Report No 41644-BI Republic of Burundi Health Financing Study September 30, 2008 Human Development Department Ministry of Public Health Sector Unit III and the Fight against HIV/AIDS Africa Region Republic of Burundi Document of the World Bank AKNOWLEDGEMENTS This Health Financing Study is a joint product of the Ministry of Public Health and the Fight against HIV/AIDS inBurundi andthe World Bank. The Burundi MOPH team was headed by Mr.Cyprien Baramboneranye (Directeur Ge'ne'raldes Ressources), and was composed of Mr Sosthkne Hicuburundi (Economiste, Cellule de Planfzcation),and Ms.Anne-Marie Niyonzima (Directrice du Budget, DGR). From the World Bank side, Ms. Maria Eugenia Bonilla-Chacin coordinated the over-all work. The World Bank team was composed of Mr.Marc Nene, Ms. Laurence Lannes, Ms.Veronique Hubert, Ms.MonsterratMeiro-Lorenzo, and Mr.Pamphile Kantabaze. Invaluable comments and suggestions were received from peer reviewers: Ms. Maureen Lewis and Ms. Agnes Soucat as well as members of the Burundi country and the World Bank Human Development teams including Mr.Alassane Sow, Mr.Jean-PascalNganou, Ms.Carolina Monsalve, and Ms.Hannah Nielsen. The team worked closely with Ms. N. Rabarijohn (UNICEF), Ms. S. Villeneuve (UNICEF), and Mr. N. Walelign (UNICEF). The study was done under the guidance of Lynne Sherburne-Benz, Sector Manager, John Elder, Acting Sector Manager, and John McIntire, Country Director. The study also counted with the collaboration of many donors and NGOs supporting the health system in Burundi as they provided the data that made possible this study. Finally, the study benefited from comments received during two workshops in Bujumbura (October 29th-November 2nd, 2007 and September 18th, 2008) where the results of the study were presentedanddiscussed. This study counted with the financial support of the Belgian Poverty Reduction Partnership Program(BPRP). 3 TABLE OF CONTENT Aknowledgements ............................................................................................................... 3 Executive Summary.............................................................................................................. i Background................................................................................................................... ..i Introduction ......................................................................................................................... HealthExpenditure..................................................................................................... 11 1 Background.......................................................................................................................... 3 Health Outcomes ............................................................................................................. HealthPolicy andPlans................................................................................................... 3 4 Access andUtilization of Health Services ...................................................................... 7 Health System Constraints............................................................................................... 9 HumanResources........................................................................................................ Availability o f Drugsand Supplies ........................................................................... 9 12 13 HealthExpenditure.... ....................................................................................................... L o w Quality ofthe Services Offered......................................................................... 15 15 External Assistance to the Health Sector................................................................... Government Expenditure on Health.......................................................................... 24 Private Expenditure on Health................................................................................... 33 Total Expenditure on Health...................................................................................... 35 Public Expenditure Management....................................................................................... Fiscal Space for Health.............................................................................................. 37 3 9 BudgetPlanning ........................................................................................................ 3 9 40 Resource Management Issues at Health Facility Level............................................. Budget Execution and Monitoring ............................................................................ 42 Health Services Offered to Children UnderFive and Women DuringDelivery...............44 Simulation of the Cost and Impact o fthe Presidential Measure to Eliminate User Fees for ;................................................ 44 References ......................................................................................................................... Results ....................................................................................................................... 45 Methodology............................................................. 60 Annexes ............................................................................................................................. 62 5 DISCLAIMER: This volume is a product of the staff of the International Bank for Reconstruction and Development/ The World Bank. The findings, interpretations, and conclusions expressed inthis paper do notnecessarilyreflect the views ofthe Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other informationshown on any map inthis work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptanceof suchboundaries. COPYRIGHT STATEMENT: The material in this publication is copyrighted. Copying and/or transmitting portions or all o f this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development/ The World Bank encourages dissemination of its work andwill normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA, telephone 978-750-8400, fax 978-750-4470, http://www.copyri ght.cod. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433,USA, fax 202-522-2422, e-mail pubrights@worldbank.org. 4 ACRONYMS ACT Artesunate-Amodiaquine CombinationTreatment AIDS AcquiredImmuneDeficiencySyndrome CAM Carte d 'AssuranceMaladie CAMEBU Centrale d 'Achatdes Me'dicamentsEssentiels du Burundi CBO CommunityBasedOrganizations CHUK UniversityHospitalCenter ofKamenge CNLS ConseilNational de Lutte contre le SIDA COSA Comite'de Sante' COGE Comitb de Gestion cso Civil Society Organization CPSD Cadre de concertationdes Partenairespour la Sante' et le De'veloppement DAC DevelopmentAssistanceCommittee(OECD) DFID UKDepartmentfor InternationalDevelopment DG Directeur Ge'ne'ral DGR Direction Ge'ne'raledes Ressources DPS Direction des Programmesde Sante' DHS DemographicandHealth Survey ECHO EuropeanCommissionHumanitarianAid Office EPI EnlargedProgrammeo fImmunization FBU Franc Burundais FED Fonds Europe'ende De'veloppement FP FamilyPlanning GAVI GlobalAlliance for Vaccines andImmunization GDP GrossDomesticProduct GF GlobalFund GNI GrossNationalIncome GVC ItalianCooperation HIPC HeavilyIndebtedPoor Country HIV HumanImmunodeficiencyVirus ICRC InternationalCommitteeofthe RedCross IDA InternationalDevelopmentAgency IMR Infant MortalityRate IMF InternationalMonetary Fund INSS Institut National de Se'curite'Sociale LFI Loi de Finances Initiale LFR Loi de Finances Re'vise'e MDG MillenniumDevelopmentGoals MICS Multiple IndicatorCluster Survey MinFin Ministry ofFinance Min SIDA PresidentialMinistryfor the FightAgainstHIVIAIDS 6 MMR Maternal Mortality Rate MOPH Ministryof Public Health MPDR Ministtre de la Planijkation, du Dtveloppement et de la Reconstruction MSF Mddecinssans FrontiBres MSP . Ministtre de la Santk Publique et de Lutte contre le Sida MTEF Medium Term ExpenditureFramework NGO Non-GovernmentalOrganization ODA Official DevelopmentAssistance OECD OrganizationofEconomic Co-operation andDevelopment OOP Out-of-pocketpayment ORS Oral RehydrationSolution ORT Oral RehydrationTreatment OTB Ordonnateur Tre'sorierdu Burundi PATSBU Programmed 'Appui Transitoireau Secteur de la Sante'Burundais PEMFAR Public Expenditure Managementand Financial Accountability Review PETS Public ExpenditureTracking Survey PHB ProvincialHealthBureau PHC Primary HealthCare PSPII Projet Sante'et Population 11 NHDP Plan National de De'veloppementSanitaire QUIBB Questionnaire Unife' des Indicateurs de Base du Bien-&tre QUD Questionnaire Unijit sur les Indicateurs de Ddveloppement REGIDESO Rkgie de Production et Distribution d'Eau et dElectricite' SSA sub-SaharanAfrica SWAP Sector-WideApproache TB Tuberculosis UN UnitedNations UNAIDS Joint UnitedNations Programmeon HIV/AIDS UNDP UnitedNations DevelopmentProgram UNFPA UnitedNations Fundfor PopulationActivities UNHCR UnitedNations HighCommissionerfor Refugees UNICEF UnitedNations Children'sFund USAID United States Agency for InternationalDevelopment WB World Bank WDI World DevelopmentIndicators WHO World Health Organization 7 EXECUTIVESUMMARY During the conflict, public expenditure on health drastically decreased weakening the health system at a time when needs were growing. With increasing peace and stability in the counm, foreign aid and public resources allocated to the sector have started to increase. However, health authorities stillface the challenge of mobilizing extra resources to reverse the damage to the health system and to progress towards achieving the Millennium Development Goals (MDGs). Nevertheless, more resources allocated to the sector will not necessarily improve health outcomes, especially for the poor, if weaknesses in public expenditure management and ineficiencies in the allocation of resources for the sector are not reduced. These challenges become more acute in the context of the presidential measure of eliminating userfees for deliveries andfor child health services. Background 1. Despite recent improvements, the health status of the populationhas not reached the levels prevalentbefore the 1993 crisis. One out of every six infants does not survive his first birthday and one out of every five his fifth. In addition, more than 40% of children are chronically malnourished. Maternal mortality is also high, much higher than the average in sub- Saharan Africa of 920 maternaldeaths per 100,000 live births. 2. Communicable diseases such as malaria, diarrhea, respiratory infections, and the compoundingeffect of malnutritionare the major causes of mortality and morbidity in the country, especially among children. These diseases can be prevented or cured at a relatively low cost. HN/AIDS also constitutes a serious health and development challenge inthe country, with an adult prevalence rate ofabout 3.3%. 3. About a third of the populationdoes not receivehealthcare when in need, mainly due to financial barriers to access health services. During the conflict, both foreign aid and government expenditure in the sector decreased. Due to these expenditure cuts and in order to increase funds, the government initiated a cost-recovery mechanism in all health facilities. This policy allowed these facilities to finance some of their running costs. However, as household income also decreasedand as exemption and waiver mechanisms were not functioning, financial barriersto access health services remainthe main barrierto access health services. 4. With increasedpeaceand stabilizationinthe country, the Governmentis committedto reverse the damage that more than a decade of conflict generated in the health of the population and to ensure a sustainable development of the health system. This commitment i s reflected in the country's Poverty Reduction Strategy Paper (PRSP) and in the National Health Development Plan (NHDP) whose main objectives are to: (i)reduce maternal and neonatal mortality; (ii)reduce infant and child mortality and morbidity; (iii)reduce the prevalence of communicable and non-communicable diseases; and (iv) reinforce the performance o f the nationalhealth system. 1 5. In accordance with the NHDP, in May 2006 the President announced the elimination of user fees for health services offered to children under five and women during delivery. This presidential decreewas aimed at improving maternal and child health by increasing financial access to health care. This measure has indeed increased utilization of health services among the beneficiaries but delays in the reimbursement to facilities that offer these services have created some difficulties. 6. Despite this measure, the coverage of many high impact child and maternal health interventions (e.g. ITN, full vaccination, ORT in case of diarrhea) remains low. Many of these interventions were provided "free" of charge before May 2006, but they have remained underfunded. 7. With the currentlevelof healthindicators, it is unlikelythat the countrywill reach the health related MDG and NHDP goals without significant additional efforts and resources to overcome major challenges of the health system such as insufficient and unequally distributed health care personnel, stock-outs o f drugs and supplies, and the low quality of the services offered. 8. The shortage of qualified medical personnel is one of the main bottlenecks in the deliveryof healthservices. Not only are qualified healthcare personnel scarce, but they are also unevenly distributed favoring Bujumbura and leaving poorer areas underserved. About 80% o f doctors and 50% of nurses are thought to be concentrated in Bujumburawhich is the region with the lowest poverty level and where only 10% of the population lives. During the conflict, many qualified medical personnel left the rural areas or the country altogether for security reasons. But at the moment, the main reasons behind these shortages are low production and an inadequate remuneration system for these personnel. 9. The quality of health services is not only affected by the insufficient number of qualified personnel, but also by common stock-outs of pharmaceuticals and other medical supplies. There is also evidence of poor technical quality o f the services offered. Finally, health services are the public services with the lowest level of satisfaction; most o f this dissatisfaction i s due to the cost of services. HealthExpenditure Sources of Fundsfor Health 10. Expenditure on health in Burundi remains low, mainly due to very low Government expenditure in the sector. Based on data collected for this study, in 2006, total health expenditure (THE) in Burundiwas estimated to be in a range o f US$ 14.5- 18.5 per capita. This estimate is less than halfthe sub-SaharanAfrica (SSA) average and lower than THE inother post conflict countries like Rwanda. Most of this expenditure i s financed by foreign assistance which is the source o f about 50% of it. In2006, government expenditure on health' was only about US$ 1 per capita, one o f the lowest in SSA. The private sector financed the rest, mainly through household out-of-pocket (OOP) expenditure. 1Government expenditure on health is defined as the expenditure managed by the Ministry of Public Health, the Presidential Ministry for the Fight against HIV/AIDS, and the expenditure for the University Hospital Center ofKamenge, which is managedby the Ministry ofEducation. *. 11 11. During the conflict, government expenditureon health drastically decreased, not just due to lowergovernment revenuesbut also due to a decreased share of the budget allocated to the sector. By the year 2000, less than 3% of government expenditure was allocated to health care, representing about US$ 0.58 per capita. In 2006, thanks to the enhanced Heavily Indebted Poor Countries (HIPC) Initiative, government resources allocated to the sector increased significantly, representing about 7% of the total budget or about US$2 per capita. 12. Despite this large increase in the health sector budget, government expenditure on health(commitment based) did not increase as much in 2006. This was mainly due to delays in the disbursement of HIPC resources and to budget execution problems at the Ministry of Public Health' (MSP in French) level. As consequence, that year health expenditure only represented 4% of total Government expenditure, about US$ 1per capita; remaining one o f the lowest ifnot the lowest in SSA. 13. In 2007, the health sector budget was reduced by about 23%. That year budget execution progressed significantly, partly due to simplifications in the mechanism to reimburse facilities for the "free" services offered to children under five and women during delivery. In addition, an important share of these funds was usedto pay commitments made in2006. By July 2007, many budget lines had already been used. More resources for the sector were expected during a budget revision, but this revision did not take place creating serious difficulties to the MSPto fulfill some of its responsibilities. 14. Foreign assistance has financed an important share of the health system. In 2005, donors' overall assistance to the health sector was about US$9.05 per capita (US$7.85 per capita without emergency as~istance)~. In 2006, this assistance was US$ 8.65 per capita (US$ 7 per capita without emergency assistance). In2007, this assistance reachedUS$ 8.99 per capita. 15. As pre-payment mechanisms only cover a very small percentage of the population,in Burundimost privatehealthexpenditure is out-of-pocket expenditure. Based on a household survey (QUIBB)implemented in early 2006, before the presidential measure on user fees was announced, household OOP expenditure represented about 5.6% of total household expenditure or about US$ 8.6 per capita. This large OOP on healthtogether with high poverty rates resulted in large financial barriers to access healthcare. It is not clear how this expenditure changed after the presidential measure o f May 2006 as user fees in public and confessional facilities are only part o f total household OOP health expenditure. Inaddition, as children under five and women during delivery are the only beneficiaries of this policy, OOP should still represent an important share of total health expenditure. 16. A sustainable and steady increase in government resources allocated to the health sector is fundamentalto improvehealthoutcomes in the country. An increase intotal public funds can finance the sector in a more predictable and sustainable manner than other sources of funds. This increase can also leverage more external resources for the sector. Releasing the budget in a timely manner will also be essential as an increase in the budget alone will not be sufficient ifthese resourcesare not spent. 2While the study was completed, there were two ministries (Ministry ofPublic Healthand Ministry for the Fight Against HIV/AIDS). Afterwards, these two ministrieswere mergedinto one, the MSPLS. Inthe data collectedon foreign assistance, it was not always possibleto differentiatecommitments fiom actual expenditure. Therefore, these numbers should be considered an upper bound for this expenditure. Assuming that the execution rate of these resources is similar to that in neighboring Rwanda (76%), the estimatedlowerboundfor foreignaidwouldbe US$6.6 per capita. ... 111 Efficiency and Equity in the use of resources 17. Public resources for the health sector have been used, in general, according to the prioritiesset in the PRSP and the PNDS; but there is room for improvement.A large share of the budget goes to high level hospitals and in particular to one hospital inthe capital city. In a country where the majority o f the population lives in rural areas and where the main causes of illness and deaths can be prevented or treated at lower levels o f care, this i s not the most equitable and efficient use of very limited resources. Similarly, in 2006 about 16% o f the health sector budget went to the fight against one disease, HN/AIDS. This epidemic is a real health and development challenge for the country and resources will still be needed to fight this epidemic. However, without a functioning health system there will not be as much progress in the fight against this disease. Thus, as adult prevalence i s 3.3% and given the large resource constraints and the epidemiological profile o f the country, a discussion i s neededto ascertain whether this i s the best use of the limitedresources available for health. All this said this situation changed after 2007 when the funds for HIV/AIDS dramatically decreased, as the country was not selected for the seventh round o f the Global Fund. It is important to continue funding the fight against this epidemic but in a more predictable and sustainable manner to avoid losing the gains already achieved. These resources whenever possible should be usedto strengthen the health system so it can better support the fight against HIV/AIDS. 18. Most foreign aid for the sector is extra-budgetary which makes it difficult to keep track of the levels and uses of these funds. There are many challenges surrounding foreign aid whichlimits its effectiveness. For instance, donor funds are unpredictable andhighlyvolatile; it is therefore difficult for the sectoral ministries to plan and budget activities. Similarly, this unpredictability and volatility coupled with the low maturity o f external assistance to health also introduce major challenges to the sustainability of activities that are being financed as i s the case of HIV/AIDSfundingdiscussedabove. 19. Inequality in the distribution of donor funds is also a major issue in the sector. The majority of donors and NGOs work in a limited number of provinces and as a result funding for the sector has been distributed very unequally across provinces. In2006, while Ruyigiprovince received US$ 6 per capita in foreign aid, Mwaro received only US$ 0.1. Donors and NGOs usually finance very different programs and thus foreign aid has also been unequally distributed across strategic objectives. This i s mainly due to single-issue funds and single-issue projects; for instance, in 2006 a third o f all donor funds to the sector were allocated to the fight against HIV/AIDS. This last tendency might be changing, but not as a consequence of a better distribution of funds for the sector but due to a large drop infunds for HN/AIDS. 20. All these issues raised by foreign assistance highlight the urgent need for better coordination in the sector to increase the efficiency of external assistance. By financing a strategy for the entire country and not for specific provinces, donors can ensure a better distribution of resources for the sector. Similarly, by financing the sector in a more predictable manner and for longer periods, donors can helpthe sector achieve better outcomes. Inthis regard, the Government and partnersare working towards a sector wide approach (SWAP)to support the health sector. This process i s based on the premise that all partners will support a single and comprehensive health strategy, will agree to a common plan of action, and will work together towards harmonizing procedures. All these activities will lessen some o f the inefficiencies and inequities inthe distribution o f donor support to the sector. 21. Finally, to avoid inefficienciesin the internaldistribution of resources for the sector it will be also essential to have a transparent account of all funds earmarked for health. iv Regardless of whether all resources are included in the SWAp, it is important to have a transparent account o f all funds financing the sector as well as the uses ofthese funds to allow the Government and all its partners to achieve an optimal allocation o fresources. Fiscal Spacefor Health 22. More resources for the sector will be needed for the country to significantly progress towards the MDGs. However, the options to increase fiscal space for health are limited. Debt reliefoffers the largest prospects, especially after the country reachesthe HIPC completion point. These resources are meant to be usedfor PRSP priority sectors, including health. But.there i s no pre-set amount earmarked for any sector and thus the share that will go to healthwill depend on the Government's commitment to this sector, not just reflected in the original budget allocationbut also inthe total amountsthat will be finally disbursed. Public Expenditure Management at theMinistry of Public Health 23. Increasing resources to the sector will not necessarily improve health outcomes, especially for the poor, unless the inefficiencies in the allocation of resources, mentioned above, and the weaknesses in public expenditure management (e.g. weaknesses in budget planning, monitoring, and execution) are lessened. To improve budget management in the MSP a General Directorateo f Resources (DGR inFrench) was created in2006 intime to manage HIPC funds. The DGR has strengthened the capacity o f the MSP to manage resources but weaknesses still remain. 24. The MSP faces many challenges concerning budget planning. Budget preparation i s fragmented. There i s thus a need to consolidate the budget preparation not just in terms o f recurrent and capital expenditure but also interms o f sources o f funds: HIPC, ordinary resources, and eventually donor funds. It is also important for planning purposes to generate a consolidated account of all donor funds supporting the sector and progressively institutionalizeNationalHealth Accounts. The development o f a Medium Term Expenditure Framework (MTEF) and the entire SWAP will facilitate this process. Finally, there i s a needto build capacity for budget planning at differentlevels ofthe MSP, particularly at the provincial level. 25. Concerning budget execution and monitoring, the MSP faces also many challenges. First, the long delays o f the Ministry o f Finance to disburse HIPC funds limit considerably the execution of the budget. Second, delays in the reimbursement to facilities offering the "free" package of services have been drastically reduced but not completely eliminated and thus many facilities still experience difficulties to pay suppliers and the salary o f contractual staff. In this regard, an estimate o f the real cost of the policy to offer "free" services to women and children under five is urgently needed. Otherwise it is not possible to know ifthe budget allocated to this policy is sufficient to reimburse all facilities. It i s also necessary to define more clearly the packageo f services according to priorities and resource availability. Third, the ministryhas some difficulties in the preparation of procurement documents; this i s partly due to cumbersome and centralized procedures but also to weak capacity to processall these documents. Finally, there is a need to improve the capacity of the DGR to monitor all expenditure benefitingthe health sector and notjust expenditure financed out of domestic resources. 26. On the positive side, to help guide budget execution, the MSP prepares a budget execution plan for HIPC resources. This plan could be improved by including all sources o f funds, by prioritizing expenditures, and by including a commitment plan with a chronogram for the fiscal year. V Expenditure Management at the.Health Facility Level 27. The management of resources at the health facility level, and especially at the level of health centers, varies considerably across provinces. In general, facilities do not produce regular budget or budget execution reports. Most facilities manage two bank accounts, one for revenues collected through services and another for revenues collected through the sale of pharmaceuticals. Facilities in general have flexibility inthe management o f the first account. The resources from this account are used for small recurrent costs such as payment of contractual personnel, maintenance, etc. The management o f the pharmaceutical account is more restricted and often escapes from the health center management. Some provinces had functioning drug revolving funds; however, many of these funds were affected by the loss of revenue that followed the presidential measure of 2006. Although drugs were donated to respond to this policy, some o f these drugs were not adaptedto the targeted beneficiaries. 28. I n many instances,part of the revenues collected by facilitieswas sent to the Provincial Health Bureaus to cover some of their running costs. This practice reduces the benefits the facilities can obtain from collecting own revenues as it limits the amount o f resources they can reinvest to improve the quality o f services. In any case, this practice diminished after the presidential measure of 2006 as facilities' revenues decreased and the reimbursement from the central level was largely delayed. Implications of the PresidentialMeasure to Eliminate User Feesfor health services offered to children underfive and women during delivery 29. One of the main policy changes in the health sector to improve maternal and child healthand to diminishfinancialbarriersto access healthcare was the presidentialmeasure of May 2006. Although there has not been an evaluation o f this measure, it is thought to have significantly increased health service utilization among the beneficiaries. However, to maintain this higher level of utilization and to increasethe effective coverage of these services to a larger share o f the population will require additional efforts and additional human and financial resources. This makes more urgentthe needto confront the challenges describedabove. 30. This study estimated that to achieve a 60% effective coverage by 2010 o f highimpact health interventions included in this "free" package o f services about US$ 10 per capita would be needed. This represents about 70% o f total health expenditure in 2006 or about 1.3 times total public expenditure on health (government expenditure plus foreign aid). Increasing the effective coverage of these interventions to 80% o f the targeted population by the year 2010 will require US$ 13 per capita. These results have thus the following implications: . The presidential measure o f May 2006 was a measure to stimulate the demand o f services among women during delivery and children under five. However, the measure was taken before the services were available for the entire population as not all medical personnel or the equipments and supplies needed were in place. As consequence, the increase in effective coverage o f this package o f services cannot take place immediately. This will take time and additional resourcesto not only stimulate demand but also to create the needed supply o f services (e.g. graduate new qualified medical personnel, re-train existing personnel, invest in refurbishing and equippinghealth facilities, etc.). Decisions should then be taken regarding the phasing o f this measure. These decisions need to take into account the government's budget constraint and its capacity to produce and hire new qualified personnel. This will also require a vi prioritization o f expenditure to decide what needs to be done first given the resources currently available. There is also a need to find different ways to mobilize more resources, both internally and externally, as well as improving the execution rate o fthe resourcesalready available. Infrastructure and equipmentrepresent more than 50% of the estimated costs of increasing the effective coverage o f the "free" package o f services. This high cost o f infrastructure is a consequence o f the poor condition o f the majority o f existinghealthfacilities which often lack the necessaryequipment to effectively deliver these services. The remuneration o f health care personnel, in contrast, represents about 12% of the total costs. Simulations of the extra cost o f increasing salaries o f health care personnel show that in certain instances the cost o f increasing salaries i s relatively low although not insignificant. For instance, doubling the salary o f general practitioners will increase the total cost o f offering the "free" package of health services to 60% o f the beneficiaries by an additional US$ 0.06 per capita. Doubling the salary o f nurses in the same scenario implies a cost increase o f US$ 1.06 per capita. As health care personnel represents a small percentage of the total government officials4, improving their remuneration, especially increasing incentives to work in underserved areas, might not represent a heavy fiscal burden. However, as this might create demands inother sectors5,this measure i s politically difficult. The "freeyy package of services includes essentially individual clinical care interventions. Changing maternal, newborn and child health and nutrition outcomes inBurundiwill require an emphasis on preventive actions. Individual clinical care i s important to avoid maternal and child deaths. However, the biggestgains interms of child survival can be made through actions to improve infant and young child care and feeding practices and care of sick children at household and community level accompanied by delivery o f high impact preventive interventions such as immunization, vitamin A supplementation, deworming, and insecticide-treated nets (ITNs). The large attention given to the "free" package o f clinical services might lessen attention to these other interventions that can be provided at the community level or through outreach services. Many o f these interventions are already provided "free o f charge" to the beneficiaries butthey have beenunderfunded andthus the coverage of some o f them i s still very limited. This study estimated that a community based package of interventions coupled with an outreach package will have the same impact as the "free" package of service on child survival but would cost less than half of it. Unfortunately, in the case o f maternal survival the former packagehas only a marginal effect. Health care personnel represent less than 5% of the country's civil servants and less than 3% of all governmentofficials(includingpolicy andmilitary). 'Teachersforinstancerepresent 34% ofgovernment officials and 64% of all civil servants. vii MainRecommendations 1. Total Government allocation to the health sector needs to increase in a sustainable manner for the country to significantly improve health outcomes. 2. But more resources might not result in better outcomes if these resources are not fully spent. It is therefore more important to ensure that the entire budget allocated to health is fully and regularly disbursed bythe Ministryof Finance. 3. Continued efforts to strengthen financial management at the sectoral ministries will also help improve the execution o f the budget. At the moment, weaknesses in procurement and public expenditure managementhave also limited the execution ofthe budget. 4. To ensure the effectiveness in the use o f donor funds, their unpredictability, unequal distribution, and volatility need to be reduced. The recent coordination efforts led by the Ministry of Public Health and the work towards a sector wide approach are important steps inthis regard. This will ensurethat all partners support an integrated health strategy for the entire country. Similarly, an MTEFto support the SWAPi s expected to lessenthe variability and unpredictability of donor funds by making more transparent the sources and uses of funds inthe sector in a specific time period. 5. Given the burden of disease in the country and given that most o f the population lives in rural areas, more funds should be allocated to basic services, mainly primary health services and first referralhospitals. 6. At the moment, as more resources are flowing directly to the health facilities, there is a needto increase community participation intheir management to ensuretransparency and accountability inthe use of these funds. This could be achieved by revitalizingthe Health Committees (COSA) and Management Committees (COGE) to help manage resources at the healthfacility levelandensuresome degree oftransparency. 7. As the supply o f health services was not completely assured before the May 2006 measure took place, its implementation will need to be phased taking into consideration budget and personnel constraints. 8. The package of services offered free o f charge to women during delivery and children under five includes essentially clinical interventions. However, the biggest gains interms o f child survival can be madethrough actions to improve infantandyoung child care and feeding practices and care of sick children at household and community level accompanied by delivery of high impact preventive interventions such as immunization, vitamin A supplementation, deworming, and insecticide-treated nets (ITNs). More funds and attention should go to these interventions. ... Vlll INTRODUCTION 31. The purpose of the BurundiHealth Financing Study i s twofold: to support the Government's strategy to transition from emergency humanitarian response to a sustainable development o f the health system; and to serve as the analytical base for a new World Bank support program in the health sector. These two elementsare developed below: 32. Support the Government's strategy to transition from the emergency humanitarian response to a sustainable development of the health system. The Government aims at developing a sector wide approach with its development partners. As part of this approach it is necessaryto develop a mediumterm expenditure framework agreedby all partners. The first step will then be to have a better account of all flow o f funds to the system as well as their sourcesand uses. This study aims precisely at collecting and analyzing all needed health budget/expenditure data. It also aims at simulating the cost o f alternative packages o f health services, particularly the cost o f the "free" package of services for women during delivery and children under five. This cost simulation will help assess the different implications of the presidential measure of eliminating user fees, not just financial implications but also in terms of human resources and others. In addition, as new resources in the health sector will not be able to provide the desired impact if these resources are not completely absorbed, the study also aims at identifyingbudget management bottlenecks that will need to be reduced or eliminated. 33. Serve as the analytical base for the new World Bank project in the sector. At the moment, a World Bank team is working on a new program to support the health sector in Burundi.Among other things, the programaims at supporting the delivery ofan essentialpackage of health services. This study aims at clearly identifyingthe sources of funds that are already financing some of these services, andtherefore it would help identifythe financial gaps. 34. Based on this purposethe BurundiHealth FinancingStudy's main objectives areto: ... Identifyall sourcesoffunds to the healthsystem bothpublic andprivate. Analyze the efficiency and equity ofthe allocation of funds for the sector. . Identifypublic expenditure managementbottlenecks that hinderthe execution o fthe budgetand recommend mechanisms for reducing or eliminatingmajor bottlenecks in expenditure management; and Simulate the financial cost o fthe package o f services offered free of charge to children under five andwomen duringdelivery and assess the implications ofthispackage. 35. This study i s composed of four chapters. The first chapter gives an overview of the health sector in Burundi: national health plans and policy; health outcomes; health care access and utilization; and system constraints regarding human resources, availability of drugs and supplies, and the quality of the services provided. The second chapter identifies the different sources o f funds for the sector and analysesthe equity and efficiency inthe distribution o f these fknds. This chapter is followed by a chapter that evaluatesbudget expenditure management at different levels o fthe MSP. The last chapter simulates the cost o f extending the coverage o f different packageso f services in particular the package of services, included in the May 2006 presidential measure. 1 This final chapter also examines the financial, human resources, and other implications of this presidential measure. 2 BACKGROUND HEALTH POLICY AND PLANS 36. Reversing the damage that more than a decade of conflict generated in the health of the populationand assuring a sustainable development of the healthsystem is a priority of the Burundianhealthauthorities. The Forum of the General States o f Health (Forum des Etats GBnCraux de la SantC) that took place in Bujumbura from May 31st to June 4th, 2004 started a reflection process betweenthe Government and its development partners on the main weaknesses and challenges of the health system and possible ways to ,overcome them. This process culminated with the adoption of a new National Health Policy 2005-2015 adopted in October 2004 and a National Health Development Plan 2006-2010 in December 2005. These policy documents set the health development strategy o f the country and pave the way for the transition from emergency response activitiesto a sustainable development o f the system. These documents take into account international commitments taken by the government such as the Millennium Development Goals for health. 37. These documents are anchored inthe country's Poverty Reduction Strategy adopted in2006 as well as the InterimPoverty Reduction Strategy completed inNovember 2003. Two of the four strategic prioritiesof the PRSP are directly relatedto the healthsector. These four priorities are to: (i)improve governance and security; (ii) achieve equitable and sustainable growth; (iii) develop human capital by improving the quality of social services; and (iv) reinforce the fight against HIV/AIDS. 38. In accordance with the PRSP, the vision of the National Health Policy is that "by the year 2015 Burundi will enjoy sustainable peace, socio-political stability, and economic growth that will allow all citizens to have access to basic health services through different individual and community mechanisms and under the strengthenedstewardship ofthe Ministryof Health". 39. To implementthe national policy, the NationalHealth Development Plan was drafted with ample participation. The main objectives o f this plan are to: (i)reduce maternal and neonatal mortality; (ii)reduce infant and child mortality and morbidity; (iii)reduce the prevalence o f communicable and non-communicable diseases; and (iv) reinforce the performance ofthe national healthsystem. The NHDP specifies a number o f strategies and indicators for each of these objectives; however, the baselines for many o f these indicators are not available and in some cases the goals or targets to achieve by the year 2010 are not clear (Annex 1 shows the list of NHDP objectives, strategies, and indicators). In addition, the action and strategies in the NHDP are not prioritized making difficult the use o f this document for planning purposes. To solve this issue, the Government is currently working on a more focused health program for 2007-2009. 40. The NHDP aims at developinga sector wide approach to achieve its goals and a better coordination among all health partners. To implement a SWAP the Government and its partners would not only have to agree to a comprehensive sector strategy such as the NHDP but also to an expenditure program that captures most or at least all significant contributions to the 3 sector: a Medium Term Expenditure Framework. At the moment, there i s a high commitment from both the donor community and the Government towards this sector wide approach. Much progress has been achieved to support a common strategy to operationalize the NHDP through a more focused healthprogram. However, work i s still neededto account for all external support, to draft an operational MTEF, to design a monitoring and evaluation strategy, and to harmonize processesamong partners. HEALTH OUTCOMES 41. Ten years o f conflicts brought devastating effects on the health o f the population. With increasing peace and stabilization there have been improvements, but many health indicators have not reached the levels prevalent before the conflict. This section gives an overview o f the health status ofthe population inview o fthe health objectives set by the NHDP. 42. Limited evidence shows that maternal mortality in the country is high. A model estimate from 2005 shows that the maternal mortality rate in the country was about 1,100 per 100,000 lives births (WHO, UNICEF, UNFPA and WB, 2008). This rate is larger than the estimated average for sub-SaharanAfrica o f 920. However, it i s lower than the rates inother post- conflict situation countries such as Rwanda and Sierra Leone, and lower than other neighboring countries such as Tanzania (see Table 1). Without reliable estimates of maternal mortality it is difficultto measureprogress inachieving this goal. However, the percentageof births attended by skilled personnel, one of the MDGs indicators for maternal mortality, can give us an idea of the progress achieved so far. The dark line in the graph below shows the trend inthe percentage o f birthsattended by skilledpersonnel and the dashed line the paththat the country neededto follow to achieve the MDG target o f 100% skilled attended births. As seen in the graph, the percentage o f skilled attended births has increasedbut it is far from the path neededto achieve the MDGby the year 2015. InMay 2006, the Government eliminated user fees for deliveries. This measure is thought to have increasedthe percentageof skilled attended births; unfortunately, data are not yet available. Figure 1: Percentageof birth attended by skilledhealthpersonnel,Burundi, 1987-2005 120 1 ithe 0 I 1985 1990 1995 2000 2005 2010 2015 2020 Source: DHS 1987, MICS 2000, MICS 2005. 43. The risk o f a maternal death increases the more children women have. Burundihas one of the highest total fertility rates in SSA. Burundian women have on average almost 7 children duringtheir reproductive years (Table 1). The country's contraceptive rate is low; the MICS 2005 found that only 9% o f marriedwomen or women inunions usedany type o f contraceptive method andonly 7.5% a modern method. 4 44. Child mortality increasedsignificantly during the conflict and although there are signs of improvements,the current mortality rate is still much higher than that prevalent before the conflict of 1993. One o f the objectives of the NHDP is to reduce infant mortality and morbidity. As seen inthe graphs below, infant mortality drastically increased with the start of the conflict. In2000, about one out of every six infants did not survive his first birthday and one out of every four his fifth. By 2005, one inevery eight infants died before his first birthday and more than one in every six his fifth. Despite these improvements child mortality inthe country is still very highwhen compared to the SSA average in 2003 (172 per 1000)6and, as seen inthe graphs, the country is still far from track ofmeetingthe MDGby the year 2015. Figure2: Trends ininfantand child mortalityrates,Burundi Infantmortalityrate Child Mortalityhte 120 /'% 200 - 120 150. I 100, 500 1985 1990 1995 203 2w5 2010 2015 2020 1985 1990 1995 2000 2005 2010 2015 2020 Source: DHS 1987, MICS 2000, MICS 2005. Note: The MDG baseline is 1990, but the only data available was from 1987. The data on infant and under five mortality for 1987 correspondto the period(1982-1987). 45. Infant and child mortality in the country are higher than the SSA average although they are lower than in other post-conflict situation and poor countries. As seen inthe table below, in 2005 infant and child mortality in the country were comparatively higher than other t countries in SSA; however, these mortality rates were lower than in other post-conflict situation countries such as Rwanda, Sierra Leone, DRC, and Liberia. Table 1: Healthoutcomesindicatorsin SSA countries U5M IMR per MMR per Moderateand Total Fertility Der 1000 100,000 live severe stuntina rate [20071 1000 births % 0-59 months Benin 150 89 840 31 5.46 Burkina Faso 191 96 700 39 6.38 Chad 208 124 1500 29 6.66 DRC 205 129 1100 38 6.71 Ghana 112 68 560 30 3.90 Kenya 120 79 560 30 4.97 Liberia 235 157 1200 39 6.78 Madagascar 119 74 510 48 4.94 Niger 256 150 1800 40 7.56 Rwanda 203 118 1300 41 5.25 Sierra Leone 282 165 2100 34 6.48 SSA 169 101 920 38 Tanzania 122 76 950 38 4.51 ~~ http://www.childinfo.org/areas/childmortality/ 5 Uganda 136 79 550 39 7.11 Source: www.childinfo.org. Note: data for U5M, MMR and IMR is for 2005; stunting data is for more recently available data (in the case of Burundi data on stunting is from MICS 2000). Data on total fertility rate is from UNFPA: State of World Population 2007. 46. Communicable diseases and the compounding effect of malnutrition are the major causes of mortalityand morbidity in the country.Malaria, respiratory infections, and diarrhea are the main causes of deaths and illness for children under five. These diseases can be prevented or cured at a relatively low cost as will be discussed inthe last chapter. 47. Malnutrition rates are also very high, although there are signs of improvement.Even before the start of the crisis, malnutrition rates in Burundiwere high. In 1987, about half o f all children under five were stunted, a measure of chronic malnutrition. As seen in the graph, the percentage o f children stunted increasedthroughout the 1990s. The last survey from 2005 shows an improvement on children's nutritional status although still two out of every five children are stunted. Often malnutrition is not caused by insufficient availability o f food but by household behaviors and practices that can be changed through information, education, and communication campaigns. Some of the healthy practices that can prevent malnutrition are very limited in the country. For instance, only 37% of children under five with diarrhea receive any oral rehydration treatment (MICS, 2005). Figure3: Trends in chronic malnutrition, Burundi HeighVage 1985 1990 1995 2000 2005 2010 2015 Source: Health Sector Note 2006 48. As seen in Table 1,Burundiinthe year 2000 had the highest rate of malnutritionin all SSA countries. Although the situation has improved in the last years malnutrition rates are still very high inabsolute terms and when compared with other similar countries. 49. The NHDP also aims at reducingthe burdenof communicable and non-communicable diseases, particularly, malaria, TB, and HIV/AJDS. Although there are no reliable estimates, malaria i s thought to be the main cause of mortality and morbidity in the country. Data from the EPISTAT 2004 show that more than a third of all children and adult consultations in health centers were due to suspected malaria, similarly about a third of all registered deaths in hospitals were also causedby it. 50. HIV/AIDS also represents a difficult healthand development problem in the country. The last national sero-prevalence survey o f2002 found overall adult prevalence o f 32%;inurban areas the prevalence is 9.4%, in semi-urban 10% and in rural areas 2.5% (MPDR and UNDP 2003). UNAIDS estimations for 2006 indicate an adult prevalence (15-49 years) rate of 3.3%. 6 Additionally, these estimatesshow that there are about 150,000 people livingwith the disease and about 120,000 orphans inthe country due to the epidemic. Finally, tuberculosis, often associated with HIV/AIDS, has progressed steadily in the last years and today is responsible for an important share of adult deaths inthe country. Table 2: HIV and AIDS estimatesfor 2005 1.Estimated 2. AIDS 3.0rphans 4. Trend of HIV prevalence (%) numberof Deaths due to AIDS in young (15-24) pregnant people living women in capital city with HIV Adults and Deaths in Orphans (0- children 2005 adults and 17) currently children livina 2005 " 2005 Global 38 600 000 2 800 000 15200 000 Sub-SaharanAfrica 24 500 000 2 000 000 12000000 Burundi 150000 13000 120000 2000 12.6 2004 8.6 DRC 1 000000 90 000 680000 ... ... ... ... Kenya 1 300000 140000 1100000 ... ... ... ... Rwanda 190000 21 000 210000 2001 9.8 ... ... Uganda 1000000 91 000 1000000 2000 8.5 2005 5.2 United Republicof 1400 000 140000 1 100000 2000 7.5 ... ... Tanzania Source: 2006 Reporton the globalAIDS epidemic, UNAIDSMO, May 2006. ACCESS AND UTILIZATIONOF HEALTH SERVICES 5 1. About a third of the populationdoes not receive health care when in need. Data from the QUIBB2006 survey indicate that 28% ofthe people reporting an illness or injury inthe four weeks beforethe interview did not visited a healthcare provider. The poor were the least likely to receive care when sick or injured.As seen in Figure 4, among the poorest 10% of the population 39% of those sick or injureddidnot receive health care; in contrast, among the richest 10% o f the population only 21% of those with an illness or injurydid not. Figure 4: Percentage of people ill or injured in the last four weeks that did not receive health care when in need, across consumption deciles, Burundi, 2006 45.0 1 40.0 35.0 30.0 25.0 20.0 15.0 10.0 5.0 0.0 dtcile 1 dbcile 2 dbcile 3 d6cile 4 dbcile 5 dbcile 6 dtcils 7 ddcile 8 dbcile 9 dbclle 10 Source :QUIBB2006 7 52. Duringthe conflict Government expenditure on healthand foreign aid drastically decreased. Dueto these large expenditure cuts and in order to increasefunds the government initiated a cost- recovery mechanism inall health facilities. This policy allowed these facilities to finance some of their running costs. However, as household income also decreased and as exemption and waiver mechanisms were not functioning, financial barriers remained the main barrier to access health services. As seen in Table 3, the cost of health services is the main reason why 28% of people with an illness or injury did not receive health care when inneed. Indeed, 78% of the people with an injury or illness indicatedthat the cost of health services was the main reason for non-utilization. Only about 2% of the population mentioned the distance to the facility as a reasonfor non-utilization. Table3: Reasonsfor not receivinghealthcare when in need,Burundi,2006 Has been illor injured (%) not necessary 11.1 too expensive 77.9 too far 1.8 longwaiting time 0.7 not well trained personnel 0.4 lack of drugs 1. I Other 7.0 Total 100 Source: QUIBB2006 53. To improve maternal and child health through improving financial access to health care, in May 2006 the Presidentannounced the elimination of user fees for health services deliveredto children under five andwomen during delivery.This measure is applied inpublic health facilities and in facilities managed by faith-based organizations that have an agreement with the MSP. The exact list o f health care interventions that are meant to be delivered free o f charge is not completely clear. For children under five the measure includes the exoneration o f payment for the following services offered in health centers and first referral hospitals: all outpatient visits, drugs included inthe national list of essential drugs prescribed and distributedin the health facility, lab test prescribedand giveninthe facilities, and all other nursing, medical and chirurgical services. For women duringdelivery the measure includes the exoneration o f payment for the following services offered in health centers and first referral hospitals: all health care linked to delivery, accommodation, tests, drugs included in the national list of essential drugs prescribed and distributed by the healthfacilities, C-sections, and hospitalizations. 54. Although there has not been an evaluation o fthis measure, it is thought to have substantially increasedthe utilization of health services among the beneficiaries. 55. Even before the May 2006 measure was announced, some high impact maternal and child health interventions, mainly preventive ones (e.g. vaccinations, ITNs, and others), were not subject to user fees but, as they have remainedunderfunded, their coverage is still low. For instance, as seen in Table 4 only 8% of children under five sleep under an insecticide treated bed net, only 23% of children with diarrhea receive Oral Rehydratation Treatment (ORT) or increased fluids and continued feeding, and only 45% o f children are fully vaccinated. In addition, the poor are less likely to be covered by any o f these health interventions. For instance, while only 5% of childreninthe poorest 20% of households sleep under an ITN, 19%o f children leaving in the richest 20% o f households do. Similarly, while 73% o f mothers in the poorest wealth quintile are protected against tetanus, almost 80% of mothers in the richest quintile are 8 protected. Antenatal care i s the only interventions in the table with large coverage and no difference among income levels; however, as will be described later on, the quality o f this service is low. Table 4: Coverage of different highimpact healthinterventionsacross wealth quintiles,Burundi, 2005 Full ReceivedORT or Children % of mothers % of pregnant Vaccinations increasedfluids sleeping protected women receiving AND continued under an ITN against tetanus ANC one or more feeding times during pregnancy Poorest 40.5 21.9 4.6 73.2 96.7 II 47.1 18.9 5.3 78.0 95.9 Ill 41.7 19.8 5.8 73.5 96.4 IV 51.7 29.8 7.6 76.3 96.7 Richest 45.6 26.9 19.1 79.1 95.9 Total 45.3 23.1 8.3 76.0 96.3 Source: MICS 2005 HEALTH SYSTEMCONSTRAINTS' 56. To improve the performance of the health sector, the fourth goal of the NHDP, the health authorities face important challenges. The conflict and the continued underfunding of the sector weakened the delivery system. The chronic deficit of qualified medical personnel and stock out of drugs and supplies were (and still are) common. This section will look at some o f these challenges by examining the status of health human resources, the availability o f drugs and supplies, and other issuesrelatedto health service quality. HumanResources 57. The shortage of qualified health personnel is one of the main bottlenecks in the deliveryof quality services. This scarcity affects different types of health personnel. Inthe year 2003*, there were 343 doctors (22 gynecologists, 27 surgeons, 12 internists, and 13 pediatricians), 2101 nurses and assistant nurses, 157 lab technicians, 99 sanitation technicians, 14 dentists, and 68 pharmacists. Other sources of data show even lower numbers of some qualified medical personnel. For instance, the Department of Human Resources' (Direction des Ressources Humaines or DRH) o f the Ministry of Public Health estimates that there are 221 doctors, 946 qualified nurses, 80 pharmacists, 45 midwives, 98 lab technicians, and 1557 auxiliary nurses. With three doctors and thirteen qualified nurses per 100,000 people, the number o f doctors and nurses per population is among the lowest in the region. There i s also a shortage o f public health specialists and managers o f health care services which might hinder in the short term the Government's health reform efforts. The National Health Development Plan reported a deficit of about 26 public health specialists, and 58 health service managers. There is also a large deficit of pharmacists as well as laboratory technicians. 'This section borrows heavily from the Ministkre de la Sante Publique and World Bank 2006 Burundi Health SectorNote. * The sourceo fthese numbers is the Observatoire de 1'Action Gouvernementale2003. 9 Informationpresentedby Director of the department of Human Resources during health partners' joint missioninBujumburainOctober24", 2007. 9 58. Not only are health personnel scarce, but they are also unevenly distributedfavoring Bujumbura and leavingthe poorest areas underserved. In 2003, it was estimated that 80% of the doctors and 50% of the nurses were concentrated in Bujumbura". Similarly, data from the DRH show that 42 out of 87 doctors (52%) working in the public sector work are in administrative positions at the central level of the Ministryof Public Health inBujumbura. Partial data on health care personnel also indicate large disparities in the distribution of health care personnel across provinces. The figure below compares the number of nurses per capita across provinces with an estimated percentage of poor people in each province. As can be seen in the figure, there are large regional differences inthe availability of nurses, the city ofBujumbura has the largest number of nurses per 100,000 people. Similarly, often the poorest regions are the ones with the lowest number o f nurses per population. For instance, Kirundo, Kayanza, and Muyinga, the provinces with the fewest nurses per capita are also some of the poorest provinces. This regional imbalance i s more marked inthe case of doctors. Figure 5: Number of nursesper 100,000 inhabitantsacross provincesand estimated percentageof peopleliving in poverty across provinces 160.0 -- DNurseper100,000+Poverty level rn Source: Data on nurses comes from the MOPH Department of HumanResources data base on personnel 2006 and it also includesnurses working on administrativetasks. This data is not exhaustiveas it includesmainly personnelunder the ministry payroll. Dataon poverty levels is basedon an analysis ofthe QUIBB2006 survey. 59. During the conflict, many qualified health personnel left the rural areas or the country altogether for security reasons.In addition,the country is not producingenough physiciansto meet its needs. Until recently, the Medical School in Bujumbura was the only institution preparing general doctors. This school had produced, on average, about 20 doctors per year, which is not sufficient to meet the needs. The situation i s partly improving as the number of students has recently increased, although not the number of professors. Despite these improvements, the NHDP estimates that there is a deficit o f 250 physicians in the country, including both general and specialized. At the current rate o f graduation this deficit will not be eliminated inthe near future. At the moment, three private universities(Ngozi, Mwaro, and Great Lakes) have started to train doctors; however, their first graduates are expected in about three to four years. More importantly, the diplomas o f these future graduates have not been accredited or validated so far. Although the deficit o f qualified nurses i s larger than that o f doctors, the situation in terms o f production is not as critical. There are three public and five private schools 10 preparing auxiliary nurses (A3); three public schools preparing qualified nurses (A2) and one preparing licensed nurses (Al). However, there is only one school preparing laboratory technicians and thus it will take many years to prepare all neededlab technicians. 60. However,the current shortage and the unequal distributionof personnel is mostly due to an inadequate remuneration system, as incentives are not provided to hire and retain personnel, particularly in underserved areas. These incentives are not only limited to salaries but also to career development, working conditions, and others. Concerning the remuneration level, by 2005 the salary of a general doctor had received only three nominal increases in more than a decade (CBtC, 2005). Between 1989 and 2001 the starting salary (base salary + housing benefits) of a general doctor increased about 50%. However, during that period inflation increasedabout 374%. This situation has started to change inpart due to two civil servant salary increases one o f 10% in 2001 and another o f 16% in 2006. In addition, a 34% increase in allowances will become effective in2008. 61. Despite recent salary increases, high skilled health personnel can find better remunerationoutside the country or working for either donors or NGOs. Donors have paid and in some cases continue to pay salaries and/or allowances to some public service doctors working outside Bujumbura. This has often created disincentives as personnel doing the same work received different remuneration. For instance, doctors receiving civil servant salaries worked side by side doctors receiving much higher salaries paid by donors. To eliminate the scarcity some donors have also sent doctors, such as China, Cuba, Nigeria, and Egypt. 62. Salaries for healthcare personnel are low in absolute numbers but they are similar to other poor countries in the region as multiplesof GDP per capita. As seen in the table below after the 34% increase in allowances, the salary o f qualified health personnel in Burundi as multipleofper capita GDP will be similar to that of countries inthe region. Table 5: Health care personnelsalariesas multipleof GDP per capita in SSA countries Country General practitioner Diploma Nurse Burkina Faso 7.3-23.5 4.2-13.5 Burundi 10.8 4.2-9.6 Cameroon 4.7 1.7 Chad 10.3-18.8 5-10.6 DRC 1.8-2.40 Ethiopia 18.3-30 11.7-27 Kenya 17 6,9 Mauritania 5.67-9.45 3.2-5.7 Niger 10.6-20.8 5.3-12.0 Zambia 22,6 5,4 Source: Burundi calculations were based on total remuneration o f health care personnel including the increase in allowances that will become effective in January 2008. The other countries data are from WE3 Human Development, Africa Region, and Country Status Reports. For Zambia and Kenya : WDI, Country case study on health worvorce financing and employmentin Kenya (forthcoming) and Zambia report on human resources for health (forthcoming). 63. Increasing the remuneration level might not change the distribution of health care personnel if no incentives are given for personnel to work in underserved areas. The Government i s currently trying to generate incentives to all medical personnel through 11 performance based agreements and through payments o f 'primes d'kloignement" for doctors that accept to be posted insome district hospitals. 64. Nevertheless, after the 34% increase in allowances, changing the remuneration level, including increasing incentives to serve outside Bujumbura, will be challenging. With this new increase in allowances, the wage bill will represent about 11% o f GDP, one o f the largest wage bills in the region (PER, forthcoming). This leaves the authorities very little room for maneuver. Nevertheless, health care personnel represent a very small share of the total civil servants in the country and some incentives or increased compensation in this sector could be possible if these measures do not produce pressures in other sectors. The personnel o f the Ministry ofHealth represent less than 5% of all civil servants inthe country and less than 3% of all government officials, including the police and the military. A limited increase in the compensation might not represent, in relative terms, a large fiscal burden. However, if this increase creates pressures in other sectors this will not be fiscally sustainable. For instance, teachers represent more than 34% of all government officials (64% o f public servants); while the police and the military represent about 17% and 26% respectively. A possible way around these issues is through the use of contractual staff that could serve in underserved areas; linking these contracts with performance targets might also bringother quality improvements. Availability of Drugs and Supplies 65. The continued supply and distribution of pharmaceuticalproducts to health facilities i s not guaranteed. In 2005, a survey" that randomly sampled 20% o f health care facilities in 10 provinces found a large percentage of facilities with stock-outs of pharmaceutical products. For instance, in only 39% of facilities sampled gentamicine was available. Health facilities managed by faith-based organizations or "ugre'e'"were more likely to have the drugs and supplies available than public facilities and private for profit facilities (Table 6). Table6: Percentage of healthcare facilitieswith pharmaceuticalproducts available across type of facility ownership, Burundi2005 Public Agree Private TOTAL Artesunate+Amodiaquine 77.4 91.3 40.0 75 Quinineoral 82.3 95.7 60.0 82 Les deux medicaments 69.4 91.3 33.3 69 Benzylpenicilline 66.1 60.9 46.7 62 Quinine injectables 59.7 82.6 73.3 67 Chloramphenicole 43.5 43.5 40.0 43 Gentamycine 33.9 39.1 60.0 39 Serum glucose 67.7 69.6 73.3 69 Serum physiologique 32.3 43.5 46.7 37 Lactatede Ringers/lV liquides 48.4 47.8 66.7 51 Diazepam 46.8 73.9 80.0 58 Source: Ministkre da la SantC Publique, 2005. ((Evaluation de la prise en charge des cas de paludisme dans les structures de soins confomkment au nouveauprotocole de traitement antipaludiqueD. 66. After the presidential measure of May 2006, large stocks of pharmaceuticals were donated to both public facilities and those managed by faith-based organizations. An evaluation made on January 2007 on the utilization and distribution o f the drugs donated by ~~ "Ministere de la SantC Publique. 2005. ((Evaluationde la prise en charge des cas de paludisme dans les structures de soins conformement au nouveauprotocolede traitement antipaludique D 12 ECHO also found stock-outs of some drugs and supplies (WHO and MOPH, 2007). In hospitals, this evaluation found stock outs of Amoxicycline (30% of hospitals), syringes (30%), gloves (30%), serum glucose (30%). In public health center the same evaluation found the following stock outs: Amoxicycline sirop (59% of health centers), ORS (30%), Metronidazole sirop (28%) tablets (22%), Paracetamol250 mg(26%) and lOOmg(24%), and Promethazine tablets (20%). Low Quality of the Services Offered 67. Other indicators also corroborate the low quality of the services offered. In 2003, Me'decins sans FrontiBres (MSF) surveyed 1,383 patients o f Primary Health Care (PHC) services in rural areas, stratifying the sample by type of user-fee policy in place.12Data from the three samples o f patients indicate that only one in every ten patients received a clinical examination when visiting a PHC facility; less than 45% was offered a diagnostic; and in less than 40% o fthe cases their temperature was taken. The same survey found that in all samples, on average, a consultation lasted less than seven minutes. Similarly, the MICS 2005 survey found than among women that received antenatal care only 8% had a blood sample taken, 47% had their blood pressuremeasured, and only 6% had a urine sampletaken. Figure6: Satisfactionwith public healthservices, Burundi, 2004 Police Maternity 39.2 Health center 39 1 49.8 Secondary schools Primary schools Public transportation 0 10 20 30 40 50 60 70 80 Inon-poorIpoor Source:PovertyAssessment forthcomingusingdata from perceptionofpoverty survey 2004 68. Health services have the lowest satisfaction rates among all public services in the country. The Perception o f Poverty survey 2004 showed that less than 50% o f Burundians were satisfied with primary health care centers and less than 40% with maternity hospitals. These are the lowest rates received by any other public service on which information was collected. In the case of primary health care centers, the poor are less likely to be satisfied with them, while inthe case of maternity hospitals the poor are slightlymore likely to be ~atisfied'~. 69. The cost of services is the major cause of dissatisfaction with health services, followed by ineffective treatments, long waiting time, and lack of drugs. The QUID 2002 survey collected information on utilization o f health services, satisfaction with these services, and reasons for dissatisfaction. About 60% o f patients that visited any health facility were satisfied with the services received. Among those users that were not satisfied, three out o f every five l2 Health facilities were selected randomly (with probability proportional to population size in the surroundingarea) andhouseholdswithin 5 kmwere sampled. l3This survey did not collect information onthe reasons why peoplewere not satisfiedwith services. 13 mentioned high cost as the main reason for their dissatisfaction. The situation should have changed since May 2006 as both childrenunder five and women during childbirthare exempted from any payment. One out of every five complained about the ineffectiveness of treatment and lack of drugs in the facilities. A similar percentage of people complained about the longwaiting time. The latter is not surprising; the 2003 MSF survey reported that the median waiting time in the three differentsampleswas about an houranda half. 14 HEALTHEXPENDITURE 70. This section examinesthe trends, sources, and uses o f funds for the health sector inBurundi. In particular, it examines the distribution of funds for the sector and issues concerning each source of funds: Government, donors, and private out-of-pocket expenditure. Government Expenditure on Health 71. This section examines trends in Government's health expenditure from 2001 to 2007. The analysis is mainly focused on the Ministry o f Public Health and to a lesser extent on the Presidential Ministry for the Fight against HN/AIDS (Min SIDA in French). In this analysis, public expenditure on health has been defined as all expenditure managed by the MSP, Min SIDA, and expenditure linked to the University Hospital Center of Kamenge (CHUK in French) which is under the responsibility of the Ministryo f Education. These ministries are not the only financing agents managing government's resources for health. However, it was not possible to identify with the data available the resourcesthat other ministries allocate exclusively for health activities. 72. This section analyzes public expenditure from government own resources. Although a portion of donor funds0 for the sector are registered in the budget, the Ministry o f Finance (MinFin)does not monitor this expenditure andthus only information on projections is available. The present analysis is basedon data from the Ministryo f Finance but the data are complemented with information from the MSP, the World Health Organization (WHO), and the International Monetary Fund(IMF). Trendsin Governmentexpenditureon health 73. Government expenditure on healthdrasticallydiminished during the years of conflict as government revenues and the share of the budget allocated to health decreased. The political crisis starting in 1993, the three year embargo imposed after 1996, and the fall in the price of coffee brought large reductions in the country's GDP. As seen in table below, the country's GDP decreased more than 25% between 1992 and 1996. Public expenditure on health was then drastically reduced; but not only due to the government's reduced revenue base but also due to a smaller share of the budget allocated to the sector. Before the crisis, more than 4% o f total government expenditure was earmarked for the health system. Since 1996, less than 3% of government expenditure went to health care. By 2000, the government spent less than one dollar per capita on health; less than halfthe per capita amount that it usedto spendbefore the crisis. 15 Table 7: Governmentfinancial indicators,Burundi, 1992-2000 1992 225.6 258.1 38.2 43.7 19.4 22.2 4.4 2.6 3.0 2.17 1993 227.9 245.1 39.8 42.8 20.7 22.3 4.4 3.0 3.2 2.10 1994 233.3 233.3 42.2 42.2 5.7 5.7 3.6 2.1 2.1 1.38 1995 249.9 216.7 44.1 38.2 9.1 7.9 3.1 2.0 1.7 1.30 1996 272.6 198.7 42.3 30.8 6.1 4.4 2.5 1.9 1.4 1.oo 1997 337.3 199.4 46.5 27.5 10.3 6.1 2.9 2.1 1.2 0.93 1998 400.2 212.4 68.5 36.3 3.6 1.9 2.6 2.5 1.3 0.85 1999 455.5 210.2 74.0 34.1 11.4 5.3 2.3 2.7 1.2 0.72 2000 511.1 208.3 98.3 40.1 15.8 6.5 2.3 2.8 1.2 0.58 Source:IMFand WB staffcalculationsbasedon IMFandWB data. Note: the health expenditure in this table includes expenditure that was included in the nationalbudget, regardless of the sources ofthe resources.For instance, expenditureon IDA grants and credits are included. 74. By 2002, Burundiwas the sub-SaharanAfrican countrywith the lowest expenditure on health as percentage of total Government expenditure. Inthe first years o f the period under examination (2001-2005), the budget and e~penditure'~ health as percentage of GDP and as on percentage o f total Government budget and expenditure remained constant at about 1% and 3% respectively (see Table 8 and Table 9). As seen in the graph below, this was the lowest expenditure on healthout oftotal government expenditure inthe region. Figure7: Public expenditure on health out of totalGovernmentexpenditure,2002 20 18 16 14 12 10 8 6 4 2 0 f Source: WHO statisticalinformationsystem 75. Governmentexpenditure on health markedly increased after 2005 but it is still among the lowest in the region. With peace and stabilization in the country, things have started to change; especially since 2006 when the country started to benefit from the enhanced Heavily Indebted Poor Countries Initiative. That year the budget allocated to health more than doubled reaching 3% o f total GDP and about 7% o f total Government budget (Table 8). The actual l4Information on actuals is based on commitments as information on payments was not fdly available. 16 expenditure on health increased but not as much as the budget due to some difficulties in executing that year's budget. However, despite all these increases, the total resources allocated and spent on health are still very low. In that year, budget allocations to the health sector out of domestic resources reached about US$2 per capita, twice as much as the average o f the previous decade. However, actual expenditure in2006 was only US$ 1 per capita (Table 9). 76. The increased allocations to heath care showed the Government's commitment to two of the PRSP priorities: improving the quality o f social services and reinforcing the fight against HIV/AIDS. The total budget allocation for the health sector increased by 277% in real terms, while the total Government budget increased only by 36% (see Table 8). The HIPC allocation increased the Min SIDA budget by 2511% that year, and the MSP budget by 246%. The MSP received 75% o f all HIPC resources for health and MinSIDA the remaining 25%. Table 8: Trends in Government budget for health (real and nominal), 2001-2007 2001 2002 2003 2004 2005 2006 2007 Total health budget, in 4288 4753 4724 6077 6326 24613 19865 millionsof FBU (current prices) Total health budget, in 4288 4673 4164 4943 4414 16624 12850 millions of FBU (constant 2001 prices) Total health budgetas % 3% 3% 3% 3% 2% 7% 6% of total Government budget Total health budgetas % 1% 1% 1% 1% 1% 3% 2% of GDP Annual rate of growth of 9% -11% 19% -11% 277% -23% health budget (real) Annual rate of growth of 11% -1% 29% 4% 289% -19% health budget (nominal) Annual rate of growth of 26% -7% 13% -2% 36% -8% total budget (real) Annual rate of growth of 0% 10% 13% 3% 109% -2% social sector budget (real) Health budget in FBU per 635 668 578 673 589 2,176 1,649 capita (constant2001 prices) Annual rate of growthof 5% -13% 16% 12% 269% -24% I per health budget (real) Per capita health budget in 1 1 1 1 1 2 Nd US$. Total budget in millions of 138,942 177,405 184,172 225,443 256,788 360,975 347,917 FBU (current prices) Total budget in millionsof 138,942 174,429 162,324 183,379 179,174 243,808 225,048 FBU (constant 2001 prices) Note Is: GDP in billionsFBU 521.5 544.7 538.1 564.1 569.2 598.4 631 (constant 2001 prices) GDP in billionsFBU 550 584.6 644.2 731.5 860.8 934.6 1,029.4 (market prices) 0 GDP Annual rate of growth 4% -1% 5% 1% 5% 5% (realterm) GDP deflator (base = 105.5 107.3 119.7 129.7 151.2 156.2 163.1 l5Ministry of Finance inBurundiand IMF, World Economic Outlook data base, April 2007. 17 Populationin millions 6.75 7 7.2 7.344 7.491 7.641 7.794 Exchanae rate FBUlUS$ 830 931 1083 1101 1082 1029 Nd (periodlaverage) Source: Ministry of FinanceandWorld Bank estimation Table 9: Trends of public expenditure on health (commitment base), 2001- 2006 2001 2002 2003 2004 2005 2006 Public expenditure on health as % of total public 3% 3% 3% 2% 2% 4% expenditure. Public expenditure on healthas % of GDP 1% 1% 1% 1% 1% 1% Rate of growth on publicexpenditureon health (real) 6% -9% 14% -10% 85% in %. Annual rateof growth on publicexpenditureon health 8% 2% 24% 5% 91% (nominal), in %. Annual rateof growth of publicexpenditure(real), in 13% -2% 22% -8% 15% % Public expenditure on health in per capita FBU 630 642 570 640 562 1,018 (constant 2001 prices) Public expenditure on health in per capita US$ 0.8 0.7 0.5 0.6 0.5 1.o Annual rateof growth publicspendingon health per 2% -11% 12% -12% 81% capita, realterm. Public expenditureon health MSP and AIDS in FBU 546 553 498 570 500 920 per capita (current prices) Public expenditureon healthMSP and AIDS per 0.7 0.6 0.5 0.5 0.5 0.9 capita in US$ Total governmentexpenditure in billions FBU (current 138.9 160.0 174.5 231.2 247.1 293.6 prices) Total governmentexpenditure in billions FBU 138.9 157.3 153.8 188.0 172.4 198.3 (constant 2001 prices) Source: Ministry ofFinanceandWorld Bank estimations 77. The resources allocated to the health sector also increased as a proportion of total expenditure to the social sectors.16Duringthe crisis, the resources allocated to health decreased not only as share of total expenditure but also as a share of the social sectors expenditure. However, this trend also changed after 2006 thanks to HIPC resources. As seen in Table 10, the total budget of the MSP increased from an average of 13% of the social sectors budget in the previous five years, to 18% in 2006. In contrast, the budget from Min SIDA increased from almost nothingto about 4%. Table 10: Share of allocated.resources in social ministries out of total budget in the social sector, Burundi, 2001 - 2007 2001 2002 2003 2004 2005 2006 2007 Min NationalEducation 74% 82% 84% 83% 85% 75% 80% Min SIDA 0% 0% 0% 0% 4% 2% Min Public Health 14% 14% 12% 13% 11% 18% 15% Ministereactionsocialeet de la promotion de la 1% 2% 1% 1% 1% 0% 0% femme Min21la &insertion et a la &installation des 7% 0% 1% 1% 1% 0% 0% deplaceset des rapatnes Min. of Youth, Sports, and Culture 2% 1% 1% 1% 1% 1% 2% Minartisanat, de I'EnseignementdesMetierset 0% 2% 2% 2% 2% 0% 0% de I'alphabetisation l6Inthis analysisthe social sector budget is the budget of the following ministries: Education, Youth and Sports, Reinsertion and Repatriation, National Solidarity, Fight against HIV/AIDS, Public Health, Handicraft(Artisanat), andCraft andAlphabetization. 18 Min. of NationalSolidarity 0% 0% 0% 0% 0% 2% 2% Budgetallocationto health as percentageof 16% 17% 14% 14% 12% 23% 18% budget allocationto all social sectors Budget allocation to health, real annual rate of 9% -11% 19% -11% 277% -23% growth Budgetallocationto social sectors, realannual 0% 10% 13% 3% 109% -2% rate of growth Source:Ministry of Finance andWorld Bank estimations Agencies managing Governmentfundsfor health 78. The MSP is the main financing agent managingGovernment resources to the health sector but not the only one. As there are no National Health Accounts in the country, it is not possible to have a complete estimate of total public resources allocated to the sector. With the data available we can only identify as health expenditure the expenditure managed by the MSP, Min SIDA, and a small part of MinEducation expenditure. These three ministriesmanage most Government resources for health; however, this is still an underestimation as other ministries such as Defense, Interior, Public Administration, and National Solidarity" also have expenditure on health. Nevertheless, based on the data available, MSP manages more than 80% o f the funds allocated to health, followed by Educationuntil 2005 and by Min SIDA after that year. The MSP i s likely to manage the large majority of funds going to health, but of course the actual share will be smaller once expenditure of the remainingministries are taken into account. Table 11: Health sector budget across ministries,shares, 2001-2007 2001 2002 2003 2004 2005 2006 2007 Min Public Healthbudget as 87% 83% 85% 87% 87% 80% 80% share of total health budget Min SIDA budget as share of 3% 3% 2% 2% 16% 12% total health budget Min Education(CHUK) 13% 14% 13% 10% 11% 5% 7% budget as share of budget on health. Total 100% 100% 100% 100% 100% 100% 100% Source: Ministry ofFinance andWorld Bankestimations 79. The healthfunds managedby the Ministryo f Education finance subsidies and equipment for the University Hospital Center of Kamenge. This hospital budget represents a disproportionallylarge share of government budget to the sector, equivalentto about 12% of the total resourcesallocated to the MSP in the last years. The recurrent budget allocated to this hospital has been equivalent to more than 40% of the total subsidies allocated to the 17 hospitals under management autonomy (see table below). Ina country where the main causes o f mortality and morbidity can be prevented and treated at lower levels of care and where most o f the population lives in rural areas, this represents an inefficient and unequal allocation of resources. This unequal allocation i s also apparent when compared to health care utilization patterns. Data from the QUIBB 2006 survey indicates that the majority of the population (57%) uses public health centers when sick or injured. In contrast, only 11% of people visited a public hospitalwhile the rest visited a private provider. l7Ministry of Defense, Direction des services de sante' de I'arme'e; Ministry of Interior, Direction des services de sante' de la Police; Ministry of Public Administration and Social Security, Mutuelle de la fonction publique (MFP); Institut national de se'curite'sociale (INSS); and Ministryo f National Solidarity, Fonds Social et Culturel. 19 Table 12: Recurrent budget allocated to public hospitals (FBU million), MSP and Min Education, Burundi2004-2007 2004 2005 2006 2007 Total recurrentbudget allocatedto hospitalswith management 822 850 1,113 1,276 autonomy inthe MSP in million FBU (currentprices) Recurrentbudget allocatedto the CHUK underthe Min Educationin 497 542 775 976 million FBU (current prices) Total recurrentbudget allocatedto public hospitalsin million FBU 1,319 1,392 1888 2,252 (current prices) Recurrentbudget allocatedto CHUK as share of total recurrent 43% 38% 41% 43% budget allocatedto hospitalsin % Total recurrenthealth budget 5,373 5,655 21,422 16,443 Recurrentbudget allocatedto these hospitalsas percentageof total 25% 25% 9% 14% health recurrent budget Source: Ministry of Finance and World Bank estimations 80. The main causes of mortality and morbidity in the country and especially among children under five can be prevented and treated through high impact health interventions that can be delivered at community level or in primary health care facilities. In the case o f maternal mortality, access to basic and comprehensive emergency obstetric care is also important. This care can be provided in primary health care facilities and in first referral hospitals. However, as shown in the table above, a large share o f the budget i s allocated to higher levels of care. In Burundi, the majority of hospitals with managerial autonomy are not first referral hospitals but second or third referral hospitals. As seen inTable 12, these hospitals usedto receive as much as 25% of the total recurrent allocation to health before 2006. After that year and thanks to HIPC funds the situation started to change. In2006 and 2007, the budget allocation to those hospitals represented9% and 14%ofthe total recurrent allocation to the sector. 81. Similarly, resources allocated to the fight against HIV/AIDS represented in 2006 16% of domestic resources allocated to health care, or equivalent to 20% of the budget of the MSP.Even though resources will still be needed to fight this epidemic; without a functioning health care system there will not be as much progress inthis fight. Inconsequence, and given the large resource constraints and the epidemiological profile of the country (adult HIV/AIDS prevalence is about 3.3%), a discussion i s needed to ascertain whether this is the best use o f the limitedresourcesavailable for health. Structureof MSP budget 82. Between 2001 and 2005 almost all budget allocations to the MSP were used for recurrent costs, mainly salary payments. During this period, more than 90% of the entire budget from domestic resources was used for running costs. Most o f this expenditure was earmarked for the payment o f personnel remunerations. Salaries represented about 40% o f the recurrent budget. This i s however a large underestimation of the total expenditure on personnel. Subsidies to hospitals and other autonomous agencies (Public Health Institute, the Center for Blood transfusion) represented the second largest recurrent expenditure during this period; an important share of these subventions is also usedto pay personnel remunerations. However, there is no centralized account on how the autonomous agencies use these resources. Table 13: Evolutionof recurrentand capitalbudgetsout of Government funds, 2001-2007,in % 2001 2002 2003 2004 2005 2006 2007 Ministrv of Public Health Recurrentbudgetas percentageof total budget 92 96 96 91 92 86 81 Salaries 46 43 42 36 35 16 36 20 goods and services 27 27 27 33 19 64 30 Subventions 27 30 30 31 46 20 34 Capital budget as percentage of total budget 8 4 4 9 8 14 19 Total 100 100 100 100 100 100 100 Presidential Ministrv for the fiaht aaainst HIV/AIDS Recurrent budget as share of total budaet 24 37 40 41 95 100 Capital budget-as share of total budget 76 63 60 59 5 0 Total 100 100 100 100 100 100 Source:Ministry ofFinanceandWorld Bank estimations 83. As seen in a previous section, despite such a large share of the budget earmarked for salaries, salary levels remainedlow. As consequence, many donors paid or complemented the salaries o f some public healthworkers to improve their availability especially outside Bujumbura. 84. Despite a relatively large share of resources earmarked for pharmaceuticals, medical supplies, and the maintenance of facilities, in absolute numbers, the amounts are still insufficient. Indeed, the central level is supposed to cover the running cost o f the Provincial Health Bureaus (PHB) and of the facilities attached to them. Previously, PHB received a small budgetary allocation to cover the costs o f gas, wheels, and others. They have stopped receiving this allocationalthough now they can requestfrom the central level ofthe MSPthese resources in kind.Nevertheless, these resources are limited and do not cover the needs o f the PHB let alone those o fthe health facilities under their supervision. 85. Duringthe conflict and as consequence o f the insufficient resources allocated to the health sector, the MSP started a cost recovery policy for the large majority o f the services and pharmaceuticals provided in public facilities. The revenues collected through user fees allowed facilities to finance some small recurrent costs such as maintenance and the salaries o f support personnel. In many provinces, the PHB retained a percentage o f the facilities' resources to cover part o f the PHB runningcosts. In M a y 2006, some o f these fees were eliminated. However, the reimbursement to facilities for free services was largely delayed in 2006 due to a highly cumbersome reimbursement system. This delay created difficulties for health facilities as their revenues decreased and they were thus not able to pay for some o f their runningcosts. Similarly, facilities stopped the payments to the PHB. 86. Autonomous hospitals must cover their running cost out o f their own revenues, which includes the subsidies received from the central level. Delays in reimbursements also affected hospitals whose debts with suppliers such as CAMEBU (Centrale d'Achat des Mkdicaments Essentiels du Burundi) and the REGIDESO increased. 87. Thanks to the HIE'C initiativein 2006, the resources allocated to non-salary recurrent costs markedly increased. That year goods and services represented about 60% of total recurrent expenditure. Although part o f the HPC resources has been used to create incentive payments to health personnel and to hire needed professionals, most o f these resources have been used to increase payments for other running costs. For instance, the reimbursement to health facilities for the services offered free o f charge includes payment for the medical supplies used when offering the services. 88. The flow of pharmaceuticals follows a slightly different pattern. There is a budgetary line in the MSP to subsidize drugs and medical supplies. In 2006, that line included about FBU 700 million. But this line is not enough to cover all needs. Most facilities have financed part o f their pharmaceutical needs with revenues collected from pharmaceutical sales. Insome provinces, 21 there are functioning drug revolving funds that helped facilities maintain and improve their stocks. Inaddition, facilities have benefited from large pharmaceuticaldonations includingrecent donations from ECHO and DFID to help implement the policyto eliminateuser fees for women duringdelivery and childrenunder five. But even before this measure was implemented, donors supported the procurement o f some supplies such as vaccines (EPI/UNICEF), condoms (FP), ORS (UNICEF), ACTS(GF), andothers. 89. In some years the MSP budget includeda subsidy to the CAMEBU. The CAMEBU is an autonomousagency, but for a few years itwas notableto cover all its costs. HIPC resources 90. HIPC funds for the MSP constitute an important source of additional resources, representing about 53% of the total budget of the MSP in 2006 and 48% in 2007. These resourcesallowedthe MSP to implement the policy of offering services free of charge to women duringdeliveryandto childrenunder five. Table 14: Breakdown of MSPHIPC funds, Burundi, 2006-2007 2006 2007* Recurrent Capital Total Recurrent Capital Total spending spending spending spending spending spending MSP HlPC Budgetin 9,844 484 10,329 4,925 2,775 7,700 millions FBU MSP HlPC Expenditures- 3,093 3,709 3,130 1,193 2,096 3,289 Commitmentbase in millionsFBU MSP HlPC Execution 31% 8% 30% 24% 76% 43% rate, in % MSP execution rate 57% 4% 50% 13% 16% 14% (commitmentsbase) in % MSP HlPC budget -50% 473% -25% annual variation, nominal term in % MSP HlPC commitments -61% 5552% 5% annual variation, nominal terms in % Shareof MSP HlPC 74% 76% credit to total HlPC credit in % Shareof HlPC budget to 58% 18% 53% 38% 93% 48% total MSP budget *Source:Ministry ofFinance andWorldBank estimations Expenditurefor 2007 only includecommitments done upto May31" ofthatyear. 91. In 2006, the MSPfaced many challenges in the management ofHIPC resources.Before 2006, the budget o f the MSPwas mainly composed of personnel remuneration. Most investment expenditure was financed and oftenmanageddirectlyby donors andtherefore the MSP had little recent experience in resource management. There are several reasons that explain the low execution rate of HIPC allocations in 2006. First, the Ministry of Finance largely delayed the disbursement of these funds that year; disbursements only started after June. There were also many delays in the reimbursementsto healthfacilities for the services offered to women during delivery and children under five. Additionally, there were problems in the preparationof tender offersfor large procurements,as will be detailed inthe nextchapter. 22 92. As consequence of the delays in the utilization of HIPC resources in 2006, in 2007 the HIPC allocation to the MSP was reduced. The total HIPC allocation to the MSP was reduced by about 25% in the 2007 budget. MinFin promised to re-visit these allocations during the revision ofthe budget; however, the budgetwas never revised. 93. The execution of HIPC funds has markedly improvedthis year, 2007. The mechanism to reimburse facilities for the payment o f "free" services was largely simplified by efforts from both the MSP and MinFin.As a consequence, by July 2007 the budget line for the reimbursement of services was almost finished (98%). This high execution rate was partly due to the use o f this budget line to pay for arrears from 2006. Other budget lines have also been almost completely used but in many cases this was also due to the use of these lines to pay arrears from 2006. For instance, the investment allocation has beenused to pay advances for large commitments made in 2006. The H P C resources allocated to MSP in 2006 that were not used were not re-allocated to the ministryin2007 as their HIPC allocation inthe initialbudget was reduced. 94. As the budgetwas not revised and the MSP did not receive the extra resourcesagreed previously with MinFin, the ministry has had serious difficulties in fulfilling some of its responsibilities. As mentioned before, both the budget line for the payment o f "free" services, as well as part ofresources for investments were used mainly for the payment o f arrears o f 2006. As the revision o fthe budget never took place the ministrycurrently has serious cash constraints. In addition to these budget lines, MinFin had also agreed to add FBU 1.5 billion in the revised budget to ensurethe stock of drugs to replace DFIDpharmaceutical donation. These resources are a condition to complete the DFlDgrant. 95. I t is not clear if the resources allocated to the reimbursement of "free" serviceswill be enough to cover the cost of the entire program. Even if MinFin would have added the extra FBU 1.5 billion to a revisedbudget, it is not clear ifthese resourceswould have been enough to cover the reimbursementof all services delivered. 96. In general, the distribution of HIPC funds could be improved to better follow the priorities set by the PRSP and the NHDP. A large share of the budget was allocated to the reimbursement of facilities for services offered "free" of charge to women and children. These resources allowed an increase inutilization o f health services. Some o f these resources were used to create payment incentives and to recruit health personnel to work in underserved areas. However, some of the resources did not follow the set priorities. For instance, at the end o f 2006 a large share o f the HIPC resourcesthat were not committed was used to buy 72 vehicles instead o f beingusedto pay arrears to the facilities offering "free" services. Similarly, some o f the biases inthe allocation of resources continued, such as large expenditure in one hospital or ina disease specific program. Budget Execution 97. With the exception of 2006, the budget execution rates of the MSP have been high. Before benefiting from HIPC resources, salaries constituted the largest share o f the MSP budget. Salary execution rates were generally high. The only year when the ministry had significant problems in executing its budget was 2006 for the reasons explained above. It is however important to note that there are large differences between the data from the MSP and from the Ministry of Finance regarding the execution rate (commitment base) that year. The MSP data shows much larger execution rates. Part of the difference is due to commitments made by the MSP and signed by MinFin at the end o f the year that MinFin did not classify as commitments 23 for 2006. MinFin did this to avoid increasing domestic arrears as they considered that the execution o fthese commitments would take place in2007. Table 15: Trendsof MSPspendingexecution rate (commitment base), in %, 2001-2006 2001 2002 2003 2004 2005 2006 On MSP recurrent spending 99 97 100 96 96 57 salaries 99 97 101 101 102 97 Goods andservices 96 99 100 88 79 36 Subsidies 100 95 100 98 99 95 On MSP capital spending 102 99 66 81 82 4 On MSP total spending 99 97 99 94 95 50 Note: Total Governmentcapitalspendingexecution rate 72 81 88 95 89 96 Total Governmentrecurrentspending executionrate 102 91 95 103 96 86 Source: Ministryof Financeand World Bank estimations ExternalAssistanceto the Health Sector Donorfun& in Burundi 98. The decade of conflicts and civil war during the 1990's was characterized by a massive withdrawal of external assistanceto Burundi (see Table 7 in previoussection). From 1992 to 2000, grants to Burundiregistered in the budget decreased, in real terms, from FBU22.2 million to FBU6.5 million. 99. From 2000 onwards, with the stabilization of the country, aid flows progressively increased. Net Official Development Assistance" (ODA) to Burundi increased from US$ 227 million in2003, to US$362 million in2004 andUS$365 million in2005 (Figure 8)". Duringthe same period, Burundibenefited from rapid growth and increasing public and private financing. As a result, ODA as a percentageo f gross national income (GNI) dropped from 55.9% in2003 to 46.8% in2005. 100. In 2005, the overall ODA per capita flowing to Burundi was a little over the average for Sub-Saharan Africa countries. Burundireceived approximately US$ 48 per capita in 2005 while SSA countries received on average US$46. 101. Althoughthe countrygets more externalassistance per capita than the average of SSA countries, it gets less aid than other post-conflict countries. In 2005, Sierra Leone received US$ 62 per capita, Rwanda US$ 64, and Liberia US$ 72. As external assistance to Burundi is transitioning from emergency to development assistance, it is not clear whether this relatively high level of overall external financing will be sustainable. At the moment, half of all funds Burundireceivesfrom external assistanceare for emergencypurposes. l8Accordingto the OECD DevelopmentAssistance Committee(DAC), ODA is made of "grants or loansto countries and territories on Part Iof the DAC List o f Aid Recipients(developingcountries) which are: (a) undertaken by the official sector; (b) with promotion of economic development and welfare as the main objective; (c) at concessionalfinancial terms. Inaddition to financial flows, Technical Co-operation(q.v.) is includedinaid. Grants, Loans andcredits for military purposesare excluded". l9OECD-DACdata. Countryat aglance. 24 102. There are many donors active in social sectors as well as in infrastructure, agriculture o r emergency assistance. In 2004-2005, the main donors in Burundi were the European Union, the World Bank andthe United States o fAmerica (Figure 8). Figure 8: ODA flowing to Burundi(2003-2005) W ~ W ~ ~ ~ Receipts 2003 200-1 2005 flet ODA (USDmillion) LL8 ---, -'. 7:-L - _ _ 7; C -F? i r Bilateral share [gross ODA) c-0: 51'; " i ' C 58 Net ODA GlII 5`; i`, a >E,<: - 49 43 !let Private flows (US0million) - 1 I .. - L 30 24 23 12 11 - 11 Source: OECDDAC, country at a glance. 103. Several aid modalities coexist which makes it difficult for the Government o f Burundi (GOB)to track external resources, to planand budget activities, and to know the exact uses o f those resources. Most development assistance in Burundi i s off-budget. Aid modalities in Burundiinclude in-kindaid, official loans and grants under various modalities including budget support and donors' aid, managed directly by the donor or by autonomous agencies or disbursed via nationaland international NGOs, foundations, or churches. External assistanceto the health sector in Burundi 104. The Ministry o f Public Health worked jointly with the World Bank inthe preparation o f different instruments to collect data on external assistance to the health sector. These instrumentsincluded a questionnaire to collect information from donor agencies on the amount of resources provided to the sector (see Annex 2) and a questionnaire to NGOs on the total amount o f resources they have spent inthe sector and the sources and uses o f these funds (see Annex 3). The information collected was complemented with other sources of data such as donors and NGOs annual reports. This information was then organized in a single database. The following analysis and conclusions are basedon the data collected. 25 105. Burundi has at least 15 donors (see list in Annex 8) financing the health sector, generally disbursing funds via more than 30 NGOs (see list in Annex 9). In 2006, the main donors were: ECHO (emergency assistance), World Bank, EU, UNICEF and Global Fund; and the major NGOs working in the sector were: Medecins Sans Frontieres, Health Net International TPO, CORDAID and IMC (see Annex 4). 106. As for overall external assistance to Burundi, external assistance to the health sector is transitioning from emergency to a more sustainable assistance. As a result of peace and stabilization, donors and NGOs that formerly withdrew from Burundi in the 1990's are progressively comingback while emergency NGOs are diminishingtheir activities. 107. The share o f development assistance in overall external assistance to the health sector has been increasing since 2003. Development aid represented72% o f total external assistanceto the sector in 2003, 76% in 2004, 87% in 2005, 81% in 2006, 87% in 2007 and is projected to represent 97% in 2008 (Figure 9). Figure 9: Evolution of development assistance and emergency assistance to the health sector per capita (2003-2008) I 1 I 10.00 7 1 1 9.00 8.00 .-3 7.00 P m 6.00 0 zP 5.00 69 4.00 ` 3 v) i 3.00 2.00 1.oo 2003 2004 2005 2006 2007 2008 Source: Authors' estimations based on the data base on external financing ofthe health sector inBurundi. 108. Donors are the main source o f health financing in Burundi. Based on the data collected for this study, the rest of the world's contributionto the health sector reached US$ 8.63 per capita in 2006 (including emergency assistance)20. Development assistance on health (excluding emergency assistance) reachedUS$7 per capita21. 109. External assistance flowing to the health sector is difficult to capture as it is mainly off- budget. In 2006, 82% of external assistance (from donors and internationalNGOs) to the health 2oSource: authors' estimations. These estimations include donors and international NGOs' hnds and may be overestimated as it i s difficult to distinguish commitments ffom disbursements. "Idem. 26 sector was extra-budgetary22. Most o f the resources actually flow through national and international NGOs, foundations, churches or are in-kind aid. Technical assistance, emergency response and direct procurement o f pharmaceuticals provided by donors are examples of off- budget donor assistance to the sector that are difficult to capture. 110. Part of the external assistance to the sector is registered in the national budget. However, as mentioned before, the government does not monitor or register any expenditure financed through donor funds. If this assistance registered in the budget i s not spent, it i s then addedto the budget ofthe followingyear. 111. Inthe health sector, only IDA financing is registered inthe budget law. Table 16 shows the difference between the budget line for the Second Health Population Project (PSPII) and the amounts actually releasedby the World Bank. This informationcan illustratethe difficulties of the Government to both predict the resources they can realistically spend as well as their weakness in absorbing foreign aid. Table 16: Comparison betweenbudget line for PSPIIand effectivedisbursement from World Bank to PSPII(2003-2006) 2003 2004 2005 2006 Effectivedisbursementof WB to 489,750 2,908,653 4,928,653 1,036,861 PSPll Budget law line PSP II 36,570 8,738,486 8,621,933 Source:2003 to 2006 BudgetLaws and World Bank. 112. From 2003 to 200523,external assistance to the health sector, particularly HIV/AIDS financing (Figure lo), progressively increased. This assistance, from both donors and NGOs, decreased slightly in 2006. The share of HIV/AIDS programs in overall health sector financing i s even more important when looking exclusively at development assistance (i.e. excluding emergency assistance provided by ECHO, ICRC and MSF). According to the data collected for this study, HIV/AIDSfinancing was increasing at a more rapid pace than the other activities o f the health sector. In2005, donors' overall assistance on health including emergency assistancerepresentedUS$9.05 per capita (US$7.85 without emergency assistance) out o fwhich US$ 3.15 were for HIV/AIDS (i.e. 35% o f overall external assistance on health and 40% o f development assistance). In 2006, donors' assistance amounted to US$ 8.65 per capita (US$ 7 without emergency assistance), out o f which US$ 2.8 were for HIV/AIDS (i.e. 32% o f total external assistance on health or 40% of development assistance). This trendinHIV/AIDS funding changed drastically, as the country was not selectedfor the seventhround of the Global Fundand as the WB project closed. These two donors, the Global Fundand the World Bank, are the main sources of HIV/AIDS financing. They usually provide funds to the government's SEP/CNLS. This financing agency then sub-contracts to health services providers, mainly national and international NGOs. Out of 30 NGOs currently operating in the health sector in Burundi for 22 Off-budget resources are the resources that were not included in the Loi portant j7xation du budget ge'ne'ralrhise'pour l'exercice 2006. This increase may be related to missing data. However, OECD-DAC data for total ODA flowing to Burundigives indicationthat ODA has beenprogressively increasing inthe pastyears. 27 which data are available, 10 implement only HIV/AIDS activities and 11 implement HIV/AIDS activitiesand other healtha~tivities~~. Figure 10 : Trends in external assistance on health (donors and international NGOs) including emergency assistance 70,000,000 60,000,000 50,000,000 o 40,000,000 I 9 30,000,000 20,000,000 10,000,000 - +bternalAssistance(excludingHNIAIDS) Source: Authors' estimationsbasedonthe databaseonexternalfinancingofthe healthsector inBurundi. Projectionsof external assistance to the health sector 113. Projectionsfor the period 2007-2008 follow past trends of donors and NGOsassistance to health. Projectionsare incomplete but trends can neverthelessbe observed (Table 17). At the moment, projectionsfor 2007 are a little bit higherthanthe 2006 levelbothfor development and emergencyassistance.Accordingto available data, US$63 millionare committedfor 2007 bythe rest ofthe world, of which 8 million are for emergency assistance, but only US$ 32.5 million are committed for 2008 (of which less than US$ 1 million for emergency assistance). Finally, external assistance from the rest of the world to HIV/AIDS programs is expected to drop from 2007 onwards (from US$ 2.8 per capita in 2006 to US$ 1.4 per capita in 2007 and US$ 1.1 in 2008). Table 17 : Projections of external assistance to health (including HIV/AIDS) with and without emergency assistance 2005 2006 2007 broi.) 2008 hroi.1 Rest of the world (million US$) 63.4 54.9 60.4 49.1 63.0 55.0 32.5 31.6 Rest of the world (US$ per capita)----- - 9.1 7.8 8.6 7.0 9.0 7.9 4.6 4.5 24Authors' calculationbasedonthe databaseonexternalfinancingofthe healthsector inBurundi. 28 National NGOs (million US$) 0.0 0.0 1.9 1.9 3.7 3.7 0.0 0.0 National NGOs (US$ per capita) 0.0 0.0 0.3 0.3 0.5 0.5 0.0 0.0 Depensestotales de sante (million 63.4 54.9 62.4 51.O 66.7 58.7 32.5 31.6 US$) Depensestotales de sante (US$ par 9.1 7.8 8.9 7.3 9.5 8.4 4.6 4.5 habitant) ----- - Source: authors' calculation basedon the data base on external fmancing ofthe health sector inBurundi. Allocation of external assistanceto the health sector 114. Most external assistance from donors and NGOs flowing to the health sector is allocated to recurrent expenditures. In 2006, out o f US$ 25 million allocated to health for which we have data classified between recurrent and capital, US$ 19.5 million were allocated to recurrent expendituresand only US$ 5.5 million to capital expenditures. 115. A large share of external assistance is allocated to human resources25.Although the database on aid flowing to the health sector does not allow a comprehensive overview on uses o f funds for health, some examples o f donors and NGOs that provided rather comprehensive data, can be illustrative (Annex 10). In 2006, human resource costs represented one third or more of Belgium, European Union, MSF and CCM expenditures and half of ICRC expenditures. Training sessions, information, education and communication activities and pharmaceuticals are usually the second and third main source o fexpenditures. 116. HIV/AIDS has been the main vertical program financed by external partners in Burundi. The data available show that most health expenditures are earmarked to HIV/AIDS programs, mother and child health, nutritionprograms and emergency care to refugees. 117. Most donors and NGOs usually focus on one or two main strategic objectives. In the case of the World Bank, funds are either usedto finance HN/AIDS related activities through the Multisectoral HIV/AIDS project or to finance other health activities. DFID finances mainly HN/AIDS as well as mother and childhealthprograms; UNDP and most of the NGOs operating inthe healthsector inBurundiexclusively finance HIV/AIDSprograms. Issues raised by externalassistance to thehealth sector Geographicaldistribution 118. Donor and NGO assistanceto the healthsector is unequally distributedamong regions in Burundi. In 2006, there was an unequal allocation o f external resources to the health sector with per capita spendingvarying from US$ 0.1 per capita inthe least funded province (Mwaro) to US$ 6.3 inthe highest funded (Ruyigi)26.Inequalitiesseem to be even growing over time (Figure 11) and are not likely to disappear in the near future as many donors have made pledges to support particularregions inthe near future. 25Conclusions basedonthe analysis of selecteddonors andNGOs activeinthe healthsector inBurundi. 26Authors' calculationbasedonthe database on external financingofthe healthsector inBurundi. 29 Figure 11:Distribution of external assistance to the healthsector (2006-2007) 18.00 1 , 90 15.00 80 70 C I -; 3 12.00 60 W m 0 W ta 9.00 50 E 40 .E 0 +e 6.00 30 3 v) 0) 3.00 20 10 0 Source: authors' calculation basedon the data base on external financing ofthe health sector inBurundi. Note: This figure is based on indicative data. It takes into account available data given by donors and NGOs on the distribution ofresources imong provinces. 119. This unequal distribution is due to donors concentrating their support in specific geographical zones. Donors and NGOs are usually active in a limited number o f provinces; in addition, the type o f support varies considerably across donors. Figure 12 provides geographical distribution of external assistance of three main donors and three mainNGOs active inBurundiin 2006; representing 28% of total foreign assistance to the sector. Half o f the provinces receive nothing or almost nothing from these six main donors or NGOs while four provinces receive US$ 2 to 3 per capita. Karuzi, the highest funded province, receives US$ 6. However, as Karuzi is mainly financed by donors and NGOs providing emergency assistance, one may wonder if external assistance to Karuzi will be sustained inthe comingyears. Figure 12: Geographicaldistributionof externalassistanceto health for selecteddonors and NGOs 006) 7.00 6.00 U 3 5.00 2c 4.00 * 3.00 `3 2.00 1.oo 0.00 Source: authors' calculation based on the data base on external financing ofthe health sector inBurundi 30 120. NGOs, becauseof their limited human and financial capacities, are usually working in a smaller number of provinces than donors. CCM is working in only one province, Action against Hunger in two provinces while European Union is financing five regions, ECHO seven and WHO i s supporting all provinces. Verticalfunding 121. The distribution of donor funds by strategic objective is also affected by inequalities mainly because of single-issue funds and single-issue projects. Burundibenefits from several large disease initiatives such as the Global Fund for AIDS, tuberculosis and malaria and the Global Alliance for Vaccine and Immunization (GAVI). Despite the bias that single-issue funds might create they also offer needed additional resources to improve health outcomes. These resourceswhenever possible should be usedto strengthenthe health system ingeneral as the fight against any o f these diseases depends on a system with adequate drugs, personnel, equipment, facilities, etc. 122. The development of vertical programs and global initiatives increases the fragmentationof the sector and goes against the objectivesof harmonizationand alignment. Although these new sources of health financing represent a great opportunity to increase the resources flowing to the sector and to address health issues in the country, they also represent a burden on the country's system. Vertical financing increases absorptive capacity problems in a country in which institutions and management capacities are generally weak. It also emphasizes financing distortions as global health initiatives may prevent funds from flowing to the sector's priorities. In Burundi, in 2006, one third o f total health expenditures financed HIV/AIDS programs, one third financed other vertical programs and only one third contributed to finance the health systemas a whole. Predictability 123. Externalassistanceto the healthsector is volatileand unpredictable.Donor hnds are in general difficult to predict. At the moment, the data available indicates a decrease in funds in the coming years. However, partlythis is due to some donors' difficulty to know the exact amount o f their future support. Adding to this unpredictability, donor funds are usually volatile as consequence o f the variability in the exchange rates or due to political decisions o f the donor agency. For instance, between 1997 and 2000 the Burundian Franc depreciated on average 27% in relation to the US dollar. At the same time, the US dollar depreciated about 7% inrelation to the Euro. Partly as consequence o f these changes, donor funds in those years went from representingabout 50% oftotal funds for the sectorto lessthan 10%(Gottret et al. 2006). 124. As most donor funds are extra-budgetaryand hard to predict, it is difficult for GOBto have a clear pictureof the uses of funds in the sector and to plan and budget activitiesfor the comingyears. 125. In line with international commitments such as the one agreed by the G8 in the Gleneagles Summit, external assistance to Burundi is expected to increase in the coming years. However, forecasts available to date do not show this trend. On the contrary, given the data collected from donors and NGOs, external assistance i s expected to decrease from 2007 onwards which demonstrates that even for the current year, government and donors don't have comprehensive information on planned activities. 31 126. The lack of predictabilityand the low maturity of external assistance to health also introducemajor problems of sustainability of activities that are being financed. As most o f the resourcesgoingto the health sector can only beplanned on avery short-term basis and last for at most four years, it is not possible to know, for example; if there will be enough resources to finance the recurrent costs induced by an investment in the health sector or to sustain existing activities. Low execution rate of externalfinancing 127. Burundi faces many challenges that slow down the execution of overall external assistance or reduce the efficiency of funds disbursed. According to an OECD-DAC 2006 survey on monitoring the Paris Declaration in 34 selected developing countries including Burundi2', the country is still experiencing large capacity constraints after a decade of conflicts, particularly to absorb new finance and to design and implement pro-poor economic reforms. Although there is a system in place for coordinating external assistance that includes structures dealing with formulation, sectoral issues, implementation and reporting, the OECD-DAC survey states that "the government would do well to take more leadership over development assistance coordination". Serious donor concerns also remain over the quality o f Burundi's public financial management and procurement systems28. On the Burundian side, the uncoordinated donor missions, the lack of harmonization of donors' procedures and the lack o f alignment on the country's priorities are lamented. Finally, the OECD-DAC survey laments the weak quality o f available data andthe total absence of mutual assessment although mechanisms exist. 128. Executionof donor funds seems to be a major probleminthe healthsector inBurundi. Although there i s no available data to compare donors' commitments with their disbursements, partial information from donors and NGOs implementation reports show that disbursements are sometimes delayed29. 129. Some NGOs did not receive the resources committed by donors to implement the activities agreed upon. For example, in its 2006 annual report, ABUBEF states that out o f US$ 40,000 committed by one donor, only US$22,000 were disbursed intime, and out o fUS$ 50,000 committed by another agency, only US$ 11,000 were disbursed. On the contrary, ABUBEF received financial support from CARE and FHI although their financial assistance was not forecast inthe budget approved in2006. Necessary move towards better donorscoordination 130. As the health system of Burundi is moving from emergency response to sustainable development of the system, there is an increasing need to better coordinate external assistance and to align it with GOB priorities and national health policies. The health 272006 Surveyon monitoringthe ParisDeclaration, CountryChapter, Burundi, OECD-DAC. 28Under the World Bank's Country Policy and Institutional Assessment (CPIA) for 2005, which assesses the quality ofbudgetaryand financial management, Burundi'spublic financialmanagement(PFM) systems receive arating of 2.5. This is significantlylowerthanthe average of 3.2 for all InternationalDevelopment AssociationBorrowers. 29Accordingto the last implementationreport of the PATSBU (NEvaluationfinale duprogramme d'appui transitoire au secteur de la sant6 burundais PATSBUN Burundi, Lettre de marche n02006/126249, - 11/23/2006), 54% of the funds didn't reach the beneficiaries. At the end of the program, most pharmaceuticals and equipments were not available and construction and rehabilitation were not done. According to this report, 80% of the funds disbursed were actually used to finance salaries and bonuses. Similar issues affect WB fundingto the sector (see Table 16). 32 authorities have recently made significant progress in the coordination o f donor support to the sector. In this regard, a framework for donor coordination in the sector (Cadre de concertation des Partenairespour la Santb et le De'veloppementor CPSD in French) has been recently created and is functioning well. 131. To better coordinateaid flows and to increase the efficiency of external assistance, the Government and partners are working towards a SWAP to support the health sector. This process is based on the premise that all partners will support a single and comprehensive health strategy, will agree to a common plan o f action, and will work together towards harmonizing procedures. All these activities will lessen some o f the inefficiencies and inequities in the distribution of donor support to the sector. During October 23-31 2007, the first joint partner's mission for the heath sector took place inBujumbura. The success o f this process will be crucial to solve many issues raised inthe health sector by uncoordinated donor aid. Private Expenditure on Health 132. As public expenditure on health decreasedwith the start of the crisis, households had to increase their expenditure to offset this reduction. However, as household income levels also decreasedand existing exemption and waiver mechanisms were not functioning properly, the increase in out-of-pocket expenditure on health increased financial barriers to access these services as was explained earlier. 133. Most private expenditure on health is OOP as pre-payment mechanisms cover a very small percentage of the population. There is a pre-payment program for health services, the Carte d'AssuranceMuladie (CAM); but, due to lack of funds and design problems this program almost disappeared. The C A M was put in place in 1984 to improve access to health care among the population in the informal sector. Today many provinces have officially eliminated these cards and in those where it still operates they cover a small percentage o f the population; this percentagevaries considerably across provinces. There i s also a social insurance mechanism that covers people working in the public administration. La Mutuelle de la Fonction Publique (MFP) i s a compulsory social insurance for all public servants and thus it only covers a very small percentage of the population. The government contributes 6% of the total payroll and the employees 4%. This insurance covers all consultations, hospitalizations, and some pharmaceuticals. The person insured contributes 20% for generic drugs and 30% o f the hospitalizations and non-generic drugs. 134. By early 2006, the burdenof healthexpenditure for Burundian families was very high. The 2006 QUIBB survey collected information on health expenditure from a national representative survey of 7132 households. This survey estimated an annual per capita OOP expenditure on health of about FBU8871, equivalent to US$ 8.6 per capita. The survey data also indicate than on average Burundians spend about 5.6% of their total consumption on health. In other low income countries in the region, such as Madagascar and Ethiopia, this percentage is only As seen inthe graph below, the burdenofhealth expenditure was comparatively lower among the richest ten percent o fthe population. 30 Source: Ethiopia Health, Nutrition, and PopulationCountry Status Report (WB, 2004) and Madagascar Health Sector Public ExpenditureReview(Madagascar Ministerede la Sante Publiqueand WB, 2006). The data fiom these two countries refers to the share of health expenditure out of total household expenditure andnot out oftotalper capitahouseholdexpenditure, but the differences shouldnotbe large. 33 Figure 13: Household out-of-pocket expenditure on health out of total expenditure on health, Burundi, 2006 7.0 1 6 0 5 0 4 0 3 0 2 0 1 0 0 0 decile 1 decile 2 decile 3 decile 4 decile 5 decile 6 decile 7 decile 8 decile 9 decile 10 Source: Estimations based on data from the QUIBB 2006 survey. Households here are ranked from poorest to richest. The first decile is the poorest decile, decile 10the richest. 135. Expenditure on pharmaceuticals represents the largest share of OOP health expenditure. As seen in the graph below, on average, pharmaceuticals represent about 63% o f total expenditure on health, followed by "miscellaneous" costs (14.5%) which mainly include transport costs. Figure 14: Categories of healthexpenditure, Burundi, 2006. Source: Estimations from the QUIBB 2006 survey 136. With this pattern of health expenditure, many householdsrisk falling into poverty or deeper intoit as consequenceof catastrophic healthexpenditure.Estimates usingthe QUIBB 2006 data show that one in every three households spends between 5% and 10% o f their total expenditure on health. More worrisome still, as many as 11% of Burundian households spend between 10% and 25% o f their annual consumption on health care and about 1% more than a quarter. Poorer households are more likely, to have health expenditures representing more than 5% of their total annual expenditure. As seen inthe graph, halfo f the poorest 10% o f households 34 spend more than 5% o f their total consumption on health care; while among the richest 10% of households only a quarter of households spendmore than 5% of their total consumption on health care. 137. This situation has started to change as the country has received extra funds through debt relief and external aid.The May 2006 presidential measureto eliminate user fees for some services, financed by HIPC funds, is thought to have increased utilization o f health services. There has not been a formal evaluation o f the effect of the measure and thus it is not possible to know for sure its effect on OOP expenditure. However, if the Government i s able to fully reimburse facilities for the services offered "free" of charge, this measure might have also reduced OOP. Nevertheless, user fees in public facilities are only part o f these OOP payments which include payments for transport, opportunity costs, and payments for private providers. In addition, this measure only benefits part of the population. For instance, children under five represented about 17% o f the total QUIBBsample and were responsible for at least 19% of total OOP expenditure. Given the large proportion living under poverty level, health expenditure is likely to still representa large expenditure burdenfor households andthus it should still represent a large barrier to access health care. Total Expenditure on Health 138. As expenditure on health from both public and private sources might be overestimated, it is more appropriate to give a range for THE in Burundi. For instance, at the moment, part of the information on donor funds is based on projections and not necessarily on expenditure. If .we assume a donor funds' execution rate similar to that of Rwanda, where from 2002-2005 the execution rate of donor funds was 76.4 %31, we can then assume a lower bound o f US$ 6.6 for rest of the world's assistance. 139. Despite recent increments in domestic and foreign resources, health expenditure continues to be low in absolute numbers and when compared to similar countries in the region. In2006, public expenditure on heath and foreign aid inBurundiwas inthe range of US$ 7.6-9.732.Donors and internationalNGOsprovided the largest share of these resources (US$ 6.6- 8.7), about 90% of the total. This is one of the largest per capita oficial development assistance inthe region and it is much higher thanthe SSA average (in2003 the per capita ODA for health inSSA was only about US$ 5.833).Butas government expenditure on health is very low the total public expenditure on health (from public sources and from the rest of the world) i s not as high. As seen in Figure 15 the highest end o f our estimated THE range is lower than the average per capita total expenditure in SSA; while the lower end is also lower than that o f Rwanda. The highest end is close to the level o f Rwanda. The figure also shows the WHO estimate for that same year which is a much lower estimate. However, as there has not been any NHA in the country, this number is likely to be an underestimation as most donor funds are extra-budgetary and, before the data collected for this study, there was no account of total donor expenditure for the sector. 31 Rtpublique de Rwanda, Ministere de la Santt Publique & European Commission. 2006. Revue des Dtpenses Publiques du Secteur de la Sank Report written by Guy Scorraille, J. Nachtigal, and J. Munyenpenda. 32 Authors' calculationbasedon the database on external financingof the healthsector inBurundi 33WHO, World Health Statistics2006. 35 Figure 15: Total health expenditure per capita in selected SSA countries (2005) 50,I I 45 I 40. 35 30 25 20 15 10 5 0 Source: WHO, NHA databaseandauthors' estimationsfor Burundi. Note:the estimationfor Burundiis for 2006. 140. Concerning expenditure from private sources o f funds, there i s no data on OOP after the presidential measure to eliminate user fees for deliveries and services to children under five so it i s not clear how OOP changed. Experience from Uganda where user fees were eliminated for primary health care services shows that catastrophic expenditure among the poor did not change after user fees were eliminated but they significantly decreased among the non-poor (Xu et al. 2006). In the case o f Burundi, if we assume that the reimbursement to health facilities for the services they now offer "free" o f charge diminished total OOP, total OOP expenditure should then be below US$ 8.6. Before the presidential measure, health care for children under five represented at least 19% o f total OOP or a minimumo f US$ 1.5; it i s not clear how much OOP was due to deliveries but we will assume US$ 0.5 per capita based on a rough e ~ t i m a t i o nThis. ~ ~ can give us a possible range for OOP expenditure on health between US$ 6.6-8.6. Inaddition, in 2006 nationalNGOscontribution to the sector reached US$0.3 per capita. 141. In summary, total expenditureon healthinBurundi is in a rangeof US$14.5-18.5. 34 Assuming 372,027 deliveries that year (data from PNSR), and only 25% skilled attended and 5% of total by C-section. The OOP for normal delivery was assumedto be the mediancost of spendingtwo nights ina CDS under cost recovery (MSF data) and the cost ofa C-sectionofFBU150,000 as mentionedinCordaid. 36 Fiscal Space for Health 142. To increase fiscal space3' for health expenditure governments have few alternatives. They can increase revenues through tax measures, reallocate from other priorities, borrow resources, receive grants, or they can use their power o f ~eignorage~~ (Heller, 2005). Table 18: Internationalcomparisonson fiscal revenues, early 2000s. Total revenue as % of Tax Revenueas % of GDP GDP Region Sub-SaharanAfrica 19.7 15.9 IncomeLevel Low-incomecountries 17.7 14.5 lower-middle-incomecountries 21.4 16.3 Upper-middle-incomecountries 26.9 21.9 High-incomecountries 31.9 26.5 Source: Gottretet al. 2006. Health FinancingRevisited. 143. Burundi has a relatively high tax base and thus it will be very difficult to generate fiscal spacethrough more taxes or even throughimprovementsinthe tax administration.In the last few years, Burundi's tax revenue has been about 18% o f GDP. This tax revenue is higher than the averages in SSA and in low income countries (see Table 18 and Table 19). The tax code will berevised as the countryjoins the East African Community. This mighttemporarily decrease revenues as tax rates are likely to decrease; but, as export volumes increase, the total revenue i s expected to increase from 2008 onwards (see Table 19). Regarding improvements in the tax administration, there might be some room for increasing revenues as the tax code has an excessive number oftax exemptions (see forthcoming PEMFAR). 144. Increasing revenues in the country i s tightly linkedwith future economic growth. Although available economic projections indicate relatively high levels of growth inthe nextyears, income growth inthe recent past has been low. Inany case, per capita income growth will remain limited due to Burundi's highpopulationgrowth rate (3.7% in200637),the highest in SSA. Table 19: MacroeconomicIndicators, Burundi 2000-05 2006 2007 2008 2009 2010 2011 average Prel. Projections Real GDP growth rate 1.7 5.1 3.6 4.5 5 5 5 Inflation (GDP-deflator) 9.5 3.7 9.5 7.5 7 7 6 Total revenue (excl. grants) % 20 19 18.1 19.2 19 19 19.2 of GDP Tax revenue 18.5 17.4 16.9 17.8 17.7 17.7 17.6 Non-tax revenue 1.5 1.6 1. I 1.4 1..3 1.2 1.2 Source:World Bank, IMF,own calculations. Macro-frameworkusedfor PEMFAR (forthcoming) 35Fiscal space is the availability of budgetary room that allows governments to provide resources for a purposewithout any prejudice to the sustainabilityof its financial position(Heller, 2005). 36 The power of seignorage is the capacity of having the central bank print money to lend it to the government. 37World DevelopmentIndicators.2006. 37 145. Reallocations from other sectors in the near future will also be difficult. As consequence of the recent conflict, military spending increasedto the detriment o f social sectors. Thanks to the demobilization and reintegration of ex-combatants as well as other peace and stabilization activities, this spending has been decreasing significantly since 2000. However, expenditure inthe nationalpolice has rapidly increaseddue to higher numbers o f police staff and to the harmonization o f their salary structure with that o f the military (forthcoming PEMFAR). Nevertheless, there i s some room for reallocations in the medium term as the demobilization proceeds and the number o f police effectives decreases. For instance, by December 2006, about 21,000 soldiers have been demobilized and the objective for 2007 is to demobilized 27,000 ' soldiers. Similarly, the medium term governments' objective is to reduce the police force from about 18,000 to 15,000 people. 146. Debt relief and grants might offer the largest possibilityto increase public resources for health. More borrowing, the other side of debt relief, i s another option but again these resourceswill eventually have to be re-paid andthe capacity of the country to continue borrowing is very limited. Nevertheless, Burundi has started to benefit from the enhanced HIPC initiative and substantive debt relief will be granted once the country reaches the completion point. Inthe interim period, mid 2005-end 2006 the country benefited from about US$ 26 million per year (about US$ 36 million in 2006), but this aid will increase to an average of US$ 46 million per year from 2007-2039 (IMF and World Bank, 2005). These resources are to be used in poverty related sectors, including health care. There i s no pre-set amount earmarked for any sector and thus the share that will go to healthwill depend on the Government's commitment to this sector, notjust reflected inthe original budget allocation but also in the total amounts that will be finally disbursed. As explained before, in 2006 the budget for the sector increased significantly, but not so much the expenditures due to late disbursements of these funds and to other budget execution problems. In 2007, the budget allocation actually declined. Concerning other grants, the projections available so far do not show an increaseinextra-budgetary funds. 147. Finally, even if more resources are allocated to the sector this will not necessarily improve health outcomes, especially for the poor, unless additional measures are taken. There are many reasons why a higher budget allocation alone might not make a significant difference on healthoutcomes. First, the resourcesmight not be disbursed intime by the Ministry o f Finance or the budget might not be fully executed by the sectoral ministries as was the case in Burundi in 2006. Second, even if resources are disbursed and the budget executed, these resources might not benefitthe poor the most. Internationalexperience shows that often the better off urban population benefits more from public health expenditure. Inthe case o f Burundi, there are indications that a disproportionally large amount o f resources allocated to the sector benefit hospitals, especially one hospital in the capital city where the better off 10% o f the population lives. Third, even if resources are meant to benefit the poor the most, they might not reach their intended beneficiaries. Currently, the Government of Burundi is implementing a Public ExpenditureTracking Survey (PETS) which will allow identifyingleakages of public resources in the health sector. Finally, even if the resources arrive they might not necessarily improve the quality of the services offered if there are vacancies, if personliel absenteeism is common, or if the resources are mismanaged. The lack o f qualified medical personnel is an issue that needs immediate attention in the country as discussed in other sections. To improve the management o f resourcesat the facility level, not only would there be a need for more supervision from the MSP but also more community involvement inthe management of facilities, especially primary health care facilities as will be discussed inthe next chapter. 38 PUBLICEXPENDITUREMANAGEMENT 148. Despite recent budgetary increases, total expenditure on health remains low and its level will depend not only in the capacity of the MOPH to mobilize more donor funds but also in its absorptive capacity. For instance, the deficit of qualified medical personnel could hinder in the short run the absorption of resources intended to increase the coverage of health services. But these bottlenecks could also be related to budget management issues such as issues with budget planning and execution that could also hinder the absorption o f new resources. This chapter will examine the latter. BudgetPlanning 149. Recognizingthe need.toimprovethe management of resources,the Ministry of Health created last year a Direction Ginirale des Ressources (DGR) to manage both financial and human resources. This Direction is incharge ofthe preparation o fthe consolidated budget o fthe ministry. For planning purposes, the MSP has a Cellule de Plunifi~ution~~ supports the that Directeur Ge'ne'ralde la Sunte',the Directeur des Programmes de Sante',and the DGR. The entire team (DG, DGR, DP, and the Cellule de Plunificution) is at the moment drafting a health plan for 2007-2009. Although in general the team has the skills neededfor planning, it could benefit from training on costing o f healthprograms. 150. The MSPfaces manychallengesinthe preparationof its budget. The budget preparation starts when the Ministry of Finance sends a letter to all sectoral ministries. This letter does not include a budget ceiling to guide the ministries.Inthe MSP, all Provincial HealthBureaus and all health programs and departments are supposed to prepare their own budget plans. There is low capacity for budget preparation at these levels o f the ministry, especially at the Provincial Health Bureaus. As a result, these budget plans are often not linked to the sector strategies or objectives and budget allocations are not prioritized. Once these plans are ready, they are sent to the DGR which then preparesthe consolidated budget ofthe Ministry. 151..At the central level o fthe Ministry(DGR with some help from the Cellule de Plunificution), the consolidated budget is linkedto the sectoral policies and objectives and some expenditures are prioritized. However, the budgetpreparation at this level also confronts some challenges: 1. The budget preparation is fragmented. The budget for HIPC resources i s prepared separatelyfrom the rest of the budget. Additionally, recurrent and investment budgets are not consolidated at the sectoral ministrylevel which meansthat the projects' running costs are not always integrated into the programmatic budget3'. 38There are two economists working at the cellule. 39IMFReport07/46, January2007. 39 2. There is no consolidated account o f donor funds flowing to the sector. Although each healthprogram knows what they receive inexternalaid, the central level does not have complete information on donor funds. This limitsthe ministry's capacity to plan and prepare a comprehensive budget as the majority o f external funds are not registered in the budget law. 3. Donor funds are highlyuncertain. 4. There i s limited capacity to calculate the costs of healthplans. 5. At the DGR, the few people with the skills to prepare budgets are currently overwhelmed with other functions. For instance, the same people incharge of budget preparation are responsible to check all the invoices sent by the facilities for the reimbursement o fthe "free" services. The same people also sign and sendthe letter to the Ordonnateur Tre`sorierdu Burundi (OTB) for the release o f funds to all these facilities (there are about 500 facilities inthe country). 6. Finally, the difficulties that the Ministryo fFinance faces inpredictingannual revenues create a challenge for the MSP inthe preparation o ftheir annual budget. For instance, as seen inthe table below the difference betweenthe original budget law and the revised law can be large. In2006, the allocations for the sector increasedby about 5%, butthe allocations for capital expenditure actually diminished. Table 20: Annual rate of change on MSPspending between Initial Budget Law (LFI)and Revised (LFR) 2006 LFI LFR variation Current budget in millions FBU 15,405 16,986 1,581 Capitalbudget in millions FBU 3,296 2,666 -630 Total budget in millions FBU 18,702 19,653 951 Annual rate of change, nominalterm, in% 5% Source: MinistryofFinanceandWorld Bank estimations Recommendations 1. Consolidate budgetpreparationnotjust interms o f recurrent and capital expenditure but also interms of sourcesof funds: HIPC, ordinary resources, and eventually donor funds. 2. Generatea consolidated account of all donor funds supporting the sector and progressively institutionalizeNational HealthAccounts. The development o f an MTEF and the entire SWAPwill facilitate this process and will also lessen the uncertainty of donor support to the sector. 3. Buildcapacity at differentlevels o fthe MSP for budget planning. At the central level, there is also a needto improve capacity to cost health programs. Budget Execution and Monitoring 152. The ministry also faces the following challenges and opportunities related to budget execution: 1. Longdelays o fthe Ministryo fFinance to disburse funds and especially HIPC funds. 40 2. To better planthe use ofHIPC funds, the MSP preparesan execution planfor these resources. 3. The delays inthe payment to the facilities for the reimbursement of "free" services have beendrastically reduced. To be reimbursedthe facilities have to write invoices for the inputs usedindeliveringservicesto women and childrenunder five. When this policy was first putinplace, these invoices hadto go through avery cumbersome control systemwhere the Provincial HealthBureaus, the central level of the Ministry and then the Ministryof Finance checkedthese invoices before approving their payment. Needless to say, there were long delays and by March 2007 many facilities had only received partial payment. This situation has improved considerably as the MinistryofFinancehas startedto makeadvancedreleasesto the MSP who now makesthe final control and sends a notification to the OTB for payment o fthese invoices4'. One challenge remains; it is not clear what the total cost of this policy is andwhether there are enough resourcesto pay for it.For instance, by June 2007, more than 98% of all resourcesbudgetedfor thispolicy had already beenpaid (these resourceswere also usedto pay for last year's arrears though). 4. The ministry has some difficulties inthe preparation ofprocurement documents. The DGRdoes not havethe number of staff or the capacity to prepareall these documents. The long and cumbersome procedures inplace for large procurements also create further complications and delays. 5. The ministry has very limitedpersonnel to monitor large expenditures (e.g. construction, rehabilitation, etc.). No monitoring i s done o fthe expenditure financed and managedby donors. 6. It is not clear whether there are significant leakagesofresources inthe health sector. At the moment, the government is implementingaPublic ExpenditureTracking Survey to estimate the magnitude ofthis issue. Recommendations 1. The budget execution planthe MSPpreparesfor HIPCresourcescouldbe improved by prioritizingexpenditures and by including a commitment planwith a chronogram for the fiscal year. This execution plan should includeall resourcesand not only the budgetfor HIPC resources. 2. An estimate ofthe real cost ofthe policy to offer services "free" o fcharge to women and children under five i s urgentlyneeded. Otherwise it is not possible to know ifthe budget allocated to this policy is sufficient. It is also necessaryto define more clearly the packageof servicesaccordingto priorities andresourceavailability. 3. There is a needto improve the capacity o fthe DGR to monitor all expenditure benefitingthe healthsector and notjust expenditure financed out o fdomestic resources. 40These notifications to the OTB are signed by the Directeur Gknkral des Ressources andthe Directeur du Budget of the MSP andby the Directeur Gknkral du Budget et de la Comptabilit6 Publique of the Ministry ofFinance. 41 Resource Management Issues at Health Facility Leve141 153. Not only do primary health care facilities receive funds from upper levels of the Ministry of PublicHealth, but they also collect their own revenues from the healthservices they provide. The salaries o f most health care workers are paid by the Ministry of Public Administration, Labor, and Social Security (Ministire de la Fonction Publique et de la Se`curite` Sociale), although some contractual staff are paid by the facilities own revenues. The facilities also received some drugs and supplies from the Provincial Health Bureaus. Finally, since May 2006, after the presidential measure to eliminate user fees for children under five and mothers during childbirth, facilities are supposed to be reimbursed by the central level for the services they provide "free" of charge. In addition to these resources, facilities charge for services to patients not covered bythe presidential measure. 154. The management of resources at the PHC facility level varies considerably across provinces and often escapes from the Government public expenditure monitoring mechanisms.Facilities do not produce regular budget or budgetexecution reports. Most facilities have two bank accounts, one for medical "acts" and another for pharmaceuticals. The majority o f health centers have flexibility in the use of the funds collected in the "acts" account. These resources are often used to cover running costs, including the payment o f salaries of some contractual personnel such as the case of security staff. In contrast, the management and use o f resources from the pharmaceutical account i s more restrictive and often escapes from the health center. For instance, inNgozi only the BPS can withdraw resourcesfrom these accounts. 155. Community participation in the management of health facilities is still limited. In a case study o f three provinces (Gitega, Rutana, and Ngozi), only in one, Rutana, the community has ample participation in the management of these resources (see Annex 11). This province received some support from the EUinresource management at the facility level. Inthis province, facility revenues are managed by a Management Committee (Comite` de Gestion or COGE in French)which is part of the Health Committee (Comite`de Sant6 or COSA inFrench) where the community i s largely represented.Ineach health center there i s an administrator who i s chosenby the community and who reports to the COSA, and therefore to the community. The salary of this administrator is financed bythe health center. 156. Some revenues collected at the facility level are sent to the ProvincialHealth Bureaus or to communal structures. For instance, in certain provinces the health centers must send part of their revenuesto the PHB to cover part of their runningcosts. Inthe province o f Gitega, PHC centerswere supposedto send 10% o ftheir revenues to the BPS; inNgozi, they were supposedto send 20% of the revenues from the sale of pharmaceuticals to the BPS. This year, as many facilities were not reimbursed in time for the services offered "free" o f charge, they stopped sendingpart of their revenues to the BPS. This practice reducesthe benefits to the facilities from their own revenues as it limits the amount of resources it can reinvest to improve the quality o f services. Recommendations 1. There i s a need to increase community participation inthe management of health services. This could be achieved by revitalizing the Health Committees (COSA) and 41 This section i s basedon acase study ofthree provinces (Rutana, Ngozi, andGitega) done as background of the BurundiHealthfinancingstudy (see Annex 11). 42 Management Committees (COGE) to help manage resources at the health facility level and ensure some degree o f transparency. 43 SIMULATIONOF THE COSTAND IMPACTOF THE PRESIDENTIALMEASURETO ELIMINATEUSERFEES FORHEALTHSERVICESOFFEREDTO CHILDREN UNDERFIVEAND WOMENDURINGDELIVERY 157. The objective of this chapter i s to evaluate the implications o f the presidential measure of May 2006 of eliminating user fees for health services offered to children under five and women during delivery. As described in this chapter, the supply o f health services in Burundi is still limited; in consequence, the measure to eliminate user fees on its own will not be enough to significantly increasethe effective coverage of the targetedinterventions. Reinforcing the supply o f services as well as creating and stimulating their demand will be necessary in such a context. This chapter will precisely look at the financial, material and human resource implications o f the provision andutilization o fthe interventions included inthe presidential measure. Methodology 158. The cost and impact estimations presented in this chapter were carried out with the assistanceof a tool known as the Marginal Budgeting for Bottlenecks (MBB). The MBB is a planning tool developed by UNICEF and the World Bank that allows estimating costs and impacts of health intervention^^^. By linkinghealth inputs to outcomes, this tool helps rationalize budget decisions and thus optimize outcomes. The data used in these simulations were collected or validated with the help o f relevant program managers or departments within the Ministry o f Public Health including the Department o f Reproductive Health, the Direction of the Fight Against Infectious and Deficiency Diseases, the National Tuberculosis and Leprosy Control Program, the Expanded Program on Immunization (EPI), the Integrated Child Disease Management Program (PCIME), the National Institute of Public Health, the Department of Water and Sanitation, etc. 159. The estimated costs represent the total costs required for reinforcing the supply of healthservices and for creatingor stimulatingtheir demand. They include the investment and recurrent costs (old and new) necessaryto improve the level of performance o f the health system. The methodology used starts by identifying major bottlenecks that limit the delivery and utilization o f high impact health interventions. Shortages o f human resources, o f drugs and supplies, inadequate access to health services, as well as issues concerning household behavior are analyzed. Once this i s done, the inputsrequiredfor reducing or eliminating these bottlenecks and implementing corrective measures are estimated. The estimated costs thus reflect the costs o f obtaining an effective coverage o f the interventions included inthe "free" healthcare measure. In summary, the unit costs usedmay be classified into the following categories: (i) resources human (salaries and incentives); (ii)transport costs; (iii)materials; (iv) drugs and supplies; (v) infrastructure and equipment; (vi) demand stimulation; (vii) performance bonuses; (viii) communicationfor behavioral change; (ix) subsidies; and (x) monitoring and evaluation. 42See annex 12 for details. 44 160. The concept of effective coverage used in this chapter refers to the proportion of the target populationthat effectively receivesor utilizes a given interventionin accordancewith scientifically-accepted quality norms and standards. For example, inthe fight against malaria, the effective coverage for the use o f impregnated mosquito nets among children under five refers to the proportion o f children under five who are in possession o f an impregnated mosquito net and who actually sleep every night under this net. 161. The methodology used for estimating the impact of scaling up the package of "free" health services calculates the expected reductions in infant, child, and maternal mortality and morbidityfrom specific diseases brought about by an increase in the effective coverage of health interventions included in this package. The statistics on the effectiveness o f the interventions used in the MBB impact model derive primarily from international reviews such as the 2003 and 2005 Lancet series on child and newborn survival respectively, the Cochrane systematic reviews and the 2005 British Medical Journal series dedicated to maternal and neonatal health. Results Which interventions are included in thepresidential measure to eliminate userfees? 162. The President announced in May 2006 the elimination o f user fees for deliveries and for health services to children under five. To evaluate the financial implications o f this measure and the potential impact o fthe resulting increase in service coverage, it is important to define in detail the content o f the "free" package of services. While obstetrical services included in the measure are relatively easy to identify, this is not the case for curative care offered to children under five. The interventions in this case are not clearly defined and could extend to services to treat communicable and non-communicable diseases as well as nutritional deficiencies. 163. The package o f "free" services used for these cost simulations was not exhaustive. However, the interventions included are already provided in the country's health facilities. In addition, these interventions are effective in confronting the major causes o f mortality and morbidity amongchildren under five inthe country. Figure 16 shows the main causes of mortality among children inBurundi.These data are also usedto estimate the potential impact of the health interventions included inthis simulation. 45 Figure 16: Causes of mortality among children under five, Burundi, 2000 I Accidents, 18Yo Measles,3.0% -, i Malana 8.4% A t Severe infection 7 5% Neonataltetanus 2 3% ?onatalcauses nia 22 8% Preterm 47% Diarrhoealdiseases.0.7Yo Congenitalanomalies, 1.2% I Diarrhoealdisease,18 2% i HIV/AIDS.8 0% Source: OMS, ChildHealthandEpidemiologyReference Group(CHERG) 164. The package of maternal and child health interventions used as basis for the analysis is shown inTable 21. The table also shows the conditions that each intervention intends to alleviate. Some interventions, like assisted delivery by qualified health personnel, can improve the condition o f both children and women. Interventions Causes of child mortality Causes of maternal mortality Skilled delivery X X X X x BEOC(Basic X X X X xx xx x x Emergency Obstetric Care) Comprehensive X X X X x x x x x CEOC (Comprehensive Emergency Obstetric Resuscitation of X newborn babies sufferiw from 43TheBasic Emergency Obstetric Care (BEOC) refers to the following six basic functions: (i) parenteral antibiotics, (ii) parenteral oxytocics, (iii) parenteral sedativesianticonvulsants, (iv) manual extraction o f the placenta, (v) uterine revision, and (vi) assisted vaginal delivery. As regards CEOC, in addition to the six basic BEOC functions, this refers to bloodtransfusions and Caesareans (UNFPA, 2003). 46 Interventions Causes of child mortality Causes of maternal mortality Managementof X neonatal infections Antibioticsfor X U5 pneumonia ART for children X with AIDS Antibioticsfor X diarrhea and enteric fevers VitaminA - X treatmentfor measles Artemisin-based X Combination Therapy for children Universal x x X X emergency neonatalcare (asphyxia aftercare, managementof serious infections, managementof the VLBW infant) Source: Adapted from the MBBTechnical Guide(forthcoming) 165. Two scenarios were envisaged for the simulations: a scenario to increase the effective coverage of these interventions to 60% of the target population and another one to 80%44. Data on the current coverage o f these interventions are scarce and thus is not easy to estimate the increase in coverage required to achieve the defined targets. However, the MICS 2005 survey provides some o f the information required such as the percentage o f births taking place in health facilities (28.5%) and the percentage o f children under five suffering from pneumonia and for which care was sought from a qualified provider (37.8%). These two indicators, whose values are much lower than 60% or 80%, were used duringthe exercise as indirect indicators o f the demand for healthcare by women during delivery and by children under five. These values were used to estimate the baseline coverage o f other interventions o f the package. 44The universal coverage hypothesis (at least 95% effective coverage) of the interventions presentedhas been explored but the simulation results were not of such a nature so as to change the conclusions of the cost analyses. 47 The issue of human resources 166. The poor availability of quality human resources constitutes one of the major challenges the country faces to guarantee the effective coverage of the interventions includedinthe presidentialmeasure. 167. To estimate personnelshortages, the norms defined by the MSP4'at different levels of the health pyramid were compared with the actual number of various categories of health professionals. These estimated shortages linked to an assessment o f the potential sources o f traininghecruitment of health personnel allowed a definition of various policy scenarios to overcome these shortages andto ensure the optimal delivery o f services. The analysis below only includes health centers, district hospitals, and regional hospitals. This analysis takes the national health norms as given without making any judgment on them. However, given the current situation regarding human resources, a national discussion on the practicability of these norms and apossible revision might be needed. It is also necessary to discuss a realistic timeframe to achieve these norms. Table 22: Health personnelshortagesaccordingto the MSPnorms Professionalcategory Current number Current availability Objectivem Level of fulfillment (per 1 million (per 1 million of requirements inhabitants) inhabitants) A I and A2 Nurses 946' 126 537 24% A3 Nurses 1557' 208 703 30% Paramedics 208 21 260 8% General practitioners 112b 15 , 21 71% Specialists 95 13 24 54% Source: Director of Human Resources, presentation on human resource managementissues in Burundi in October a 2007 during ajoint donor. bNationalHealthDevelopmentPlan. 168. The data in Table 22 show the shortages in nursing and paramedical personnel (laboratory and radiology technicians, and others.) Currently, only 24% o f the requirements for qualified nurses (A1 and A2) are fulfilled, implying a need for 3079 additional staff to overcome this shortage. In the case of unqualified nurses, only 30% o f the requirements are fulfilled. The situation appears relatively better for general practitioners and specialists where the requirements, as per the MSP standards, are fulfilled in 71% and 54% respectively if we only consider the health pyramid levels mentioned earlier. However, 52% o f the doctors working in the public sector have administrative positions in Bujumbura4' and do not provide direct services to users. Therefore, our figures on the fulfillmento fthe norms regardingdoctors are underestimated. Inthe case of paramedics, only 8% o f the requirements, as established by the national norms, are fulfilled. 45 MSP, WHO (2007) Healthnorms for the implementation ofthe 2006-2010 PNDS inBurundi. 46 The document o f National Health Norms for the Implementation o f the 2006-2010 PNDS in Burundi (final version, June 2007) defines the heath norms interms o f the number o f personnel per health facility. The document also provides the target population for each health facility category. This information enablesto translate these norms interms o f personnel per million inhabitants. 47 Director of HumanResources, presentation on human resource management issues inBurundiinOctober 2007 during ajoint donor mission. 48 169. In the absence of detailed information on the production capacity o f the country's health training schools and the annual rate o f loss (retirement, death, brain drain), it is difficult to estimate the years needed to overcome this shortage if we rely entirely on training within the country. For the purpose ofthis simulation, it was assumedthat it would be necessaryto cover the human resource requirements by at least 60% in 2010, the final year of the National Health Development Plan, to achieve an effective coverage o f 60% o f the interventions included in the "free" package o f services, or 80% inthe case of the 80% scenario. Table 23: Additional personnel requirements in 2010 according to the assumptions of the achievement rates of the MSP norms Professionalcategory Current number Additional personnel Additional personnel requirementsin 2010 to requirementsin 2010 to 60% of the achievement 80% of the achievement rate rate A1 and A2 nurses 946a 1669 2541 A3 nurses 1557a 1870 3013 Paramedics 208 1109 1531 Generalpractitioners 11Zb -- 27 Specialists 95 20 59 Source: Director of Human Resources, presentation on human resource management issues in Burundi in October a 2007 during ajoint donor. NationalHealthDevelopmentPlan. 170. By 2010, the additional requirements for qualified nurses for a projected population o f 8.9 million people are estimated to be 1669 and 2541 respectively for the 60% and 80% coverage scenarios. For unqualified nurses, the additional requirements for the two scenarios are 1870 and 3013. For paramedics, to achieve 60% o f the MSP norms would entail 1109 additional staff. To achieve 80% of the norms, 1531 additional paramedics would be required. As mentioned earlier, the situation i s less alarming inthe case of doctors. Inthe case of general practitioners, currently 71% of the norms are already achieved and thus no additional personnel are required with exception of those needed to replace outgoing personnel so as to maintain the current level. For the second scenario, the need for general practitioners is estimated at 27. The figures for specialists are 20 `and 59 respectively. These figures on doctors will need to be increased by an additional 42 to overcome the shortage created by the number o f doctors working in administrative positions inBujumbura. 171. In summary, it would be necessaryto triple the number of qualified nurses available by 2010 to achieve60% of the MSPnormand quadruplethis numberto achieve 80% of the norm. Similarly, it would be necessary to double or triple the number o f unqualified nurses. Regarding paramedic, it would be necessaryto multiply the current number by eight or by ten to achieve 60% and 80% o f the norm. For doctors, it will be necessary to increase the current number of general practitioners by about 25% by 2010 to achieve 80% of the norm and the number of specialists by about 20% and 60% respectively to achieve 60% and 80% of the MSP norm. WillBurundi have thesepersonnel available by 201O? 172. Several solutions are conceivable to improve the availability of health personnel and to ensure the delivery o f the services included in the presidential measure, in particular, assisted delivery, and basic and comprehensive emergency obstetric care. 49 173. The Government has envisaged the retraining o f unqualified nurses (A3 category) to become A2 nurses. The methods for this retraining have not yet been clearly defined. The analysis that follows does not take any judgment on this measure. However, given the current situation regarding human resources, a national discussion on the practicability of this measure mightbe needed. 174. The above measurewould imply the retrainingand conversion of more than 1500 A3 nurses into A2 nurses, significantly increasing the pool of qualified nurses and would virtually allow the achievement of almost 60% of the MSP norm by 2010. This measure will bring about a progressive elimination of A3 nurses but this gap will needto be also covered for the country to reach the national norms. The country has several training schools for unqualified nurses whose courses would also need to be adapted to train A2 qualified nurses. According to the Department of Human Resources, Burundihas around twelve paramedical schools (5 public and 7 private including the Universities o f Ngozi and Mwaro). These establishments produce around 600 unqualified nurses each year4*. However, the availability of teaching staff and the investment requirements could constitute a major bottleneck to the rapid and effective adaptation o f the programs. Nonetheless, without prejudging the absorption capacities o f trained personnel and the dynamics o f the labor market, the country would have sufficient production capacity for qualified nurses to satisfy the specified norms. While the due date o f 2010 may seem somewhat ambitious, the year 2015 seems to be more reasonable ifthe requiredinvestments are made early enough and ifthe methods o f retraining are clearly definedand followed. 175. The situation appears to be far more complex regarding laboratory and health technicians. According to the Health Sector Note4', the country trains approximately 40 laboratory and healthtechnicians eachyear. All things being equal, Burundiwould need approximately27 years to reach 60% of the MSP norm and 38 years to meet 80% of the norm. The country cannot rely on its internal production capacity to overcome this shortfall by 2010. To make the health facilities fully functional, the country must find the means to solve this situation. An alternative would be retrainingone category of existingpersonnel to rapidly reduce this shortfall. 176. According to the 2006-2010 PNDS, the Bujumbura Faculty of Medicine trains around 20 general practitioners each year and offers some specialization cycles in conjunction with several French universities5'. If we assume that four extra years will be required to train doctors in gynecology-obstetrics, pediatrics, general surgery or internal medicine, and that halfthe outgoing class follows these specializations, it would take five years to train 20 specialists and to cover 60% of the needsspecified by the MSP norms. 177. Apart from producing new doctors, brain drain remains a major issue to be resolved to makethe district and regionalhospitalsfunctionaland to assist them in ensuring efficient care. Inthe case of doctors, the shortfall could be rapidly absorbed ifappropriate incentives are in place to hire and retainthe personnel inunderserved areas, to reverse the tendency to migrate, and to facilitate the return of expatriate personnel. The financial implications o f a salary increase for medical personnel are discussed inmore details later on. 48Health SectorNote, 2006. 49Idem. 50 Private universities have also been training doctors and soon the first graduates will be available. However, their diplomas have not been certified and thus the analysis was based on the graduates ffom Bujumburaonly. 50 178. The country has also benefited from donor programs that have brought foreign doctors to work in the country. For example, the Gitega regional hospital has several Chinese specialist doctors financed mainly by the Chinese Government. All these efforts can help the country to have the needed doctors to reach at least 60% o fthe norms by 2010. 179. The estimations presented above are based entirely on the norms established by the Ministry of Public Health and do not judge their adaptability to the current country context. These norms are necessary for planning and they can guide the national health policy. However, the design o f these norms should take into account a realistic assessment o f the needs in terms o f human resources (personnel profiles and level o f services), the work load o f the personnel, and the capacity o f the country to generate the conditions needed to ensure that the personnel is hired and retained inhealth facilities. Cost and impact of the different assumptions on coverage Table 24: Expectedimpacts of each scenario and the per capita cost Reductionof Annual per capita neonatal mortality Reduction of infant-juvenile maternalOf mortality cost Scenario I(60%) 32% 30% 61% 10.8 US$ Scenario II(80%) 43% 40% 82% 13.0US$ 180. Increasing the effective coverage of the package of maternal and infant health interventions defined earlier to 60% would potentially reduce maternal mortality by about 6l%, neonatalmortalityby 32% and childmortalityby 30%. This, at a total cost o fUS$ 10.8 per capita per year by 2010, of which US$ 0.7 per capita per year would be for administrative services and technical support. Achieving 80% effective coverage would imply a potential 82% reduction in maternal mortality, a 43% reduction in neonatal mortality and a 40% reduction in child mortality. This, at a total cost o f US$ 13.0 per capita per year by 2010, of which US$ 0.9 would be for administrative services and technical support. These estimated costs represent the expenditure levels to be achieved by 2010 to reach the specified coverage level. They represent the recurrent costs andthe necessary investments annualized over the planningperiod 2007-2010. 181. In absolute terms, these figures imply a total expenditure of US%234 million over the next five years to achieve 60% coverage of the interventions included in the "free" package of services. This excludes the financial resources required to maintain, or even increase the coverage o fpreventive interventions such as children's vaccinations, the use o f insecticide-treated mosquito nets by children and pregnant women, prenatal care, etc. To attain the objective o f 80% effective coverage, the financing requirements are estimated at about US$ 279 million during the 2007-2010 period. 182. Estimates from previous chapters indicate that total expenditure on health in Burundi in 2006 was between US$ 14.5 and US$ 18.5 per capita. Public expenditure on health (donors and Government) was estimated between US$7.6 and US$9.7 per capita. These figures suggest that it would be necessary to allocate the total Government budget and all donor support to activities linked to the "free" clinical package to achieve about 60% coverage of the beneficiariesand achieve a potential30% reduction in childmortalityand a 61% reduction in maternal mortality between 2007 and 2010. The cost of achieving 80% coverage o f the 51 "free" package i s significantly greater than the public resources available. The Government will therefore need significant additional resources to increase the coverage of this package o f services. 183. Phasing the investment required over a longer period of time, beyond 2010 or 2015, will be necessary given the current resource constraint. The total amount to rehabilitate the system and to cover a larger proportion of the target population is only indicative of the investment and runningcosts required. The implementationperiod may thus be re-defined by the national experts. The definition o f the implementation period should take into account the macroeconomic situation of the country and its capacity to mobilize more resources for the sector. Table 25: Allocation of the financing requirements by expense category (inUS$ thousands) Category Scenario I Scenario II (60%) (80%) Antimalarials 7 403 10961 Other medicationand consumables 23 936 34 612 Transport 5 659 7 589 Infrastructureand equipment 125244 138 264 Humanresources(salaries, incentives, benefits, etc.) 29 013 30 094 Promotion+ creation of demand 43 51 Basictraining 8 592 10 525 Follow-up,evaluationand technical and institutionalsupport 33 663 46 689 GrandTotal 233 553 278 786 184. Table 25 shows the estimated costs of delivering the "free" package o f services disaggregated across expenditure category. Infrastructure and equipment representthe most significant expense item in the two scenarios, with approximately 54% o f the expenses (US$ 125 million) inthe 60% coverage scenario and about 50% o f the expenses (US$ 138 million) in the 80% scenario. As indicated earlier, the expenditure profile which is dominated by infrastructure and equipment i s explained by the need to bring the health system to the level neededto make the services available to beneficiaries across the country. The limited information obtained from a health facility survey carried out in 2006 (health map) suggests that less than 40% o f the health centers have the necessary equipment for providing effective and quality BEOC". The data for hospitals were not available but anecdotal evidence o f the condition of hospital infrastructure suggests a need for considerable rehabilitation o f some hospitals. These costs do not only include the investment costs for improving equipment and infrastructure, but also the operating and maintenance costs which accompany the runningo f these establishments. We assume that the revitalization of the existing health establishments would be sufficient to meet the demand generatedby the presidential measure. Therefore these costs do not include the construction of new healthfacilities. 51 A set of equipment deemed essential to BOEC management was identified and the health centers in possessionof halfthis set were considered to be satisfactory. The availability of the equipment analyzed includedan aspirator bulb, an episiotomy kit, a delivery kit, an obstetrical stethoscope, etc. Approximately 37.5% ofthe health centers analyzedwere inpossessionofhalfo fthis equipment. 52 rigure 17: Relative weight of each expense category for each scenario of effective coverage ScenarioI (60%) Scenario II (80%) Other Follow-upand evaluationAnti-malanab med,cat,Dnand FI,llow-up and evaluation 3 93% medicationand Other and technicalsuppait 3 17% consumables anid technicalsuppolt Basictraining Basic Stimulationof 3 68% 1KiinlnQ demand 3 78% 0 02% Stimulationof demand - 0.02% Salanes. Salaries, Incentives incentives, beneffis etc benefits. etc. 12 42% 10.79% Healthequipment and matenals and materials 53 63% 49.60% 185. Monitoring and evaluation activities as well as technical support from the provincial level to the districts and health facilities, or from the district to the facilities constitute the second largest expense item. This category represents 14% and 17% o f the total cost o f the implementation of the "free" package o f services at 60% and at 80% effective coverage respectively. The cost of these activities i s estimated at about US$34 million inthe 60% scenario and about US$ 47 million inthe 80% coverage scenario. The relatively significant weight of this expenditure item can be explained by the need for training and for raising the standard o f health personnel particularly in BEOC and CEOC but also in clinical IMCI which i s still in an experimental phase. The reinforcement o f the EPISTAT health information system forms an integral part ofthis category. 186. Drugs and supplies, including antimalarials and antiretrovirals for children, also represent a significant cost of the implementation of the "free" healthcare measure. Maintaining drug revolving funds for essential drugs in health facilities represents an important part of the system's reinforcement. The cost of the initial training o f healthcare providers appears relatively marginal in relation to the total cost o f the "free" healthcare package. This cost i s estimated at just over US$ 8.5 million if we wish to meet 60% o f the norms regarding nursing staff and at around US$ 10.5 million to meet 80% of the norms. However, these costs do not include investment costs neededto reinforce the production capacities o fthe training institutes. 187. Lastly, the salaries, incentives and various benefits to medical personnel represents the fourth largest expense item, with just over 12% of the total costs in the 60% scenario and under 11% in the 80% scenario. The remuneration costs are relatively low when compared to other expenditure items such as infrastructure and equipment. The level of salaries and various bonuses used for the purpose o f this simulation are markedly different from the actual average remuneration currently received by heajth care personnel. The base salary used for this cost estimation was provided by the Director of Human Resources of the Ministry o f Public Health. To this base salary, the level of allowances announced to take place starting January 2008 was added. As indicated in Table 26, the increase in the various bonuses and benefits contribute substantial additional charges which, in the majority of cases, constitute almost two times the 53 base salary52.These figures do not include family allowances, which amount to approximately US$ 1.4 per child per month, up to a maximum of 3 children and US$ 1.8 per month per spouse, nor travel allowances of around US$ 8.0 per official per month. Table 26: Basesalaries and various bonuses and allowances use for the cost estimations Professional Base salary53 Housing Risk premium: Basic Category allowance 30% of the base allowance Total (monthly in US$) in (monthly in salary (monthly in US$) (monthly in US$) US$) US$) AI Nurses 30 45 9 14 98 A2 Nurses 21 27 6 10 63 A3 Nurses 15 18 4 6 43 General practitioners 39 45 12 14 110 Specialists 48 45 14 14 121 Source: Department of HumanResources ofthe MSP. 188. The personnel remuneration cost of delivering the "free" package of services is estimated at US$29 million inthe 60% coverage scenario and at US$30 millionin the 80% scenario. The salary amounts used for this cost estimate are already higher than the salaries received by medical personnel in Burunditoday. However, they appear relatively insufficient to effectively accompany the redeployment effort and the retention o f medical personnel. For instance, the remuneration level inRwanda i s two to four times higher than inBurundidepending on the professional category concerned (see Table 27). To increase the incentives for,qualified personnel to work in underserved areas, the Government has already allocated bonuses o f about US$450 to doctors posted in 14 district hospitals outside Bujumburathat remainedclosed. This bonus is for a period o f one year and i s beingfinanced through HIPC funds. 189. To assess the financial implicationsof a salary increase of medicalpersonnel, we have carried out various simulations by doubling, tripling or quadrupling the salaries. The amounts in question are relatively small compared to the total cost o f the package but implementingsuch a measure is politically difficult. The Ministry o f Health workers are part of the civil service. As a result, it is difficult to increasethe salaries of one category o f civil servants without creating demands from other sectors such as that of education or the army where the number of civil servants is much higher and where a wage increase would have a significant impact on the overall wage bill. One option to solve this issue would be to have contractual staff and sign performance contracts. Another option would be to sign performance contracts with civil servants. The basic salaries of these personnel would fall within the jurisdiction o f the civil service and would follow the same salary scales. However, the portion o f salaries concerning performance bonuseswould be administered by the Ministryof Health. Table 27: Comparisonof health care personnelremuneration in Rwanda and Burundi ProfessionalCategory Burundi Rwanda Ratio Rwanda / Burundi AI Nurses 98 296 3.0 A2 Nurses 63 176 2.8 General Practitioners 110 491 4.5 Specialists 121 707 5.8 52The levelofallowances andpremiumsbefore January2008 was muchlower. 53Includesthe 15% increase appliedsince July 2006. 54 190. As indicatedby the simulations in Table 28, an increase in the salaries of doctors by a factor of 2,3 or 4 to accompany the policyof personnel retentionand their redeployment in the field would generate a marginal increase in the total cost of scaling up the "free" maternal and child health package. To double the salaries o f general practitioners and specialists, as estimated in Table 26, would only require additional financing o f about US$ 0,06 per capita beingjust under US$ 2 million for the 2007-2010 period inthe 60% scenario and US$ 0.07 per capita beingjust over US$ 3 million in the 80% scenario. When considering the most optimistic scenario which seeks achieving 80% of the MSP personnel norms and an 80% coverage of the interventions, tripling doctors' salaries would require additional US$ 0.11 per capita or US$3.5 million for the 2007-2010 period. Tripling the salaries of general practitioners and quadruplingthose o f specialists was estimated at around US$O.l3 per capita orjust over US$ 4.1 million underthe most optimistic scenario. Table 28: Financial implications of an increase in doctors' salaries (US$ thousands) 6O%Scenario 80%Scenario US$ per (US$ US$per (US$ capita thousands) capita thousands) Human resources (salaries, incentives, benefits, etc.) 0.89 29 013 0.93 30 094 Additional resources requiredto doublethe salaries of doctors o.06 1 973 0.07 2 078 Additional resources requiredto triple the salariesof doctors 0.10 3 350 0.11 3 503 Additional resources requiredto triple the salariesof general practitionersand to quadruplethose of 0.12 4 008 0.13 4 166 specialists 191. Regarding nursingstaff, the amounts in question are higher even though the salaries are lower than those of doctors. The numberof staff affected by such measure (in excess of 2,500) is considerable when compared to that of doctors (barely more than 200) such that the financial needs rapidly become significant. To better understand the financial implications of an increaseinnurses' salaries, the analysis was carried out including inthe package o f services preventive services which were "free" of charge before the presidential measure such as vaccination, the distribution of insecticide-treated mosquito nets, and others. Inthe case o f nurses, the simulations ought to have extended beyond the "free" clinical healthpackage as this category of personnel i s used at various levels and across a range o f service delivery methods. A detailed analysis o f the delivery o f these diverse services in addition to the "free" package o f curative services i s detailed later. As indicated in Table 29, under the less optimistic scenario (60% effective coverage and a 60% achievement rate o f personnel norms), to double the salaries o f qualified and unqualified nurses would, in addition to the expenditure required to cover the salaries presentedinTable 26; requires about US$ 0.69 per capita or just over US$22 million for the 2007-2010 period. To triple the salaries of nursing staff under the same scenario, the Government will need to spend an additional amount o f about US$ 1.24 per capita or about US$ 40 million between 2007 and 2010. Under the most optimistic scenario, these figures will be in the region of US$ 0.74 per capita or US$ 24 million and US$ 1.32 per capita or US$ 43 million respectively to double and triple nurses' salaries. The salary simulations were done separately for doctors and nurses for methodological reasons. However, it would not be advisable to increase the remuneration of one category o f personnel and not the other; especially given that nurses provide most ofthe services available and given their scarcity. 55 Table 29: Financialimplications of an increase innurses' salaries (US$ thousands) 60% Scenario 80% Scenario US$ per (US$ US$per (US$ capita thousands) capita thousands) Human resources (salaries, incentives, benefits, etc.f4 1.06 34 507 1.11 35 851 Additional resources required to double the salaries of nurses o.69 22 442 0.74 24 198 Additional resources required to triple the salaries of nurses 1.24 40 393 1.32 42 951 Community andpopulation health interventionsand child health 192. The major component of the "free" package o f services relates to health care for children under five. As detailed earlier, the provision o fthe childhealth interventions included inTable 21 at a 60% effective coveragewould reduce child mortality by 30% at a cost of US$ 10.8per capita per year between 2007 and 2010. Effective coverage o f 80% for the same interventions would reduce childmortality by 40% at a cost o fUS$13 per capita per year over the same period. 193. I t would be important to know if similar or better results can be obtained with a different package of services given the high estimated costs of the "free" package. Indeed, the knowledge accumulated through experience in several low and middle income countries suggests that some community, household, and population health interventions would enable a substantial reduction inchild mortality usingrelatively inexpensive technology. The Lancet series on child and neonatal survival offer an overview of different effective health interventions that can be offered at relatively low cost in a development context. Most o fthese interventionscan be provided through outreach or scheduled clinical care or through community-based interventions as is the case ofvaccinations, ITNs, exclusive breastfeeding, and others. 194. At the moment, Burundidoes not have a community healthdevelopment strategy but the Government is aware of the approach and intends to make it an essential pillar in its healthsystem. The community health workers (CHWs) are generally only usedwithin the scope of the extension of clinical activities at community level such as the fight against tuberculosis or HN/AIDS. These activities are not well coordinated and create considerable distortions in the healthcare system. As no formal and standard remuneration for the CHWs exists, the more lucrative programs benefitfrom their service indetriment o f less well-financed ones. 195. At the moment, some population based interventions offered through outreach or through scheduled clinicalservices have a relatively high coverage in Burundi. Although, as explained in a previous chapter, their quality is not always assured. For instance, the MICS 2005 survey indicates that 67.9% o f children have beenvaccinated against measlesand at least 96 % o f birthshave benefitedfrom one or more pre-natal care visits. However, these services which were free of charge before May 2006 have remained underfundedand thus some o fthese services have very low effective coverage. For instance, the same survey indicates that only 45% o f children in the appropriate age are fully vaccinated. Similarly,the contraceptive prevalence rate remains low; only 7.5% o f marriedwomen 15 to 49 years use a modern contraceptive method. 54Representsthe human resource expense items (salaries, incentives, benefits, etc.) as part of a cost estimate which includes the fiee maternal and infant healthcare package, as well as other preventive interventions already provided free o f charge prior to the measure of free health care. 56 196. As in the "free" package of services, the estimated costs of the community and population-based package of services include the investment costs required to develop a community health program and the recurrent costs requiredto reach60% to 80% coverage levels. The training o f the CHWs is included in the cost simulation. An integrated approach was assumed for this training where various fields of responsibilities are taught. For instance, the training sessions on the management o f child diseases are assumed to be integrated. The same applies for training sessions on communication activities for behavioral change such as advice on nutrition or the use of impregnated mosquito nets. The costs associated with the motivation and supervision of the CHWs are also included. Table 30 lists the community and population-based interventions used for this simulation. This package is not exhaustive and may be revised and adapted in accordance with Burundi's specific context and with the national policy trends. For instance, whether or not to authorize CHWs to administer cotrimoxazole to children within the scope of community management o f pneumonia and dysentery, which are two o f the major causes of child mortality in Burundi.This i s a decision that the Government must take regarding the community health development. Table 30: Community and population-based healthcare package Community-basedhealthinterventions Use of long-lasting impregnatedmosquito nets by children under 5 and pregnantwomen Handwashing by child caretaker Clean delivery and cord care Early breastfeedingup to 5 months Extendedbreastfeedingup to 24 months Complementaryfeedingfrom 6 months Oral rehydrationtherapy (new formula) in the case of diarrhea Artemisin-basedcombinationtherapy for malaria management Managementof acute respiratoryinfectionsand dysenterythrough the use of antibiotics Population-basedhealthinterventions Family Planning Pre-natalcare Tetanus vaccinationsto pregnantwomen Intermittent PresumptiveTreatment (IPT) for pregnantwomen PMTCT (testing and counseling,AZT + sd NVP and infant feeding counseling) Cotrimoxazoleprophylaxisfor HIV + mothers ExtendedProgramon Immunization(measles, BCG, pentavalent,polio) Vitamin A supplementation 197. As indicated in Figure 18, 60% coverage for each of the interventions in the community and population-based package would potentially reduce child mortality by about 31% at a cost of US$4.9 per capita per year from 2007 to 2010 (of which US$0.7 per capita would be for management costs and technical support). However, this package would not have a significant impact on the maternal mortality ratio, just slightly over a 6% decrease. In the 80% coverage scenario, the community healthpackagewould reduce child mortality by about 40% at a cost of US$ 6.3 per capita per year (of which US$0.9 would be for management costs and technical support). Maternal mortality would only decrease by 8% in this scenario. Indeed, high-impact interventions for maternal survival such as BEOC and CEOC are not included inthis package and cannot be delivered at this level since they require the presence of qualified health personnel such as nurses, obstetricians, etc. 57 Figure 18: Impact and cost of family-oriented community-based services $7.0 40.0% - $6.0 35.0% f $4 9 c $5.0 30.0x- PC 25.0% - $4.0 .s E F E 20.0% - Q 2 C $3.0 0 1 s -g 8 1s.w- $2.0 10.0% . 5.0% 1 $4.0 0.W c Scenario 60% SCE.*IIO80K fMortallt6 neonatale 63 Mortalite des enfants de rnoins de 5 an5 1L Mortalite rnaternelle IICootoarhabitantIan 198. We can conclude, based on a comparison between the results o fthe free package o f services and that ofthe community basedpackage, the following: a. Regardingchildrenhealth, the community and population-based package appears to be more cost-effective than the "free" packageof services. Inthe two simulation scenarios (effective coverage o f 60% and o f 80%) for the two service packages (community/population and clinical) we can expect a similar impact on child survival. As mentioned earlier, under the most optimistic effective coverage scenario, both packages will bring about an approximate 40% reduction in child mortality. However, the costs o f implementing these two packages are very different. The cost of implementing the "free" package is estimated at about US$ 13 per capita for the 80% coverage scenario, whereas the community/population package would cost about US$ 6.3 for the same level o f coverage. These results suggest that the development o f a community health program together with improvement in population based services i s the best alternative for substantially reducing childmortality at a reasonable cost. b. The impact of the community/population package on maternal mortality is insignificant even if the cost of this package is less than that of the "free" clinical package. As mentioned earlier, and as recaptured in Table 31, the "free" clinical package would potentially enable a 61% reduction in maternal mortality in the 60% effective coverage scenario and of 82% in the 80% effective coverage scenario, this is mainly due to BEOC and CEOC which are included inthis package. Table 31: Comparison of the "free" clinicalpackageand the communityand population-based child mortality Maternalmortality Per capita cosffyear 60% "free"clinical package 30% 61% 10.8 US$ 60% Communitypackage , 31% 6% 4.9 US$ 80% "free" clinical package 40% 82% 13.0 US$ 80% Communitypackage 40% 8% 6.3 US$ 58 199. The above results suggest that Burundi could, at a relatively low cost, undertakethe implementation of a community health program that would facilitate rapid progress in child survival. This approach would ensure the management o f the principal causes of death in children at community level. However, the need to reduce maternal mortality requires not losing sight o f the reinforcement of the healthcare delivery system that could take place gradually by first revitalizing Health Centers and district hospitals to enable them to practice caesarians and to cope with the demand createdby the community-based activities. Conclusionandpolicy implications 200. To achieve an effective coverage of 60% of the health interventionsincluded in the packageof "free" services, the country would need to allocate about US$ 10 per capita per year in a four year period. This can result in a 30% reductionin child mortality and a 60% reduction in maternal mortality. However, this represents about 70% o f total health expenditure in the country in 2006 and about 1.3 times total public expenditure on health (including foreign aid). Ina more optimistic scenario of reaching an effective coverage of 80% in 2010, the Government wound needto spend about US$13 per capita per year to achieve an infant mortality reduction of about 40% and a maternal mortality reduction o f about 80%. 201. Giventhe large resourceconstraints the countryfaces, it is necessaryto re-examine the implementation period of the investments necessary to revitalize the health system to achievethese objectives.A horizon larger than 2010 and even 2015 will be needed. This phasing decision should take into account the Government's budget constraint and the capacity o f the country to produce, hire, and retainthe neededhealthcare personnel. Additionally, the capacity to mobilize additional resources (both internal and external) and to utilize them will also be important to ensure the effective implementationof the measureto eliminate user fees to children under five andwomen duringdelivery. 202. The large amounts o f estimated resources needed to increase the effective coverage o f the "free" package o f services are partly based on the personnel norms defined by the Ministry o f Public Health. These cost estimations did notjudge the pertinence o f these norms but given the scarcity o f personnel, they might need some revision. This revision should be based on a realistic assessment o f the system's needs in terms of human resources (personnel profiles and level o f services), the work load o f the personnel, and the capacity o f the country to generate the conditions neededto ensure that the personnel is hiredand retained inhealthfacilities. 203. Infrastructure and equipment represent about half of the total estimated cost of increasing the coverage of the package of "free" services. In contrast, the remuneration o f personnel representsabout 12% of the total cost. The simulations suggestthat an increment of the remuneration of the health personnel in certain cases can represent only a small share o f the total cost of the package. However, as a measureto increase remuneration could generate pressures in other sectors to also increasethem this measure can be politically difficult to implement. 204. The "free" package of services is essentially a clinical package of services. Nevertheless, the most important gains in terms of child survival can be achieved with community-based interventions (e.g. ITN usage) combinedwith population-based ones (e.g. children vaccinations). These interventions could potentially have the same impact that the clinical package but at a much lower cost. However, this package only has a marginal impact on maternal survival. 59 REFERENCES Atim C. (2005) "Plan de Financementdu PlanNational du DCveloppementSanitaire (NHDP) du Burundi (2006-2010) : Besoins, Scenarios et StratCgies de Financement." Darmstadt GL, Bhutta ZA, Cousens S, Adam T, Walker N, de Bemis L for the Lancet Neonatal Survival Steering Team. 2005. "Evidence-based, cost-effective interventions: how many newbornbabies can we save?, TheLancet Neonatal Survival Series 365: 977-88 ECHO, Dr.J Shepherd-Barronand F Fadiga. 2005. "Evaluation of ECHO'SFinancedActions in Burundi", EvaluationReport.Germany. European Commission, J-P Louis, A Halajko and T Hatorimana. 2006. ((Evaluation finale du programme d'appui transitoire au secteur de la santC burundais PATSBU D. Rapport Provisoire.Lettrede march6No2006/126249.Belgique. Gottret, P. and G. Schieber. 2006. "Health Financing Revisited: A Practitioner's Guide". The World Bank, Washington, DC May 2006. Heller, Peter. 2005. UnderstandingFiscal Space. "IMF Policy Discussion Paper". International Monetary Fund, WashingtonDC, March 2005. ISTEEBU and UNICEF. 2001. Enquete d'Evaluation des Conditions de vie de 1'Enfant et de la Femme au Burundi(ENECEF-Burundi 2000). RapportFinal. Bujumbura :Mars 2001. Jones G, Steketee R, Black R, Bhutta Z, Morris S, and the Bellagio Child Survival Study Group. 2003. "How many child deaths can we prevent this year?, The Lancet Child Survival Series 362: 65-71 MCdecins Sans Frontibres. 2004. AccBs am soins de s a d au Burundi: RLsultats de quatre enqugtes Lpide`miologiques.May 2004, study financedby ECHO. Ministbre de la Planification du DCveloppement et de la Reconstruction (MPDR) and UNDP.2003. ((Rapport National sur le DCveloppement Humain au Burundi: L e VWSIDA et le DCveloppement Humain au Burundi D. Bujumbura. MinistryofPublic Healthofthe RepublicofRwandaandEuropeanCommission, JNachtigal and JMunyenpenda.2006. Revue des dCpenses publiquesdu secteur de la santC (( D. Ministere de la SantC Publique : Projet SantC et Population 11. 2005. Assistance technique en gestion des ressources humaines. Documentpreparedby Andre C6tC. Ministbre de la SantC Publique : Programme de Lutte contre les Maladies Transmissibles et Carentielles. 2005. Enquete dans les Structures de Soins : Evaluation de la Prise en (( 60 Charge des Cas de Paludisme dans les Structures de soins conformement au nouveau Protocole de Traitement Antipaludique D. 24 Janvier-23 FCvrier2005. Ministry of Public Health, WHO .2007. Health norms for the implementation of the 2006-2010 PNDS inBurundi. Final version, June 2007. OECD DAC. 2007. "2006 Survey on Monitoring the Paris Declaration, Country Chapters, Burundi.An OECD Publication. Organisation Mondiale de la SantC et Ministere de la SantC Publique. 2007. Evaluation de la (( distribution et de l'utilisation des mCdicaments donnCs par ECHO dans le Cadre de la Mise en (Euvre de la Mesure de GratuitC des Soins en Faveur des Enfants de Moins de Cinq ans et des AccouchementsD. Mars2007. RCpubliquedu Burundi. 2004. Projet de Politique Nationale de SantC (( D.Bujumbura, septembre 2004. Republiquedu Burundi, Ministbre de la SantC Publique.2005. PlanNational de DCveloppement (( Sanitaire 2006-2010)). Bujumbura, 4 octobre2005. Soucat A, Van Lerberghe W, Diop F, Nguyen S, Knippenberg R. "Marginal Budgeting for Bottlenecks: A new costing and resourceallocation practice to buy health results: Using health sector's budgetexpansionto progress towardsthe MillenniumDevelopmentGoals in Sub-Saharan Africa" The World Bank Policy and Sector Analysis Support Team- Africa, Washington. UNAIDS. 2006. "2005 Report on the Global AIDS Epidemic: A UNAIDS 1 0 Anniversary ~ SpecialEdition". A UNAIDSPublication.Geneva. United Nations Population Fund (UNFPA). 2003. Maternal Mortality Update 2002: A Focus on EmergencyObstetric Care.New York: UNFPA. UNFPA. 2007. "State of World Population2007" WHO. 2008. Maternal Mortality in 2005: Estimatesdeveloped by WHO, UNICEF, UNFPA and World Bank. World Health Organization, Geneva. Xu K, EvansaDB, KadamaaP, NabyongabJ, Ogwalc PO, NabukhonzodP, Aguilar AM. (2006) "Understanding the impact of eliminating user fees: Utilization and catastrophic health expendituresinUganda," Social Science & Medicine 62: 866-876 61 ANNEXES 62 eC C 'FE 'FE ,2+ i; ..a i us n 6 5) z7ucc U a ? e 6+e fa i j $ 6 C C eC .L ; F D EE 4 -rIt el U Q .. .3 Y sf .3 P 82 3 3E z" 0 v) 0 (Y I .. CI v) Q 0 Pa -cma P 5 ,-v) h s E -m P0E U c Q tiEa u W I .I -& a a I 5a) BBw a EM 0 `a) bo .I Y P1 .I a c, h m Pi m 0) 3 E 3 .3 IA 3 W 3 3Ya z .3 m m W 3 o m o r xm pi c, -1 v1 .n v) c, m .. 3 rA r r .-C 0 a2. r m 3. ' m 0 e -e m Y 8 8X E z 0 f Lh i!E rA 1 .. U 4B X 0 uorr Cia 0 0 0 a a a 0 8 0 c) I. .. 8 m G? ' f 'I rA m P (9 L c, 0 V c, 0 M e .I V c 9e I m E8 Y 8 E s 0 r, c, m 51 m 58E t0 5a2 t U m U .. v) aJ n 4 E .-c 0 -.-E 0 Q, 0 N 0 .-aiVE+2a .-C0 .-Q c 0 Y E 0 W 8 .-C0 -.-E c 0 0) 0 b hl 0 0 W 0 N .-C I m e c) Q) 0X s 0 E P c) m m m f8 t z 8 c) .. \o 4LX Annex 7: References used for the construction of the data base on external assistance to the health sector in Burundi 1. Donors 0 World Bank Aide-Me'moirede la Mission de suivi de la Banque Mondiale du 14 octobre au 21 octobre 2005. Second Health and Population Project; Preparation o f the Third Health and Population Project; Preparation of the health sector note. Aide-Mimoire de la Mission de la Banque Mondiale du 6fe'vrier au 4 mars 2006. Preparation of the Implementation Completion Report o f the Second Health and Population Project (PSP 11); Identification and Preparation of the Third Health and Population Project; Preparation o f the health sector Note for Burundi. Projectfor Health andPopulation Project II.Final Report. March2006. Bujumbura. Aide-MPmoire de laMission d'appui de la BanqueMondiale au secteur de la santi dejuin 2006. Implementation CompletionReport No ICR 0000332 on a Credit and a Grant in the total amount of SDR 21.5 million to the Republic of Burundi for a Second Health and Population Project. January 2007. AFTH3 andAFCC2, Africa Region. Finance lawsfrom 2003 to 2007. Belgium Questionnaire sent bythe MSP and the WB and completed by PNLT. Projections provided by the Ministryof Public Health o f Burundi. Switzerland Questionnaire sent bythe MSP and the WB andcompleted by the Swiss Cooperation. DFID Questionnaire sent bythe MSP and the WB and completed by DFID. DFIDBURUNDI: Assistanceto the Health Sector 2002-2007. Annual reports o fthe CNLS. ECHO Humanitarian Aid to vulnerablepopulation groups in Burundi. Global lans 2004,2005, 006. European Commission. List of commitments/contributions andpledges toprojects not listed in theAppeal as of July 11, 2007. Compiled by OCHA for 2005 and 2006 on the basis of information providedby donors and appealing organizations. FinancialTracking Service, Tracking HumanitarianAid Flows. UNFPA Questionnaire sent by the MSP and the WB and completed by UNFPA. Country Programfor Burundi 2005-2007.DP/FPA/CDP/BDV6. Executive Board of the UNFPA. October 2004. Global FundsMalaria Re'alisations du Fonds Mondial Paludisme 2004, 2005 et 2006. Support to the Initiative ((Roll Back Malaria D. Global Fundsto fight HIV/AIDS Questionnaire sent by the MSP and the WB and completed by the Global Funds to fight HIV/AIDS. Annual reports o fthe CNLS. - 75 - 0 Global FundsTuberculosis Questionnairesent by the MSP andthe WB andcompletedby PNLT. 0 France List of commitmentshontributions andpledges toprojects not listed in the Appeal as of July II, 2007. Compiled by OCHA for 2005 and 2006 on the basis of information provided by donors and appealingorganizations. Financial Tracking Service, Tracking HumanitarianAid Flows. 0 GAVI Questionnairesent by the MSP and the WB and completedby GAVI. Situationfinanci2re des comptes du Programme Elargi de Vaccination.Ministryof Public Health inBurundi, Direction GBnBrale de la Sante Publiqueet Direction des Services et Programmesde SantB, ProgrammeElargi de Vaccination. March2007. Bujumbura. e Japan Projectionsprovidedbythe Ministry ofPublic Health of Burundi. 0 World Health Organization Questionnairesent by the MSP andthe WB and completedby WHO. Projections providedbythe Ministry ofPublic Health of Burundi. 0 UNDP Questionnairesent bythe MSP andthe WB and completedby UNDP. Annualreportsofthe CNLS. Appui du PNUD h I'intensiJication de la Iutte contre le Sida. Node projet 39514. May 2006. Bujumbura. 0 UNICEF Questionnaire sent by the MSP andthe WB and completedby UNICEF. Aspect du Financementpar Donateur 2006, 2007. November2007. Bujumbura. Financementen cashdes ONG2006-2007. November2007. Bujumbura. Work plans and action plans ofUNICEF 2004,2005,2006. 0 EuropeanUnion Questionnaire sent by the MSP andthe WB andcompletedby the EuropeanUnion. Projectionsprovidedby the MinistryofPublic Healthof Burundi. 0 USAID USAIDData Sheetfor FY 2005andFY 2006. Reconstructionfor Developmentin Burundi. Guiding criteria and selected key issues. USAID. April 2006.USA. - 76 - 2. NGOs Association Burundaise Dour la Prise en Charge des Malades VulnCrables du VWSIDA (ABC MAV) Action Plan 2007. Alliance Burundaise contre le SIDA (ABS) Annual report 2006 and action plan2007. January 2007. Bujumbura. 0 Association Burundaise pour le BienEtreFamilial (ABUBEF) Annual report 2006. February2007. Bujumbura. 0 Association de Cooperation et de Recherchepour le Developpement (ACORD) Annual report2006. January 2007. Bujumbura. 0 Action Contre la Faim(ACF) Annual report2006. 0 Adventist Developmentand ReliefAgency (ADRA) Report on 2006 activities. April 2007. Bujumbura. 0 Association Jeunesse Chretienne contre le SIDA (AJC-SIDA) Reporton 2006 activities and action plan 2007. March2007. Bubanza. 0 CARE Annual report 2006. February2007. 0 Comitato Collaborazione Medica (CCM) Questionnaire sent bythe MSP and the WB and completed by CCM. Report on 2006activities. March 2006. 0 CED Caritas Burundi Annual report 2006. February 2007. 0 Communauti des Eglises de PentecGte(CEPBU) Annual report 2006 and action plan 2007. 0 InternationalCommittee o fthe RedCross (ICRC) Questionnaire sent by the MSP and the WB and completed by ICRC. 0 Catholic OrganizationFor ReliefAnd Development Aid (CORDAID) Annual report 2006. January 2007. Programmationde I'intewention CORDAIDau Burundi. Firstdraft. March 2005. 0 Concern Worldwide Burundi Questionnaire sent bythe MSP and the WB and completed by Concern Worldwide Burundi. 0 Ensemble pour une Solidarite Therapeutique EnRCseau(ESTHER) Questionnaire sent bythe MSP and the WB and completed by ESTHER. 0 Familv Health International(FHI) - 77 - Questionnaire sent by the MSP and the WB and completed by FHI. 0 Force des Universitaires Contre le SIDA CFUCOS) Report on 2006 activities. 0 GruupoVolontaria Civil (GVC) Annual report 2006. February 2007. 0 HealthNet InternationalTranscultural Psvchosocial Organization(HNITPO) Annual report 2006. December 2006. 0 HandicapInternational Annual report 2006. InternationalMedical Corps (IMC) Questionnaire sent by the MSP andthe WB and completed by IMC Kirundo and IMC Muyinga. Annual report 2006. February 2007. Bujumbura. 0 Jesuit Refugee Service (JRS) Annual report 2006 for health activities. 0 Association Internationale des Volontaires La'ics (LVIA) Questionnaire sent by the MSP and the WB and completed by LVIA. MCdecins Sans FrontiBresBelgique (MSF Belgique) Questionnaire sent by the MSP and the WB and completed by MSF. 0 Nouvelle EspCrance Annual report 2006. Organisation d'Appui A 1'Auto-Promotion (OAP) Annual report 2006. March2007. Bujumbura. 0 Pouulation Services International(PSI) Report on 2006 activities and activitiesplanned for 2007. 0 Association des femmes africaines face au SIDA (SWAA) Annual report 2006. February 2006. Bujumbura. 0 World Vision InternationalBurundiCWVIS) Questionnaire sent by the MSP and the WB and completed by World Vision International Burundi. - 78 - Annex 8: Donors listed in the databaseof external resourcesflowing to the healthsector in Burundi World Bank Belgium Switzerland DFID ECHO UNFPA GlobalFundsMalaria GlobalFundsHIV/AIDS GlobalFundsTuberculosis France GAVI Japan World HealthOrganization UNDP UNICEF EuropeanUnion USAID -79- Annex 9: NGOslisted in the database of external resources flowing to the health sector in Burundi Association Burundaise pour la Prise en Charge des Malades VulnCrables du VWSIDA (ABC MAV) Alliance Burundaise contre le SIDA (ABS) Association Burundaise pour le BienEtre Familial (ABUBEF) Association de Cooperation et de Recherchepour le DCveloppement (ACORD) Action Contre la Faim (ACF) Adventist Developmentand ReliefAgency (ADRA) Association Jeunesse Chretienne contre le SIDA (AJC-SIDA) CARE Comitato Collaborazione Medica (CCM) CED Caritas Burundi CommunautCdes Eglises de Pentecdte(CEPBU) ComitC Internationalde la Croix-Rouge (CICR) Catholic Organization For ReliefAnd Development Aid (CORDAID) Concern Worldwide Burundi Ensemble pour une Solidarite ThCrapeutiqueEnRCseau(ESTHER) Family Health International(FHI) Force des UniversitairesContre le SIDA (FUCOS) Gruppo Volontaria Civil (GVC) HealthNet InternationalTranscultural Psychosocial Organization(HNITPO) Handicap International InternationalMedical Corps (IMC) Jesuit Refugee Service (JRS) Association Internationale des Volontaires Lajics (LVIA) MCdecins Sans Frontikres Belgique(MSF Belgique) Nouvelle EspCrance Organisation d'Appui A 1'Auto-Promotion (OAP) Population Services International(PSI) Association des femmes africaines face au SIDA (SWAA) World Vision InternationalBurundi(WVIB) - 80 - Annex 10: Health evpendituresfor selected do Drs and NGOs in Burundi (2006) r- i Belgium (2006) European Unien i MSFBelgium CCM Personnel Training, IEC o Pharmaceuticals uMaintenance Contrstruction, rehabilitation E3 Medical equipement Other equipement Source: authors' calculation based on the data base on external financing of the health sector inBurundi. - 81 - Annex 11:Case study: budget managementat different levels of the Ministry of Health A. Introduction a. Context After 1993, Burundi's public expenditure on health decreased sharply, negatively affecting the functioning o f the health system. Households had to increase their expenditure to offset this reduction. However, as household income levels are low and existing exemption and waiver mechanisms were not functioning properly, the increase in out-of-pocket expenditure on healthhas created large financial barriers to access these services. With peace and stabilization inthe country, the government started to receive additional resources. The international community has returned and the country has benefited from debt relief through the Heavily Indebted Poor Countries Initiative. Thanks to HIPC resources the budget ofthe MSP increasedsignificantly after 2005; in2006 alone the budgetincreasedby 80%. Before benefiting from the HIPC initiative, salariesconstituted a large share of the MSP's budget. Thanks to HIPC funds, the MSP must now managea muchhigher level o fresources. This creates a challenge to the MSP who is currently facing difficulties to execute its budget. This is partly due to weaknesses in planning and management o f resources, particularly in the procurement o f goods and services. Inaddition, the recent presidential measure to eliminate user fees for services provided to women during delivery and children under five has created more management difficulties for the MSP. Recognizing the need to improve the management of resources, the Ministry of Health created a Direction Ge'ne'rale des Ressources (DGR) but the insufficient number of human resourceshas limited its efficacy. b. Objectives This case study was conducted inthe context described above and has three objectives: (i) understandtheflowofresourcestodifferentlevelsoftheMinistryofPublicHealth; (ii)to to assessthe constraints faced in the management o f these resources; and (iii) to propose corrective measuresto reduce these constraints. c. Methodology These case studies were based on guided interviews55with health personnel in three provinces: Rutana, Gitega, and Ngozi (see the questionnaires and the list o f persons interviewed). Ineachprovince, the provincialMe'decin-cheJ;the FinancialandAdministrative Directors ofthe Bureaux Provinciaux de Sante' (BPS), the Me'decin Directeur of the Regional Hospital or District Hospital, and the titular of a health center were interviewed. These interviews took place the last week o f March2007. The specific objectives o f these interviews were to better understand: (i)the flow of resources at provincial and health facilities levels; (ii) the planning process and the preparation of the budget at provincial and health facilities levels; (iii)constraints in executing the budget at these levels; and (iv) the impact o f the presidential measure to eliminate user fees for children 55These interviews were conducted by Ms Euphrasie Ndihokubwayo from MSP, Ms Maria Eugenia Bonilla-Chacin and Mr Marc Nene ffom the World Bank. This study was made with support ffom Mr CyprienBaramboneranyeand MrJean-PaulNyarushatsi from MSP. - 82 - under five and women during delivery. The information collected during the visits to the three provinces was complemented with guided interviews with key personnel in the Ministry o f Finance and at the central level o fthe MSP. The selection of the provinces was based on the need to identify differences in resource management processes between provinces receiving some external support and those not receiving any. One province, Rutana, received support through an EUproject; a second province, Gitega, has not received any major support; and finally, Ngozi has recently started to receive support from the Swiss Cooperation and had previously received support from the World Bank financed PSP I1project. B. Resourceflow There are four main resource flows from the central level o f government to the BPS and health facilities. These four resource flows are used for the payment o f personnel, non-salary recurrent costs excluding drugs, investment costs, and pharmaceuticals and medical supplies. All these four flows follow the government's public expenditure processes. a. Salaries and bonuses The salaries of civil servants are paid directly by the Minist2re de la Fonction Publique. Besides civil servants, health facilities recruit contractual staff for technical, administrative, or support purposes. Usually, the salaries of contractual staff are paid out o f the health facilities' own revenues. Hospitals with managerial autonomy receive a budget allocation from the Ministry o f Finance for their functioning. These allocations are usually small and mainly cover salaries. The salaries o f health care personnel paid by the Fonction Publique are low and have proven insuficient to provide incentives for personnel to work outside Bujumbura, especially in remote areas. As consequence, partners to the health sector have paid for salaries and bonuses to healthcare workers. However, these salaries and bonuseshave beencontroversial. They are often many times higher than the salaries paid in the public sector and have thus created a large disincentive for personnel paid by the Fonction Publique and working along side those receiving salaries and bonuses from partners. These salaries and bonuses varied considerably between partners and between provinces and even within a province. For instance, in the provincial hospital of Rutana, doctors paid through external assistance received about US$ 3,000 per month while those paid by the Fonction Publique received less than US$ 100. In Ngozi hospital, specialists paid through external assistance earn about Euro 1,500 and general practitioners Euro 300 while civil servants receive much less. To compensate physicians paid by the Fonction Publique, this hospital gives them bonuses but these bonuses are not sufficient to cover the difference. b. Non-salary recurrent costs BPSs receive resources from the central level to cover their runningcosts as well as the runningcosts of the health facilities under their supervision. These funds are usually insufficient inrelationto the needs. EachBPS submits a requestto the MSP for office supplies, commodities, maintenance, etc. but it often does not get what it had asked for. - 83 - Until recently the Bureaux Provinciawc de Sante` received a small budget allocation for recurrent costs but they had limited flexibility inthe use of these resources as they could only be used for specific budget categories (fuel, generator, wheels, etc.). Since 2006, these resources are not directly allocated to the BPS; the MSP now has to provide all the resources needed in kind. Additional resources from HIPC funds are now available for the functioning o f BPSs and health facilities but they are entirelymanagedat the central level o fthe Ministryo f Health. Hospitals with managerial autonomy must cover their running costs with their own resources. c. Investmentcosts Investment costs of all the health facilities, including hospitals, are mainly financed through donors' support. Part of these resources are directly managed by donors or executing agencies, usually international NGOs. Another part o f this support is managed by Project Execution Units (Unit& d'Exe`cutionde Projet) inthe MSP; as was the case of the PSP I1project financed by the World Bank. Investment costs financed by donors often escape the government's expenditure chain and mainly follow donors' monitoring procedures. d. Pharmaceuticals There is a specific budgetary line in the Ministry o f Health budget to finance pharmaceuticals and medical supplies through CAMEBU. In2006, this budgetary line reached FBU700 million. However, this line is not sufficient to cover all the needs ofthe health facilities inthe provinces; therefore, BPSs buy some pharmaceuticals from the CAMEBUwiththe revenue collected by healthcenters. Health facilities also receive drug donations. For instance UNICEF has distributed essential drugs. It has also distributed, thanks to ECHO financing, drugs, delivery kits and caesarean section kits to respond to the presidential decree to eliminate user fees for children under five and women during delivery. However, some of these pharmaceuticals do not correspond to needs. For instance, pharmaceuticals adapted to children under five were not provided in the kits. Although CAMEBU was responsible for the distribution o f all these drugs, the first stock of drugs and kits donated by ECHO was managed separately from other drugs and supplies. Healthfacilities hadto managetwo different drugstocks, one for deliveries and children under five and another for all other purposes. The management of these two stocks was difficult to handle and inefficient. DFID has also provided large quantities o f pharmaceuticals to respond to the May 2006 presidential measure; by March 2007 these drugs were starting to be distributed to health facilities. The management of DFID donation has been integrated with the rest of the pharmaceuticals. Hospitals with managerial autonomy finance most o f their pharmaceuticals with their own revenuesbutthey also benefitfrom ECHO and DFIDdonations. In the visited provinces, the pharmaceutical chain, the flow of resources to buy pharmaceuticals and the way revenues from the sale o f pharmaceuticals are managed are neither standardized nor clear. Health facilities buy pharmaceuticals to the BPS when they are not provided by UNICEF/ECHO or DFID donations. The BPS replenishes its stock thanks to the - 84 - MSP budgetary line for pharmaceuticals and to the revenues collected selling these pharmaceuticals to health facilities. However, there is no harmonized price policy for services provided or for pharmaceuticals. The BPS sells pharmaceuticals to health centers with some benefit, varying from a province to another. For instance in Rutana, where drug revolving funds have been established, the BPS buys pharmaceuticals from the CAMEBU at 100% of their price and sells them back to the health centers with a 5% margin. Finally, health centers sell the drugs to users with a 15% margin. This system is supposedto ensure the sustainability ofthe revolving fund for pharmaceuticals. However, now that some services are provided "free" o f charge, these drug revolving funds have experienced some difficulties. e. Revenuesof healthfacilities To compensate for the decrease in health facility revenue caused by the presidential decree of May 2006, the central level of the MSP instituted a mechanism to reimburse facilities for services delivered "free" of charge. The reimbursement i s made by the MSP who checks all the invoices sent by each health facility. The reimbursement is financed by HIPC funds and i s directly sent to the health facilities' bank account. The government's delay to reimburse these invoices causedsignificant difficulties to health facilities as they could hardlycover their running costs, such as salaries of contractual staff, maintenance, supplies, and, in hospitals with managerial autonomy, utilities. By March 2007 many hospitals had large and increasing debts with REGIDESO, the water andelectricity company. Parallel to this top-down flow from central to lower levels, there i s a bottom-up flow o f resources from health facilities to higher levels o f MSP. Part o f the revenues collected by health centers, essentially from drugs and services provided, goes to the BPS or to communal structures. However, as mentioned above, the management and the use of these resourceswidely vary across provinces and escape from the government's monitoring mechanisms for public expenditure. For instance, in some provinces, health centers must send part o f their revenues to the BPS to cover part of the BPS running costs. In Gitega province, health centers used to send 10% o f their revenues to the BPS. In Ngozi, the BPS used to retain 20% of the revenues health facilities collected from the sale of pharmaceuticals. With the presidential measure to eliminate some user fees, these practices have stopped as the facilities' revenues markedly declined. Ingeneral, facilities' own resources are managed more transparently in health centers where communities are largely involved in the management process through Health Committees (COSA) and Management Committees (COGE). In the three visited provinces health facilities usually have two bank accounts, one for the revenues collectedfrom the sale o f drugs and another for the revenues collected from the services or "acts". Most health centers have flexibility in the use of the funds collected in the "acts" account. These resources are often used to cover running costs, including the payment of salaries of some contractual staff such as security personnel. In contrast, the management and the use of resources from the pharmaceutical account are more restrictive and often escape the control o f the health center. For instance, inNgozi, only the BPS can withdraw resources from this account. InGitega, the revenues from the pharmaceutical bank account are managed by the Titular Nurse of the health center and the Communal Administrator without any community participation. Before, inthis province, all these revenues were sent to the communal government. Among the three provinces visited, Rutana, which benefited from EU support, is the only province where the community i s fully involved in the management o f the revenues collected by health centers. Inthis province, the revenues o f facilities are managed by a Management Committee which is part of the Health Committee in which the community is - 85 - largely represented. Ineach health center, an administrator elected by the community reports to the COSA, and therefore to the community. The salary o fthis administrator is paid by the health center. The revenues collected on the pharmaceutical bank account constitute part o f the facility's drug revolving fund; this fund was functioning relatively well before the presidential measure of 2006. C. Budget preparationprocess The overallbudget preparationprocessis centralizedbut fragmented across type of expenditure and source of financing. The budgetary process starts in June-July when the Ministryof Finance sends a letter to all sectoral ministries. As the letter sent by the Ministry of Finance does not provide a budget envelop, the preparation of a realistic budget is difficult at best. Furthermore, recurrent and investment budgets are not consolidated at the sectoral ministry level which meansthat the projects' runningcosts are not always integrated to the programmatic budget56.DuringJuly and August the budgetary bargainingacross sectors takes place. At the level of the MSP, each programme, department, BPS and autonomous hospital preparesa budgetplan for non-salary runningcosts and sends itto the Direction du Budget (In the Direction Gbne`raledes Ressources) which is responsible for the preparation o f the consolidated MSP budget. However, these budget plans are seldom taken into account by the Ministry of Finance and the budget from the previous year is often reproduced with a 10% increase. As far as HIPC resources are concerned, budget planning takes place after the budget allocations are known. Inother words the MSP preparesan expenditure plan for HIPC resources once the budget i s approved. The MSP faces several challenges during budget preparation. At lower levels o f the ministry, especially at BPS level there is low capacity for budget preparation. As a result, their budget plans are often not linkedto the sector's strategiesor objectives and budget allocations are not prioritized. At the central level o f the Ministry (DGR with some help from the Cellule de Planzfrcation),the consolidated budget is linkedto the sectoral policies and objectives and some expenditures are prioritized. However, the budget preparation at this level also faces challenges: 1. Budget preparation is fragmented. The budget for HIPC resources is prepared separately from the rest o f the budget. Additionally, recurrent and investment budgets are not consolidated at the sectoral ministry level which means that the projects' running costs are not always integratedto the programmatic budget57. 2. There is no consolidated account o f donor funds flowing to the sector. Although each health program knowswhat it receives from external assistance, the central level does not have complete information on donor funds. This limitsthe ministry's capacity to plan and prepare a comprehensive budget as most external assistance i s not registered in the budget. 3. External assistance i s unpredictable. 4. There i s limitedcapacity to estimate the cost of health plans. 5. At the DGR the few persons with skills to prepare budgets are overwhelmed with other functions. For instance, the same people in charge o f budget preparation are responsible to check all the invoices sent by the facilities for the reimbursement o f "free" services. The same people also sign and send the letters to the Ordonnateur Trborier du Burundi 56 IMFReport 07/46, January2007 57 IMFReport 07/46, January2007 - 86 - (OTB) for the release of funds to all these facilities (there are about 500 facilities in the country). D. Budgetexecution Budget execution is low due to the centralization of public procurement and to delays in the disbursement of funds. The MSP receives monthly allocations of funds from the Ministry of Finance for its recurrent expenditure. Expenditure larger than FBU2 million must go through public procurements managed by the Ministry of Finance. Sectoral Ministries only prepare the procurement documents but the entire process is driven by the Ministry o f Finance. The MSP, as other sectoral ministries, also has some difficulties to prepare procurements documents. The DGR does not have sufficient staff with skills to prepare all these procurement documents. Because o f the difficulties and delays causedby this cumbersome procedure for large procurements, the responsibility to manage the procurement process i s being decentralized towards sectoral ministries. Another challenge the MSP confi-onts when executing its budget i s the limited number of personnel to monitor large expenditures (e.g. construction, rehabilitation, etc.). No monitoring i s done o fthe expenditure financed and managed by donors. Finally, in2006 the delays of the Ministryof Finance to disburse HIPC funds also negatively affected the budget execution o f the MSP. For instance, in2006, HIPC funds were only allocated to the MSP in July causing large payments delays. The expenditure chain within theMSP is submitted to several controls and faces many constraints leading to large delays in the budget execution process. Until recently, a large part of the MSP budget was used to pay salaries leaving very little for other expenditures. The MSP therefore has limited experience in the management of public resources. Since 2006, with the HIPC funds, the budget o f the ministry has significantly increased. That year the MSP had difficulties inexecuting the new increasedbudget. The low budgetexecution in2006 was then mainly due to the following: 0 The new procedures designed to manage HIPC funds took a long time to be implemented and as consequence the disbursement o f these funds were markedly delayed. 0 The expenditure plans are prepared very late in the year. For instance, the expenditure plan for HIPC funds in 2006 was finished in April. In 2007, the expenditureplanwas finalized inMarch. 0 In2006, procurement documents were givenby the MSPto the Ministry of Finance inJunebutthe latter took a longtime to processthem due to changesinthe Direction des Marchb Publics. In the end, procurement documents were approved in December. In order to better plan the use of HIPC funds, the MSP prepares an execution plan for these resources. This budget execution plan could be improved by prioritizing expenditures and by including a commitment plan with a chronogram for the fiscal year. To improve resource management, the MSP created last year a Direction Gknkrale des Ressources to manage both financial and human resources. This Direction i s in charge o f the preparation ofthe consolidated budget ofthe ministry. Its efficacy remains limitedas it lacks - 87 - human resources. For planning purposes, the MSP has a Cellule de Planification that supports the Directeur Ge`ne`ralde la Santd,the Directeur des Programmes de Sante`,and the DGR.The entire team (DG, DGR, DP, and the Cellule de Planification) is at the moment drafting a health plan for 2007-2009. Although in general the team has the skills neededfor planning, it could benefit from building capacity to cost health programs. A system was put inplace by the Ministryo f Health to compensatefacilities for the loss o f revenue generated by the presidential initiative to eliminate user fees for women during delivery and children under five. For a facility to be compensated for these "free" services it, needs to send as "proof" an invoice for each o f the services provided and a photocopy o f the identity document ofthe beneficiary. UntilMay 2007, the procedure to check the invoices sent by health facilitiesfor the reimbursement of "free" services was cumbersome and led to long reimbursementdelays. Each health facility had to send to the BPS invoices for the inputs usedto deliver these services. The BPS checked these invoices and consolidated them by health facility before sending them to the central level ofthe Ministry for further control. After the MSP had checked all the invoices, they were sent to the Ministry of Finance for final approval and payment. This cumbersome procedure explains the large delays in payment to facilities. By March 2007, only a few hospitals were reimbursed for the expenditures made during the three months following the presidential measure. Many health centers hadn't received any payment and were severely affected by the decrease inrevenue this measurecreated. Three propositions were formulated to further alleviate the procedure: Each BPS could send one note to the Ordonnateur Trborier du Burundi for payment o f all facilities in the province instead o f a note for each health facility sent by the MinistryofFinance. 0 The Ministry of Finance could make regular advance releases to the MSP and each BPS could withdraw funds from these releases when presenting a declaration o f credit. Change from payment o f inputs ex-post to pre-payment of services. Each health facility could receive a small initial allocation based on estimations o f the quantity o f services delivered to women and children. This amount would be renewed regularly after the facilities present invoices justifying the use o f the funds. This procedure could be difficult to implement as most health centers do not have much experience . inbudget management. The government o f Burundichose to combine the first two propositions presented above. The Ministry of Finance decided to provide advance releases to the MSP for the payments o f services offered to children under five and women during delivery. As consequence, by May 2007, the situation had improved considerably and delays in the payment of invoices were drasticallyreduced.It was agreed that the Ministryof Finance would send monthly advances o f FBU 150 million and that the MSP would send, for each health facility, a note to the OTB that would be signed by the Directeur Ge`ne`ral des Ressources, the Directeur du Budget and the Directeur Ge`ne`ralde la Comptabilite`Publique inthe Ministry o f Finance. This has considerably facilitated the payments to health facilities even if some payment arrears remain. For instance, there are still invoices in the Ministry o f Finance waiting to be paid and the MSP i s still waiting for old invoices fiom health facilities. - 88 - Inorder to acceleratethe payment ofthese invoices, the MSP sent to each BPS amanager exclusively responsible for the reimbursement o f "free" services. Despite these improvements, the MSP still needs to resolve the issue o f the sustainability of the measure of "free" services. Although we still ignore the financial cost o f this measure, it seems necessaryto better define the services that can be provided "free" o f charge and to estimate their cost. In addition, the monthly releases from the Ministry of Finance to the MSP won't be sufficient to cover the cost o f all the invoices sent by the health facilities as there are large arrears from the previous year. By the end o f May, 80% of the budgetary line for the payment o f these services was already used. E. Province specificobservations a. Rutana province There are two hospitals in Rutana province and thirty two health centers, twenty five public, four private for profit, and three private non-for-profit known as "agre`b'7.These health centers are located in the six communes o f the province, each commune having on average five health centers. Each health center is in charge o f a population varying from 4,000 to 12,000 inhabitants and has either four or two nurses depending on whether the health center provides housingaccommodation or not. Rutana province benefited from the European Union support through PATSBUproject that was financed by the 9` European Development Fund(EDF). Resourceflow andmanagement in Rutana's BPS The Provincialhealth bureau inRutana receives funds from several sources including the national budget, development partners such as the European Union, GTZ, the NGO International Medical Corps (IMC), and from the sale o f pharmaceuticals to the province's health centers. As far as the budget preparation and execution are concerned, the BPS in Rutana, as all other BPSs, i s supposed to have managerial autonomy but in reality it does not have direct control over its budget. The BPS orders equipments to the MSP for its functioning and receives them in kind. Underthe PATSBUproject financed bythe 9' EDF, the BPSwas preparingan actionplan with a budget but now it only preparesan annual actionplan. Salaries, bonuses and other running costs The salariesofthe BPS staffare paiddirectlynby the Fonction Publique. Inaddition to the payment o f salaries, the BPS usually receives a three-month allocation o f FBU 625,000 from the Ministry of Finance to cover its runningcosts but these funds are sent irregularly. For instance, they hadn't received these funds for almost one year. Furthermore, these funds are allocated to very specific budgetcategoriesand reallocations bythe BPS are not allowed. As mentioned above, the BPS also receives small equipment and supplies from the MSP for its functioning. These equipments and supplies must also be sent to the health centers upon their request. Development partners also provide suppliesto the BPSandhealthcenters. - 89 - Pharmaceutical chain The BPS in Rutana has a pharmacy whose initial stock was financed by the European Union through PATSBU project. The BPS buys pharmaceuticals to the CAMEBU at 100% o f their price and sells them back to the province's health centers with 5% benefit. The benefit collected selling pharmaceuticals i s usedto replenishthe stock at the BPS level. In addition to these drugs, the BPS also receives drug donations. The BPS distributes themto healthcentersto cover the needs of children under five and women duringdelivery inthe context o f the implementation of the presidential measure on "free" services. However, these pharmaceuticals are not always adapted to the needs o f the beneficiaries. For instance, pharmaceuticals in pediatric form are often missing in the kits. As a consequence o f this inadequacy between needs and available supplies, the BPS sometimes accumulates debts with the CAMEBU. Resourceflow and management in Rutanaprovincialhospital The provincial hospital in Rutana has managerial autonomy. It has one general practitioner, who is also the Mddecin Directeur; thirty two nurses; five A2 and twenty seven A3; eight administrative staff, two contractual staff; and twenty four support staff, thirteen contractual. The communities' involvement in the management o f hospital activities is led by COGE whose members are: the physician in charge o f the BPS, the hospital Director, the staff association and community representatives. Underthe EUPATSBUproject the hospitalhad inaddition two general practitionersand one surgeon. These physicians were directly paid by external assistance with a salary o f about US$3,000 per month. The hospital prepares an annual action plan with budget estimates and receives a global budget from the government to cover its non-salary running costs. Except for the salaries o f contractual staff that are paid with hospital's own resources, salaries are directly paid by the Fonction Publique. The hospital own revenue is also usedto provide incentives to the personnel through bonusesvaryingacross staff categories. In the case of pharmaceuticals, the drug revolving fund created under the PATSBU project was working relatively well and has allowed the hospital to avoid drug stock-outs. Stocks are replenished through the CAMEBU. However, the hospital at the moment o f the visit seemed to have financial difficulties to buy some pharmaceuticals. The provincial hospital o f Rutana also benefits from drug donations from development partners. Effect of thepresidential measurefor `pee''services According to the Mddecin Directeur o f the hospital, after the presidential measure the hospital, occupation rate has beenabove 100%.The human resource shortage is even more acute now that the demand for health services has drastically increased. Drugdonations to confront the increasein demand are often irregular and not adaptedto the needs o f beneficiaries. The hospital is facing important financial difficulties due to the decrease in revenue produced bythe May 2006 measure and to large delays inthe reimbursement o f the invoices. - 90 - Resourceflow and managementin Gitabahealth center Gitaba health center has four A3 nurses, one aide-soignante,a COGE manager chosen by the community, and four support staff. As all other health centers in the province, Gitaba health center has a health committee and a Management Committee. It i s worth noticing that this health center did not benefit from EDF support to the province. Salaries and other running costs The nurses and the aide-soignante are directly paid by the Fonction Publique. The salary o f the COGE manager i s paid by the health center out of the facility's own funds. The health center's revenue is also used to cover some non-salary runningcosts. The support staff is paid by the commune following an agreement between the Governor and the BPS. The health center also receives, ifrequested, small equipment and supplies from the BPS for its functioning. Revenues andpharmaceuticals The health center buys its pharmaceuticals from the BPS which sells them with a 5% margin compared to the indicative price given by the CAMEBU. The health center then sells them to the population with a 15% margin above the CAMEBU indicative price. The health center therefore retains a benefit o f 10%. Pharmaceuticals for children under five and pregnant women are sent for free by the BPS upon request. In the health center, the manager is responsible for the pharmacy. The titular nurse provides health services and writes prescriptions for patients who then visit the COGE manager to buy or, insome cases, to get pharmaceuticals free o fcharge. The health center has two bank accounts: one "acts" account for the revenues collected for the services offered and one account for revenues collected selling pharmaceuticals. As said above, the revenues from the "acts'y account are used to pay the salary ofthe COGE manager and to cover other running costs. The cost of services offered to children under five and women during delivery is supposed to be reimbursed by the MSP but until March 2007 Gitaba health center had not receive any payment. EfSect of thepresidential measurefor `Yee''services The number of children under five receiving care in the health center drastically increased since the M a y 2006 presidential measure but the health center's revenues significantly decreased as children under five and women during delivery represent the large majority of the patients attending the health center. As other health centers in the province, Gitaba health center has financial difficulties caused by the delays in the reimbursement o f the "free" services. In some health centers, the salary o f the COGE manager hasn't been paid for months. According to the heads o f the health center, the returns o f refugees, of people formerly displaced and o f former combatants, as well as the large number o f poor people have also decreased the facility revenue. -91 - b. GitegaProvince Gitega province has four health sectors (secteurssanitaires); each of them has a hospital. In addition to these hospitals, the province has a regional hospital and forty four health centers. The province effectively moved towards the district system in June 2007 with a Mdecin chef responsible for each district. As said before, Gitega province didn't receive any significant support from development partners but since November 2006 with a contractual scheme the NGO HealthNet TPO has provided direct support to some health centers inKibuye sector; this support has includedrehabilitation, drugdonations, and performance basedbonuses. Resourceflow and management in GitegaBPS There are thirty three people working in the BPS in Gitega. However, according to the Me`decin Directeur there is a surplus of about 20 people. The main team of the BPS would therefore be made of the Director, the Manager, the three provincial coordinators (health, reproductive health, and PNLT), four Heads of Sector, one person in charge o f the health information system, one nutrition focal point and one managerfor the pharmacy. The BPS receives resources from the government, from development partners such as ECHO, UNICEF, WHO, UNFPA, the NGO Health Net TPO and from the province's health centers. As far as the budget is concerned, the BPS in Gitega has an operational plan for 2007 with a budget prepared after the MSP had sent a note in December 2006 providing the budget framework. Before this year, the BPS hadan operational plan from 2003. Salaries, bonuses and other recurrent costs The salaries of the thirty three people working in the BPS in Gitega are paid by the Fonction Publique. In addition to the payment of salaries, the BPS receives subsidies from the government to cover non-salary running costs. However, between April 2006 and December 2006, the BPS didn't receive any financial resourcesbut only inkindresources. The MSP directly pays for some running costs such as invoices for telephone, water, electricity, car insurances for the BPS's vehicles andfuel. In addition to the support from the government, the BPS receives resources from the health centers that send to the BPS 10% o f the revenues collected. These resources are used to cover the BPS's running costs. However, after the presidential measure on "free" services, the healthcenters have not beenable to sendthese funds any more. Development partners such as UNICEF and WHO provide support to the province, either through projects such as the community based Nutrition Pilot Project financed by UNICEF, or through drug donations. Resourceflow andmanagement in GitegaHospital Gitega hospital is a regional hospital with 250 beds, 188 of which are functional. It has managerial autonomy since 1996. It has ten physicians, three from Burundi and seven from China; seventy three nurses, thirty two A2 and forty one A3; and one hundred ninety two - 92 - administrative and support staff. The hospital infrastructure and equipment are in very poor condition. The hospital has neither an ambulance nor a resuscitation service and most of the buildings are in urgentneedof rehabilitation. On the administrative side, the hospital has a Board of Directors (Conseil de Direction) and an Administrative Board (Conseil dYdministration) appointed by presidential decree. The Administrative Board meets every three months and is composed o f one President who i s the Mbdecin-directeur of the BPS, one Deputy-President from the MSP, one Secretary who is the HospitalDirector, one community member elected by the community administrators, one hospital staff, one person from the local administration, and one medical representative. As far as the budget is concerned, the heads of the hospital prepare a budget planwhich must be approved by the Administrative Board before being sent to the MSP and then to the Ministry of Finance for budget bargaining. However, these budget plans are often not taken into account and the budget is generally the same as the previous year with a 10% increase. However, duringthe budget execution, the additional 10% are not always made available. Salaries,bonuses and other running costs Gitega hospital receives an annual budget of FBU 142 million from the government to cover its running costs. This subsidy is essentially used to pay the salaries o f Burundian physicians. The amount of these salaries, which increased by 20% with the move towards managerial autonomy, isn't fixed by the hospital but by the Fonction Publique. The amount of bonuses givento the personnel is also determined by the government. The revenues from service delivery are also used to pay runningcosts and part o f the salaries. The salaries of the Chinese physicians are directly paid by the Chinese government but the government of Burundipays for their housing. Gitega hospital faces important financial difficulties since its managerial autonomy in 1996. These difficulties were intensified by the decrease in revenue caused by the measure on "free" services and the delays in the reimbursement of invoices. By the end of March 2007, the hospital had only been reimbursed for the services offered "free" o f charge between May and August 2006. This exacerbated previously existing hospital debts. For instance, the debt accumulated bythe hospital to the REGISESO since 1996 reachesalmost FBU60 million. Pharmaceuticals Gitega hospital mainly buys pharmaceuticals from the CAMEBU. If some pharmaceuticals are not available inCAMEBU, an authorizationfrom the MSP is requiredto buy pharmaceuticals from other suppliers. To help inthe implementationo f the presidential measure on "free" services, the hospital received an initial stock o f drugs of FBU25 millionthrough the BPS but this stock is insufficient to face the demand for services from children under five and pregnant women. Impact of thepresidential measure on `pee" services The measure on "free" services was followed by a drastic increase in demand for health services although the hospital's capacities were not strengthenedto face it.The staff (for instance inthe maternityward) and infrastructures (available beds) are insufficientto meet the demand. As - 93 - said before, the delay for the reimbursement of invoices increased the hospital's financial difficulties and debts. In addition to its debt to REGIDESO, Gitega hospital owes about FBU 50 millionto several pharmaceutical suppliers, including CAMEBU. Resourceflow and management at the health center level in Bukirasazi During the mission's visit the health center was distributing insecticide-treated nets to children under five and pregnant women. The large number o f beneficiaries and the work burden the distribution represented for all the staff didn't allow a complete interview with the titular nurse. The mission then visited a health center connected to a hospital managed by the Methodist church. This healthcenter seemed to work well and was waiting for an agreement with the government to implement the measure on "free" services. The health center had stopped implementing it as the MSP was not reimbursing the invoices for the services delivered. In addition, this hospital was supposed to start playing the role of a district hospital a few weeks after the mission's visit. c. NgoziProvince Ngozi province i s divided in three health districts that are already functional. The province has four hospitals, including the Ngozi regional hospital, and forty seven health centers, thirtythree public, sevenprivateand seven "agrdes". Since September2006, the province benefits from the support o fthe Swiss cooperation for the decentralization process and the development o f healthdistricts. The staff inthe districts i s made o f one district Mkdecin-chef(in the three districts these posts are currently occupied by nurses), one nurse coordinating the health information system, one nursemanagingthe district pharmacy and two support staff. Resourceflow andmanagement at theBPS level in Ngozi The BPS inNgozi i s composed of the Me`decinDirecteur of the province, a manager, two persons in charge of the pharmacy, one provincial coordinator, one coordinator for reproductive health, one coordinator for PNLT, one coordinator for the health information system, two secretaries and four security staff. The BPS receives resources from different sources, namely the government, development partners such as WHO, UNICEF, DFID, Swiss Cooperation and health centers o f the province. As far as budget is concerned, the BPS prepares an annual action plan with objectives and a provisional budget and sends them to the MSP but there i s no budget bargaining before the budget is allocated by the central level. Salaries, bonuses and other running costs As in other BPSs, the salaries of the personnel o f the BPS in Ngozi are directly paid by the Fonction Publique. Inaddition to the payment o f salaries and to the payment o f the invoices for water, electricity, and telephone, the government gives to the BPS a subsidy of FBU325,000 every three months. However, according to the Mkdecin Directeur of the BPS, this subsidy - 94 - wasn't received last year. It was given in 2007 but through specific budgetary lines for expenditures such as generators, fuel, and others. Besides the three-months budgetary line ofFBU 325,000, the MSP mainly provides in kind support for the functioning o f the BPS such as office supply, wheels, record books, etc. These supplies are furnished once the BPS has submitted a request but often the BPS does not receive all the items requested. In addition to the support from the government, the BPS generally receives resources from the province's health centers. They send 20% of their revenues from the sale o f some pharmaceuticals to the BPS. However, since the presidential measure on "free" services, health centers don't sendresourcesto the BPS any more. Pharmaceuticalchain There are no drug revolving funds in Ngozi. According to the head o f the BPS, the pharmacy o f the BPS buys pharmaceuticals to the CAMEBU thanks to an annual subsidy provided by the government. Then, health centers furnish themselves at the BPS after submitting a request for all pharmaceuticals, including those covered by the presidential measure of May 2006. As mentioned above, health centers send to the BPS 20% o f their revenues earned selling pharmaceuticals. The other 80% were originally supposed to go on a bank account where they would accumulate at the end of each year to help create a revolving fund in health centers in the future. However, as the subsidy for pharmaceuticals sent by the MSP was insufficient, these funds were partially usedto replenishthe BPS's stock o fpharmaceuticals. To helpwith the implementationo fthe presidential measure for "free" services, UNICEF provided drug donations financed by ECHO to the BPS to cover the last three months o f 2006. Drugdonations fi-om DFID are also expected in 2007. To face the increase in demand, the BPS sometimesneeds to runinto debts with the CAMEBU. Impact of thepresidential measurefor `pee''services At the level of the BPS, the control and compilation of the invoices sent by health facilities take a long time for the staff and also require additional office supplies. However, the delays for the reimbursement of these invoices are mainly due to the slow administrative process at the central level. Resourceflow and management at thehospitallevel in Ngozi The hospital in Ngozi is a regional hospital with 222 beds. The hospital provides the following services: surgery, gynecology and obstetrics, pediatrics, neonatology, resuscitation and intensive care, emergency care, dentistry and otorhinolaryngologic services. There are fourteen physicians working inthe hospital, six o f them are specialists (pediatrician, obstetrician, surgeon, cardiologist, dentist, otorhinolaryngologist); 103 nurses, three with a license, two A1 nurses, twentythree A2 nurses, sixty nineA3 nursesand six nurses trained at work. Ngozi hospital was built in 1935 and rehabilitated in 2005 thanks to'the second Health and Population project financed by the World Bank (Projet Sante`Population Io.It now benefits from the support of the Pro-Africa Foundation financed by the Italian Cooperation which pays most o f the salaries and some equipment. On the administrative side, the hospital has a Direction Committee,and a Management Committee but it has no Administrative Board. The Ngozi hospital sends a monthly report to the accounting authority (Cow des Comptes) and to the Ministry o f - 95 - Finance. Similarly, in order to have the MSP replenish its bank account, the hospital has to send a reportto the MSP with a copy to the Ministryof Finance. The hospital has a financial deficit, mainly due to debts generated by the delays in reimbursements for the services provided to the Beneficiaries of the Carte d'Assurance Maladie and ofthe presidential measureo f 2006. Salaries, bonuses and other running costs The hospital in Ngozi receives funds from several sources. It benefits from FBU 7 million monthly allocations from the MSP. The hospital has autonomy in the management of these resources. As mentioned above, the replenishment o f the bank account is made after the monthly report is submitted. In addition to this subsidy, the hospital earns revenues from its activities and is supported by the Foundation Pro-Africa. It also benefited from drug donations from other development partnerssuch as UNICEF, ICRC and DFID. The salaries of the hospital staff are paid by the MSP, the Foundation Pro-Africa and the hospital itself. For instance, the salaries o f the six specialists are paid by the Foundation whereas four general practitioners are paid by the Fonction Publique and four directly by the hospital. However, to reduce the differences in remuneration, the hospital provides incentives and night duty bonuses to the staff o f FBU 32,500 and FBU 30,000 respectively. The Foundation also provides a monthly bonus o f FBU 20,000 to all the staff. Specialists and general practitioners paid bythe Foundation earn respectively Euro 1,500 and Euro 300. Inthe context ofthe implementationofthe presidential measurefor "free" services, the hospital inNgozi benefited from the reimbursement o f invoices for services delivered to children under five and women during delivery from May 2006 to September 2006. The delays in the reimbursemento fthese invoiceshave hada negative effect onthe functioning o fthe hospital. Eflect of thepresidential measurefor `pee" services The hospital inNgozi i s facing a significant increase in demand for health services since the presidential decree on "free" services. To meet this demand, as the invoices are not reimbursed in time, the hospital sometimes runs into debts with the CAMEBU and other suppliers. Stock-outs are frequent since May 2006. Although the hospital had a previous debt with the REGIDESO, this debt increasedconsiderably after the presidentialmeasure on user fees. SinceMay 2006, the hospital has an FBU3 milliondebt. Resourceflow and management at the health center level in Rukeco The Rukeco health center is located 12 kilometers from the BPS. This health facility has three nurses, one A2, who i s the titular nurse, and two A3; the health center also has 3 support staff. On the administrative side, the health committee was recently established. Before this, the titular nursemanageddirectly all facility resources. Salaries and other running costs Except for the salaries of the two security personnel, the staff salaries are directly paid by the Fonction Publique. The health center also receives support from the BPS such as small materials for its functioning. In addition, the health center finances its running costs with the - 96 - revenue collected through the delivery o f some services. The facility's own revenue is mainly used to pay for the salaries of the security personnel and some maintenance costs. The health center keeps all the revenuescollected from services or "acts". However, for any withdrawal from its "acts" account, the health center needs first to have an authorization from the Midecin Directeur o fthe BPS. Invoices for services deliveredto children under five and pregnant women are sent each month to the BPS. However, when the interview took place in March 2007, the health center hadn't received yet any payment. Pharmaceuticals As mentioned above, the health center submits to the `BPS all request for pharmaceuticals. Pharmaceuticals that are not concerned by the presidential measure are sold to patients at the CAMEBU's price and all revenues collected are sent to a bank account. The BPS withdraws 20% from this bank account to finance part of its running costs. The other 80% are accumulated each year and are supposed to support the creation o f a drug revolving fund in the future. In this health center, drug stock shortages are more related to the limited capacity to manage the stocks than to financial difficulties. The Titular Nurse pointed out that it was difficult to monitor the drug stock as they had too much work to do. Eflect of thepresidential measurefor `pee" services This measure was followed by a decrease in revenues and therefore the facility had difficulties to cover some running costs such as maintenance. - 97 - List of persons interviewed Health Province Institution visited List of persons interviewed Function Ministryof Finance M. DonatienBwabo President Cellule d'Appui Mme. BeatriceHamenyayo aux reformes des finances publiques Directeurdu Budget Ministryof Public Mme. Anne Marie Directeurdu Budget Health Rutana Provincial Bureau in Dr Antoine SINDAYIGAYA Directeurdu BPS Rutana M. NDlTlJE Melchiade Gestionnairedu BPS Rutana Hospital Dr NIYONGABO Directeurde I'h6pital M. BUTOYIJean Marie Gestionnairede I'hbpital Gitaba HealthCenter M. NlJlMBERE Patrice Titulaire du CDS Gitega Provincial Bureau in Dr Georgette Directeurdu BPS Gitega NDIH0KUBWAYO Gitega Hospital Dr Fidele NIYONKURU Directeurde I'h6pital Kibuye Hospital M. BlGlRlMANA Pascal Titulaire du CDS et de I'h6pital Mrne NlYONlZlGlYE Helene Gestionnaire Mrne NDIKUBWAYOFranpise Bukirasazi Health Titulaire du CDS Center Ngozi Provincial Bureau Dr InnocentNKURUNZIZA Directeurdu BPS M. NGENZEBUHOROJean Gestionnairedu BPS Ngozi Hospital Marie M. MBONIMPAJoris Gestionnairede I'hdpital Mme MBONIMPAGonzague Chef de Nursing Rukeco Health M. NDAYISABA Libere Center Titulaire du CDS 98 Questionnairesused duringthe interviews Indicativequestionnairefor the casestudies: Bureauxprovinciauxde santC : BPS : 1. CharacteristicsoftheBureauxprovinciauxdesantC: 1.1. Number of health facilities under the responsibility ofthe BPS. How far on average are the centres de santC (CdS) from the BPS? Are there healthdistricts already functioning inthisprovince? 1.2. How many people are there working inthe BPS; brieflywhat are their main responsibilities? 1.3. Who is the person or persons incharge o fthe planning, preparation, and execution o fthe BPS budget? What i s hisher level o f education and training? Have they recently received any training on resourceibudget management? 2. Flow of funds to the BPS: 2.1. Which are the sources of funds managedbythe BPS: government budget, donors, NGO? Not needto know the exact amounts butjust an idea ofwho pays for what. 2.2. Does the BPS managea bank account? 2.3. Does the BPS managetheir entire budget? (or parts o f it, such as salaries, are already set bythe central level). 2.4. What is the responsibility ofthe BPS interms o f ensuring the distribution o f pharmaceuticals to the CdS? 99 2.5. What are the uses o fthe resources managed at the BPS (e.g. own functioning, to buy pharmaceuticals and supplies, transfers to the CdS). 2.6. Incase the BPS transfers resources to the CdS, how are these resources, both financial and in-kind(salaries, drugs, supplies, etc.), sent to the facilities? D o personnel from the BPS goes to the CdS to deliver the goods? D o personnel from the CdS collect it at the BPS? I s there any other mechanism? Does the BPS have a vehicle to do this? Dothe CdS have vehicles? 3. BudgetPlanningand Preparation 3.1. Does the BPS prepare its own budget? 3-2. H o w i s the budget for the BPS prepared? What information i s taken into account inthis preparation? I s there a pre-set mechanism to prepare it? Or do the BPS basesthe budget estimations on previous budgets? 3.3. How i s the total amount o f resources going to each CdS, ifany, determined? I s there a pre-set formula? Or is the budget determined by previous budget levels? 3.4. Does the BPS receive inputs from the CdS inthe preparation o ftheir budgets? 3.5. Does the central level send the BPS a "ceiling" for their budgets? What type o f information ifany i s sent to the BPS from the center to prepare the budgets? Incase o f questions is there somebody from the central level or regions they can contact? 100 3.6. What i s the calendar for the preparation ofthe budget: when does the BPS starts preparingthe budget? When is the first draft sent to the central level? H o w has the system changed since the introduction o f program budgets? 3.7. Are the BPS draft budgets taken into account inthe preparation o fthe national budget? Dothey receive any feedback from the center regardingtheir budget and whether or not they received the amount they have asked for? 4. Budgetexecution 4.1. Could you describe the "chaine de d6penses" du BPS et CdS? 4.2. H o w i s the budget execution monitored? 4.3. What are the problems the BPS confronts regardingbudget execution? What are the constraints they face? 4.4. Have these constraints changed since the introduction o fthe presidential decree to eliminate the user fees for children under five and mothers? 4.5. Intheir views, how can these constraints be eliminated? 101 Indicativequestionnaire for the case studies: HBpital H6pital : 1. Questionsparticularto the hospital 1.1. How many people are there working inthis hospital? How many doctors? How many nurses?How many administrative workers? How many of others? 1.2. Are there vacancies?How many and for how long have there beenvacancies? Why? 1.3. What services are provided inthis facility? 1.4. Has the facility had any stock-out o f drugs and supplies inthe last month? Which drugs and/or supplies? (e.g. anti-malarial, vaccines, syringes, etc.) .1.5. Does the facility have all neededequipment? 1.6. Does the facility have runningwater, electricity, phone or radio? 1.7. Does it have a car or motorbike? 1.8. Notes on the facility condition: 2. Flow of fundsto the hospital: 102 2.1. Which are the sources o f funds received bythe hospital: central level, donors, own resources, others? Who pays for salaries, for drugs and supplies, for equipment, for maintenance? 2.2. Are the payments for salaries on time? How does the staff receive their salaries? 2.3. Are drugs, supplies, and equipment received on time? Why or why not? 2.4. What are the constraints faced by the facility to receive payments and supplies intime? H o w do they think these problems could be solved? 2.5. Does this hospitalmanage its own budget or does it receive all payments inkind. I s this a hospital with administrative autonomy? 2.6. Who manages this budget? Is the community involved inthe management o fthe budget? 2.7. What are the resources inthis budget use for? H o w much flexibility the facility has to use this budget? 2.8. Does this hospital manage its own bank accounts? 2.9. H o w has the flow o f funds changed since the presidential decree to eliminate user charges? H o w have the sources o f funds changed? H o w has the management o f these funds changed? What are the new constraints the facility faces? 2.10. Has this facility received any support on resource management? From which agency? 103 Indicativequestionnairefor the case studies: Centrede SantC(CdS) Centrede SantC : 1. Questionsparticularto the CdS 1.1. How far is it from the BPS? 1.2. How many people are there working inthis CdS? 1.3. Are there vacancies? How many and for how longhave there been vacancies? Why? 1.4. What services are provided inthis facility? 1.5. Has the facility had any stock-out o f drugs and supplies inthe last month? Which drugs and/or supplies? (e.g. anti-malarial, vaccines, syringes, etc.) 1.6. Does the facility have all neededequipment? 1.7. Does the facility have runningwater, electricity, phone or radio? 1.8. Does it have a car or motorbike? 1.9. Has the facility received any supervisory visit from the BPS inthe last month? When was the facility last supervised? 104 1.10. Noteson the facility condition: 2. Flow of funds to the CdS: 2.1. Which are the sources of funds receivedby the CdS: central level, BPS, donors, own resources, others?Who pays for salaries, for drugs and supplies, for equipment, for maintenance? 2.2. Are the paymentsfor salaries on time? How doesthe staff receivetheir salaries? 2.3. Are drugs, supplies, and equipment receivedontime? Why or why not? 2.4. What are the constraintsfaced bythe facility to receivepaymentsand supplies intime? How do they think these problemscould be solved? 2.5. Doesthis CdS manage its own budget or does it receiveall payments inkind. 2.6. Who managesthis budget?I s the community involved inthe managementofthe budget? 2.7. What are the resources inthis budget use for? How muchflexibility the facility hasto use this budget? 2.8. Dothe CdS managebank accounts? 2.9. How hasthe flow offunds changed since the presidentialdecree to eliminate user charges?How have the sourcesof funds changed?How has the managementof these funds changed?What are the new constraintsthe facility faces? 105 2.10. Has this facility received any support on resource management? Fromwhich agency? 106 Annex 12: MBBMethodology5* The MBB approach, supported by a spreadsheet, was designed by UNICEF and the World Bank and assists in streamlining the programmatic and budgetary choices in the health sector. As atool, it makes it possible to link the inputsto the impacts inthe health sector. The approach used starts with the identification of systemic obstacles limiting the transition at the level o f health interventions. Six major dimensions o f the health systems are analyzed in order to determine the constraints: (i)the logistics and supply chain for medication and supplies; (ii)human resources; (iii)physical accessibility to the health services; (iv) the demand for services by the populations; (v) adherence to the healthcare system; and (vi) the technical and organizational quality o fthe health services. The hypothesesofthe removal ofthese obstacles, which are informed by the strategies to be implemented, enable the evaluation o f the effective level o f coverage o f the interventions that may be expected from the reinforcement o f the health system. The effective coverage o f an intervention being defined as the proportion of the target population of this intervention that effectively uses the interventionconcerned inaccordancewith the accepted quality standards. a. Identification of high-impact interventions The second article5' o f the 2003 Lancet series dedicated to child survival carried out a review o f the state of scientific knowledge on the interventions enabling the reduction o f mortality attributable to each o f the principal causes of death amongst children under five (diarrheal diseases, acute respiratory infections, measles, malaria, HIV/AIDS, asphyxia among newborn babies, preterm, neonatal tetanus and neonatal infections). The authors' analyses focused in particular on the preventative or curative interventions, which tackle the immediate causes of infant-juvenile mortality and which may be implemented through the health sector in low- or middle-income countries. In total, 23 interventions were identified and classified according to their proven effectiveness. The effectiveness data o f these interventions were derived from prior meta-analyses or systematic reviews carried out by the authors and members o f the Bellagio Study Group. The secondarticle ofthe 200560Lancet seriesdedicated to the survival o fnewbornbabies enabled the identification of nineadditional interventions to improve the survival o f babies during the first 28 days o f life, thus increasing the number o f high-impact neonatal and infant health interventions to 32. As regards maternal health and within a logic of continuity of healthcare from intra- uterine life up until adulthood, a set of interventions with proven effectiveness on the principal causes o f maternal death (post-partum hemorrhage, puerperal infections, Eclampsia, distocia, abortion complications, malaria, anemia, tetanus, non-specific causes and others) were identified from the database on the Cochrane randomized controlled trials which i s available on line, the WHO library on reproductive health and the 2005 British Medical Journal series dedicated to 58This annex i s basedon MBB technical guide (forthcoming). 59 Jones G, Steketee R, Black RE, Bhutta ZA, Morris SS, and the Bellagio Child Survival Study Group. How many child deaths can we preventthis year?Lancet 2003; 362: 65-71. 6oDarmstadt GL, Bhutta ZA, Cousens S, Adam T, Walker N, de Bernis L.Evidence-based, cost-effective interventions: how many newbornbabiescan we save?Lancet2004; 365: 977-88. 107 cost-effective interventions to achieve the MDGs. Table 21 presents these high-impact interventions and the causes o f mortality which they are able to reduce. We note from Burundi's health strategy and policy documents that several o f these interventions were retained by the country to improve the state o f health of the population. Table32 summarizes some of these interventions identified on a global level and which are in place in Burundi. These interventions were organized according to their method o f delivery and insub-packages to optimize their delivery. Table32: High-impact interventions organized by service delivery method Interventions Inplace in Burundi? 1. Family oriented community based services 1.1Family PreventivelWASHServices Insecticidetreated mosquito netsfor under five children Yes Quality of drinkingwater Yes Use of latrines Yes Washing of hands by mothers Yes 1.2 Family neonatal care Clean delivery and cord care Yes Earlybreastfeedingand temperaturemanagement Yes Universalextra community-basedcare of LBW infants No 1.3 Infant-juvenile feeding Breastfeeding for children0-5 months Yes Complementary feeding Yes 1.4 Community management illnesses Oral Rehydration Therapy Yes Community basedmalariatreatment of children with ACT No Antibioticsat community levelfor pneumonia No 2. Population oriented schedulable services 2.1 Preventive care for adolescents and adults Family planning Yes Preconceptualfolate supplementation No 2.2 Preventive pregnancy care Antenatal care Yes Detectionandtreatment of asymptomaticbacteriuria No Prevention and treatmentof sideropenicanemia during pregnancy Yes Intermittent PresemptiveTreatment (IPT) for pregnant women Yes 2.3 Prevention and treatment of HlVlAlDS PMTCT (testing and counseling,AZT + sd NVP and infant feeding counseling) Yes Condom use Yes Cotrimoxazole prophylaxisfor children of HIV + mothers Yes 2.4 Preventive infant-juvenile care Measles immunization Yes Hib immunization Yes HepatitisB immunization Yes Zinc preventive No 3. Individual oriented clinical services (constantly available) 3.1 1Clinical primary level skilled maternal (L neonatal care Skilleddelivery care Yes Antibiotics for PretermlPrelaborRuptureof Membrane (P/PROM) Yes Detection and management of (pre)Eclampsia(Mg Sulphate) Yes Management of neonatalinfectionsat PHC level Yes 3.2 Primary Health Centre care Antibiotics for U5 pneumonia Yes Vitamin A treatment for measles Yes Artemisinin-based CombinationTherapyfor children Yes ART for children with AIDS Yes ART for pregnant womenwith AIDS Yes DOTSfor TB Yes 3.3 Clinical first referral illness management 3.3.1 Clinical maternal and neonatal second referral illness management Basic emergency obstetriccare (B-EOC) Yes Clinicalmanagement of neonataljaundice Yes 3.3.2. Care at first referral level 108 Management of complicatedmalaria(2nd line drug) Yes Management of first lineART failures Yes 3.4 Clinical first referral illness management 3.4.1 Clinical maternal and neonatal second referral illness management Comprehensiveemergency obstetric care (C-EOC) Yes 3.4.2. Clinical second referral illness management Management of 2nd line ART failures Yes Managementof multi-drug resistant TB Yes b. Identijication of the systemic obstacles and definition of the limits of the coverage Identification of the systemic obstacles The methodology usedby the MBBto identify the systemic obstacles and to evaluate the coverage attainable after havingremovedthese obstacles, is based on the Tanahash?' framework. This framework for analysis uses five or six indicators which each measure a critical dimension of the health service delivery process and which reflect the complex interactions between the supply and the demand for health services. This process starts with the evaluation o f the availability of medication and basic supplies at different levels of the health pyramid and concludes with an evaluation o f the effective coverage o f a given intervention. This effective coverage is defined as the proportion of the target population of an intervention which Identification of Bottlenecks effectively uses this intervention according to the I required standards and quality with the I expectation o f a biological impact. Between Adequate Coverage contlnulty - I I these two phases, the following are also analyzed ,,,, 9I I \ I ina successivemanner: the availability ofhuman Uttiization 1st contactof muiti.contact - ', I II resources compared with the coverage standards services \ \ I defined by I the country, geographical Accessibility Physicalaccess of services - I I accessibility, also in comparison with the I standards, the initial use or first contact o f a availability critical healthsystem Inputs - \\\\ II I service which requires the user's compliance for , II 8, Targn Populatlon Y the expected impact, and lastly, the ongoing use Adaptedfrom TanahashiT. Bulletinof the WorHHealth Organhation. 1978, 56 (2) thereof (adequate coverage). Mtp:/iwhqlibdoc.who.i~ull~ti~1978Nolbulletin~1978~56(2)~295-303.pdf In a perfectly efficient system, each input is translated into an output. This assumption thus enables the assessment of the performance of the healthcare systems by measuring their capacity to translate the inputs into outputs. The limiting factor to the performance o f this system, still referred to as a "bottleneck", is the weak link in the chain o f factors determining effective coverage. 61Tanashi T. "Health service coverage and its evaluation", Bulletin of the World Health Organization, 1978. 109 Utilization of health care services which can be proxied by the first use of multi-contact service (e.g. first antenatal contact or measles immunization). Utilization indicates the members o f the catchment population actually usingthe service when it is available; Continuity (adequate coverage) in utilization o f services or adherence: this determinant indicates the extent of achievement compared to optimal contacts and services (e.g. the percentage o f children receiving three doses o f Diphteria-Pertusis-Tetanus vaccine or the percentage of women receiving three antenatal contacts). Thus, this indicator documents the continuity o f care and compliance; Quality (effective coverage) o fthe service providedor received: this indicator measuresthe quality ofcare by assessingthe skills ofthe health workers, their ability to examine the beneficiary, diagnose, provide the requisite interventions, use the equipment appropriately and advise appropriately. The quality coverage also means that potential usersare usingthe services ina correct and effective matter. These determinants are sequential and the health care delivery system bottlenecks are identified by examiningthe gaps among the five determinants and fmdingthe weakest linkinthe service delivery chain. The exercise is conducted for each mode of service delivery and for each sub-package identified in Table32 in order to identify the bottlenecks and to propose corrective measures to improve the efficiency o f the healthcare system, which in turn enables the transition of essential health interventions. The figure below illustrates the case of Burundifor the use of impregnated mosquito nets by children under the age of five, the vaccination of children and the control of acute respiratory infections amongst children at the level of health centers. The graph indicating the use of mosquito nets suggests that at community level, the bottlenecks are located at the level of availability of commodities and utilization thereof. The underlying causes of these bottlenecks, which include financial barriers, the lack of information and certain beliefs o fthe populations, are fully analyzed in the Note Sectorielle de la Sante'. For the advanced strategy activities and the clinical care provided at the level offormations sanitaires, the graphs suggest that the major bottlenecks are the human resources constraints and the poor demand for care although these are available and physically accessible. The data from the QUIBB survey suggests that only 65% of the population live within a reasonable distance from aformation sanitaire, which may constitute a constraint if the objective it to achieve universal coverage of health services. The loss rate between the first and second doses of DTC and the poor rate of comprehensive vaccine coverage, suggest significant constraints on the healthcare demand side. 110 Figure 19: Determinants of coveragefor three interventions Utilizationof LLlN by under-fivechildren Childimmunization @X0% 1 90.0% - 50.0% 50.0% 80.0% 80.0% - * 73.5% 70.0% - 61.1% 60.0% - 50.0% - 40.0% - 34.6% 30.0% - 20.0% - , 10.0%- 0.0% 7 ITNs in stffik Acces to Households Hwsehdds Utilization of vaccines in Availability Access to DPTl DPTB distribution withnet withiTN UlN by U5 stffik nurses EPI (pcp < 1 points hour) 70.0% 1 65.01 Managementof ARI 80.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% Stocks fw Nurses 5 km fmm ARVfever ARl/feva Prdessionai ED PHC treated treatedby trainedin wtside home professional iMNCl Definition of the limits of the coverage The limits of the coverage are defined as the maximum level o f effective coverage that would arise from the removal o f one or several identified constraints, taking into account the actual level of efficiency o f the healthcare production system. The objective o f this approach is to capture the various elasticities62which link the different dimensions of the healthcare system mentioned previously and to anticipate the variation of one dimension in relation to another. In a practicalway, this approach seeks to anticipate by what percentagewould, for example, the use o f impregnated mosquito nets amongst the population increase, if accessibility were improved. This approach enables the evaluation o f the levels o f coverage that one could obtain if the constraints relatingto demand and supply were removed. By assumption, the MBB uses elasticities of 1 between the different dimensions of the healthcare system. This impliesfor example, that a 25% increase inthe geographical accessibility of a service (it could be a question of increasing the portion o f the population living within a radius of 60 minuteso f aformation sanitaire from 65% to more than go%), would lead to a 25% passive increase in the initial use of this service, which would then lead to a 25% increase inthe continuous use thereof and finally to effective coverage of 25%. This chain reaction will result in achieving the effective coverage objectives expected at the close o f an investment program for reinforcement of the health system by improvingthe critical dimensions such as human resources, 62 In economics, elasticity is a measure ofthe relative variation o f one variable compared with another. 111 the availability o fworking capital for medication, the demand for healthcare, the technical quality o f the services, etc. For each of the packages presented above, two cases or scenarios were applied. Scenario IorScenario60%correspondstoaconservativereductioninthebottlenecksidentified(between 50% and 70%) in order to bringthe effective coverage o f the interventionsto at least 60%. Then, the more optimistic Scenario I1 or Scenario 80% corresponds to a reduction in the bottlenecks identified(between 70% and 90%) in order to bring the effective coverage o f the interventions selectedto at least 80%. c. Impact assessment The method for assessing the impact o f increased coverage o f the interventions utilized by the MBB is based on the model developed by the authors o f the second article of the 2003 Lancet series dedicated to child survival. The method for calculating the impact o f the interventions makes use o f the following five principal parameters: (i) the actual coverage o f the intervention concerned, (ii) the objective o f coverage or limits of coverage, (iii) effectiveness the of the intervention on the causes o f mortality o f the population concerned, (iv) the affected portion o fthe population concerned, and (v) the portion attributable to death due to this condition inthe populationconcerned. The use o f this method requires knowledge of the principal causes o f death among children under five and among pregnant women and the respective loading attributed to each o f these causes. The figures on the causes o f death amongst children under five in Burundiand their allocation were obtained from the Child Health and Epidemiology Research Group (CHERG) o f the WHO. The data on the causes of death among pregnant women in Rwanda was used as a substitute for Burundi. The followingbox gives an example of how the MBBestimates the impact on under-five mortality ate of an increased utilization of Long-Lasting Insecticidal Nets (LLIN) by children underthe age o ffive impactof lon&iastlnginsectlcldalnetsutilizationon under-flve moralltyrate (IUS,~): Formula: IUSMR = [(Coverage Increase x Effectiveness) / (I-Baseline coverage x Effectiveness)] x % USMRdue to malaria Effectiveness =Eficacy x Affectedfraction Parameters: Efficacy o f long-lasting insecticidal nets: 75% Affected fraction: 100% Baseline coverage o f long-lasting insecticidal nets: 8.39'0 Target coverage: 60% Proportion o fUSMRattributable to malaria: 8% The WHO CHERG data suggest that, inBurundi, malaria is responsible for 8% ofthe death o f children underthe age o f five. The MICS 2006 data also suggestthat 8.3% o funder-five children regularly sleep under a LLIN.For the simulations, we assumed an increase to 60% o fthe effective coverage inorder to reach the Abuja target. Since the LLINare effective in 75% o f the cases for any child under the age o f five, the affected fraction is 100%. Using the formula above, we estimate that LLIN utilization by under-five children could reduce the under-five mortality rate by 3%. The impact (mortality reduction) o f several interventions implemented simultaneously is estimated in a residual way in order to avoid double counting. 112 d Cost estimatiod3 The costing module o f the MBB tool normally yields additional resources required for removing a set o f health system bottlenecks that are considered to hinder health service delivery to the population. It is based on the premise, that while a basic package of effective interventions can improve health, the cost estimate should reflect the cost o f eliminating the constraints or bottlenecks that hinder its expansion (Figure20). Figure20: MBB costing methodology Marginalcost of overcoming Humanresource Logistics Barriersto Bottleneckswith Gaps in accessto bottlenecksfor bottlenecks& utilization& technical & packages implementationof marginalcosts of stimulating organizational packages supplies demandOf packages quality of packages' The MBB model was slightly modified to produce the total cost o f scaling up health interventions and not just the additional cost. Thus, as previously mentioned the cost estimated reflect not only the investment needs but also the total recurrent cost necessary to sustain the healthproduction function. Overall, the methodology used by the MBB to estimate the additional cost can be summarized by the following generic formula. The additional cost of an input i (MC,) i s calculated as follows: Inthe aboveformula, MC, isthe additionalcost ofinputi;Poiisthe unitprice ofinput iinthescenario;Qoiisthequantityofinputiperunitofoutput,orperserviceproductionunitin the scenario (in other words, the amount of the input needed to produce one unit of a given service/output); Soi i s the SPU per 1million population o f input ifor the objective coverage (in other words the amount o f service/output i for producing the coverage/outcome for 1 million people) ; Pbi is the baseline unit cost of input i; Qbi is the quantity o f input i per service production units for baseline coverage; S,, is the SPUper 1million population o f input ifor the baseline coverage; and n is the population. The additional cost estimates are then aggregated over all inputs i and subsequently aggregated over all determinants of coverage, and service delivery modes, allowing an estimate o f the total cost o f overcoming bottlenecks in the health system. The aggregated amounts show the cost of removing each bottleneck, as well as the cost of scaling up interventions through the three service delivery modes. The productiono fhealth services involves a variety o f inputs including human resources, infrastructures, drugs and equipment, vehicles, etc. The types, combinations and numbers o f the inputs vary across types of health services, service delivery modes, and regions with their 63MBBTechnicalNotes, forthcoming. 113 different population densities. To help simplify the unit costing, the MBB uses the term Service Production Unit (SPU) to identify the cost center related to the scaling up o f health services or overcoming bottlenecks. The SPU i s to be estimated on the basis of the capacity (population covered) of each health facility or outreach team. Therefore, an SPU can be a functional health center, physicians, nurses or midwives, community workers, drug and vaccine kits, a mobile team, or a hospital on the supply side; it could also be household behavior or utilization of health services on the demahd side. The above formula was usedto estimate the cost o f all additional investment required to upgrade the health system. The secondpart ofthe formula was removed to estimate the total recurrent cost inorder to reflectboththe current recurrent cost andthose impliedbythe additional. 114