43718 REACHING THE POOR 2008 WITH HEALTH SERVICES Cambodia Exempting the Poor from Hospital User Fees The Issue cost, most would have faced formidable challenges in devel- oping systems to verify the accuracy and honesty of facility Cambodia's Health Equity Funds seek to address the classic reimbursement claims, and to transfer funds promptly to the limitation of efforts to exempt the poor from user fees: the facilities concerned. loss of income that such exemptions have typically repre- Cambodia was one of many countries confronted with sented for service providers, and the providers' resulting this problem. Its health system had been virtually destroyed reluctance to grant them. during the Khmer Rouge regime of the 1970s and the civil This problem dates from the late 1980s, when developing war that followed. Reconstruction had proven difficult, with country health ministries increasingly began to introduce the ministry unable to provide adequate and reliably steady user fees in their facilities, on the advice of leading interna- financial support to its outlying facilities. Thus, user fees had tional agencies and as a way of helping deal with their dire been introduced in 1997, with almost all of the revenues financial situations. On paper, the ministry plans usually retained by facility staff in order to supplement salaries and included a provision for exempting the poor, and sometimes help cover running costs. The government had issued a these were effective. But often, they were not. Implementa- decree exempting the poor, but few exemptions had been tion often consisted of little more than circulars to facilities issued. The resulting cost to the poor had become a major that front-line providers tended to ignore or interpret very concern. narrowly, especially when a significant portion of user fee revenue was retained at facilities and represented a badly- needed source of health worker income. In such settings, few The Approach exemptions were granted, leaving poor people to bear the brunt of user fees along with everybody else. In response, some of the many external non-government The most obvious solution to this problem--that is, to organizations active in Cambodia began experimenting with reimburse service providers for the income they lost in alternative approaches to health financing in the areas where attending to poor clients--suffered from equally obvious dif- they were working. Among the approaches emerging from ficulties. One was cost. As noted, Governments had been this experimentation was that of a "health equity fund" attracted to user fees in large part because of their potential (HEF)--a fund operating independently of the health system, for revenue generation, but to the extent that fee income whose staff identified people in particular need of financial remained at the facility level, the resources of central health assistance for health services, especially hospital care, and ministries remained the same. To make reimbursement pay- paid health service providers for the service provided to the ments to facilities for the income they had foregone in grant- needy. ing exemptions represented an additional expense that few By late 2006, there were twenty-six HEFs in operation. ministries felt they could afford. Also important were logisti- Each of the twenty-six projects is autonomous. All share the cal challenges: even had ministries been willing to bear the basic defining characteristic noted above--that is, all feature www.worldbank.org/wbi/healthandaids an independent organization identifying poor people and ing to the definition used, the HEF representative arranges reimbursing service providers on their behalf. However, there for the provision of program benefits for that person. HEF is also considerable variation among them. staff members also often visit the facility wards to determine if there are other patients who qualify for benefits--say, Administration. Although a few HEFs are administered patients who were able to pay the admission fee only by bor- directly by international agencies, most are managed by rowing heavily or by selling important productive assets. Cambodian non-governmental organizations (NGOs). Many Where possible, HEF staff visit recipients of support in their different types of NGOs are involved. For example, some are homes after discharge, in order both to double-check their pre-existing national organizations; others are local, created financial status and to provide social support. specially to operate the HEF. In at least one case (Kirivong), Active identification, also known as pre-identification, Buddhist monks from local pagodas play a prominent role in involves surveying a district's population, in order to deter- the administering agency. mine in advance who is poor enough to qualify for HEF assistance. Those who qualify are issued some sort of identity Definition of Eligible Population. All HEFs define the people document that they produce upon arrival at a participating eligible for support in terms of specified household character- health facility. In some cases, the survey is conducted by out- istics. However, the procedures used in determining which side investigators using a formal questionnaire. In others, the characteristics apply are usually considerably less formal procedure is considerably simpler, drawing on the local than the proxy means testing procedures (like those described knowledge of respected community leaders. in the Colombia and Mexico Progresa Reaching the Poor The two approaches are not mutually exclusive, and some briefs) which use statistical analyses of household survey data HEFs use some combination of the two. This is because each to determine which characteristics are most closely associ- approach is recognized as having both advantages and limi- ated with poverty. More often, the Cambodia HEF proce- tations. Advocates of the passive approach cite its simplicity dures draw on the views of knowledgeable local observers and thus low cost, its acceptability to people in the areas cov- concerning the characteristics that most clearly denote ered, and the possibility of implementing it without the poverty. The characteristics most commonly identified extended period often needed to establish a pre-identification include the occupation and marital status of the household system and the considerable effort to keep it up-to-date. At head, number of dependents, land ownership, housing con- the same time, they acknowledge that the passive strategy struction materials, and possession of productive assets. risks missing the many poor people not aware of the finan- After a list of characteristics is drawn up, people lacking a cial support available and reluctant to come for service. Prac- certain portion of them are deemed eligible for HEF subsi- titioners of active or pre-identification argue that the greater dies. In some cases, more than one poverty level is estab- complexity and cost of this approach are more than justified lished, with those deemed "very poor" being eligible for by the greater accuracy and coverage among the poor that it greater benefits than those categorized as "poor." permits. See Table 1 for information about the identification procedures and criteria used to identify beneficiaries for four Identification of Eligible Population. Once the criteria for hospital-based HEFs in Cambodia. HEF support have been defined, those people meeting those criteria have to be identified. This is done in two different Proportion of Population Found Eligible. Application of the ways: passively and actively. procedures just described has led to varying percentages of Passive identification features a determination of the the population found to be eligible for HEF support, reflect- poverty status of people once they arrive at the health facil- ing in part differences in economic conditions among the dis- ity. This is typically done by HEF staff members stationed in tricts covered, and no doubt also in part the result of differ- the participating health facilities. If a patient reporting to the ences in the definition procedure used. In general, the facility reception lacks the money to pay the facility's admis- percentage of the population found eligible ranged from sion fee, the receptionist refers her or him to an HEF staff approximately 12% to 25% of the total. These percentages member, who asks a series of questions about the household. are usually well below the 35­75% of the population in the Should the responses indicate that the patient is poor accord- areas concerned living below the $1/day poverty line. Table 1. Procedures and Criteria Used in the Four HEFs to Identify the Poorest. Svay Rieng Pearang Kirivong Sotnikum Identification process Identification method Household assessment Household assessment Household assessment Household assessment Selection place Household Household Village Hospital NGO office Selection time Ex-ante Ex-ante Ex-ante At the illness episode Selection process Pre-identification (proxy Pre-identification (proxy Pre-identification Passive identification (proxy means testing) means testing) Approval by Chief Monk means testing) at episode of Verification Verification Edition of entitled list illness, at hospital, by local (Data entry (database) (Data entry (database) NGO staff Selection tool Formal scored Formal scored Informal list of criteria for Informal. questionnaire questionnaire community-based targeting Non-formalized interview. Entitlement document Equity certificate Equity certificate Voucher (non-permanent) None (except records in the Database Database Entitled list books of the NGO) Alternative process Passive identification at Passive identification at Certification letter None episode of illness, at episode of illness, at signed by the pagoda hospital, by hospital staff hospital, by NGO staff chief monk Criteria Household Occupation of Occupation of No. dependents Marital status characteristics household head household head (alt. criteria) No. disabled members Marital status Marital status No. dependents No. children <18 years No. dependents No. children at work No. elderly dependents Health status Length of severe illness Chronic disease in during the previous year household Productive assets Type of housing Roof and wall and Type of housing Size of land/rice fields and belongings Transport means m2/person Size of farmland Productive assets Size of land Size of productive land Transport items No. cows, buffalos Electronic items (alt. criteria) and pigs Transport means Farm animals Farm assets and livestock (alt. criteria) Power supply Electronic items Quantity of rice harvested (alt. criteria) Income/expenditures Cash income/expenditures Household income Lack of food security Health expenditures during the previous year Others Appearance and social capital Scoring Score/criteria Score/criteria None None and threshold and threshold Source: Noirhomme et al., Health Policy and Planning, 2007 Services Covered. The services supported by the HEFs are By the end of 2004, the latest period for which data are avail- delivered through government facilities. Most HEFs deal able, HEF support recipients varied between less than 10% primarily or exclusively with fees charged for services in dis- to over 50% of all patients in the hospitals concerned. trict hospitals, which are considerably higher than those Data on hospitalizations for HEF beneficiaries and non- charged for primary care at lower-level facilities and thus beneficiaries in four hospital-based HEFs (Sotnikum, Svay further beyond the means of poor patients. Some of the Rieng, Pearang, and Kirivong) between the third quarter of HEFs have begun to extend the scope of their activities to 2000 and the last quarter of 2004 can be seen in Figure 1. include costs to patients of primary care provided at lower- Increases in patients at three of the HEFs took place after level facilities. these were launched (Sotnikum and Svay Rieng) or after household equity certificates were distributed (Pearang). Benefits provided. As indicated earlier, the most common According to data collected for the latter three hospitals, it benefit offered by HEFs is payment of user fees on behalf of seems that the HEF-supported patients are new clients who those found eligible for support. Often, the payment is for previously could not access the services for financial reasons. the full amount of the fees concerned, but other patterns While it can be assumed that the creation of the HEF affected exist. In some cases, for example, the proportion of the fees the patient numbers for these three hospitals, the proportion covered varies according to just how poor the patient is. of patients using the hospital at Kirivong was too small to Some of the HEFs go beyond this by also providing trans- draw any conclusions about HEF impact on hospital access portation, food allowances, and other benefits when needed. in that area. Although the HEFs have yet to be as rigorously evalu- Cost. On average, the annual cost of operating the HEFs ated as some other health equity projects, they appear to be appears to be on the order of $US 0.50 per person found eli- achieving their objective of reaching Cambodia's lowest gible for support. However, this figure varies widely depend- economic groups. In one rural HEF (Sotnikum), over 90% ing upon the benefits provided, the approach employed to the patients it supported were either poor or very poor, identify beneficiaries, and other considerations. A recent compared with 60% in Cambodia as a whole and 75% in WHO study suggests a range from $US 0.10­0.15 to nearly the district where it operated. In another (Kirivong), HEF $US 2.00 per eligible person. beneficiaries were significantly poorer than other district residents with respect to all of the dimensions measured: Source of Funds. The funds required to cover these costs occupation, literacy, income, land ownership, and others. A come primarily from external agencies. They and interna- further suggestion, or at least hint, that this is the case tional NGOs receiving their support have also played a lead- comes from data indicating that the number of paying hos- ing role in introducing HEFs and the broader reforms of pital users has remained steady as the number of HEF ben- which they have been a part. A notable exception to this eficiary users has risen, implying that better-off people have funding pattern is the Kirivong HEF, referred to above, where continued to use their own funds rather than draw upon the Buddhist pagodas helping administer the project also HEF resources. raised a part of the required resources by soliciting donation The twenty-six HEFs established by late 2006 cover about from better-off families in the project area. a third of Cambodia's 76 health operational districts. The HEF concept has received strong governmental support, and has become part of the government's poverty reduction strat- The Initial Record and Future Prospects egy. As a result of this and continuing donor interest, the During their initial two or three years, most of the HEFs with number of districts covered by an HEF is expected to rise sig- available data have recorded significant increases in the num- nificantly over the coming years. ber of people receiving hospital services with HEF support. Figure 1. Hospitalizations for HEF beneficiaries and non-beneficiaries in the four HEFs. 1600 1400 Sotnikum 1200 1000 Start of the HEF 800 600 400 200 0 1600 1400 Svay Rieng Start of the HEF 1200 1000 800 600 400 200 0 1600 1400 Pearang 1200 Equity certificate distribution 1000 800 Start of the HEF 600 400 200 0 1600 1400 Kirivong 1200 Start of the HEF 1000 800 600 400 200 0 00 00 10 10 10 10 02 02 02 02 03 03 03 03 04 04 04 terra terra terra re rta terra terra terra terra terra terra terra terra terra terra terra terra terra qu qu qu qu qu qu qu qu qu qu qu qu qu qu qu qu qu 3rd 4th 1st 2nd 3rd 4th 1st 2nd 3rd 4th 1st 2nd 3rd 4th 1st 2nd 3rd Paying patients HEF beneficiaries Source: Noirhomme et al., Health Policy and Planning, 2007 This brief is intended to summarize good practices in reaching the poor Insights from Kirivong Operational Health District in Cambodia" (vol with health, nutrition, and population services. It is based primarily on 21, 2006, pp.27­39); amd Wim Hardeman et al., "Access to Health a 2006 series of three articles published in the journal Health Policy and Care For All? User Fees Plus a Health Equity Fund in Sotnikum, Cam- Planning: Mathieu Noirhomme et al., "Improving Access to Hospital bodia", (vol. 19, 2004, pp. 22­32). Without wishing to implicate them Care for the Poor: Comparative Analysis of Four Health Equity Funds any way for the brief's contents, the editors thank Frederic Bonet, Bart in Cambodia" (vol. 22, 2007, pp. 246­62); Bart Jacobs and Neil Price, Jacobs, Bruno Meessen, Mathieu Noirhomme, Ir Por, and Wim Van "Improving Access for the Poorest to Public Sector Health Services: Damme for information povided through personal communications. www.worldbank.org/wbi/healthandaids