Documentof The World Bank ReportNo.: 48569 PROJECT PERFORMANCE ASSESSMENTREPORT ERITREA HEALTHPROJECT (CREDIT NO. 30230) HIV/AIDS,MALARIA, STD AND TUBERCULOSIS CONTROL PROJECT (CREDIT NO. 34440) June 2,2009 Sector Evaluation Division IndependentEvaluation Group, WorldBank Currency Equivalents (annual averages) Currency Unit =Ethiopian Birr (ETB) and Eritrean Nakfa (ERN) Health Project (as of November 17, 1997, date of PAD) (as of June 16, 2005 date of ICR) $1 = ETB 7.10 $1 =ERN 13.5 ETB 1=$0.141 ERN 1= $0.074 HIV/AIDS,Malaria, STDs and tuberculosis(HAMseT)ControlProject (as of November 27, 2000 date of PAD) ' (as of Octobe;26, 2006, date of ICR) $1 =ERN 9.77 $1 =ERN 15.00 ERN 1= $0.102 ERN 1= $0.067 Abbreviations and Acronyms AIDS Acquired Immuno-deficiencySyndrome MOH Ministry of Health ANC Antenatal clinic MOLHW Ministry ofLaborandHumanWelfare ART Antiretroviral therapy MOTC Ministry of Transport and Communication BCC Behaviorchange communications NATCoD National HIV/AIDS/STI and TuberculosisControl Division CAS CountryAssistanceStrategy NCEW National Confederationof EritreanWorkers CMHRP Community-managedHAMSeT ResponseProgram NMCP National Malaria Control Program DDT Dichloro-Diphenyl-Trichloroethane NRS NorthernRed Sea DHS Demographic and HealthSurvey NTCP National Tuberculosis Control Program DOTS Directly ObservedTherapy-shortcourse NUEW National Union of EritreanWomen ELISA Enzyme-LinkedImmuno-sorbentAssay NUEYS National Union of EritreanYouth and Students ESMG EritreaSocial MarketingGroup PAD Project Appraisal Document FBO Faith-basedorganization PER Public Expenditure Review FHI Family HealthInternational PHARPE Public Healthand RehabilitationProgram in Eritrea FY Fiscalyear PHC Primary health care Global Fund Global Fundto fight AIDS, malariaandtuberculosis PLWHA People living with HIV/AIDS HAMSeT HIV/AIDS, malaria, STIs and tuberculosis PMU Project ManagementUnit HBC Home-basedcare POP Persistentorganic pollutants HIV Human Immuno-deficiencyvirus QER Quality EnhancementReview HMIS HealthManagementInformation System RGBIS HIV/AIDS Risk Groups and Risk Behavior Identification Survey ICR ImplementationCompletionReport RRI RapidResults Initiative IDA InternationalDevelopmentAssociation SRS SouthernRed Sea IEG IndependentEvaluationGroup STI Sexually transmitted infection IRS Indoor residualspraying TB Tuberculosis ITN Insecticidetreated bed-net UNAIDS Joint UnitedNationsProgramon HN/AIDS LQAS Lot Quality Assurance Sampling UNICEF United Nations Children'sEmergencyFund M&E Monitoring and evaluation USAID United States Agency for InternationalDevelopment MOD Ministry of Defense VCT Voluntary counseling and testing MOE Ministry of Education WHO World HealthOrganization Fiscal Year Government: January 1-December 31 World Bank July 1-June 30 Director-General, Evaluation : Mr.Vinod Thomas Director, Independent Evaluation Group (World Bank) : Ms.Cheryl Gray Manager, Sector EvaluationDivision : Ms.Monika Huppi Task Manager : Ms.Gayle H.Martin 1 IEGWBMission: Enhancing development effectiveness through excellence and independence inevaluation. About this Report The Independent Evaluation Group assesses the programs and activities of the World Bank for two purposes: first, to ensure the integrity of the Bank's self-evaluation process and to verify that the Banks work is producing the expected results, and second, to help develop improved directions, policies, and procedures through the dissemination of lessons drawn from experience. As part of this work, IEGWB annually assesses about 25 percent of the Bank's lending operations through field work. In selecting operations for assessment, preference is given to those that are innovative, large, or complex: those that are relevantto upcoming studies or country evaluations; those for which Executive Directors or Bank management have requested assessments; and those that are likely to generate important lessons. To prepare a Project Performance Assessment Report (PPAR), IEGWB staff examine project files and other documents, interview operational staff, visit the borrowing country to discuss the operation with the government, and other in-country stakeholders, and interview Bank staff and other donor agency staff both at headquarters and in local offices as appropriate. Each PPAR is subject to internal IEGWB peer review, panel review, and management approval. Once cleared internally, the PPAR is commented on by the responsible Bank department. IEGWB incorporates the comments as relevant. The completed PPAR is then sent to the borrower for review; the borrowers' comments are attached to the document that is sent to the Bank's Board of Executive Directors. After an assessment report has been sent to the Board, it is disclosed to the public. About the IEGWB Rating System IEGWB's use of multiple evaluation methods offers both rigor and a necessary level of flexibility to adapt to lending instrument, project design, or sectoral approach. IEGWB evaluators all apply the same basic method to arrive at their project ratings. Following is the definition and rating scale used for each evaluation criterion (additional information is available on the IEGWB website: http://worldbank.org/ieg). Outcome: The extent to which the operation's major relevant objectives were achieved, or are expected to be achieved, efficiently. The rating has three dimensions: relevance, efficacy, and efficiency. Relevance includes relevance of objectives and relevance of design. Relevance of objectives is the extent to which the project's objectives are consistent with the country's current development priorities and with current Bank country and sectoral assistance strategies and corporate goals (expressed in Poverty Reduction Strategy Papers, Country Assistance Strategies, Sector Strategy Papers, Operational Policies). Relevance of design is the extent to which the project's design is consistent with the stated objectives. Efficacy is the extent to which the project's objectives were achieved, or are expected to be achieved, taking into account their relative importance. Efficiency is the extent to which the project achieved, or is expected to achieve, a return higher than the opportunity cost of capital and benefits at least cost compared to alternatives. The efficiency dimension generally is not applied to adjustment operations. Possible ratings for Outcome: Highly Satisfactory, Satisfactory, Moderately Satisfactory, Moderately Unsatisfactory, Unsatisfactory, Highly Unsatisfactory. Risk to Development Outcome: The risk, at the time of evaluation, that development outcomes (or expected outcomes) will not be maintained (or realized).Possible ratings for Risk to Development Outcome: High Significant, Moderate, Negligible to Low, Not Evaluable. Bank Performance: The extent to which services provided by the Bank ensured quality at entry of the operation and supported effective implementation through appropriate supervision (including ensuring adequate transition arrangements for regular operation of supported activities after loanlcredit closing, toward the achievement of development outcomes. The rating has two dimensions: quality at entry and quality of supervision. Possible ratings for Bank Performance: Highly Satisfactory, Satisfactory, Moderately Satisfactory, Moderately Unsatisfactory, Unsatisfactory, Highly Unsatisfactory. Borrower Performance: The extent to which the borrower (includingthe government and implementing agency or agencies) ensured quality of preparation and implementation, and complied with covenants and agreements, toward the achievement of development outcomes. The rating has two dimensions: government performance and implementing agency(ies) performance. Possible ratings for Borrower Performance: Highly Satisfactory, Satisfactory, Moderately Satisfactory, Moderately Unsatisfactory, Unsatisfactory, Highly Unsatisfactory. ... 111 Contents PRINCIPAL RATINGS ............................................................................................................................... vi1 KEY STAFF RESPONSIBLE .................................................................................................................... Vlll PREFACE ................................................................................................................................................... IX SUMMARY .................................................................................................................................................. XI 1. INTRODUCTION .............................................................................................................................. 1 Backgroundand Context ............................................................................................................... 1 Government Health Strategy ......................................................................................................... 4 World Bank and other External Support to HNP: 1994-present ............................................... 4 2. HEALTH PROJECT ......................................................................................................................... 6 Objectives and Design ................................................................................................................... 6 Implementation ............................................................................................................................... 9 Achievement of Objectives ......................................................................................................... 11 Objective: Improvement in the delivery of health care services .............................................................. 11 Objective: Improvement in health status................................................................................................. 14 Project Ratings ............................................................................................................................. 16 3. HAMSET CONTROL PROJECT ................................................................................................... 19 Objectives and Design ................................................................................................................. 19 Implementation ............................................................................................................................. 23 Achievement of Objectives ......................................................................................................... 28 Objective: reduce the mortality and morbidity due to HIV/AIDS and STls .............................................. 28 34 Objective: reduce the mortality and morbidity due to malaria ................................................................. Objective: reduce the mortality and morbidity due to tuberculosis .......................................................... 37 Project Ratings ............................................................................................................................. 40 t 4. CONCLUSION ............................................................................................................................... 44 Lessons ......................................................................................................................................... 44 Value-added .................................................................................................................................. 45 Remaining Challenges ................................................................................................................. 46 IIJanuar$Februa& This reportwas preparedby Gayle H.Martinwith the assistance of MarthaAinsworth. who assessedthe project in 1 2008 Marie-jeanne Ndiayeprovidedadministrativesupport and assistance with data entry . . iv REFERENCES ................................................................................................... .........................................49 ANNEX A. HEALTH PROJECT DATA...... ......................................................... ........................................53 ANNEX B. HAMSET CONTROL PROJECT DATA .............................................................. ..................... 57 ANNEX C. PERSONS INTERVIEWED ..................................................................................... ..................65 ANNEX D. TIMELINE OF WORLD BANK SUPPORT TO ERITREA'S HEALTH SECTOR ..................... 69 ANNEX E. ANALYSIS OF APPROVED CMHRP SUB-PROJECTS ............................................. ............73 ANNEX F. SECTORAL DATA AND COMMUNICABLE DISEASE OUTCOMES ..................................... 77 Tables Table 1-1.World Bank involvement in the health sector Table 2-1. Planned versus actual costs, by component,,,..,,.,,,,,.,,,..,111111111..11.(1..,.........,,.......,.....................................,.., Table 2-2. Utilization of clinic services, 1995-2002 Table 2-3. Malaria morbidity and mortality rate, ........................................................................ 15 Table 2-4. HIV prevalence in screened blood, 2003-2005 ............................................................................................. Table 2-5. Health Project: Summary IEG Ratin Table 3-1. Planned versus actual costs, by component Table 3-2. Distribution of approved subproject Table 3-3. Distribution of approved sub-projects for HlVlAlDS and (joint) HAMSeT diseases, by target group Table 3-4. HIV prevalence survey results, 19 Table 3-5, Tuberculosis morbidity, 1999-2007. .............,.*.. Table 3-6. Bed-net ownership and utilization, 2 Table 3-7. HAMSeT Control Project: Summ Table 3-8. Average annual project supervision costs, nominal US$. Figure 1-1. IMR and child mortality rate relative to Sub-Saharan Africa, 1970-2006 1.1.111.1.11.1111...1....11.11.111111..1,,...,.,...... Figure 1-2. Ranking of malaria morbidity rate in 1998 Figure 1-3.External assistance to the HNP Sector, 1996-present Figure 2-1. In- and out-patient caseload in Gash Bark Figure 2-2. Trends in immunization-preventable diseases, 1998-2006 Figure 2-3. Malaria interventions and malaria inciden Figure 3-1. Condom distribution, 2000-2 Figure 3-2. Public sector STI treatment, ............................................................................. 32 Figure 3-3. HIV prevalence among antenatal clinic attendees, 2003-2007 ................................................................ Figure 3-4. Tuberculosis case detection rate and cure rate relative to global "Stop TB" targets for TB control, 1999- Figure 3-5. Malaria morbidity rate by zoba, 1998-2006 .................................................................................................................. 39 V Boxes Box 1-1. Malaria in Eritrea,,.,,,,, Box 2-1. Health Project: components and intended activities........... ....................................... Box 3-1. HAMSeT Control Project: components and intended activities .................... Box 3-2. Efficacy of voluntary counseling and testing..................................... Box 3-3. The epidemiology of Tuberculosis in Eritrea-some lingering questions ... ................... vii PRINCIPAL RATINGS ICR * ICR Review* PPAR Health Project (Credit 30230) Outcome Unsatisfactory Moderately Moderately unsatisfactory unsatisfactory Institutional Development Impact** Substantial Substantial Riskto Development Outcome Moderate Sustainability*** Likely Likely Bank Performance Satisfactory Satisfactory Moderately satisfactory Borrower Performance Satisfactory Satisfactory Moderately satisfactory HIVIAIDS, Malaria, STDs and TuberculosisControlProject (Credit 34440) Outcome Satisfactory Moderately Moderately satisfactory satisfactory Institutional Development Impact** Substantial Modest Riskto Development Outcome Moderate Sustainability* * Likely Likely Bank Performance Satisfactory Satisfactory Satisfactory Borrower Performance Highlysatisfactory Satisfactory Satisfactory * The ImplementationCompletionReport (ICR) is a self-evaluationby the responsibleBank department.The ICR Review is an intermediateIEGWB productthat seeks to independentlyverify the findings of the ICR. **As o f July 1, 2006, InstitutionalDevelopmentImpactis assessedas part of the Outcome rating. ***As of July 1, 2006, Sustainabilityhas beenreplacedby Riskto DevelopmentOutcome. As the scales are different, the ratings are not directly comparable. ... V l l l KEY STAFF RESPONSIBLE Project Task Division Chief/ CountryDirector ManagerLeader Sector Manager Health Project (Credit 30230) Appraisal DavidDunlop Arvil Van Adams Oey Astra Mesook Sundararajan Srinivasa Gopalan Supervision DavidBerk MontserratMeiro- Lorenzo EvaJarawan ChristopherWalker Completion ChristopherWalker DzingaiMutumbuka ColinBruce HIKAIDS, Malaria, STDs and TuberculosisControl Project (Credit 34440) Appraisal EvaJarawan Arvil Van Adams Oey Astra Mesook Supervision EvaJarawan ChristopherWalker SonNamNguyen Completion SonNamNguyen DzingaiMutumbuka Colin Bruce ix Preface This is the Project Performance Assessment Report (PPAR) for the Health Project andthe HIV/AIDS,Malaria, STDs and Tuberculosis (HAMSeT) Control Project. The HealthProject (FY1998-2005), the first World Bank supported project inthe country's health sector, was financed through IDA Credit No. 3023 inthe amount o f $18.3 million (SDR 13.4 million), a grant fromNorway of $2.7 million (NOK25.0 million) and planned government contribution o f $3.5 million. The credit was approved on December 16, 1997, became effective on May 28, 1998, and was 96 percent disbursed when it closed on December 31,2004, 18 months after the original closing date. The HAMSeT Control Project (2001-2006) was financed through IDA Credit No. 3444 inthe amount o f $40.0 million (SDR 31.4 million), with planned government contributions of $10.0 million. The credit was approved on December 18,2000, became effective on March 1,2001, and closed as plannedinMarch 2006. A follow-on project, the HAMSeT I1Project financed by an IDA credit o f $24.0 million, i s currently being implemented. This PPAR was prepared by an IEGteam consisting o f Gayle H.Martin(Senior Evaluator and Task Manager) and assisted by Martha Ainsworth (HumanDevelopment Cluster Coordinator), who visited Eritrea inJanuary/February 2008. The mission met with representatives from the MinistryofHealth(MOH) and five other ministries, donors, non-governmental entities, faith- based organizations as well as people with HIV/AIDS. The mission visited four of Eritrea's six regions (zobas) and visited community-based HAMSeT sub-projects and health facilities (including the two hospitals built under the Health Project). Other sources o f evidence consulted include: (a) interviews with relevant World Bank staff, (b) World Bank project files, (c) project- related reports, (d) economic and epidemiological data, studies, surveys and research, and (e) primary analysis o f the community-managed sub-project data for the HAMSeT Control Project (see Annex E). This PPAR will contribute to a forthcoming evaluationby IEGo fthe World Bank's support to health, nutrition and population outcomes. As such, more material has beenpresented inthis "enhanced" PPAR than is the IEGstandard. The IEGteam gratefully acknowledges all those who made time for interviews and provided ' documents and information. Following standard IEGreview procedures, copies o f the draft PPAR were sent to the relevant government officials and agencies for their review and feedback. However, no comments were received. xi Summary This Project Performance Assessment Report assesses the development effectiveness oftwo projects-the Health Project and the HIV/AIDS, Malaria, STDs and Tuberculosis (HAMSeT) Control Project-in the context ofthe World Bank's overall support to Eritrea's healthsector duringthe period 1997-2006, as well as other Bank-financed products such as economic and sector work. When Eritreajoined the Bank in 1995 the country was emerging from three decades o f war. Eritrea is one o f the poorest countries inthe world with an estimated per capita GDP o f $200. Throughout project implementationEritrea was either under conflict or simultaneously a post- conflict and fragile state. Following strong growth performance inthe early 1 9 9 0 the~ ~ macroeconomic situation progressively worsened after border hostilities resumedin 1998. O f the country's 3.24.9 millionpeople, two thirds live inpoverty. Healthexpenditureis low (between $8 and $13 per capita) and highly donor dependent with external assistance accounting for more than two thirds o f total public sector health spending. Furthermore, the health sector faces serious human resources constraints. Over the past decade malaria, HIV/AIDS and tuberculosis have ranked among the top causes o f mortality and morbidity. The objective o f the Health Project (FY1998-FY2005) was to improve the health status o f Eritreans, and it mainly financed: (i) the construction o ftwo hospitals, (ii) refurbishmento f the 30 clinics, drugs and medical supplies, (iii) theNational Malaria Control Program, and (iv) the establishment o f the National Blood Transfusion Service. The HAMSeT Control Project (FY200 1-FY2006) objective was to reduce mortality and morbidity from the HAMSeT diseases. The project mainly supported the HAMSeT disease control programs inthe Ministry o fHealth, disease control efforts in selected non-health sectors, and community-based disease control efforts. A follow-on project, the HAMSeT-I1Project, i s currently beingimplemented. The Health Project had mixedresults. Project implementation coincided with significant improvements inthe health indicators (e.g., infant and child mortality). Many o f these improvements could be linked to project-financed outputs such as clinic refurbishmentand the provision o f drugs and medical supplies. There were, however, other indicators that the project sought to influence but was less successful (e.g., malnutrition). The hospital investment expanded the sector's physical infrastructure-through the construction o f the Barentu and Mendefera Hospitals-but the returns to the investment havenot yet beenfully realized because the expanded capacity i s under-utilized. In2007, a year after their opening, the two hospitals had among the lowest bed occupancy ratios inthe country. Both projects financed malaria control activities and substantially contributed to the reduction in malaria morbidity inexcess o fthe target o f "80 percent reduction" set by the Minister o f Health in 1999. The interventions-prompt diagnosis and treatment o f malaria cases, insecticidetreated bed-net distribution and re-impregnation, indoor residual spraying, larvaciding and source reduction-were significantly associated with lower malaria incidence, even after controlling for fluctuations inrainfall. Eritrea was the first country inAfrica to achieve the Abuja targets for Roll Back Malaria. The MOH's evidence-based planning and performance monitoring i s best practice inmalaria control. The Bank was the largest single source o f fundingto the malaria xii control program, although other important sources o f funding for malaria were: USAID and the Italian Cooperation, and inrecent years, the Global Fundto fight AIDS, Malaria and Tuberculosis. Inthe area oftuberculosis control, the HAMSeTControl Projectfinancedtuberculosis drugs, information dissemination, capacity building for the National Tuberculosis Control Program, health worker training, procurement o f diagnostic microscopes and related medical supplies in support o f the expansion o f the directly observed treatment-short course (DOTS) program. Information dissemination had limited success, and in2005 a tenth o f women did not know any tuberculosis symptoms, and knowledge o f the most distinguishingfeatures o f tuberculosis was extremely low. Case detectionrates continued to leave room for improvement, but there was greater success at expansion o f the DOTS program and treatment o f identified tuberculosis cases. There has been a downward trend intuberculosis morbidity inrecent years largely thanks to the zoba-level services. At the national level, there were staffing problems and lagging performance o f the tuberculosis control program. It i s a missedopportunity that there has not been greater learning, sharing and adoption o f practices betweenthe malaria and tuberculosis control programs inthe areas o fplanning, monitoring and evaluation. The HAMSeT Control Project financed a range o f HIV prevention activities including: extensive awareness-raising among the general populationand among key risk groups, HIV/AIDSand life- skills education inthe schools, blood safety and encouraging voluntary counseling and testing among the general population as well as highrisk groups. Implementationwas through seven ministriesincluding the defense force, non-governmental entities and community structures through the Community-managed HAMSeT Response component. The extent o f coverage o f risk groups (other than the military) was hard to ascertain, and information on behavior change was limitedand particularly challenging to interpret inthe context o f a general population that already had relatively low HIV risk. The almost universal coverage o f HIV prevention inthe military was an important success given that nearly every young adult spends some time inthe military (because o f the national conscriptionpolicy) and the fact that the demobilized military was the most important source o f HIV risk at appraisal. Interventions targeting sex workers were implementedinall major urbanareas. Despite limitedbehavioral data, the declining trendin HIV prevalence among 15-24year olds as well as declining STI prevalence suggests important successes inHIV prevention. The project was also successful at extending care and support for people with HIV/AIDS,support to orphans and vulnerable children, and stigma reduction. IEGrates the outcome of the HealthProject as moderately unsatisfactory. The objective to improve the health status o f Eritreans was consistent with the post-conflict situation, and remains relevant giventhe country's prioritization o f human development investments. The contribution o fthe investmentsto the project objective i s constrained by under-utilization o fthe project hospitals which accounted for about two thirds o f the total project costs. It is, however, possible to make strong links between some of the other project investmentsand health outcomes, notably inthe areao fmalariacontrol andimprovements inblood safety. Bank and borrower performance are rated moderately satisfactory. There were weaknesses inquality-at-entry on the Bank's side, and on the borrower side, the numeroushospital designchanges well into hospital construction and some initial weaknesses inoverseeing construction caused substantial delays. There were also other factors-economic and security-related-that were beyond the Bank and borrower's control. ... Xlll The overall outcome ofthe HAMSeT Control Project is ratedmoderately satisfactory based on the following disease-specific ratings. Malariacontrol is ratedhighly satisfactory becausethe sustained and targeted malaria control efforts resulted incontinued declines inmalaria morbidity even inthe face o f increases inaverage rainfall, as in2004. HIV/AIDS/STIs control is rated moderately satisfactory because o f near universal coverage o f the militarywith HIV prevention interventions, the country's most important HIV risk group, targeting o f other risk groups such as sex workers, expandedvoluntary counseling and testing, coupled with evidence o f declining HIV trends inyoung adults and declining STItrends. There was some inefficiency inthe implementation o f the community-managed program, due to some weaknesses incoordination with sectoral interventions. This component did however substantially contribute to reducing stigma and extendingcare and support to orphans and people with HIV/AIDS. Tuberculosis control i s rated moderately unsatisfactory largely due to continued low case detectionrates and because the persistent programmatic weaknesses raised doubts about attribution o f the downward trends intuberculosis morbidity to the project. The sectoral participation inthe HAMSeT Control Project was prioritized based on sectors' comparative and strategic advantage in HAMSeTdisease control, andbenefited from strong leadership by MOH. The designofthe community-managed program, however, lacked detail and caused implementation to suffer early intheproject, althoughafter the mid-ternreviewmanyweaknesses were addressed. Against the backdrop o fpersistent security challenges, worsening economic conditions and continued human resource constraints, Bank and borrower performance are rated satisfactory. While several factors bode well for the sustainability o fthe two projects' achievements (e.g., the highlevel of government commitment andownership, increasingutilizationofhealth facilities and the highly successful malaria program and strong reliance on community involvement) the persistent "no war no peace" situation continues to pose economic and security risks. Inthe face o fthe food and fuel crisis, the government's difficult fiscal position will likely constrainthe sector's future resources. The risk to development outcome for bothprojects i s therefore rated moderate. The most important lessons are: In post-conflictsettings engagement is a means to an end, and needs to be accompaniedby a sustainedpolicy dialogueto ensure that developmentgains are realizedinthe mediumterm. Inthe late 1990sthe Bank's dialogue with Eritrea's health sector was highly contentious, but despite the initial technical disagreements the Bank engaged the sector. While the Health Project was not fully successful, it i s unlikely that the development gains inthe decade that followed under the HAMSeTprojects would have been realized inthe absence o f the early engagement inthe sector. The content o f the sectoral dialogue that followed the initial engagement was able to steer the sector from the post-conflict "reconstruction mode" to a "development mode." The reorientation o f the sectoral priorities was only possible with sustained and sometimes challenging dialogue underpinned by analytical work. Inthe area of HIV/AIDS a demand-drivenmodelof communitysub-projectsis moreappropriatefor service deliveryinterventionsthat are responsiveto local communityneeds (such as home-basedcare, supportto orphans andvulnerable children)than for preventiveinterventionstargetingstigmatizedrisk groups or stigmatizedbehavior. The community component was based on a demand-driven xiv model that i s a part o f many HIV/AIDS projects. Inthe HAMSeT Control Project the community component financed a combination o f HIV/AIDS prevention, care and support activities. The project experience demonstrated that community demand(on which the demand-driven approach i s based) i s sub-optimal as a basis for allocation for programs aimed at stigmatized risk groups or behavior. Community activities that are strategically planned and coordinated with the local health authorities can provide an important complement to health facility-based disease control efforts. Community outreach and community-based activities proved to be key elements o fthe highly successful malaria control efforts. Incontrast, the strong reliance on facility-based interventions was a key factor underpinningthe low case detectionrate inthe less successful tuberculosis control program. Disease control projects can be complementary to-and need not undermine- cross-cutting health system functions. Inthe HAMSeT Control Project various cross- cutting functions and systems were strengthened(e.g., health promotion, disease surveillance, laboratory service, drug distribution) because the sector's leadership sought to achieve programmatic efficiencies across individual disease control programs instead o f duplicating these systems for each disease control program. This experience i s particularly important giventhe debates inthe internationalhealth community about the negative impacts o f disease-specific projects on health systems. Multi-sector projects, such as HIV/AIDS projects, achieve better results if the sectors involved are strategically chosen according to their comparative advantage indisease preventionand control. The HAMSeT Control Project prioritized the participation o f sectors based on the comparative advantage o f each sector inHAMSeT disease control, allowing the country's disease control efforts to balance comprehensiveness with selectivity inorder to achieve maximal disease impact. Vinod Thomas Director-General Evaluation 1 1.Introduction Backgroundand Context 1.1 Eritrea i s one o fthe poorest countries inthe world with an estimated per capita gross domestic product (GDP) o f $200 in2006.' Two-thirds o f the population live inpoverty2and in 2007 the country ranked 157thamong 177 countries inthe HumanDevelopment IndexV3After three decades o f war and defacto independence in 1991, the country inheriteda shattered economy, devastated infrastructure, and neglected social sectors. A third o f the population was displaced. By the late 1990s the country appeared on the way to economic recovery--evidenced by 10.9 percent average annual (nominal) growth rate4-and extensive reconstructionand rehabilitation. However, the 1998-2002 border hostilities with Ethiopia, and the consequent and on-going `no war no peace' situation have marredeconomic performance. Ayerage annual growth slowed to 3.6 percent for 2003-2005 and in2007 the economy contracted. Throughout project implementation Eritrea was either under conflict or was simultaneously a post-conflict and fragile state. The country currently faces unsustainable fiscal deficits and precariously low foreign exchange reserves. The timeline inAnnex D summarizes some o fthe key events inthe country and the health sector's history. 1.2 Trends in HNP indicators. Since the 1990s the improvements ininfantand child mortality have exceeded the pace o f improvement inother countries inSub-Saharan Africa (Figure l-l).5 these gains, some indicators continue to lag (e.g., maternal mortality Despite ratio)6 and the improvement inhealth status has been uneven, especially among the poor (Figure F- 1 inAnnex F). Infectious andparasitic diseases such as malaria, tuberculosis andHIV/AIDS accounted for the single largest source o f death inEritrea (44 percent). Other important causes o f mortality were: respiratory infections (16 percent) and non-communicable diseases (22 percent) (Table F-2 inAnnex F). Eritrea has the full range o f malaria endemnicity-from Gash- Barka and Debubzobas with hyper-endemic transmissionto the plateau (including Asmara) that has much lower levels o f transmission, to the South Red Sea zoba that has very few cases. The country experienced an'exceptionalincrease inmalaria morbidity and mortality in 1997-1 998 due to unusually highrainfalls (Figure 1-2 inBox 1-1). Tuberculosis, an air-borne disease, is 1. Gross national income (GNI) per capitahas fallen from $220 to $170 between 1998 and 2005; in international purchasingpower parity (PPP) terms, GNI dropped from $1,220 to $1,O 10 over the same period (World Bank 2008a). 2. World Bank (1996a). The country has never had apopulation census since independence. The Ministryof Development Planning maintains that the population size is about 3.2 millionwhile accordingto other sources, includingthe international community, it as high as 4.9 million-a difference ofmore than athird. This uncertainty influences the sampling of demographic, economic as well as health surveys, the estimation ofpopulation-based indicators and the interpretation of any change inindicators over time. 3. UNDP 2007. 4. World Bank 2008b (pii). 5. Infant mortalityrate (IMR) decreased from 72 to 48 deaths per 1,000 live births, and the under-five mortality rate, dropped from 136to 93 deaths per 1,000 live births between 1995 and 2002 (Macro International 1995,2002). 6. In2000 the maternalmortality ratio was estimatedat 630 deaths per 100,000 live births (WHO 2006). 2 found throughout the country, with the highest morbidity and mortality rates in Southern Red Sea (SRS), Maekel and Gash Barka. The first AIDS case was reported in 1988 inthe southern port, Assab. The AIDS epidemic can be characterized as a low prevalence epidemic that i s concentrated inspecific risk groups (sex workers, active and demobilized military conscripts, truckers, as well as the sexual partners o f these groups). National sentinel surveillance in2001 revealedan HIV prevalence rate o f 2.8 percent among antenatal clinic attendees, butwithhigher infection rates among specific groups-female bar workers (22.8 percent) and military personnel (4.6 percent) (Table 3-4). Heterosexual contact is the main form o f HIV transmission. InEritrea the overwhelming ma ority o f menare circumcised, an important protective factor insexual transmission o f HIV. ?i InfantMortality Rate Under 5 Mortality Rate I 1970 1975 1980 1985 1990 199520002005 1970 1975 1980 1985 1990 1995 2000 2005 I Source: World Bank 2006a. 1.3 Health Expenditure. Health expenditure is low and highly donor dependent. Per capita total public (government and donor) health expenditure in2005 was between $8 and $13 and external assistance accounted for more than two thirds o f total public sector health spending (Table F-3 inAnnex F).' The country's difficult economic and fiscal position will likely constrain future increases ingovernment health spending. Cost recovery through user fees has been an important source o f funding to complement the relatively modest government health spendingo f about $3 per capita (between 1996 and 2006; see Table F-3 inAnnex F).9 Consequently household spendingon health care i s high; in2000 poor households spent more thanatenth ofhouseholdconsumptionexpenditure onhealthcare. 7. The rate of malecircumcision inEritrea is estimatedat 95 percent, accordingto Williams et al. 2006. Three randomizedcontrolledtrials have confirmedthat malecircumcisionis associatedwith a reductioninfemale-to-male transmissionof HIV by 50-60percent(Auvert et al. 2005, Bailey et al. 2007, Gray et al. 2007). 8. Author's calculationbased on governmentexpenditurereportedinTable F-3 (WorldBank 2008b) andpopulation estimates usedinthe HMIS. The 1993 WorldDevelopmentReport estimatedthat a basic package of servicescost $12 (WorldBank 1993). In2001the MacroeconomicCommissionon Healthestimatedthat $34 is neededin low incomecountries(LICs) to implementa basicpackage ofessentialhealthservices(WHO 2001). The main differencebetweenthe two estimatesis that the Commissionexplicitlytook into accountsome healthsystem investments, andusedmore detailedcost analyses. 9. In2002, user fees accountedfor morethan a tenth (US$1.2million) of governmentrecurrenthealthexpenditure ($10.7 million) (WorldBank 2008b). 3 1.4 Health service delivery i s largely through government-ownedhealth facilities through a system o f national referral hospitals, zoba referral hospitals, hospitals, health centers and health stations. Currently 62 percent o f all health facilities are government-owned, andjust under a fifthis ownedby faith-based organizations (FBOs) and afifthbyprivate industry. Inrecent years the private health sector has expanded-mainly inAsmara andthe largetowns.lo Box 1-1.Malaria inEritrea The distribution ofmalaria risk acrossthe country is strongly influenced by climatic, altitude and rainfall differences (see map).a Malaria transmission is highly seasonal and mainly inthe south and western parts o fthe country during September-November, while the eastern coastal zones have highest transmission between January and April. There are, however, some areas where malaria risk persists throughout the year, mainly linkedto the existence o f man- made water sources. In2006 Gash Barka accounted for three quarters o fthe country's malaria morbidity and more than half of the country's malaria mortality. While Debubhad the second highest number o f malaria cases, it has the second lowest malaria mortality (most likely due to better access to health services).b According to a prevalence survey in2000/01, houses with mudwalls were associated with higher malaria risk (after adjusting for rainfall and altitude). Chloroquine resistance is widespread inEritrea. Duringthe epidemic outbreak between 1997 and 1998 the number of malaria cases increased from 171,200 to 254,100. InAnseba the caseload increased by 94 percent and inNorthernRed Sea (NRS) by 105 percent. In Gash Barka and Debubthe reported cases increased by a third and a half, respectively. Figure 1-2. Ranking of malariamorbidityrate in 1998 Lowest ranking Source: Adapted from MOH 2007a Source: Graves 2004; Nyarango et al. 2006; MOH2004a. a. The central highlands run north-south (altitude 1,500-2,000 meters) and descend on the east to a coastal desert plain (1- 1,000m). on the northwest to undulating terrain and on the southwest to flat-to-rolling plains and lowlands (600-1,500m). b. Another possibleexplanation is immunity to malaria in high-transmission areas which the other areas do not have. However, this does not explain the difference inmortality between Gash Barkaand Debub zobas. 10. MOH2006a (p22). 4 Government Health Strategy 1.5 Eritrea has few natural resources and shortly after independence the country's approach to development strongly emphasized human resource development, as reflected inthe Government o f Eritrea's (GOE) macroeconomic policy and development strategy released in November 1994, just before joining the Bank in 1995." The objectives o f the health sector are to reduce and eventually eliminate death from easily preventable diseases, and to enhance awareness o f good health practices inorder to improve the productivity o f the work force. This i s to be achieved by: giving priority to primary health care in controlling major health hazards, encouragingprivate sector participation, community and beneficiary contribution inhealth finance, and promotion o f healthy practices. 1.6 The MOHwas initially focused on rebuildingthe health infrastructureinthe 1990s; the next five years was characterized by a focus on communicable disease control and child health; and inrecent years on addressing the sector's human resource shortages andmaternal health. Since 1991 13 hospitals, 19 health centers and 112 health stations have beenconstructed. l2A substantial part o fthe M O H expertise was derived from experience with runningthe military health services duringthe war o f independence. For this reason, the health sector benefited from a small but experienced cadre o f health professionals andpara-professionals to provide leadership to the development o f the sector. Nonetheless, humanresources were a key constraint facing the health system-in 1999 the ratio o fphysicians per 1,000 people was 0.02, well below the regional average o f 0.17 per 1,000. l3The humanresource shortages continue to be one o fthe most important challenges facing the Eritreanhealth system. 1.7 Giventhe low level o fhealth expenditure, therehas beenan emphasis onthe most cost- effective disease control andpublic health interventions, underthe leadership o fnational disease control programs and implementedthrough a decentralized primary health system at the zoba level, World Bank and other External Support to HNP: 1994-present 1.8 Between 1997 andthe present the World Bank committed approximately $82.3 million in IDA credits and grants to support three health sector operations inEritrea (Table 1-1). The Bank's support to the sector was informed by an Information Sheet on HealthNutrition, Population and Poverty (1999) and a Health Sector Report in2003/04. Inaddition, IDA financing has beenmade available through two multisectoral operations: (i) Integrated Early the Childhood Education Project which financed clinic- and school-based delivery o f child health and nutrition services; and (ii) Community Development Project which financed the the construction o f 31 health facilities. Other key sources o f external support to Eritrea's health 11. In 1994the GOE issuedapolicy documentcalled "Macro-Policy.'' Inadditionto astrong emphasison human development, the other mainthrusts ofthe strategy were: swift transitionto a market economy; introductionof a liberaltrade policy; and a central role of the private sector as an engine of growth (World Bank 1996b, p4). 12. MOH2006g (p3). 13. World Bank 2004b (pp82-3). 5 sector were: the U.S.Agency for International Development (USAID),I4the ItalianCooperation andthe various UNagencies, notably the World Health Organization (WHO), the United Nations Population Fund(UNFPA), the Joint UnitedNations Program on HIV/AIDS (UNAIDS) andthe UnitedNations Children's Fund(UNICEF). Two projects intersected directly with the areas covered by the Bank's projects: the Environmental Health Project financed by USAID and the Public Health andRehabilitation PrograminEritrea (PHARPE) Project financedby the ItalianCooperation. Table 1-1. World Bank involvementinthe healthsector World Banksupport Implementation Actual Credit Period S million HealthSector Projects 87.1 HealthProject 1211997-1212004 17.5 HAMSeT ControlProject 12/200(M3/2006 45.6 HAMSeT Project I1 06/2005-on-going 24.0a Projectsclosely relatedto the HealthSector CommunityDevelopmentProject 02/1996-12/2001 16.1 IntegratedEarly ChildhoodDevelopmentProject 07/2000-03/2007 41.6 Economic and Sector Work Eritrea HealthSector Reportb - 2003 Source: World Bankdata. a. This amount is the IDA commitment as the project is still under implementation. As ofJune 2008, $14.7 million (60 percentof the credit)has beendisbursed. b. This piece of EWS was formally publishedin 2004 as "The HealthSector in EritreaWorld Bank Country Study." 14. Inresponse to a request by the GOE on July 26,2005, USAID offices closed and development assistance programs to Eritrea ceased on December 3 1,2005. 6 Figure 1-3. External assistanceto the HNP Sector, 1996-present a e Health Proiect (1997-20041 H A M S e T IProject (2000-2006) . H A M S e T I1Project (2005-ongoing) USAID(1996-2005) Italian Cooperation (1997-2007) 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 1.9 Inaddition to Bank-financedtechnical assistanceto the MOH, USAID's Environmental Health Project financed studies on the vectors o f malaria transmission and the efficacy o f malaria control methods in2001,technical assistance for malaria surveillance, a malaria prevalence survey conducted in2000/01,epidemic preparedness and program strengthening which concluded in2004.l5The P H A W E Project was implementedby WHO over the 1997-2007 period, and coveredthe following areas: (i) human resource development, (ii) and malaria tuberculosis control, (iii) surveillance, and (iv) infrastructure and maintenance support to health centers and hospitals (especially inGash Barka). 1.10 The country i s politically isolated, and compared to other countries there are very few donors inthe health sector, and the country ingeneral. While the result i s severe resource limitations, the health sector did not suffer from the distortions caused by competing donor priorities often seen elsewhere. 2. Health Project Objectives and Design 2.1 The health sector dialogue formally started shortly after Eritreajoined the Bank in 1995. There was post-independence euphoria throughout the country, and a "can do" attitude permeated the agencies o f government. The President o f Eritrea made the political commitment to deliver a hospital ineachzoba, and the support o fthe Bank was sought specifically to finance 15. Shililu2001a,b; Sintasath 2004; and Graves 2004. 7 "two 200-bed, two-storey hospitals." The objective stated inthe Development Credit Agreement was "to contribute to the improvement o f health care delivery services and health status."'6 2.2 Therewas disagreement between the Bank andthe GOE as to how best to achieve this objective. The GOE made a legitimate argument that referral hospitals are an integral part o f the primary health care system interms o f their referral function, and the technical support and oversight they provide to lower level health facilities. The government, through the Eritrean Liberation Front, had gained considerable experience with runningbasic health services during the war and needed assistance with the hospital sub-sector. Furthermore, at the time the M O H was receiving substantial external support and technical assistance for clinic services from bilateral donors and UNagencies, and the sector had benefitted from clinic reconstruction and renovation implementedunder the Bank-financed Community Development FundProject during 1992-1997. l7 2.3 The two proposedhospitals were to replace a 102-bed hospital inMendefera(inzoba Debub) and a 41-bed hospital inBarentu (inzoba Gash Barka). The Bank correctly questioned the size of the two proposed hospitals, their affordability, and their impact on the already constrained human resource situation. Eritrea's bed capacity was withinthe ran e o f other low income countries inthe region, and utilization o fthe existing hospitals was low. F* Itwas estimated that the two hospitals alone would imply additional per capita recurrent expenditure o f between $1.O and $1.5 annually. l9A further concern was the displacement o f staff from basic health services inorder to staff the referral hospitals. Inanticipation o f future economic growth and associated government revenues, the government argued that the recurrent costs were within affordable limits. To address the human resource concerns the M O H agreed to develop a human resource plan for the hospitals. 2.4 Inadditionto financing hospitalinfrastructure, the project financed clinic refurbishment, drugs and medical supplies, blood safety, and capacity building. The components andactivities, as anticipated inthe project appraisal document, are summarized inBox 2-1. These activities address some o f the challenges facing the sector-lack o f physical access to health services of acceptable quality, particularly by the poor and rural population, weak institutional capacity at the national and zonal levels-although the financial barriers to health care access was not addressed. 16. According to the project appraisal document the objective o f the HealthProject was: "to improve the health status o fthe people of Eritrea thereby enhancing quality o f life and the ability o f Eritreans to participate inthe country's socio-economic development." 17. The project built23 health stations, 1 health post and 7 health centers (World Bank 2002). 18. Eritrea's bed capacity was within the range o f other low income countries in the region: bedpopulation ratio for Eritrea was 0.76 beds per 1,000; low income countries in SSA region: inter-quartile range for the period 1995-2000 was 0.71 to 1.49 (World Bank 2007a). At the time of project preparation, the bed occupancy rate nationally was 48.1 percent; 32.2 percent in Mendefera hospital, and 52.9 percent inBarentu hospital (World Bank 2004a,b). 19. World Bank 1997. 8 Box 2-1. Health Project: components and intendedactivities 1.Strengthening Health CareServices Ia.ExpandingAccess toSecondary Referral Health Care in twoRegions($12.7million; 60.2percent of appraisal cost): constructing, equipping and staffing two referral hospitals inBarentu and inMendefera. lb. Strengtheninghealth servicesnationally ($4.2 million; 19.9percentof appraisal cost): (i)supporting 18 health centers and 12 health stations,aby providing equipment, furniture, essentialdrugs andvaccines, training providers, and improvingtheir management by better training, communication and supervision, and (ii) expandingthe national blood bank service including the construction of anationalblood bank inAsmara and strengtheningthe network of blood banks. 2. Capacity-building 2a. Program management and sustainability ($2.0 million; 9.5percent of appraisal cost) Provision oftechnical advisory services, training programs, study tours and studies to strengthenmanagerial capacity at all levels inthe MOH, to improve mechanismsto decentralize decision-making within the hospitals and rural health facilities, and to test local-level mechanismsto raise, retain and utilize additional revenuesfor health ' services. 2b. Project management and implementation ($1.0 million; 4.7percent of appraisal cost) This sub-componentcoveredcapacity building for project management and implementation inthe Ministry of Health and inthe zonalhealth offices. Source: World Bank (1997). a. These facilities were all beingbuilt at appraisal or constructedwithin the preceding3 years largely by donor and community contributionsimmediately after independence but remainednon-operationallackingequipment and furniture. 2.5 The project rationale included reference to the lack o f health service access particularly , among the poor.20 The project designindirectly sought to address the needs o fthe poor by the choice o f health centers and health stations for refurbishment and the choice o f hospitals for reconstruction, inparticular Barentu hospital which i s located inone o f the deep ruralzobas. 2.6 Reparationtook 21 months from project concept to approval inDecember 1997.*' Despite a protractedproject preparation periodthere were important shortcomings: (i) project the sought to influence total fertility rate, prevalence o f female genital mutilation, and malnutrition yet there were no complementary interventions to specifically achieve these health outcomes,22 (ii) wasnosystematichumanresourceplanningdespiteaverydetailedanalysisoffinancial there sustainability, and (iii) safeguard policies were not complied with regarding the medical waste producedby the hospitals. 2.7 Implementation arrangements. The implementing agencies were the M O H and the zonal health offices. A project management unit (PMU) was established inthe MOHunder the supervision o fthe office o f the Minister. Project implementationmade use o f existing M O H 20. "lack ofphysical and financial access to health services of acceptable quality to the people ofEritrea, particularly to the poor and to those living inremote rural areas" (World Bank 1997, p4). 21. Project Concept Documentwas datedMarch 1996.Project preparationwas supportedby a PHRD grant and a Project PreparationFacility. 22. Key performanceindicators: infant mortality rate; under five mortality rate; maternalmortality ratio; total fertility rate; prevalenceof malnutrition inchildren under five and inwomen; prevalence oftuberculosis, malaria, acute respiratory infections, sexually transmitted diseases and acquiredimmune deficiency syndrome(AIDS); percent of publicly provided health services financed by local and community resources. 9 coordination mechanisms, and implementationwas coordinated with the relevant divisions and national programs inthe MOH, the zonal health offices, and other pertinent structures (e.g., Pharmecor, the parastatal pharmaceutical manufacturer). 2.8 Risks. The project appraisal document identifiedriskspertaining to utilization, implementation capacity, and the appropriateness and maintenance o f facilities and equipment. Riskmitigation measures were identifiedbutit is unclear who was responsible for their implementation. For example, poor utilization due to, amongst others, financial barriers to access was identified, but no specific remedyor assignment o f responsibility was proposed. Giventhe Bank's valid concerns at appraisal about the affordability o fthe hospitals' recurrent costs (as raised inthe economic and financial analysis), it i s surprisingthat these issues did not feature more prominently inthe risk assessment. 2.9 M&E design. While the links between the outcome and impacts identified inthe logframe were plausible, there was a disconnect with the project outputs for selected indicators (e.g., total fertility rate, prevalence o f female genital mutilation, and malnutrition as mentioned in paragraph 2.6). The indicators were all national inscope, while a large share o f the interventions was regionally focused. No targets were set for the key performance indicators (KPI). A combination o f sector-level and project-level indicators were proposed, with the health management information system (HMIS) beingresponsible for collecting the former and the P M U beingresponsible for the latter. Very little detail was provided inthe project appraisal document and the project implementationplan on how project data (as opposed to sector data routinely collected through the HMIS) would be collected at the zoba level. A householdhealth utilization and expenditure survey (EHHUES) was to be expanded from the existing two zobas to include all six zobas inthe country to provide baseline data. Implementation 2.10 Planned and actualexpendituresby component.The HealthProject becameeffective on May 28,1998, was implemented over aperiod o f sevenyears, and closed on December 31,2004,18 monthsafter the original closing date. The reasonfor the extension was the substantial delays inthe hospital construction. The actual project cost was $22.6 million, 107 percent o fthe cost estimated at appraisal. Uponproject closingthe credit was 99 percent disbursedand SDR 3,405 was cancelled. Borrower contributionwas 68 percent o fthe appraisal estimate. The actual project cost included a grant from Norway o f $2.8 millionfor hospital construction that was notplannedat appraisal. 2.1 1 Followinga malaria outbreak and inresponse to humanitarian needs following border hostilities-both occurring in 1998 shortly after effectiveness-the Bankagreed to reallocate $2.8 million from the training and study budgetto the malaria control program and $1.2 million to supportpost-conflict emergency programs. The Bank also agreed to reallocate funds to nurse training institutions under Component 2 (Capacity Building) inorder to support the country's strategy to train additional health staff. The latter was justified by the severe human resource constraints heightenedby the post-conflict situation. There were no revisions to the project objectives or key performance indicators. The mid-termreview missed the opportunity to fix the disconnect between the project outputs and the KPIs. 10 Table 2-1. Planned versus actual costs, by component Appraisal Share of Component Actual appraisal cost ($ million)a estimate ($ million) (percent) 1. Strengthening Health Care Services 17.8 19.41 109.0 la. Expanding access to Secondary Health Care 13.1 13.07' 99.8 1b. Strengthening health services 4.70 6.37 132.8 2. Strengthening Institutional Capacity 3.30 3.22 97.6 2a. Management 2.10 1.28 61.0 2b. Project Management 1.20 1.94 162.0 Total Project Cost 21.10 22.63a 107.2 Source: World Bank 1997. a. Not reflected inthe project costs is the cost overrun of $5.3 million for the two referral hospitals that was reallocated from and accounted for underthe HAMSeT Control Project. 2.12 Constructiono f the hospitals started three years into the project's lifespan and was not completed by the delayed closing date. Although the hospitals were said to be 95 percent completed at project closure, the two referral hospitals became operational several years later in 2006, following an additional allocation o f $5.3 million from the HAMSeT Control Project. Taking into account this outlay, the total construction costs for the two hospitals came to $18.4 million, 145 percent o fthe appraisal estimate. Constructionwas plagued by cost overruns and substantial delays, due to: (i) under-estimationo f construction costs at appraisal;23 (ii) in delay finalization o f design and designalterations well into construction; 24 (iii) shortagesafter labor adults were re-mobilized to the war front in 1998; (iv) shortages o f goods and supplies; (v) foreign currency restrictions causing delays inpayment o f suppliers; (vi) bureaucratic delays in getting approval from the maritime, customs and port authorities at Massawaport; and (vii) damage to the Barentu Hospital duringthe border hostilities. Many o f these factors were beyond the control o fthe ministry andthe Bank. Initiallythe PMUlackedexperience with managing civil works and with Bank procedures in general, although this improved over time. According to the ICR, an independent assessment o f procurement inprojects inEritrea rated the procurement performance o f the HealthProject as satisfactory. 2.13 Implementation o f activities by the National Malaria Control Program (NMCP) and the National Blood Transfusion Service (NBTS) progressed well. The reallocationsmentioned earlier detracted from funds for capacity buildingand institution-strengthening. Furthermore, civil works was the largest expenditure category, accounting for half o f total project costs. An additional nine percent o f project cost was devoted to hospital equipment. Duringproject supervision the project team expressed concern about the imbalance between investments in hardware and software. The dominance of the project by the construction o f the two hospitals, offered the team limitedroom to introduce sectoral and policy dialogue. The team utilized the 23. The issue o f cost overruns was raised as early as the project's launch mission. 24. Examples o f design alternations include: expanding scope o f medical gas to all wards; adding elevators; changing the layout o fnurses' stations in wards to accommodate an electronic nurse call system. 11 opportunity ofproject supervision to engage the MOH leadership inpolicy dialogue on broader sectoral issues-for example the country was encouragedto develop aNational Health Strategy. Policy dialogue regarding management and improvement of the efficiency of the hospital sub- sector was, however, largely limitedto requiring the MOH to develop a hospital staffing plan. 2.14 The reallocation for malariameantthat the NationalMalaria Control Program also became a key part of the project's implementation arrangements. The project procured malaria control drugs and supplies, and training and workshops. The project-financed inputs accounted for the largest share of external funding to the NationalMalaria Control Program over the period 1998-200 1.25 In 1999the MOH convenedthe first NationalMalaria Conference, andthrough the Mendefera Declaration on Malaria Control in Eritrea committed the country to reducing malaria morbidity andmortality by 80 percent from the 1999 levels. 2.15 Safeguards. The project was subject to an environmental assessment becauseof hospital medical waste. An environmental assessment was not completed at appraisal and the medical waste planwas only drafted inMay 2003, nearly five years after project effectiveness. Although the supportto the NationalMalariaControlProgram was added after appraisal, the same safeguardrequirementshad to be met for the procurement of insecticides, inparticular dichloro- diphenyl-trichloroethane(DDT). No evidenceof measures taken to comply with safeguard policies could be found. 2.16 M&E implementation and utilization of data. The preliminary results ofthe household health and expenditure survey for selectedzobas were available in 1997andthe full survey was completedinSeptember 2002, four and a halfyears after effectiveness and undermining the intention that it would serve as abaseline survey. The delay was mainly due to border hostilities in 1998-2000 causing substantialdelays inthe fieldwork. There was no follow-up survey and hence no trends could be assessed. There is, furthermore, only limitedindicationthat the survey data were usedto inform project-level implementation or sectoral decision-making. The project benefitedgreatly from the HMIS, although very few ofthe HMISinputswere financedby the project. Supervision missions repeatedly stressed the importance of strengthening the project's M&E systemandthe appointment of anM&E specialist inthe MOH. Constrainedby human resource availability an M&Eunitwas establishedinthe MOH only in2007. Achievement of Objectives 2.17 The assessment of the achievementof the project objective-to improve health care delivery services andhealth status-is divided into two parts: (i) improvement inthe delivery of health care services; and (ii)improvement inhealth status. The outputs by component are summarized inAnnex A. OBJECTIVE: IMPROVEMENT CARESERVICES IN THE DELIVERY OF HEALTH 2.18 While service delivery has improved following the opening o f the 144-bedBarentu Hospital and the 168-bedMendefera Hospitalin2006, bothhospitals were functioning belowthe capacity intendedat appraisal. More than a year after the opening of the hospitals the bed 25. World Bank2005b. 12 occupancy rate was 27 percent for Mendefera hospital and for Barentu it i s estimated by IEG at about 30 percent.26 InBarentuHospital (inzoba Gash Barka) there has only been a small increase inutilization since opening (Figure 2-1a) while in-and out-patient data for Mendefera Hospital show an increasing trend inutilization (Figure 2-lb). 2.19 The two hospitals were builtas referralhospitals, providing specialist services, for example surgical services, and receiving referrals from health centers and other hospitals inthe zoba. The numberofsurgical procedures inMendeferaHospitalhas nearlytripled-from 780 in2005 to 2,834 in2007. Referrals from health centers to Mendefera Hospital nearly doubled between 2005 and2007 (Table F-1inAnnex F), andreferrals from other hospitals increasedfrom 109to 238 between2004 and 2007. For BarentuHospitalthe picture is slightly different. While referrals to Barentu Hospital from health centers inthe zoba increased substantially between 2005 and 2007, referrals from other hospitals inthe zoba declined (Table F-1inAnnex F). Barentu Hospital essentially h c t i o n s as a zoba hospital serving the healthcenters inthe sub-zobas inits catchment area, and not as a referral hospitalto the zoba hospitals. The under-utilization o fboth hospitals and the lack ofprovision ofsome specialist services inBarentuhospital contribute to the overall inefficiency o fthe hospital investment. A mitigating factor i s the re-use o fthe replaced hospital facilities as nursetraining facilities, meeting an important need inthe country. 2.20 Utilization o f health centers and health stations i s assessedusingtwo examples- immunization and antenatal services-with due recognitionthat these are not the only services provided by these facilities, and that not all the improvements can be attributed to Bank-financed input^.^' There have been significant increases inimmunization coverage: full immunization coverage expanding from 41 percent in 1995 to 76 percent in2002 (Table 2-2). Despite some improvement inantenatal clinic service access, the share o f deliveries attended by a health worker has not shown similar increases. The share o fpregnant women visiting a healthfacility for at least one antenatal visit increased from 44 percent in 1998 to 64 percent in2005.28 Of those who attended at least one antenatal clinic visit, a tenth failed to complete the required number of antenatal clinic visits.29In 1999 delivery service coverage was 17 percent with only a modest increase by the end o f the project (Table 2-2). 26. MOH2007a. Bedoccupancy rate for Barentu Hospital was reported at 70 percent, the highest inthe country. But, this basedon a60-bedhospital, the same as the hospital that was replaced. 27. Data on the indicators o f health facility access and utilization proposedinthe PAD (World Bank 1997, Annex 1 p2) were not available, and alternate measuresor slightly modified indicators that could be obtained from the HMIS have beenused instead. 28. MOH2007a (p37). 29. MOH2006a (p38). In2000 and 2003 the drop-out rate was 11.1 percent and 10.6 percentrespectively. 13 Figure2-1. In- and out-patient caseload in Gash Barka and Debubby hospital,1998-2007a Gash Barka 35,000 Hospital opened 30,000 25,000 20.000 15.000 10.000 5,000 1998 2000 2002 2004 2006 --cBarentu --c Aoordat 4 ) Tessenei Debub 35,000 30,OqO 25,000 20,000 15.000 10,000 5,000 1998 2000 2002 2004 2006 -cMendefera --cAdi-Kei ~h. Dekemhare Source: MOH2008a. a. The factors that contribute to the low utilization are: Design.While there were several positive aspects to the hospitaldesign,30 some design features constrainedutilization: e.g., the porous floor material intheaters and some wards made sterilization impossible, and mosquito nets were not fixed to the beds inBarentu hospital, inthe zoba with the highest malaria prevalence. Many of the design alterations that resulted in considerabledelays are still not functional-zg., the central medical gas supply to all wards is currently not operational; the ward layout was changed in order to install a nurse calling system that is still not functional; an elevator was added to the Mendefera hospital for better access to the second storey with administrative offices, is not utilized. Energy supply. InBarentu Hospital there are 3 hours of electricity inthe day and 3 hours at night." Solar power seems to be the only plausible solution, but was consideredtoo costly. 30. Air-conditioning was limitedto the operatingtheaters and the pharmacy; the extensive use ofnatural light; the use of natural water collection through construction of boreholes, water towers and undergroundwater run-offto ensure water self-sufficiency of the facilities; landscapingwas designedto facilitate natural cooling; design allowed for revenuegenerationthrough rental of conferencefacilities and other hospital facilities. 3 1. The source of power i s inMassawaon the Red Sea, and GashBarka i s on the country's westernborder adjacent to Ethiopia and Sudan. Energy supply was always constrainedbut the current situation is particularly acute due to diesel shortages and high gasoline costs. 14 Table 2-2. Utilization of clinicservices, 1995-2002 Service 1995 1998 1999 2000 2001 2002 2005 Immunizationcoveragea BCG 61 91 DPT3 49 83 ........................................................................................................................................................................................................................... Full immunization 41 76 ANC coverage (% pregnant women with at least 1ANC 44.2 40.5 38.4 45.3 51.4 64.1 ................................................................................................... visitIb " ....................................................................................................... Share of births attended by skilled health personnelc 17.4 16.6 18.9 22.7 26.2 Source: Macro International 1995, 2002; MOH2007a. a. Macro International 1995,2002. b. The project appraisal document uses the indicator '% of pregnancies with at least 3 ANC visits' but this indicator was not available from the HMIS reports(MOH 2007a p37). c. MOH2007a (p41). OBJECTIVE:IMPROVEMENT IN HEALTH STATUS 2.21 Project implementation coincided with significant improvements inthe health indicators (e.g., infant and child mortality) but attribution i s confounded by several factors. These include: (i) havebeensignificantimprovementsinsocioeconomicconditionsafterindependence, there (ii) wereothercomplementaryactivitiesthatalsoaddressedthisobjective(e.g., the there Integrated Early Childhood Development Project, and inputs financed by other donors: UNICEF, WHO, USAIDand Italian Cooperation), and (iii) project hospitals are functioning well below the the capacity intendedat appraisal, and impairingthe functioning o fthe referral chain. Nonetheless, there i s some evidence o f reduction inmorbidity from immunization-preventable diseases between 1998 and 2005 consistent with the expandedimmunization coverage shown in Table 2-2 -neonatal tetanus, diphtheria and whooping cough but not for measles and polio (Figure2-2). Figure2-2. Trends in immunization-preventablediseases, 1998-2006 1 , +Neonatal Tetanus +Other Tetanus 1-0- Whooping Cough +Measles A Hepatitis B 1 -6Diphtheria rr AFP 900 1998 1999 2000 2001 2002 2003 2004 2005 2006 ' 1998 1999 2000 2001 2002 2003 2004 2005 2006 2.22 All indicators o fmalaria morbidity and mortality decreasedsignificantly between 1998 and2003 to below pre-outbreak levels. 32 Malaria morbidity rate decreased by 74 percent, 32. The endyear of 2003 is usedherebecause when the HAMSeT ControlProject startedthe support to the malaria controlprogramwas mainly derived from the new project. 15 malaria mortality rate by 85 percent (Table 2-3), andmalaria case fatality rate (Le., the percentage of malaria deaths out of health facility-admitted malaria patients) decreased by 78 percent (from 3.6 to 0.8). Similarly, bed occupancy due to malaria also decreased byjust under a tenth (9.1 percent). Table 2-3. Malaria morbidityand mortalityrate, by zoba, 1998-2006 Malaria morbidity rate (per 100,000) Malaria mortality rate (per 100,000) Zoba 1998 2003 2006 1998 2003 2006 Anseba 5,544 336 83 0.2 Debub 7,857 2,059 258 0.1 Gash Barka 15,544 5,617 2,199 4.5 Maekel 986 436 164 0.3 Northern Red Sea (NRS) 7,460 804 90 0.4 Southern Red Sea (SRS) 3,601 634 137 0.0 National 7,546 1,945 568 15.8 2.3 1.4 Source: IEGcalculations based on MOH2007a. 2.23 A substantial share of the reduction inincidence canbe ascribedto the NationalMalaria Control Program's programmatic achievements. While the trends inmalaria interventions and malaria incidence are high1 suggestive of a programmatic success inthe country's malaria control efforts (Figure 2-3) a key question is: how muchofthis success can be ascribedto the l 3 varying rainfall pattern? In 1997/98the rainy season lastedfour months insteadof the usual two `' months. Between 1999 and 2003, more than half of the variance inmalaria incidence can be explained by variance inrainfall.34Furthermore, in2001 rainfallexceededthe level in 1998,yet mortality and morbidity in2001 was betweenathird and a quarter of the 1998 level. Figure2-3. Malariainterventionsand malaria incidence,1998-2006 - 140 wb,& ITNs + bednets reimpregmated 1 :: 1 (lO.000) 100 Breedingsites 3088 T filled and treated (1.000) !sumNumber of houses sprayed (1.000) 40 -Incidence rate 20 .-&- Averageannual 0 ramfall 1998 1999 2000 2001 2002 2003 2004 2005 2006 Source: Nyarango et al. 2006 (incidence and rainfall data); MOH2004a, 2006b (intervention data). 2.24 Until2001the Bank was the major source offunding to theNational MalariaControl Program. The program also benefitedfrom technical assistancefrom bilateral donors, notably the US AID-fundedEnvironmentalHealth Project and assistancefrom the Italian CooperatiodWHOthrough the PHARPE Project, as well as Bank-financed technical 33. Given the overlap betweenthe HealthProject and the HAMSeTControl Project, the results reported inthis section shouldbe viewed together with those reported in Chapter 3. 34. Author's calculations. 16 supervision. These inputswere highly complementary to the project-financed programmatic support to the NMCP. 2.25 Blood screening for HIV, hepatitis and syphilis was introduced following the establishment o fthe National Blood Transfusion Service, and by 2003 half o fthe country's safe blood supply needs were being met. Table 2-4 shows the decreasing HIV prevalence rate in screened blood, suggesting the success inthe quality control mechanisms implementedby the National Blood Transfusion Services3' Table 2-4. HIV prevalencein screened blood, 2003-200536 Year Units of Blood Collected HIV prevalence (percent) 1999 4.60 2003 4,245 0.50 2004 4,088 0.50 2005 4,455 0.34 Source: MOH2007b (p65). PROJECT RATINGS 2.26 The outcome o f the HealthProject i s rated moderately unsatisfactory, and i s based on the ratings for relevance, efficacy and efficiency as they apply to the two parts o f the objective (summarized inTable 2-5 and discussed below). Table 2-5. HealthProject:SummaryIEGRatingsby Objective IDevelopment Objective I& k e , v , ~ ~ ~ ~ ~Efficacy I d 1 Efficiency I Outcome Design Improvethe delivery o f health Moderately services Modest Substantial Modest unsatisfactory Improvethe health status of Moderately the peopleof Eritrea Substantial Substantial Modest satisfactory 2.27 The overall relevance of the project objective and design for the first and second sub- objectives is modest and substantial, respectively. The relevance o f the objective to improve 35. The ICRestimatedthat ifonly the impacton HIV i s considered, the screeningmethodsresultedinsaving o f 163,440 HealthAdjustedLife Years (HeaLYs) andthat for eachprojectdollar spent onbloodsafety 56 healthydays were obtained(WorldBank 2005b). The assumptions underpinningthis analysis, inparticular the epidemiologic modeling are not knownand it was not clear from the analysiswhat the number ofHIV infections avertedwas. 36. The decliningprevalencein screened bloodhas incorrectly been interpreted in some reports as evidence o f declining or stabilizingHIV prevalence(e.g., MOH2007b (p64) andWorldBank 2006b) failingto recognizethe selectionbias imposedby the National BloodTransfusionService screeningmethods. Insteadthis declining prevalencetrend shouldbe interpretedas evidenceofthe quality controlmechanisms implementedbythe National BloodTransfusionService. 17 health care delivery services and health status o f Eritreanswas consistent with the post-conflict situation, and remains consistent with the priority placed on human development by the GOE and the MOH. The establishment o fthe blood bank was also an important priority. The relevance o fthe project design had some shortcomings. While respondingto the dire need for reconstruction, the design failed to account for the results chain ensuringthat the project outputs (especially the construction o f two hospitals) are linkedto all o fthe intended impacts as reflected inthe performance indicators (e.g., total fertility rate, prevalence offemale genital mutilationand malnutrition). While the hospitals'had several positive design features, the designhas to be criticized for the excessive size o fthe Barentu hospital. The addition o f the malaria sub- component enhanced the relevance o f the design, especially o f the second part o fthe objective- to improve health status-as the project outputs could be directly linked to a major source o f morbidity and mortality among adults as well as infants and children. 2.28 The efficacy o f the first sub-objective, to improve the delivery o f health services, i s substantial. While the contribution o fthe hospital constructionto the project objective is constrained by the under-utilization o f specialist services inBarentuhospital, the HMIS data suggest increasingtrends inservice delivery. Furthermore, the establishment o f the blood bank was an important achievement. The efficacy o f the second sub-objective, to improve health status, i s substantial. The investments inclinic services contributed to improvements inclinic utilization and health outcomes. It is also possible to make stronger links between some o f the other project investments and outcomes, e.g., investments inthe blood bank and the successful malaria control program. Some performance indicators that the project sought to influence did not improve substantially: malnutrition andprevalence o f female genital mutilation. 2.29 Project efficiency i s modest. The investments inmalaria control and inPHC services were highly efficient as it focused on low cost and highly efficacious interventions. Onthe other hand, the under-utilization o f the two hospitals (interms o f bed occupancy and out-patient utilization) and the large share o f specialized equipment that was not operational, especially at BarentuHospital, detracted from overall project efficiency. 2.30 The project's risk to development outcome i s moderate. Several factors bode well for sustainability: the highlevel o f government commitment and ownership; the highutilization o f clinic services; the reasonable likelihood o f increasing utilization o f MendeferaHospital; and the highly successful malaria program with its strongreliance on community involvement that has proven to be sustainable. However, the government's difficult fiscal position may constrain the sector's future resources, and border hostilities continue to pose some level o f risk to the economy. 2.3 1 As mentioned, the project's duration coincided with the start of border hostilities andthe country effectively moved from a post-conflict situation to a country inconflict, creating many implementation challenges facing the Bank and the borrower. 2.32 BankPerformance. The Bank's overall performanceis moderately satisfactory. Pre- project dialogue was characterized by disagreement betweenthe Bank and the borrower regarding the need for and affordability o f two 200-bed hospitals. Some interviewees commentedthat at the very beginningo f the dialogue the Bank did not fully appreciate the political economy o f the post-conflict situation, and initially adopted an overly technocratic 18 approach. Eritrea was said to be ina "reconstruction mode" whereas the Bank was inits usual "development mode." The GOE felt the Bank was "too ideological" at the time o f project preparation, and the notion o f doing a hospital project was going against the "received doctrine" o f supporting primary health care. The GOE contended that it fully appreciated the highreturns to investments inprimary health care, but that they were adopting a much longer term perspective and was "building a country for the future." After a protracted project preparation the Bank agreed to finance the project. Ifone takes a long-term view and considers the sectoral engagement inthe decade that followed under the HAMSeT projects, the Bank was correct to engage the sector eventhough there were technical disagreements. 2.33 Quality-at-entry was, however, moderately unsatisfactory for the following reasons: (i) excess capacity inhospital designs; (ii) human resource planningdespite very detailed no analysis o f financial sustainability during appraisal, and (iii) failure to complete the environmental assessment for medical waste. Shortly after the start o f the project, support for malaria control was added, but no evidence o f any measures taken to comply with safeguard policies regarding DDT procurement could be found. 2.34 continuity inproject oversight was maintained despite some turnover intask team leader^.^' Quality o f supervision was satisfactory. Supervision missions were regular and Early supervisionwas not sufficiently results-oriented, although the project team did well to engage the government through ESW (that was formally publishedinthe Health Sector Note in 2003); a health expenditure and utilization survey (Phase Iin 1997 and Phase I1in2002) and informal analytical work during ~upervision.~~ Bank was successful inmoving the sector The dialogue from a "reconstruction mode" to a "development mode." There could, however, have been greater engagement with the sector on hospital management to increase the effectiveness o f hospital investments. 2.35 Borrower's performance was moderately satisfactory overall. Government performance was moderately satisfactory despite the country being a new member o fthe World Bank and the difficult security conditions. The performance o fthe MOH, the implementing agency, was moderately satisfactory. There were numerous and considerable delays inhospital design and construction. Most contributing factors mentioned inparagraph2.12 were beyond the control o f the borrower. However, a key factor was-numerous design changes that were introduced by the ministry well into the hospital construction and resulted indelays in construction and installation o f equipment that i s currently not beingutilized, contributing to hospital inefficiency. Implementation o f other aspects o f the project, notably the Malaria Control Program and the Blood Bank, was highly successful. While initially inexperienced, the P M U steadily gained experience and successfully managed fiduciary matters. The P M U had very little staffturnover thereby maintaining the capacity for the project that followed. 2.36 Monitoring and evaluation was modest. The M&E design included sector-level and project-level indicators, and a baseline household health utilization and expenditure survey. The 37. There were four task-team leaders over the project's lifespan, 38. For example, the MOHwas encouragedto develop a NationalHealthStrategy, andhumanresourceplans of the hospitals sub-sectorsettingthe scene for engagementon humanresources for healththat is currentlyunderway. 19 project's indicators were all national, while a large share o f the interventions was regional in scope. The implementation o f the second phase household survey was delayed due to border hostilities, and underminingits purpose as a baseline survey. It i s unclear to what extent data from this costly undertaking was usedto informproject-level implementation or sectoral decision-making. The appointment o f an M&E specialist inthe M O H was delayed and only in 2007 was an M&E unit established inthe MOH. 3. HAMSeT Control Project 3.1 At the time o fproject preparationEritreawas emerging from another period ofwar (1998-2000). While health and other socio-economic indicators continued to improve over that period (refer back to Figure 1 -l),communicable diseases accounted for a significant share o fthe disease burden. Malaria incidence had declined following the 1998 outbreak, but malaria remained a significant source o f morbidity andmortality. The threat o f HIV/AIDS was an emerging area o f concern. Objectivesand Design 3.2 The objectives, as stated inthe project appraisal document, were: "to reduce the mortality and morbidity ofthe Eritreanpopulation due to HIV/AIDS,malaria, sexually transmitted diseases and tuberculosis (HAMSeT) through an increase inutilization o f quality, effective and efficient health services for HAMSeT prevention, diagnosis and treatment, supported by healthy practice^."^^ The choice o f diseases was drivenby the disease burdenand the large positive externalities associated with controlling these communicable diseases. The malaria interventions were intendedto build on the early achievements o fthe Health Project by financing malaria control activities inother sectors (e.g., the Ministry o f Defense) and by greater emphasis on the role o f community outreach and involvement, facilitated by the malaria workers already inplace. The rationale for the emphasis on HIV/AIDS was that, while Eritrea's HIV prevalence was relatively low, early intervention would avert future disease burden. There was also a concern that the demobilized defense force and the large post-war displacedpopulation could exacerbate the spread of HIV. Tuberculosis was not only a significant communicable disease, butalso a major AIDS opportunistic infection. The emphasis on communicable diseases remains justified-in 2005 HIV/AIDS,tuberculosis and malaria were among the five most important 39. The objective, as statedinthe credit agreement has a bit more detail: "to assist the borrower in: (a) increasing knowledge and awareness of HIV/AIDS, malaria, STIs andtuberculosis(HAMSeT) amongthe populationofthe Borrower's territory; and(b) providingthe saidpopulationwith increasedaccess to preventionmeasuresandbasic early treatment, through: (i)increasingthe effectivenessandefficiency of the Borrower's policies and interventions aimedat the reductionofthe spreadofHAMSeTdiseases; (ii) enablingcommunities, households,and individuals to: (A) learn more about the practicesthat facilitate or minimizethe spread ofHAMSeTdiseases, and(B) have access to affordable preventivemeasuresand early treatment services; (iii) improvingthe quality of basichealthcare by providingdrugs andmedicalmaterials; (iv) reducingenvironmentalimpact ofvector control activities; and(v) identifyingaffordable community-managedhome-basedcare for AIDS patients." 20 causes o f m~rtality.~'Furthermore, a disproportionate burden o f malaria and tuberculosis continued to fall on the poor. 3.3 A multiple disease project design was chosen because o fthe country's extremely limited financial and human resource base, and inanticipation o f efficiency gains from elements that disease control programs have incommon: (i) healthpromotion, (ii) surveillance, (iii) logistics, and (iv) zonal-level service delivery. Improving the institutional infrastructure to perform these functions for the chosen diseases would, furthermore, serve many other disease control efforts. Box 3-1. HAMSeT ControlProject:componentsand intendedactivities 1. Collectand analyzeinformation on HAMSeT ($5.1 million; l0.2percent of appraisal cost) (a) improvingHAMSeTsurveillancetechniques, (b) establishingan epidemic forecastingpreparedness system; (c) improvingthe country's capacity to carry out operationalresearchfor identifyingchanges in HAMSeT, (d) introducingmethods to linkthe results ofresearch andM&E to policy formulation, and(e) strengtheningmanagementof communicablediseases at the MOH. 2. Multi-sectoral controlof HAMSeT transmission ($11.2 million; 22.4percent of appraisal cost) (a) Promotehealthybehaviorsthroughmulti-levelcommunication,coordinate communicationactivities of all implementingpartnersandbuildcapacity, (b) Promotehealthy lifestyles through the education system, promotinggoodhealthandpreventingthe spreadofHAMSeTdiseases through the Ministry of Education schoolhealthprogram, (c) Enhanceaccess to preventive, diagnostic, andtreatment services for conscripts, by promotinghealthybehaviorsthroughmultiplechannels o f communication,strengtheninghealthcare servicesfor conscripts,promotingcondoms use andinsecticide-treatedmaterials,and(d) Promote environmentallysound and cost-effectivetechniques for malariavector controlthat would (i) identify,test, validate, and introducesafe, cost-effective chemicalsto replace DDT, (ii) validate malariabiologicalvector control, (iii)develop a strategy for pesticideuse andcontrol, (iv) test community acceptance of validated methodsand'techniques,and (v) replicate socio-environmentally validated malariavector controlmethods. 3. StrengthenHAMSeT diagnostic,health care, and counseling($20.7 million; 41.4percent of appraisal cost) (a) Establishsafe bloodbanks inzoba hospitals,(b) Improvediagnostic,treatment, andcounselingof HAMSeTthrough in-service andon-the-jobtrainingon HAMSeTprevention anddetection, case management, syndromic and laboratorydiagnosisofHAMSeT, as well as pre- andpost-voluntary counseling andtesting; and(c) Improveavailability ofbasic medicalmaterialsand drugs requiredto diagnose andtreat HAMSeTinhealthfacilities. 4. Community-managedHAMSeT responseprogram ($9.9million; 19.8percent of appraisal cost) (a) Community counselingandsupport groups, which aimedto strengthencommunity support services providedby the Ministry of Labor andHumanWelfare andto provide counselingandestablishsupport groups for HIV/AIDS affectedpeople, and(b) Community-managedresponse, whichaimedto test the capacity of communitiesto usetheir own structuresand socio-cultural fabric to (i) respondto technical informationabout HAMSeTfor their prevention,care andcure, (ii) organizetheir internal mobilization, discussion, anddecisionmechanisms on the supportthey deemnecessaryto assess andotherwise manage the diseases, (iii) identify and inputtheir grassrootsandsocio-cultural contribution to HAMSeTmessages, prevention,care and cure methods, and available support services, and(iv) identify, decide on, and implementsub-projectsto preventor mitigatethe diseases andrelatedimpactsinthe community. 5. ProjectManagement andEvaluation ($2.7 million; 5.4percent of appraisal cost) to strengthenthe existingProjectManagement Unit inthe MOH(alsomanagingthe healthproject) to beresponsiblefor planningandbudgeting, procurement,and financial management. Source: World Bank 2000a. 40. MOH 2006b (p65). 21 3.4 The credit was approved inDecember 2000. The project had five components, summarized inBox 3-1. A fifth o f project cost at appraisal was for HAMSeT prevention and control activities implementedby non-health sectors. 41 The rationale for the involvement o f these sectors was that the HAMSeT diseases arise from the interactiono f health and non-health factors, and that addressing these factors required action inmultiple sectors. The sectors were prioritized based on their strategic and comparative advantage indisease control. The multisectoral collaboration betweenthe M O H and other ministriesbuilt on past, less formal collaboration. For example, there were links between the National Malaria Control Program and the MinistryofTransport and Communication around the rainfall data from the meteorological sites; there were links between the Ministryo f Education (MOE) and M O H around school health and the life skills program; and the Ministry o f Defense and M O H collaborated inthe area o f military health services. With the HAMSeT fundingthe M O H was able to strengthen the coordination and provide stronger motivation for action from other sectors. 3.5 The rationale for the community-managed component was to increase community awareness o f the HAMSeT diseases andto mobilize communities for prevention through affordable mechanisms and drawing strongly on the community structures. The component had a demand-driven designwhereby communities would identify disease control activities to be implementedas sub-projects to be implemented by the community. The design was particularly appropriate for awareness raising and stigma reduction as well as community-based care and support services for people with HIV/AIDS and orphans. The demand-driven design had limitation^^^ for prevention interventions targeting risk groups or behavior that are highly stigmatized and the interventions targeting these groups were implementedmainly through the multisectoral interventions and the HealthPromotion Unit. 3.6 The HAMSeTControl Project was part o f the Multi-country AIDS Program (MAP).43 The project design and implementation structure had some similarities with other MAP projects, for example, a community-managed component and a multisectoral component. There were also some differences. The project was managed by the MOH, not by a National AIDS Commission located outside the health sector. This was a source o f debate at the time o f project preparation, but the Minister ofHealth insistedthat the coordination structure be located withinthe MOH. This project was also one o fthe first M A P projects to finance the control o fmultiplediseases. 3.7 Implementation arrangements. The MOHwas the lead implementing agency, andthe project benefited from the project management skills accumulated under the HealthProject. The 41. For example, the Ministry of Health; Education;Labor andHumanWelfare; the Ministry ofTransport and Communication;the Ministry o fTourism; the Ministry ofInformation; andthe Ministry o fLocalGovernment. 42. Communitydemand for a preventionprogramtargeting for example, sex workers, their clients and others with multipleconcurrentpartnerswill likely to be sub-optimalbecauselocaldemand is less likely to internalize all the benefitsofHIV preventionamongthis risk group. 43. The eligibility criteria for the MAP projects are: (i) Evidenceof a strategic approachto HIV/AIDS, developedin a participatory manner, or a participatory strategicplanningprocess underway,with a clear roadmapandtimetable; (ii)Existenceofahigh-levelHIV/AIDScoordinatingbody,withbroadrepresentationofkeystakeholdersfromall sectors, includingpeople livingwith HIV/AIDS; (iii) Governmentcommitmentto quick implementation arrangements, includingchannelinggrant funds directlyto communities, civil society, andthe private sector; (iv) Agreementby the governmentto use multipleimplementationagencies, especiallyNGOs andCBOs. 22 implementationstructure was, however, more complex than the precedingproject. It included seven ministries at national and regional levels, several non-governmental entities,44and civil society structures. A National HAMSeT Steering Committee, chaired by the Minister o f Health, had responsibility for strategic direction and policy guidance. The committee included representation from ministries involved with implementation as well as the six zoba governors. Additional project implementation and coordination structures included: the National HAMSeT Technical Committee and six Zoba HAMSeT Technical Committees. At the cabinet level the Ministero fDevelopment Planningheldthe ministersaccountable for their contribution to the project objectives. 3.8 Risks.The project appraisal document identifiedanumber o frisks, for example, poor coordination o f external actors, lack o f coordination among implementingagencies leading to slow disbursements, poor implementation, andpossible lack o f efficacy of some interventions. All o f these risks were rated as "modest.'y However, the complexity o fthe project design(withmultiple components, sub-components, and implementers at different levels o f government and civil society), particularly giventhe relatively weak implementation experience under the previous project, risked slowing implementation andposed complex monitoringissues. The mitigationmethod(by strengthening the PMU) was insufficient. The risk o fduplication, andthe weak coordination betweenthe activities implemented by the national andzoba-level structures o fthe line ministries andnon-governmental entities andthe CMHRP was not addressed, nor was the risko flow capacity at the community level for identifying, implementingand evaluating CMHRP activities. 3.9 M&E design.According to the project appraisal document, the M&Eplanwas supposed to put a strong evaluation framework inplace early inproject implementation, but the provisions for carrying out this planwere not clearly specified. The proposed M&E mechanisms identified inthe project appraisal document didnot go beyondthe usualrequirements for project ~upervision,~~ andthe design ofthe M&Eplanwas deferred to after project appraisal. While K P I s were identified inthe project appraisal document, no targets for the impact indicators were set. 46 Some o f the KPIs could have beenmore specific-for example, HIV/AIDS and STI interventions were to target high-riskgroups, but the indicators didnot reflect these risk groups. Given the limitations o f adult HIV prevalence as a measure o fprevention impact, indicators o f behavior change among high-riskgroups could have assisted with the interpretation o f HIV prevalence data. Lastly, the project was implementedas a "process project" and a "learning-by- .doingproject" but there was insufficient allowance indesign for: on what basis learning would 44. Three non-governmentalentities, largely government funded, were involved: the NationalUnion of Youth and Students (NUEYS), National Union of Eritrean Women (NUEW) and the NationalConfederation of Eritrean Workers (NCEW). 45. ProposedM&E included: (i) supervisionmissionsand annual progressreviews; (ii) regular quarterly meetingsof the Project Central SteeringCommittee; (iii) semi-annualprogressreports based on implementation targets defined inthe Annual Work Planand Budget; (iv) mid-termreview ofthe project no laterthan 30 monthsafter effectiveness to identify project successes and issues to be addressed; and (v) baseline and follow-up surveys of beneficiaries. 46. Key performanceindicators identifiedinthe PAD: stabilization of HIV sero-prevalenceamong adults aged 15- 24 years; reduction in malaria death rate among children under 5 years and pregnant women; increase inthe proportion of diagnosedand successfullytreatednew smear-positiveTB patients; andreduction inthe prevalenceof severe anemia inwomen of child-bearing age (World Bank 2000a, p34). 23 take place; what type o f information would be collected to specifically inform learning; and how the designor implementationwould berevisedto incorporatethe lessons from learning. Implementation 3.10 The HAMSeT Control Project became effective on March 1,2001, andclosed as planned inMarch2006. Actual project cost was $51.4 million, 103 percent ofthe cost estimated at appraisal. A mid-termreview was done inNovember 2004, rather late inthe project's life (after 47 months out o f the 63 months) and consequently some o f the important changes implemented thereafter were sub-optimal intheir impact. Uponproject closing the credit was fully disbursed and the borrower's contribution was 68 percent o f the appraisal estimate. 3.11 Plannedand actual expenditure by component.The components were not revised although there was substantial reallocation across components (Table 3-1). The CMHRP accounted for a third o f actual project costs (as opposed to the intended 20 percent at appraisal), while only about 60 percent ofthe planned expenditure on collection of data andthe multisectoral activities (components 1 and 2) was actually expended. In2001 $5.3 million was reallocated from goods and consultant services to civil works to fund the cost overrun o f the hospital construction under the Health Project. Table 3-1. Plannedversus actualcosts, by component Appraisal Component estimate Actual Share of appraisal (%million) ($ million) cost (percentage) 1. Collectand analyze informationon HAMSeT 5.14 2.88 56.0 2. Multisectoral control of HAMSeT transmission 11.16 6.75 60.5 3. StrengthenHAMSeT diagnostic healthcare and counselingservices 20.69 22.04 106.5 4. Community-managed HAMSeT response (CMHRP) 9.89 16.93 171.2 5. Projectmanagement 2.72 2.79 102.6 Total Cost 50.00a 51.39 102.8 Source: World Bank 2000 (p. 12); World Bank 2006b (Annex 2). a. Includes$0.4 million for projectpreparation facility re-financing 3.12 Collect and analyze information on HAMSeT diseases (Component 1). See discussion under monitoring and evaluation. 3.13 Multisectoral control o f HAMSeT transmission (Component 2):47 a) Promote healthy behaviors through multi-level communication. The behavior change communications were guided by a Communications Strategy. The activities included: development o f materials for behavior change communications, training o f outreach workers inbehavior change communications (including development training materials), 47. See Annex B for more detail on the interventions implemented by the multisectoral implementing agencies: Ministry o fLabor and HumanWelfare, MinistryofDefense, Ministry o fEducation, religious institutions, non- governmental entities etc. 24 audiovisual equipment, and implementation o f communications activities including procurement o f airtime and newspaper space. The MOH's Health Promotion Unit provided support to the various sectors to ensure technical accuracy o f the information being disseminated. The interventions targeted specific risk groups, and there was evidence o f division o f responsibility based on the particular ministryor non-government entity in~olved.~'However, the division o fresponsibilities between the nationaland zoba-level structures-especially inthe case ofthe non-government entities-was not always clear. The mid-term review reiteratedthe need to refocus the HIV/AIDS activities on highrisk groups (e.g., sex workers), and to conduct mapping o f other high risk groups, but the mapping was not completed by project closing. The team discouraged allocations for income-generating activities among sex workers following doubts about their effectiveness as an HIV prevention strategy. b) Promote healthy lifestyles through the education system.A school-based health curriculum (including life-skills education) was developed and, although with some delay, was implemented injunior and secondary schools. The curriculum development was complemented by investment inteacher training and associated materials, reading materials etc. Materials were also developed for adults, and media was developed to target adult education students. c) Enhance access to preventive, diagnostic, and treatment services for conscripts. The program targeted the army hierarchy, all conscripts as well as the families o f army personnel. The interventions included: behavior change communications, promotion o f VCT and investment inVCT facilities and equipment, promoting condom use, the innovative incorporationo f a `condom pouch' inthe military uniforms (for storage and easy condom access), and promotion o f insecticide-treated bed-netuse. The M O H provided training programs for the military health staff inVCT, diagnosis andtreatment o f the HAMSeT diseases, including TB case detection and treatment. As mentioned in Annex Bythe project procured equipment, drugs and suppliesfor prophylaxis, diagnosis and treatment of HAMSeTdiseases. d) Promote environmentally sound andcost-effective techniques for malariavector control. The HAMSeT Control Project built on and expandedthe earlier successes o fthe NMCP. A key addition was the implementation o f a Pesticide Management Plan, which includedprudent use o fDDTfor indoor residual spraying. This was part o fa very detailed Environmental Assessment prepared for the project, as discussedinparagraph 3.17 under Safeguards. 3.14 Strengthen HAMSeT diagnostic healthcare and counseling services (Component 3): a) Establish safe blood banks inzoba hospitals. The plan to establish zoba-level regional blood banks was constrained by human resource limitations. Instead zoba hospitals were provided with facilities for storage o f blood units supplied by the Central Blood Bank in 48. For example, the Ministry o f Education and the National Union o f Eritrean Youth and Students targeted the youth, respectively focusing on in- and out-of-school youth; the NationalUnion o fEritrean Women targeted women and the Ministryo f Labor and HumanWelfare and the MOH's HealthPromotion Unitfocused specifically on sex workers; the Federation o f Eritrean Workers targeted workers and the Ministryo f Transport targeted truckers in particular. 25 Asmara. Thanks to project financing, the latter i s well-equipped, with well-trained staff and applying with up to date quality control mechanisms. The acute resource constraints facing the health system will pose important challenges especially inthe more remote zobas facing highdistribution costs and electricity outages. b) Improve diagnostic, treatment and counseling o f HAMSeT. The project financedtraining inHAMSeTprevention, casedetection, syndromic and laboratory diagnosis andVCT. It was not practicalto integrate all the training for the HAMSeT diseases, although there were instances where integrationcould have been better. Training was also provided for clinicians inthe treatment o fpeople with HIV/AIDS. The project procured equipment, medical supplies (including diagnostic tests) and drugs associated with prophylaxis, diagnosis and treatment and HAMSeT diseases. 3.15 Community-managed HAMSeT Response Program (Component 4). The implementation o f the CMHRP was substantially delayed; the first sub-projects were implemented only in December 2002,22 months after the project became effective. The delays were inpart due to the lack of detailed planningwhen the project was approvedand late completion ofthe Operational Manual.49 The project overestimated the ability o f communities to assess needs, design, implement and evaluate sub-projects. Inorder to expedite implementation, the project team launched the Rapid Results Initiative (RRI), which implementedresults-oriented projects over a period o f 100 days, and the Community Capacity Enhancement Process (CCEP) which provided training inproposal development, pro'ect management, and reporting to help communities define their needs and pri~rities.'~ The overall implementation experience (particularly inthe first half o f the project) revealed substantial confusion, and lack o f coordinationbetween activities implemented by the line ministries and non-governmental entities at the national andzoba levels and the CMHRP sub-projects. Despite the early implementationweaknesses, the result was an extensive network o f peer-educators and peer- facilitators that reached down to the village level. The opportunity o fthe mid-termreview in November 2004 was usedto address some o f the implementation challenges: a) The low capacity o f communities often ledto reliance on zoba-level line ministriesand non-governmental entities as implementers. At the mid-term review there was concern that sub-projects were not sufficiently community-driven. It was agreed that only sub- projects actually managed by the community would be funded. b) A tentho fthe approved CMHRP sub-projects andnearly a quarter o fapproved grantfunds (roughly $1.8 million) were for activities like renovationo f sports fields, construction o f community buildings, and libraries(Table 3- 2 and Table E-3inAnnex E). These activities hadbeenjustifiedby the notionthat by providing alternate activities (sport and libraries) 49. This was inpart due to the unfortunatedeath of the consultantwhile on mission. 50. The RRIis an implementationandmanagementtool that was introducedto enhance the implementationof CMHRP sub-projects(as well as other aspects ofthe project). The CCEP was implementedto increase the ability of communitiesto successfully develop and implementCMHRP sub-projects. Out ofthe 8,283 individuals receiving CCEP training, 5,908 were community facilitators and2,375 were community management team members. According to the independentevaluation 50 sub-projectshadbeengeneratedthroughthe CCEPtool (Kerouedan and Appaix 2006 inMOH2006f). 26 sexual debut among the youth mightbe postponed, and that the meetingvenues provideda location for awareness raising events especially aimed at the youth. Table 3-2. Distribution of approvedsubprojectsand grants by disease Approved sub-projects reviewed` Approved grant Disease Projects Percent Nakfa (000) Percent AIDS 495 58.6 50,576 43.9 HAMSeT 67 7.9 16,051 13.9 Tuberculosis 69 8.2 9,920 8.6 Malaria 97 11.5 9,2 13 8.0 Diseasenot specifieda 23 2.7 2,504 2.2 Non-HAMSeT 93 11.0 27,042 23.5 Total 844 100.0 115,307b 100.0 Source: IEG analysis o fzoba-levellists of aporovedCMHRP oroiects. .. . " a. Inthese cases, there was an intervention, like"sensitization" or "training", but it was impossibleto determine which of the diseases it pertainedto, from the list. b. This is the total amount approvedfor CMHRP grantsfor the 844 projectsreviewedby IEG. Accordingto the PMU, a total of 98,822,995 Nakfawas actuallydisbursedfor the 908 projectsthat were completed. c. The distribution of approvedsub-projectsby disease differs considerablyfrom the distribution reported inthe ICR, which attributedall approvedprojectsto one ofthe three main diseases (HIV/AIDS,tuberculosis, malaria)and did not show any sub-projects attributableto the HAMSeT diseasescollectively. c) At the mid-termreview the team expressed concern that too many sub-projects were not directly linked to HAMSeTdisease control. The team recommended that only activities that could be directly linked to the HAMSeTdiseases be funded, andthat the maximum amount per sub-project be reduced from $30,000 to $5,000. Inthe last year o f the project the size and orientation o fthe CMHRP sub-projects reflected these changes. 3.16 Financial managementandprocurement. The project was relatively complex-with numerous implementing partners at various levels o f government (national andzoba-level) as well as at the community level. Nonetheless, the project finances were well managed, thanks to experience gained duringthe implementationo fthe previous project, development o f financial management manuals (for the national and zoba-levels), recruitment o f accountants for zonal PMUs, training and a computer-based accounting systemfinanced by the project. All annual audit reports were submittedas required and on time, and the external audits provided unqualified opinions on the annual financial statements for the project's entire duration. While there were some problems with delays inprocurement, all procurements were carried out in accordance with the Development Credit Agreement. As remarkedinthe ICR, the Bank shares credit for this achievement because o f a realistic procurement implementationplan and close supervision. 3.17 Safeguards. The Programhad a category B rating for environmental safeguard purposes. An environmental assessment was conducted to assess the impact o fthe malaria control interventions, the main focus beingthe use o f DDT for indoor residual spraying. The assessment concluded that only a small share o f houses at risk would be sprayed with DDT; that the method o f spraying was consistent with WHO guidelines and the POPs e~ernption;~` and recommended that malaria surveillance should be improved andthat DDT should gradually be replaced by 5 1. The "POPs (persistent organic pollutants) exemption" restricts DDT use andproduction to disease vector control only (not agriculture) and requirescountriesusingDDT to follow WHO guidelines for disease vector control. 27 alternatives. A Pesticide Management Plan (PMP) was prepared following the recommendations o f the environmental assessment. 3.18 M&E implementation.The M&Eplan was not approveduntillate inthe life o fthe project and consequently the project lacked an M&Eframework and M&E plan for most o f its duration. The intention to collect baseline indicators by the end o f FY02 was not fully accomplished. Untilvery late inthe project there were no staff inthe PMUwith dedicated responsibility for M&E. The analysis o fthe CMHRP sub-projects by IEGrevealed many inconsistencies and inaccuracies raising questions about the detail and quality o f monitoring and supervision o f sub-projects at zoba-level as well as at the central level. 3.19 A number of surveys were conducted and while they generated important information, there were some weaknesses. Several surveys were only single cross-sectional surveys with no baseline for com arison (the TB prevalence survey, the Lot Quality Assurance Sampling (LQAS) Survey ), and some variables across multiple surveys had only limited comparability pz ' (e.g., malaria data inthe DHS in 1995 and 2002, the Bed-net survey in2003, and the LQAS Survey in2006; see Table 3-6 and Table F-5 inAnnex F). The education sector conducted a needs assessment at the start and toward the end o f the project, but the Sam ling for the two surveys was not comparable so the impact could not be formally assessed. The tuberculosis prevalence survey provided some information on the disease's geographic distribution, but there were methodological concerns and disagreements on the findings (Box 3-3). 3.20 Repeated HIV sentinel surveillance surveys were conducted among women attending antenatal clinics, an important achievement (Table 3-4).54 However, monitoring o f HIV/AIDS interventions and outcomes among highrisk groups had important limitations. There was no tracking o f coverage o f risk groups, and the recommendedmapping o f high-riskgroups and hot spots for targeted interventions was not implemented. Behavioral surveys were conducted among high-riskgroups such as sex workers and truck drivers although, there were some methodological weaknesses constraining their use intracking behavior change. 3.21 M&E data utilization. The collection o fproject data from the zonal health offices remained weak for a large part o fthe project and consequently the use o f the data to inform planning and implementation suffered. Health facility-based morbidity and mortality data were collected through the HMIS at the zoba-level and was sent to national disease control programs for analysis. The development o f the HMIS-an important achievement by the MOH- 52. Basedon the Lot Quality Assurance Sampling (LQAS) methodology the country was divided into supervision areas which roughly coincided sub-zobas. Sub-zobas consist of large communities called kebabis, and basedon the total population of each kebabi, 19 intervieweesper supervision areas were randomly selected taking a probability sample proportional to the population size. The total sample size was 854 women 15-49years. There were some differences inthe phrasing of the questions complicating comparability with the DHS findings. 53. Ministry of Education and Partnershipfor Child Development undated; Ministry of Education and others 2007. 54. Inaddition, an extensive databaseon HIV surveillance (conducted every second year) exists inthe Ministry o f Defense, but this was not available for analysis or use. This i s unfortunate becausethe prevalence data, especially amongthe new conscripts, is likely the best approximation of incidence inEritrea. 28 generally occurred independently o f the project M&Emeasures envisaged inthe project appraisal document. 3.22 The use o f data to guide implementationwas particularly strong inthe malaria control program. The program had a detailed data collection system which involved malaria agents, health stations, health centers, zoba-level andnational HMIS offices and malaria control program offices. This data was usednot only to track outputs, but to track program effectiveness and inform programming decisions. Inthe other disease control programs (HIV/AIDS, STI and tuberculosis) there was only limited evidence that data were used to assess efficacy, program effectiveness or impact. Nevertheless, the opportunity o fthe mid-termreview was used to institute important changes (see paragraph 3-15), Achievement of Objectives 3.23 The achievement o fthe objective o fthe HAMSeTControl Project-to reduce mortality and morbidity due to the HAMSeTdiseases-is discussed separately for each disease. Table B- 3 inAnnex B provides greater detail on the outputs o f the ministries participating inthe multisectoral component and the health promotion activities implemented by the MOH. OBJECTIVE: REDUCETHE MORTALITY AND MORBIDITY DUE TO HIV/AIDS AND STIS55 3.24 The IDA credit was a major source of fundingfor the country's HIV/AIDS interventions and accounted for 60 percent o f the country's total spending ($33.8 million) on HIV/AIDSover the period 2001-2005. Jointly IDA and GOE accounted for two thirds (67 percent) o f the country's HIV/AIDS spending. Starting in2005, a major new source o f funding-the Global Fund-committed $3.6 million for HIV/AIDSwith a major commitment to funding AIDS care and treatment.56 outputs 3.25 HIV preventive interventions included: extensive awareness-raising among the general population and among key risk groups; encouraging and expanding voluntary counseling and testing (VCT); behavior change activities targeting pupils and their parents;57condom-use and behavior change campaigns and encouraging VCT among sex workers and truckers; condom distribution; blood safety; and vocational and income-generating activities among sex workers.58 55. STIs are not discussed separatelyfromHIV/AIDS becausethe rationale for public intervention in STI control i s integrallyrelatedto the fact that ulcerative STIs increase the risk ofHIVtransmission. Furthermore, the HIV/AIDS and STIbehavioralinterventionsandrisk groups targetedendeavor to achievesimilar outcomes:deferredsexual debut amongthe youth, partnerreductionand condomuse amonghighrisk groups. 56. MOH2007b, ~ ~ 1 2 3 - 4 . 57. Starting in2003, a number o f the CMHRPsub-projects (some ofthempart of the RapidResultsInitiative)were implemented,reaching50,000 secondaryschoolstudents, 15,000 teachers and 17,000 parentsfor HIV/AIDS awarenessandbehaviorchange. Followingseveralyears of development, anHIV/AIDSand life skills education programwas implementedinthe schoolstowardthe endofthe project during2004-05. 58. Because o f the high cost andquestionableeffectiveness o f vocational and income-generatingactivities among sex workers interms of preventingHIV, these interventionswere discouragedfollowing the mid-termreview. 29 O f the community-based sub-projects, the majority (58 percent) was for HIV/AIDS/STIs inthe form o f awareness raising, sensitization, and behavior change activities (Table 3-3). Table 3-3. Distributionof approved sub-projects for HIV/AIDS and (joint) HAMSeT diseases,by target group Approved sub-projectsb Approved grants Group' Number Percent Nakfa Percent High-risk 67 11.6 6,7 13,276 9.8 Low-risk 311 53.8 36,172,579 52.7 Service groups 95 16.5 13381,444 19.8 Unspecifiedtarget 104 18.0 12,136,785 17.7 Total 577 100.0 68,604,804 100.0 Source: IEGanalysis o fzoba-level listsof approvedCMHW projects. a. The targeted groups are providedin Annex E. Examplesof high-risk groups include: sex workers, truck drivers,tourism and hotel workers etc. Examplesof low-risk groups include: the generalpopulation,youth, women, farmers etc. Examplesof servicegroups are orphans andPLWHA. b. Includes16 sub-projectsfor 'unspecified'diseases, mostly for sensitization, from SRSzoba andexcludes one HAMSeT sub-project in Maekelzoba that was solely for malariaand tuberculosis. 3.26 Successive National HIV/AIDS/STI Strategic Plans identified risk groups. 59Coverage o f behavior change interventions and condom distribution among the military appears to have been universal, butfor security reasons data access was limited. In2006, the Ministry o f Labor and Human Welfare estimated that there were 3,500 sex workers nationwide, and that the project-financed interventions reached about 1,000 sex workers. Interventions were also implemented by the MOHHealth Promotion Unit. It i s unfortunate that coverage data was not collected.60 According to interviews duringthe IEGmission three-quarters o f all truck drivers had been reached through awareness campaigns, but again no coverage data were available for verification. Under the CMHRP component a tenth of interventions specifically targeted high- risk groups andmore than halfof sub-projects were targeted at low-risk groups (Table 3-3). 3.27 Publicly distributed and socially marketed condoms increased by 62 percent between 2000 and 2003, after which total condom distribution declined-primarily due to a two-thirds reduction incondoms distributed by the M O H (Figure 3-1). The reasons for the reduction in condom distribution by M O H were two-fold: to reduce wastage, and to encourage private purchase o f condoms. No information on private condom sales was available to assess the responsiveness o f private condom demand. An additional source o f condoms was the Ministry o f Defense; cumulatively 12 million condoms were distributed by the military over the project's duration. The EritreanDefence Force distributed 'condom pouches' as part o fthe military 59. Youth were identifiedas an important risk group, yet surveys show low average levels of sexualactivity and the lowest HIV prevalence. Medianage at first sex is between 17 and 18 years. In 1995,99.4 percentof women aged 15-19 years reported having no sexual partners and in2002 67.6 percenthad never had sexual intercourse(Macro International 1995,2002). 60. The estimationof coveragehas many methodological challenges, as coverageis a function ofprogramreach as well as intensity of exposure. While these challenges are recognized, even imperfect coveragedata couldhave provided a startingpoint for future improvementsand couldhave assistedwith the construction of aresults chain linking inputs and intermediate outcomes. 30 uniform to 90,000 members o f the armed forces. This sent a powerful (and visible) message regardingthe military's commitment to HIV/AIDS prevention. Figure3-1. Condom distribution,2000-2005 Ii j IMOHcondomdistnbution k$# Condomsocial marketmg I 12 10 E - 8 2000 2001 2002 2003 2004 2005(Sep) Source: MOH2007. 3.28 Establishing new VCT centers, purchasing HIV tests, and information campaigns emphasizing the importance o f getting tested were major project outputs. At the start o fthe project VCT was available only ina few large towns, andby 2005 the number o f V C T centers had increased to 17. The number o f VCT clients progressively increased over the life o f the project, from roughly 2,000 in2001 to over 75,000 in2006 an important achievement (Table F- 4 inAnnex F). Contributing to the increase indemand for VCT services was a socially enforced practice o f pre-marital HIV testing encouraged by all the religious bodies.61 Box 3-2 points to some o fthe continuing controversies that exist inthe literature regardingthe effectiveness o f VCT. However, duringproject implementation the expansion o f V C T access was inline with the international best practice at the time. There i s still no consensus on this issue from HIV/AIDS authorities such as UNAIDS and WHO. Box 3-2. Efficacy of voluntary counseling and testing Voluntary counseling and testingfeaturedprominentlyinthe HIViAIDS response inEritrea. The benefitsof counseling andtesting are saidto be: (i)an entry point for care, support andtreatment services for a personwho tests positiveand to learn how to live positively by avoiding infection of other individuals; and (ii) the people for who test negative-the overwhelmingmajority inEritrea-counseling andtesting i s saidto offer the opportunity for the personto "plan one's future" and "commit to avoid risky behaviors." While HIV/AIDS experts, including UNAIDS, havepromoted VCT as a key HIV preventionintervention,the more recent literaturesuggests that its benefits for preventingthe spread ofHIV remain ambiguous. The findings from a systematic reviewby Denisonet al. (2006) ofthe effect o f VCT on risk behavior found 14 studiesthat met the quality controlcriteria and concludedthe following: (i)one randomized controlledtrial showed a significantly greater decrease inunprotectedsex with non-primarypartners among individualsinthe clients receivingVCT as comparedto the clients receivingonly health and also a significant intervention. effect for couples;b(ii) studies examinedcondomuse:two showedsignificant increases inuse; four showed eight mixedresults dependingon partnertype and durationof follow-up; andtwo showednon-significantchanges; (iii) o f the studies that measuredbiological outcomes:one showed significant decreases ingonorrheaprevalence andHIV incidence amongwomen with testedpartners; two showedno significantchanges inHIV or STI incidence; one showedhigherrates of STI/HIV comparing men who did versus did not receivetheir test result. a. Odds ratio = 0.68; p450,000 no comm. CMHRP approved reviewed ("/I Mini- mum Maximum million Nakfa contri- grant/ sub- grant grant Nkfa ($30K) bution project Anseba 144 ' 144 100.0 3,577 968,200 0 2 22 75,450 Debub 238 168 70.6 15211 2,682,179 6 12 0 217,390 Gash Barka 235 210 89.4 27,310 510,000 0 1 18 83,928 Maekel 139 139 100.0 750 2,800,000 2 4 5 148,135 NRS 89a 89 100.0 9,760 4,536,133 5 11 17 812,450 SRS 94b 94 100.0 3,400 945,061 0 7 33 185,37 1 Total 939 844 89.9 750 4,536,133 13 37 95 136,619 Source: IEGanalysis of zoba-levellists of approvedCMHRP projects. a. PMU documentscounted 90 approvedprojectsinNRS. However,on review o f the projectlist, one line number was skipped and the total number approvedis actually89. b. PMU documentscounted 95 approvedprojects in SRS. However, on reviewo f the project list, one line number was skipped and the total number approved is actually 94. 110. IEG discoveredmis-counts intwo of the zobas with 100percent coverage; the number of sub-projects actually approvedseems to havebeen939, not 941. 74 Table E-2.Distributionofsub-projectsby disease and intervention Table E-2a. Distribution of approvedHIV/AIDS sub-projectsby intervention Approved sub-projects Approved grants Intervention Number Percent Nakfa Percent Behavior change communication 221 44.6 14,004,026 27.7 Awareness/sensitization/campaigns 78 15.8 8,455,201 16.7 Care and support for orphans 27 5.5 7,606,989 15.0 Vocation traininghehab for SWs 8 1.6 4,818,933 9.5 Administrative/training 59 11.9 3,494,111 6.9 Home-basedcare and HBC training 50 10.1 3,320,629 6.6' Unspecifiedcare and support 4 0.8 2,876,000 5.7 VCT promotion, construction, expansionor post- test clubs 22 4.4 2,284,097 4.5 Care, financial support, counseling for PLWHA 13 2.6 2,186,224 4.3 Prevention of infection in hospitals 1 0.2 780,83 1 1.5 PMTCT 8 1.6 510,460 1.0 Interventionunclear 2 0.4 107,638 0.2 Incomegeneration 1 0.2 64,000 0.1 Condom distribution 1 0.2 67,500 0.1 Total 495 100.0 50.576.639 100.0 Source; IEG analysis of zoba-level lists of approvedCMHRP projects. Table E-2b. Distribution of approved HAMSeT sub-projects by intervention Approved sub-projects Approved grants Intervention Number Percent Nakfa Percent Awareness/sensitization 40 59.7 13,103,43 1 81.6 .Training/assessments 19 28.4 2,342,711 14.6 BCC promotion through sports events 4 6.0 280,000 1.7 Vocational training 2 3.0 197,975 1.2 Strengthenvillage health committees for malaria& TB 1 1.5 70,000 0.4 Specific interventionnot clear 1 1.5 57,155 0.4 Total 67 100.0 16,051,272 100.0 Source: IEG analysis ofzoba-level lists of approved CMHRP projects. Table E-2c. Distribution of approved TB sub-projectsby intervention Approved sub-projects Approved grants Intervention Number Percent Nakfa Percent Food and DOTS/outreach/case detection, shelter 38 55.1 4,953,188 49.9 Specific interventionnot identified: "TB prevention & control", "outreach", "TB 27 39.1 4,732,637 47.7 project" Training of TB promoters 3 4.3 190,000 1.9 TB "promotion", sensitization 1 1.4 44,013 0.4 Total 69 95.7 9,919,838 100.0 Source: IEG analysis ofzoba-level lists of approved CMHRP projects. 75 Table E-2d. Distribution of approved malaria sub-projects by intervention Approvedsub-projects Approvedgrants Intervention Number Percent Nakfa Percent Source reduction 41 42.3 3,543,179 38.5 Sensitization 14 14.4 3,216,936 34.9 Source reductiordsanitary equipment 27 27.8 1,016,043 11.7 Source reductionandbed-net distributiordreirnpregnation 6 6.2 408,830 4.4 Environmentalcontrols 4 4.1 270,000 2.9 Planttrees 4 4.1 591,120 6.4 Workshou 1 1.o 106.354 1.2 Total 97 100.0 9,212,462 100.0 Source: IEG analysis ofzoba-levellists of approvedCMHRPprojects. Table E-3. Distribution of sub-projects by intervention that are not directly related to control of HAMSeT diseases, by intervention Approved sub-projects Approved grants Intervention Number Percent Nakfa Percent Water supply and sanitation 12 12.9 7,894,973 29.2 Renovationof buildings, libraries,sportsfields 13 14.0 7,226,768 26.7 Constructionof recreation centers & libraries 3 3.2 3,072,279 11.4 Vocational trainingfor vulnerableidestitute women 6 6.5 2,672,127 9.9 Sportingeventsiindoor games/youthclubs 8 8.6 2,019,613 7.5 Training/incentives/&E/PMU/CMHRP administrative 15 16.1 1,462,042 5.4 Antenatal care, delivery, MCH nutrition 25 26.9 1,349,455 5.0 Culturalperformances/activities 10 10.8 1,260,135 4.7 School suuulies for students 1 1.1 84.946 0.3 Total 93 100.0 27,042,338 100.0 Source: IEG analysisofzoba-levellists of approvedCMHRP projects. 0 4: 2 d 0 f: * m v; m" N-d 0 f: m W 0 f: E t sv) UE m 0 E: 9 c1 4, P h *m 2 u M N P P a *- 0 4: a 2e? 25 ffi 2 ! u 0 4: Hea - c m E 0 E 0 d .I P 2 a a9 *L v1 M 2b u ij F 77 Annex F. Sectoraldata and Communicabledisease outcomes Figure F- 1. Improvement in IMR, by wealth quintile 100 80 8 60 0- - b 40 4 20 d E O Q1 4 2 4 3 44 Q5 Total Wealth Quintile Source: Macro International 2002; Gwatkin et al. 2007. Table F-1.Referrals to Barentu and Mendefera hospitals, 2004-2007 Referrals to:' 2004 2005 2006 2007 Barentu Hospital from health centers inzoba 177 294 from other hospitals inzoba 15 46 2 8 Mendefera Hospital from health centers inzoba 315 299 384 409 from other hospitals inzoba 109 239 211 238 Source: MOH2008a. a. Data on the reasons for the referrals are not collected by the HMIS and were not available for analysis. 78 Table F-2. Burden of Disease,2001 Share oftotal Share of total estimateddeaths estimatedDALYs (percent) (percent)" Communicable, maternal, peri-natal and nutritional conditions 70 69 Infectiousandparasitic diseases 44 41 Tuberculosis 5 4 STIs excludingHIV 0 1 HIV/AIDS 16 13 Malaria 6 7 Diarrhealdiseases 6 6 Childhood-clusterdiseases 4 5 ) Respiratory infections 16 15 Maternalconditions 2 4 Peri-natalconditions 6 7 Nutritional deficiencies 1 2 Non-communicable diseases 22 22 Cardiovasculardiseases 10 3 Injuries 7 9 All causes 100 100 Source: WHO, httD://www.who.intlwhosislenl a. Disability-adjustedlife years (DALYs) is a measure for the overalldisease burden, and is designedto quantifythe impact of prematuredeath and disability by combiningmortality and morbidity into a single, commonmetric. Table F-3.Public sector health expenditure, 1996-2005 1996 1997 2000 2001 2002 2003 2004 2005 Publichealth expenditure($000) Government 12,144 15,414 10,798 10,697 11,943 11,124 13,473 12,353 Recurrent 10,515 13,244 10,209 9,813 10,768 10,173 10,055 10,862 Capital 1,628 2,170 589 884 1,175 951 3,418 1,491 Donor 26,558 21,945 35,570 31,035 Total (Government+ Donor) 38.352 33.740 49.043 43.387 Per capitapublic health expenditure($) Usingpopulationestimates in MOHHMIS 10.2 9.3 14.2 13.2 Usingthe populationestimates in World Bank PER '7.0 9.0 10.0 8.0 Source: World Bank 2008b. 79 Table F-4.VCT coverage and HIV prevalence rate at VCT sites, 1999-2007 Peopletested HIV prevalence rate(%) 1999 1.510 2000 21010 2001 2,227 2002 10,659 2003 32,292 2004 47,663 3.9 2005 69,121 3.4 2006 75,795 3.5 2007 80,706 3.2 Source: MOH2008b. Table F-5.Level of knowledge of HIV/AIDS and STIs, marital status and sexual activity among young people 15-19, prior to HAMSeT interventions Men 15-19 Women 15-19 1995 1995 2002 (n=237) (n=1,129) (n=2,001) Source of information: Radio 78.4 73.8 Newspaper 41.6 26.1 TV 34.1 23.3 School 28.0 18.6 ................................................................................................................................................................................................................................................................ Friendhelathe 31.8 30.1 Knowledge ofHIV Ever heard ofAIDS?' 89.4 82.1 97.2 AIDS can beavoided 97Sb 95.4' 92.1 Can be avoidedby condomuse 45.9b 40.4' 78.5 Can be avoidedby havingonly one partner 39.8b 46.8' 90.9d A healthy-lookingpersoncan have AIDS 71.3b 67.3' 80.3 ................................................................................................................................................................................................................................................................ Knows someone with HIV/AIDS 7.3b 40.4 Knowledge ofSTIs No knowledgeof STIs in menor women 60.0 Source: MacroInternational1995,2002. a. Promptedquestion. b. Sample size is 212 (those who have heardof AIDS). c. Sample size is 926 (thosewho have heardof AIDS). d. Includeslimit number of partners, limit sex to one partnerhe faithful to one partner, one partner who has no other partners. 80 Table F-6.HIV knowledgeand risk perception,2006 Knowledgeofhow HIV can beprevented Self-perceptionofHIV risk Abstinence faithfUl Being Condomuse No risk Highrisk National(2006) 51.4 49.1 57.3 47.2 26.6 Anseba 48.7 47.7 49.0 45.4 25.4 Debub 49.0 60.4 80.1 69.8 20.6 GashBarka 47.1 52.4 52.9 42.4 31.9 Maekel 53.8 53.3 77.1 45.6 32.9 NRS 55.0 41.8 39.7 36.2 29.7 SRS 60.6 29.7 42.1 38.4 14.6 Source: MOH 2006c. 81