pointed to premiums being unafford- to improve access to primary health care services, the able as a factor discouraging demand for insurance in government scheme in South Africa expanded the West Africa (45), Kenya (46) and India (47). Of note, a similar scheme to that studied here was implemented 7In this paper, racial categories structured through apartheid are in Botswana in 1990, with all government employees used in recognition that race remains an important social and economicfaultlineinthepost-apartheidcontext. entitled to a 50% subsidy from the government for 115 Citation:GlobHealthAction2013,6:19253-http://dx.doi.org/10.3402/gha.v6i0.19253 (pagenumbernotforcitationpurpose) VeloshneeGovenderetal. Table5. Employee contributions(percent ofmonthlysalary) towardsmonthly insurancecontributions Salarycategory(%) Benefitoption Lowerskilled Skilled Highly-skilled Supervisory SeniorManagement Low-costoption1 0.0 0.0 3.5 1.6 0.5 Low-costoption2 7.6 6.5 4.1 2.7 0.9 Mid-rangesavingsoption1 13.0 10.2 7.0 5.2 1.7 Comprehensiveoption2 13.5 10.5 8.0 5.8 1.9 Comprehensiveoption3 34.5 27.0 16.9 10.7 3.4 1US$(cid:3)7.5SouthAfricanRand. Financialcontributionscalculationsassumeafamilyofthree(member,andanadultandchilddependant).Percentmonthlycontributionis calculatedasaproportiononthelowestlevelineachsalarycategory.Incomefromotherhouseholdmembersisunknown,andhencenot includedinthecalculation. healthinsurance.Nearly70,000membershadenrolledby enrolment complexities and issues of trust as barriers to 2010(48). enrolment (38, p. 172). Similarly, in Ghana, a household Preferences and expectations of the range of services studyoftheNationalHealthInsuranceSchemeidentified and approved providers within benefit options can en- premiums, registration fees and administrative arrange- courage (or deter) enrolment. Earlier research among ments as key factors influencing enrolment and reten- households in South Africa indicated dissatisfaction and tion (49). In Uganda (38) and Tanzania (50), lack of poorperceptionsofpublichealthservices(10,24),creat- familiarity with community insurance schemes, particu- ingapreferenceforprivatehealthcare,includingprimary larly insurance principles of pooling and prepayments, andinpatientcare.Thismightexplaintherelativelypoor contributed to low levels of enrolment. However, as uptake of low-cost option 1, despite a full subsidy for Basaza et al. (38, p. 182) caution, ‘... a good under- those in the lower-salary categories. The ‘free’ low-cost standing of CHI principles, per se, will not directly option only provides members with access to basic translate into increased enrolment.’ Qualitative research outpatient services at pre-specified facilities and public can improve understanding of the ways in which quality hospitals,whichmayconflictwiththeirstrongpreference of care, benefit options, contributions and information for private primary and inpatient care. This might also shape peoples’ knowledge and views of health insurance explain the popularity across all salary categories of and their decision to enrol. comprehensive option 4, which provides access to any Beingacross-sectionalsurveyofexistingcivilservants, private hospital. the study was unable to examine the period prior to The study identified perceptions and understandings the government scheme (i.e. pre1993 when enrolment in of insurance, particularly among low-income employees, one of a few pre-determined schemes was mandatory as a barrier to enrolment in the government scheme. for some employees, or the period 1993 to 2005, where Thesepointtopeoples’underlyingunderstandingsofthe employeeswerefreetochoosewhichschemetheyjoined). potentialrolethatinsurancemightplayineitherreducing The cross-sectional design cannot examine the institu- or averting health care costs. This suggests a need for tional context within which insurance for civil servants effectivecommunicationstrategiestoenhanceknowledge has operated, changes that occurred in the scheme and about concepts of insurance to encourage enrolment how these may impact on participation. The ability to in a health insurance scheme. The findings also suggest draw conclusions is also limited by the timing of the that older people (i.e. 60 years and older), whites, those survey, which was only about three years after introduc- in higher salary categories and tertiary education who tion of the scheme. The frequent changes made to the probably have been with their current scheme for a long timemayhave‘brandloyalty’andconsumerinertia,even schemeintheperiod precedingthisstudyandthereafter, if the new scheme offers better value for money due to restrict our ability to compare the study findings with the subsidy. Further research could more clearly define outcomesofschemesinothercountriesorcontexts.Also, reasons and preferences for this. it is possible that factors influencing enrolment in the Previous research exploring low enrolment in a com- long-runvary from those described here in the relatively munity health insurance (CHI) scheme in Uganda early stages of the scheme. Moreover, data on the identified‘amixedunderstandingonthebasicprinciples influenceofperceptionsandexperiencesofpublichealth of CHI and on the routine functioning of the schemes’, services on the decision to join the government medical lack of information, affordability, poor quality of care, scheme was not collected. 116 Citation:GlobHealthAction2013,6:19253-http://dx.doi.org/10.3402/gha.v6i0.19253 (pagenumbernotforcitationpurpose) HealthinsurancecoverageofcivilservantsinSouthAfrica Conclusion curativehealthservicesinMexicobetween2000and2006.BMC PublHealth2011;11:771. Introductionoflow-costoptionswhicharefullysubsided 7. Health Insurance System Research Office (2012). Thailand’s resultedinanincreaseinmembershipamonglow-income universal coverage scheme: achievements and challenges. An public-sectoremployees.However,uptakeofmembership independent assessment of the first 10 years (2001(cid:2)2010). particularlybyyoung,blackAfricanorcolouredgroups, Nonthaburi, Thailand: Health Insurance System Research men, lower-income employees, those with no, or only Office.p.120. primary education and in rural provinces was sub- 8. Coovadia H, Jewkes R, Barron P, Sanders D, McIntyre D. The health and health system of South Africa: historical optimal, suggesting that barriers remain. Importantly, roots of current public health challenges. Lancet 2009; 374: these are the same population groups that have limited 817(cid:2)34. accesstocarewithintheexistingpubliclyfundedsystem. 9. HRH SA (2011). HRH Strategy for the health sector: 2012/ Lower-income employees were found to contribute a 13(cid:2)2016/17. Pretoria, South Africa: National Department of higher percentage of their salaries towards health insur- Health. ancethanhigher-incomeemployeesindicatinginequityin 10. McIntyre D, Goudge J, Harris B, Nxumalo N, Nkosi M. Prerequisites for national health insurance in South Africa: the government insurance scheme. The findings suggest results of a national household survey. S Afr Med J 2009;99: that financing reforms intended to move towards UC 725(cid:2)9. must also take into account geographical and adminis- 11. Karim SSA, Churchyard GJ, Karim QA, Lawn SD. HIV trative access. Improving quality of care within public infection and tuberculosis in South Africa: an urgent facilities is critical for improving public perceptions and need to escalate the public health response. Lancet 2009; 374: encouraging the uptake of insurance especially among 921(cid:2)33. 12. McIntyre D, Van den Heever A. Social or national health low-income households. Moreover, reforms need to con- insurance.In:HarrisonS,BhanaR,NtuliA,eds.SouthAfrican siderthebenefitoptionscarefully,andmustpayattention health review. Durban, SA: Health Systems Trust; 2007, pp. to the choice and geographical location of providers. 71(cid:2)85. 13. Ranson MK. Reduction of catastrophic health care expendi- tures by a community-based health insurance scheme in Acknowledgements Gujarat,India:currentexperiencesandchallenges.BullWorld HealthOrgan2002;80:613(cid:2)21. WewouldliketothankDianeMcIntyreforreviewingearlierdrafts 14. Goudge J, Gilson L, Russell S, Gumede T, Mills A. The ofthispaper,inadditiontotwoanonymousreviewers.Wethankour householdcostsofillnessinruralSouthAfricawithfreepublic SHIELDcolleagueswhoseworkcontributedtothispaper. primarycareandhospitalexemptionsforthepoor.JTropMed IntHealth2009;14:458(cid:2)67. Conflict of interest and funding 15. Harris B, Goudge J, Ataguba JE, McIntyre D, Nxumalo N, Jikwana S, et al. Inequities in access to health care in There are no conflicts of interest and this work was SouthAfrica.JPublicHealthPol2011;32:S102(cid:2)S23. supported by the International Development and Re- 16. McIntyre D. National health insurance: providing a vocabu- searchCentre(Grantnumber103457)andtheEuropean lary for public engagement. In: Fonn S, Padarath A, eds. Commission (Sixth Framework Programme; Specific SouthAfricanhealthreview.2010,Durban,SA:HealthSystems Targeted Research Project no: 32289). Trust;2010,pp.145(cid:2)56. 17. African National Congress. A national health plan for South Africa. Johannesburg, SA: African National Congress; References 1994. 18. National Department of Health (1994). Report of the health carefinancecommitteetotheministerofhealth.Pretoria,SA: 1. WHO (2005). World Health Assembly, Resolution No. 58.33. DepartmentofHealth. Geneva:WorldHealthOrganization. 19. Monitor Company, Health Partners International, Centre 2. WHO (2005). Health systems financing: the path to universal for Health Policy, National Labour, Economic Development coverage. The World Health Report 2010. Geneva: World Institute (1996). Final report of the hospital strategy project. HealthOrganization. TheMonitorCompany. 3. Lagomarsino G, Garabrant A, Adyas A, Muga R, Otoo N. 20. Restructuring the National Health System for Universal Moving towards universal health coverage: health insurance reformsinninedevelopingcountriesinAfricaandAsia.Lancet Primary Health Care: Report of the Committee of Inquiry 2012;380:933(cid:2)43. intoaNationalHealthInsuranceSystem(Broomberg,Shisana 4. IbrahimipourH,MalekiM-R,BrownR,GohariM,KarimiI, Committee).Pretoria:NationalDepartmentofHealth;1995. DehnaviehR.Aqualitativestudyofthedifficultiesinreaching 21. National Department of Health. A social health insurance sustainableuniversalhealthinsurancecoverageinIran.Health scheme for South Africa: a policy document. Pretoria, SA: PolicyPlan2011;26:485(cid:2)95. NationalDepartmentofHealth;1997. 5. McIntyreD,GarshongB,MteiG,MeheusF,ThiedeM,Akazili 22. NationalDepartmentofHealth.Whitepaperforthetransfor- J, et al. Beyond fragmentation and towards universal cover- mation of the health system in South Africa. Pretoria, SA: age: insights from Ghana, South Africa and the United DepartmentofHealth;1997. Republic of Tanzania. Bull World Health Organ 2008; 86: 23. DSD (2002). Transforming the present-protecting the future: 871(cid:2)6. reportofthecommitteeofinquiryintoacomprehensivesystem 6. Danese-dlSantosL,Sosa-RubiS,Valencia-MendozaA.Analy- ofsocialsecurityforSouthAfrica.Pretoria,SA:Departmentof sisofchangesintheassociationofincomeandtheutilizationof SocialDevelopment. 117 Citation:GlobHealthAction2013,6:19253-http://dx.doi.org/10.3402/gha.v6i0.19253 (pagenumbernotforcitationpurpose) VeloshneeGovenderetal. 24. BroombergJ(2006).Finalreport:consultativeinvestigationin 40. Eckhardt M, Forsberg BC, Wolf D, Crespo-Burgos A. Feasi- low income medical schemes. Johannesburg: South African bility of community-based health insurance in rural tropical DepartmentofHealth. Ecuador.RevPanamSaludPublica2011;29:177(cid:2)84. 25. African National Congress. ANC 52nd National Conference 41. McIntyre D, Okorafor O, Ataguba J, Govender V, Goudge J, Resolutions,Polokwane,SA,16(cid:2)20December2007. Harris B. Health care access and utilisation, the burden of 26. GovernmentofSouthAfrica(2011).Nationalhealthinsurance out-of-pocket payments and perceptions of the health system: in South Africa: policy paper. Pretoria, SA: Department of findingsofanationalhouseholdsurvey.CapeTown,SA:Health Health/GovernmentPrintingWorks;p.29. EconomicsUnit,UniversityofCapeTown;2008. 27. Government Employees Medical Scheme. GEMS and its 42. Goudge J, Gilson L, Russell S, Gumede T, Mills A. Afford- performance. 16 May 2012 Portfolio Committee on Public ability,availabilityandacceptabilitybarrierstohealthcarefor thechronicallyill:longitudinalcasestudiesfromSouthAfrica. ServiceandAdministration,2012. BMCHealthServicesResearch2009;9:75. 28. VandeVenWPMM,VanVlietRCJA.Consumerinformation 43. GEMS (2010). Annual Report. Pretoria, SA: Government surplus and adverse selection in competitive health insurance EmployeesMedicalScheme. markets:anempiricalstudy.JournalofHealthEconomics1995; 44. Carrin G, Waelkens M-P, Criel B. Community-based health 14:149(cid:2)69. insurance in developing countries: a study of its contribution 29. Carrin G. Social health insurance in developing countries: a tothe performanceofhealthfinancingsystems. TropMedInt continuingchallenge.IntSocSecurRev2002;55:57(cid:2)69. Health2005;10:799(cid:2)811. 30. KirigiaJ,SamboL,NgandaB,MwabuG,ChatoraR,Mwase 45. De Allegri M, Sanon M, Bridges J, Sauerborn R. Under- T. Determinants of health insurance ownership among standing consumers’ preferences and decision to enrol in South African women. BMC Health Services Research 2005; community-basedhealthinsuranceinruralWestAfrica.Health 5:17. Policy2006;76:58(cid:2)71. 31. Liu T-C, Chen C-S. An analysis of private health insurance 46. MathauerI,SchmidtJ-O,WenyaaM.Extendingsocialhealth purchasingdecisionswithnationalhealthinsuranceinTaiwan. insurance to the informal sector in Kenya. An assessment of SocialScience&Medicine2002;55:755(cid:2)74. factorsaffectingdemand.IntJHealthPlannManage2008;23: 32. YingX-H,HuT-W,RenJ,ChenW,XuK,HuangJ-H.Demand 51(cid:2)68. for private health insurance in Chinese urban areas. Health 47. Ranson MK. Reduction of catastrophic health care expendi- Economics2007;16:1041(cid:2)50. tures by a community-based health insurance scheme in 33. MakokaD,KaluwaB,KambewaP.Demandforprivatehealth Gujarat,India:currentexperiencesandchallenges.BullWorld insurance where public health services are free: the case of HealthOrgan2002;80:613(cid:2)21. Malawi.JApplSci2007;7:3268(cid:2)73. 48. BPMOAS (2010). Annual Report. Gaborone, Botswana: 34. Propper C. Constrained choice sets in the UK demand for BotswanaPublicOfficers’MedicalAidScheme. privatemedicalinsurance.JPublEcon1993;51:287(cid:2)307. 49. Jehu-AppiahC,AryeeteyG,AgyepongI,SpaanE,BaltussenR. 35. DPSA, Toth Resources (2003). Research report on the devel- Householdperceptionsandtheirimplicationsforenrolmentin opmentofaclosedmedicalschemeforpublicserviceemployees. the nationalhealthinsurance schemeinGhana.HealthPolicy Pretoria,SA:DepartmentofPublicServiceAdministration. Plan2012;27(3):222(cid:2)233. 36. Jehu-AppiahC,AryeeteyG,SpaanE,deHoopT,AgyepongI, 50. KamuzoraP, GilsonL.Factorsinfluencingimplementationof Baltussen R. Equity aspects of the national health insurance the community health fund in Tanzania. Health Policy Plan schemeinGhana:whoisenrolling,whoisnotandwhy?SocSci 2007;22:95(cid:2)102. Med2011;72:157(cid:2)65. 37. Chankova S, Sulzbach S, Diop F. Impact of mutual health *VeloshneeGovender organizations: evidence from West Africa. Health Policy Plan HealthEconomicsUnit 2008;23:264(cid:2)76. SchoolofPublicHealthandFamilyMedicine 38. Basaza R, Criel B, Van der Stuyft P. Community health FacultyofHealthSciences insuranceinUganda:whydoesenrolmentremainlow?Aview UniversityofCapeTown frombeneath.HealthPolicy2008;87:172(cid:2)84. SouthAfrica 39. Preker A, Carrin G, eds. Health financing for poor people: Tel:(cid:4)27214066752 resourcemobilizationandrisksharing.2004,Washington,DC: Fax:(cid:4)27214488152 WorldBank;2004. E-mail:[email protected] 118 Citation:GlobHealthAction2013,6:19253-http://dx.doi.org/10.3402/gha.v6i0.19253 (pagenumbernotforcitationpurpose)