Report No. 28963-ET Ethiopia A Country Status Report on Health and Poverty (In Two Volumes) Volume I: Executive Summary July 2005 Africa Region Human Development Ministry of Health Ethiopia The World Bank Document of the World Bank HICES HouseholdIncome, Consumptionand TGE TransitionalGovernment of ExpenditureSurvey Ethiopia HP HealthPost TT TetanusToxoid HSEP Health Services ExtensionPackage USMR Under-fiveMortality Rate HSCSR HealthSector CountryStatus Report UNESCO UnitedNationsEducational, Scientific and Cultural Organization HIV HumanImmunodeficiencyVirus UNFPA UnitedNationFundfor PopulationActivities H S Health Station WB WorldBank IEC Information,Educationand WBCAS The World BankCountry Communication Assistance Strategy IMR Infant Mortality Rate WDR World DevelopmentReport IUD IntrauterineDevice WH Weight for Height LDC Least DevelopmentCountries WHO World HealthOrganization MBB MarginalBudgetingfor Bottlenecks WMS WelfareMonitory Survey MCH MatemalandChildHealth MDG MillenniumDevelopmentGoals MEDAC Ministryof EconomicDevelopment and Co-operation EXECUTIVE SUMMARY A countryintransition With a very low resource base and rapidly growingpopulation,Ethiopia is one of the poorest countries in the world. 1. Ethiopia i s one o f the poorest countries inthe world. Its per capita o fUS$ 100 (US$668 inpurchasing-PPP-parity terms) in2002 compares poorly with the average per capita GNIof US$450 (PPP US$ 1,683) for Sub-Saharan Africa (SSA) in2002. The disparityis even morepronouncedwhenjuxtaposing Ethiopia's per capita GNIto the world average o fUS$5,080 (PPP US$7,415) (World Development Indicators, 2003).' Millions of Ethiopianscontinue to live inabsolute poverty. The poverty head-count declined only marginally fiom 45.5 percent in 1995 to 44.2 percent in2000. While the population grew by 9 million duringthe same period, the number o fpeople living in absolute poverty increased steadily. Thepolitical system is in transition, increasing decentralization to the district level. 2. Emergingfrom civil war in1991, Ethiopiareplacedthe Dergregimewith a federal structure o f government, introducing a new constitution in 1994. The country's first multi-partyelections were organized in 1995. Ethiopia now has a parliamentary federal government administering nineregional states and two administrative councils (Addis Ababa and Dire Dawa) which are sub-divided into 560 woredas (districts). Followingthe country's decentralized policy, these woredas represent the basic units o f planning andpolitical administration. Operatingwithin the jurisdiction o fthese districts are peasant associations known as kebeles. Ethiopia has progressively undertaken economic reforms in the last tenyears. 3. Ethiopiahas operated a free-market economy since 1991. The Government introducedan Economic Recovery and Structural Adjustment Programin 1992 to stabilize the macro-economic framework and encourage private sector participation. These economic measures paidpositive dividends, reversing years o fpersistent decline in real GDP which grew on average by 5.8 percent from 1992/93 to 2001/02 while population growth was about 2.7 percent over the same period. The Ethio-Eritrean Border conflict affected GDP growth rates from 1998 to 2000. In2003, the economy faced a sharp decline and a negative growth rate as a result of the drought which affected 14 million people. 'By anothermeasure of GDP per capita (constant 1995 US$), Ethiopia has only US$ 116 compared to the World average ofUS$ 5,63 1and Sub-Saharan average ofUS$ 564. -2- Although the economy remains highly dependent on the agricultural sector, the service sector has grown to such extent that it is now the major contributor to overall GDP. 4. The agricultural sector continues to be an important contributor to the overall Ethiopian economy although its contribution has decreased from 54 percent in 1982 to 40 percent in2002. Agriculture employs over four-fifths o f the country's labor force. It accounts for 90 percent offoreign exchange earnings, andthe country ishighly dependent on its major export, coffee. Agricultural productivity, however, remains low. The sector is vulnerable to external shocks such as droughts, which have occurred every three years duringthe past decade. The humanbase for agricultural development i s also largelyilliterate and inadequately equipped with modern skills, inputs, and equipment. The services sector has emerged as the major sector inthe economy, growing from 33.2 percent inthe 1980s to 47.6 percent in2002. Industryrepresents only about 11% of GDP. A young and rapidly growingpopulation isputtingpressure on agricultural lands. 5. Life expectancy at birth (42 years) i s slightly lower thanthe SSA average o f 45 years. In2003, Ethiopia's population was 69.1 million. It is the second most populous country in SSA. The population grew by an average o f 2 million annually between 2000 and 2005, representing a rate o f 2.7 percent, which i s slightly higherthan .the SSA average of 2.5 percent. The population is largely rural (83 percent). However, with an urbanpopulation growth rate o f4.1 percent comparedto a growth o fonly 1.9 percent inrural areas, the urban-rural breakdown o f the population i s slowly changing. Moreover, urbanpopulation growth is fuelled partlyby internal migration. The population i s young, with 44 percent under the age of 15. Such a structure results in a highdependency ratio as well as a future rapid exponentialpopulation growth. Ifthis growth does not decline inthe coming years, it is expected that the populationof Ethiopia will double inabout 25 years. Population density is moderaterelative to other SSA countries, although it i s twice as highas the SSA average. However, it is very highinthe highlands, and lowest inthe eastern and southern lowlands. About 23.2 percent o fthe population is concentrated in9 percent o f the land areas. This puts considerable pressure on cultivable lands and contributes to environmental degradation. Access to social services is very limited. 6. Despite efforts made bythe Government to ensure basic social services, access to services such as water and sanitationis limited. Only 15 percent o f Ethiopians have access to improved sanitation compared to the SSA average of 55 percent. Access to clean drinkingwater is slightly higher at 24 percent but i s still lower than the SSA average o f 55 percent.2 Fifty-ninepercent o f the adult population is illiterate, measuring comparatively higher than the SSA average o f 36 percent. Females have a higher rate o f illiteracy than males. The primary school enrollment rate is 49 percent, falling also The survey on Environmental Health Profile of Regions and Selected CitiedTowns in Ethiopia (2002/03) cited in the MOWPPD report (1994) indicates a lower number for access to improved sanitation (11.5 percent) but a higher figure for access to clean water (28.4 percent). -3- below the SSA average. More than 50 percent o f Ethiopians, particularly those living inruralareas, remain food insecure. A traditional society where women 3 social status is still low. 7. While the Ethiopian constitutionrecognizes the equal rights o f women and men, the traditional societal structure keeps women ina vulnerable position. Traditional harmfulpracticesare commonwith 80 percent ofwomen havingundergone circumcision. A heavy workload (on average, Ethiopianwomen work 15-18 hours per day andmanyrural domestic tasks are highlylabor intensive) andearly marriage (the average age o f women at first marriage was 17.6 years in 1998) are common occurrences. Limitedstudies andpolice andmedia reports suggest that violence against women is quite highand increasingevery year (SCGA 2004). About 25 percent o f Ethiopian women have experienced rape (W. Post 2004). Women still occupy a very small percentage ofkey governmentdecisionmaking positions: 7.7 percent inthe House o fRepresentatives and 13 percent inregional councils in2000. The Government has renewed its commitment to improve health outcomes. 8. The Government of Ethiopia has recently confirmed its commitment to accelerate progress on matemal and childhealth outcomes. A reduction inchild and maternal mortality rates is amongthe key objectives o fthe Ethiopia Poverty Reduction Strategy (PRSP) published in2002. This strategy outlines the Government's key policy objectives and strategic options for the next five years. One key PRSP strategic option for reducing matemal and childmortality i s to expand the provision o f essential health andnutrition services to the country's rural poor. Health outcomes are slowly improvingbut remain low, particularly among rural dwellers and the very poor Child mortality has declined, albeit slowly, in the last decade. 9. Starting inthe 1960s, Ethiopia has shown a slow but steady reduction inchild mortality. Infant and under-fivemortality have continued to decline over the past 25 years with a more pronouncedreduction inthe last decade. Under-fivemortality i s presently 21 percent lower thanit was five to nine years ago. Yet, overall, infant and under-five mortality rates remain very high. Between 1995 and 2000, nearly one inevery ten newborns did not survive to celebrate its first birthday, and one inevery six children died before its fifth birthday (Figure 1). -4- Compared to other countries in SSA, Ethiopia's relativeperformance in reducing child mortality is improving. 10. In1960,the Ethiopianchildmortality ratio was higher thanthe SSA average, but slowly over time, it has been improving. Ethiopia's childmortality reduction performance ratedwell duringthe 1980s and 1990swhen many other SSA countries showed stagnation or even increase ininfant mortality. Ethiopia's performance relative to its per capita income i s also favorable. Ithas lower infant and under-five mortality rates, as well as lower levels o fwealth-based inequities relative to other countries with a similar per capita income (Figure 2). Figure 1: Trends in under-five mortality inEthiopia compared to other regions of the world 300 - - S u b - S a h a r a n Africa --+-South Asia 250 - +Latin A m e r i c a a n d C a r i b b e a n * D e v e l o p i n g countries 1 9 6 0 1 9 7 0 1 9 8 0 1 9 9 0 1 9 9 5 2000 Year Source: Unicef, 2002 Figure 2: IMR and poorlrich inequalities in different countries 4.5 + Peru 4 Bolivia 4 Turkey -= 3.5 - 4 Indonesia Egypt 2 3.0 - 2.- % 4 Paraguay*B%kican Rep 2.5 - 4 India 4 CAR ul Colombia4 44Vietnam Philippines 4 Mor82ya E. -1:m 2.0 - 4 Nicaragua 4 Cameroon 4 Madagascar Mozambique 4 - 1.5 - P) Kazakhstan 4 Zimbabwe 4 Malawi 1.0 - *+ Uzbekistan 4 Namibia 4 Chad 0.5 - 0 20 40 60Population Average (IMR) 80 100 120 140 160 -5- At the current pace, however, reaching child survival Millennium Development Goals (MDGs) will be challenging. 11. Between 1990 and 2000, the rate o f decrease inunder-five mortality has been 1.9 per 1000 live births. This contrasts with the estimated rate o f decrease o f 5.2 per 1000 live birthsneededby Ethiopia to reachthe child survival Millennium Development Goals (MDGs). By 2003, it would have to reduce its under-five mortality rate by 7.4 per 1000 live births inorder to achieve the child survival MDGs by 2015. This is a tremendous challenge given past trends andthe extent to which the needs for child survival remain unmet (Figure3). Figure3: AchievingMDGfor child survival 250 in- c -Achieve MGD E r (Decrease of 5.2 .-Y - 200 percent points per 0 Year) z 0 L 150 v) -Current Trend 5m (Decrease of 1.9 I? 0) percent points per - .- 0 100 Year) 0 +Required trend (7.4 50 G percent point per 5 -0 year from 2000) 5 0 O $ c : cz c % E a % x 82 8 z % s % s N g Ng N ~ Years Moreover, inequities in U5MRs and IMRs remain signiJicant. 12. Under-fivemortality rates (U5MRs) are highinall regions andamong all socio- economic groups (Figure 5). On average, children from more affluent householdshave slightlylower mortality rates. Infantandunder-five mortality rates are 16percent and 31percent higherrespectively among children from the poorest quintile compared to children from the richest quintile. The urbdrural difference is more marked: inurban andrural areas, infant mortality rates (IMRs) are 96.5 and 114.7, respectively and U5MRs are 148.6 and 192.5, respectively. Moreover, mortality rates vary between regions. Addis Ababa has the lowest IMR (81) andU 5 M R (113), while Gambellahas the highest IMR (123) andU 5 M R (233). There do not seem to be any gender differentials in child mortality, although female children fare slightly better than male children. -6- Figure 4: Regional variation in IMR and USMR in Ethiopia I O I M R I U 5 M R I The incidence of illnesses @e.,Acute Respiratory Infection and diarrhea) contributing to avoidable deaths is higher in Ethiopia compared to other SSA countries. 13. The highrate ofmortality i s partlydue to living conditions and a highincidence o f illness. The prevalence s o f Acute Respiratory Infection (ARI), at a rate of 24 percent, anddiarrhea amongunder-five children inEthiopia is higherthan its SSA neighbors. On average, children under five years o f age experience two episodes o f serious illness per year. The difference inillness incidence rates betweenrich and poor children measures as a narrower margin inEthiopia than inother countries. Yet the prevalence of diarrhea is higherinthe poorest quintile (25 percent) comparedto the richest quintile (19percent). In Ethiopia, diarrhea andpneumonia are the main causes of early death among children, differingfrom the average SSAprofile. 14. While malaria i s the leading cause of total morbidity andmortality inEthiopia andtherefore has a detrimental impact onlabor productivity andeconomic gr~wth,~is it estimated to represent only 4.5 percent o f the causes o f child mortality. According to the recent Lancet series estimates that are validated by international experts, most deaths among childrenunder five years inEthiopia can be attributed to pneumonia (28 percent) anddiarrhea (24 percent)--disappearing causes o f death inmany poor countries. On the other hand, measles's contribution (2.2 percent) to early deaths has declined. This decline probably contributed to the reduction inchild mortality inthe 1980s and 1990s. However, HIVhas emerged as a growingcause o f early childhood death (6.2 percent). About 68% (>46 million people) of the total population is at risk o f acquiring malaria infections (Ethiopia Rollback Malaria Consultative Mission Report, 2004). Majority o f Ethiopia's population lives in the over-crowded highlands, mainly due to the high prevalence o f malaria and other dangerous tropical diseases in the lowland regions. Hence, malaria has huge negative effects on labor productivity and economic development. It also prevents Ethiopia from realizing the growth potential o f its lowlands, and contributes to the population and environmental issues in the highlands (World Bank, CEM 2004) -7- High child malnutrition rates in Ethiopia present a significant obstacle to achieving better child health outcomes. 15. Although stuntinghas declined by about 14.7 percent from 1995/96 to 1999/2000, Ethiopia still has one o fthe highest malnutrition rates in SSA andinthe world (higher than India and Bangladesh and similar to Nepal's). Moderate to severe stunting is 51percent, while severe stuntingi s 26 percent. Forty-sevenpercent o f children under five are moderately to severely underweight. U r b d r u r a l differentials are not significant butregional differencesinchildmalnutritionareprominent (Figure5). A multivariate analysis confirms that regional differentials are significant, with Tigrayhavinghigher stuntingand undeheight rates, while Somali andGambella havehigher wasting rates. Income also has a more significant impact on nutritional status than on childmortality. Severe underweight is nearly 260 percent higher among children from the poorest quintile compared to childrenfrom the richest quintile. Figure5: Regional differencesinchild malnutritioninEthiopia Rn . ElModerate to severe underweight 50 - =Moderate to severe stunting .. ..- __... 40 - 30 - 20 - 10 - 0 7 - Heavy workloads andpoor diets combined withfrequent pregnancies also have an adverse impact on women 's nutritional status. 16. About one out o f three women and one out o f four mothers o f childrenless than three years old have Body Mass Indices (BMI) that are less than 18.5 indicating that the level of chronic energy deficiency among adult women is relatively high in Ethiopia compared to other SSA countries. Among 17 countries surveyed by the Demographic and Health Survey (DHS) from 1998-2002, Ethiopia perfoms poorly, having the second highest percentage ofmotherswho fall below the BMIcut-off o f 18.5 (Figure 6). -8- -9- Maternal mortality is high, and reaching the MDGgoal is daunting. 19. Information on matemal mortality is scarce but indirect evidence suggests that the rate is very high. According to 1995 WHO estimates, the adjusted matemal mortality ratio (MMR)inEthiopia i s 1800per 100,000 live births, a highfigure compared to other SSA countries and countries with similar levels o f GDP. The MDGgoal is to reduce the MMRbythree-quarters by2015. This meansthat Ethiopiawillhaveto reduceits MMR to approximately 450 per 100,000 live births, in2015. Nearly all countries with twice as muchGDPper capita as Ethiopiahavebeenunable to decreasetheir MMRbelow 400 per 100,000. Therefore, achieving the MMR-related MDGwill be a particularly daunting challenge for Ethiopia. TheHIV epidemic has spread rapidly over the lastfew years and tuberculosis (TB) is also widespread. 20. The first AIDS case was detected inEthiopia in 1986. The prevalence of HIV remained very low inthe 1980sbut spreadquite rapidly duringthe 1990s. About 6.6 percent o fthe adult population in2002 haveHIV/AIDS, andthe epidemic is considered generalized inEthiopia. Bythe end o f 2001, there were 2.1 million children andadults inEthiopia livingwith HIV/AIDS. AlthoughEthiopia constitutes only 1percent ofthe world's population, it contributes 7 percent o fthe world's HIV/AIDS cases. Interms o fnumber of infected persons, Ethiopia ranks fifthinSSA--after South Africa, Nigeria, Kenya and Zimbabwe. Tuberculosis (TB) is also widespread. Reported TB accounts for 3.1% o f all deaths. The incidence ratio o f all forms o fTB in2000 was 397/100,000. This ratio is slightlyhigher thanthe SSA average o f 354/100,000 and greater than the average o f233/100,000 for low-income countries. Approximately 30% o f all TB cases are also HIV positive. Householdand communityfactors affectinghealth: knowledge,attitudeand practices Short birth intervals, high birth order, low birth weight, the young age of mothers, and being part of certain religious groups, are all strongly linked to high child mortality levels. 21. An analysis o fthe proximate and underlying determinants o funder-five mortality indicates the significant link between fertility (birthinterval) inaffecting child mortality. Other strong associations are found with religion (lower child mortality rate among Orthodox Christians) andmother's age (lower mortality rate among children o f older women). Low birth-weighti s also a key factor inboth infant and under-five mortality, death beingmore prevalent among smaller children.Wealth has a moderate effect on both infant and under-five mortality, while mother's education (secondary level and up) is significantly associated with infant mortality but not under-five m~rtality.~ The draft Poverty Assessment (World Bank 2004) fmds a highly significant relationship between mother's education and under-five child mortality. It specifies mother's education inyears as a continuous -10- The analysis also indicates that infants whose mothers received ANC tetanus vaccinations while pregnant have a lower likelihood o f dying. Malnutrition is largely associated with low levels of income and education, and birth intervals. 22* Similar analysis conducted on malnutrition shows that children o f educated and more affluent mothers5have a lower likelihood o f being stunted or underweight, but there are no differences inthe proportion o fwasting. Higherparity children are more likely to be underweight. While exclusive breastfeeding is relatively high, Ethiopian households lag behind when it comes to other householdpractices including use of both iodized salt and bed nets. 23. Early, exclusive andprolongedbreastfeedingcontributes largely to children's nutritional status andEthiopian mothers fare well incomparisonwith neighboring countries. Exclusive breastfeeding o f children less than 4 months old is 63 percent, among the highest inSSA. Fiftypercent of childrenare breastfedwithin one hour o f birth. Timely supplementary feeding o f6-9 monthsold childrenis 77 percent. Onthe other hand, other householdpractices are less favorable, such as, for example, the use of iodized salt, which is still very limited. Less than 30 percent o f households use iodized salt, and families from the poorest quintile and living inTigray have a lower likelihood o f usingiodized salt. Mother's exposure to media and education are positively associated with iodizedsalt intake. 24. Although approximately 68 percent o fthe total Ethiopianpopulation i s at risk o f acquiring malaria, bednets are still largely unusedinEthiopia. In2000, only 1 percent o f households owned a bednet, out o fwhich only 17.7 percent o f the nets were insecticide treated. Even inhighmalaria prevalence areas such as Afar and Gambella, only 32 and 12percent o f households, respectively had a bednet. Less than 5 percent o f women inendemic areas were sleepingunder a bednet. The use of oral rehydration therapy (ORT) is much lower than in other poor countries, largely explaining the high level of mortality due to diarrhea. 25. The percentage o f childrenwith diarrhea who receive ORT inEthiopia is one o f the lowest inthe world. Use of ORT duringdiarrhea episodes i s more than five times higher inhouseholds from the richest quintile comparedto those from the poorest variable compared to the specification used in this report which is based on education categories (no education, primary, secondary or higher). The recent Poverty Assessment (WB 2004) also finds a significant relationship between education and children's nutritional status. Inparticular, the effect o f female education is about twice as high as that o f male education, though bothhave a positive effect. The results are similar to those o f Christiaensen and Alderman (2003) which indicate that householdresources and parental education are the maindeterminants o f child nutrition inEthiopia. They also findprices to be a significant factor although we did not include them inour analysis. -11- quintile. Sixty-two percent o f children inthe poorest quintile and 36 percent inthe richest quintile do not get any home-based treatment. Fortypercent o fwomen in the poorest quintile and 13 percent o f women inthe wealthier quintile have not heard about ORT. Overall, the potential for improvement inthe use o f this low technology and inexpensive interventionis very large. Immunization rates are improving but remain relatively low. 26. Ethiopia's immunizationperformance is mixed. The percentage o f 12-23 months old infants who have received one or more o f the EPIvaccines i s highat 83 percent. However this percentage largely reflects the coverage achieved through the polio eradicationprogram. Evenwhen HMIS data for 2000/01 and 2002/03 are used, Ethiopia's DPT rates o f42 percent and 50.4 percent still place Ethiopia among the relatively low performers by SSA standards, behind Malawi, Zambia, Benin or Ghana. There i s a highdrop-out rate between the first and subsequent vaccination-- DHS 2000 indicates a drop-out rate o f about 23 percent. However, HMIS 2000/01 data show slightly lower drop-out rates for 7 regions, ranging from 5 percent inTigray to 43 percent inAfar while HMIS2002/03 data for 10out of 11regionsindicate an average drop-out rate o f about 16percent (drop-out rates ranging from 4.6 percent inTigray to 43 percent inSomali). These results suggest that while immunization rates areimproving, itis difficult for the health system insome regions inEthiopia to ensure continuity o f services. 27. Various factors, including the child's birthorder, mother's education, her partner's employment status, wealth, andregion, significantly affect a child's immunization status. Tigray andAddis are the best immunization performers. Eighty percent o f children living inAddis are immunized for DPT 3 comparedto only one percent o f children living inAfar. Around 17 percent o f children from the poorest quintile have not been immunized compared to 6 percent inchildren from the richest quintile. Muslimchildren have a greater likelihood o fbeingimmunized compared to children from other religions. Boys have a higher likelihood o f getting one to three doses o fDPT comparedto girls. Theuse of Vitamin A supplementation is high. 28. Along with a highlevel o f immunization against polio, Ethiopia's vitamin A supplementation is among the highest in SSA. Approximately 80 percent o f children are covered, with the Tigray region having the highest level o f coverage. Children from the poorest quintile have a lower likelihood o f Vitamin A supplementation. Mother's exposure to media and educational level are positively associated with vitamin A intake. Use of antenatal and delivery care is very limited even among wealthier groups. 29. Ethiopia's performance on antenatal care, including tetanus toxoid vaccination and delivery care i s one o fthe worst in SSA. Only 26 percent o f Ethiopianwomen receive antenatal care (Figure 7) and only 5 percent avail o f assisted deliveries by a skilled professional (DHS 2000). MOHH M I S data also present very low figures: -12- antenatal care coverage o f 34.7 percent in2000/01 and 27.4 percent in2002/03. Attendeddelivery rates - while higherthanthose cited inthe DHS-are also very low at 10percent in2000/01 and 9 percent in2002/03. Figure7: CountriesrankedbyANC by a medicallytrained personinthe poorestquintile 100.0 - E 1 90.0 - 6 6 80.0 - n Pe - I" I 70.0 c W 3 60.0 -73 - U 50.0 - ,i m .- 40.0 - .a v) t 30.0 - 0 g 5* 20.0 - . * IRichestQuintile 0 10.0 - Country Source: Gwatkinet al. (2002), World Bank 30. The use ofthese interventionsis low even inurbanareas and almost negligible among the poorest quintile. Wealth-based differentials are most marked for delivery care: among the poorest quintile, less than 1percent o f deliveries were attended by a trained professional. Evenwithin the richest quintile, the rate is extremely low, at 24 percent. Regional differences are striking: inAddis Ababa, 74 percent o fwomen receive tetanus toxoid duringANC compared to only 16 percent inAfar. Sixty-nine percent o f women avail o f assisted delivery care inAddis Ababa comparedto only 3 percent inAmhara. Women exposed to mass media andwith a higher education are more likely to receive matemal health care. Onthe other hand, women with a higherparity have a lower likelihood o freceiving assisted delivery care. In rural areas, there is a large level of unmet needs infamilyplanning, especially among thepoor. 31. Between 1990 and 2000, Ethiopia's contraceptive prevalence rate (CPR) increased from 4 to 8 percent among married women.6 Yet it i s still among the lowest in SSA for that period. The urbancontraceptive prevalencerate or CPR (36 percent) is nine times higher than the ruralrate (4 percent). The rich/poor differential i s also very HMISinformation for 2000/01indicates a higherrate of 18.7 percentandan ever higherrateof21.5 percentin2002/03 -13- marked, with 29 percent o fwealthier women and only 2 percent ofpoorer women using any contraceptive method. Modern contraception is little used even amongricher groups (12.8 percent). Some regions (Amhara, SNNPR and Tigray) have higher CPRs. Accordingto a 2004 r e p ~ r tthis is largely attributable to the role o f development , ~ associations, including the use o f community based distribution agents. 32. Knowledge o fmodemmethods and women's approval o f family planningare very high(81 percent), even inrural areas (85 percent o f married women). It i s striking that the national level o f unmet need, i.e. the proportion o f women who want to space (22 percent) or limit (14 percent) births but do not do so is large (36 percent). This i s the highest unmet needinSSA. It i s even highinrural areas (37.3 percent). This suggests a failure of the supply side to respond to the demandfor family planning. Large variations are found between regions. Unmet needs are higher inAmhara (41 percent), Oromia (36.4 percent), SNNPR (35.5 percent) and Tigray (28 percent). Addressingthe lack o f family planninginthe highfertility rural areas o f these regions which represent 80 percent o f the country's population appears to be a priority, inorder to reach Ethiopia's populationpolicy goal o f44 percent CPR by the year 2015. Demand sidefactors play an important role infamily planning. 33. Inruralareas, amongthepoor, the older andless educated, andinlarge households, women are less likely to approve o f family planning andmore likely to say that having five children i s ideal. Knowledge o f modem methods i s also more likely among Orthodox Christians than amongMuslims. Significant regional variations are also observed. Compared to Tigray, women inall other regions were less likely to know about a modemmethod or a source, or approve o f family planning,or have a husband who approves of it. Generally, the husband's approval seems to be a major constraint. While 60 percent ofwomen approve o ffamily planning, only 34 percent o fhusbands do. Only amongthe wealthiest 20 percent o fthe population can we find a significantly higher proportion o fmenwho approve o f family planning(close to 60percent). The involvement o fhusbands i s essential inorder to improve family planning use. Awareness of HIUAIDS has increased dramatically but remains low among the rural poor andyounger women. 34. A veryhighpercentage o fEthiopianwomen (85 percent) andmen(96 percent) have heard o fHIVIAIDS. This is largelydue to government efforts because most people cited community meetings as their main source of information on HIV/AIDS. However, urbanandwealthier women aremore aware o fHIV/AIDS than the ruralpoor. It is striking that young women are less likely to know about HIV/AIDS, although according to the latest data they are the ones most exposed to new infections. 7RoleofNGOs inProviding ReproductiveHealthServices -14- Over-all care seeking behavior is low; it is even lower among the poor andfor women. 35. W M S Data from 2000 show that Ethiopian households seek care in about 41 percent o f illness cases (table 1). While the poor to rich ratio for incidence o f illness is 1.O, the poor/rich ratio for seeking care is 0.68, rangingfrom 0.54 inAfar to 1.12 inTigray. Althoughfemales report higherlevels ofmorbidity, they are less likelyto seek care as compared to males (Figure 8). Table 1: Distributionof individualswho had a healthproblem and who consulted for treatment in differentregionsof Ethiopia Poorest 2nd Middle 2nd Richest Overall Poorto Poorest Richest average richratio Region Tigray 51 43 43 46 45 45 1.12 Afar 33 38 46 51 62 48 0.54 Amhara 20 23 26 31 31 27 0.65 Oromiya 46 46 48 48 49 47 0.94 Somali 46 46 48 48 49 40 0.94 Benshangul 55 61 60 65 78 60 0.71 SNNPR 36 39 47 47 53 43 0.68 Gambella 36 39 47 47 53 75 0.68 Harari 36 39 47 47 53 51 0.68 Addis 36 39 47 47 53 64 0.68 Abba DireDawa 36 39 47 47 53 46 0.68 Total 36 39 47 47 53 41 0.68 Figure 8. Illnessinthe last two months, and percentof those illseekingcare by gender 0Hadillness HConsulted 50 44 45 41 40 38 35 - 30 -26 C 25 n 0) 20 15 10 5 0 Male Female Total Source: WMS 2000 -15- Care-seekingfor children is very low by international standards, even among the richest groups. 36. When looking at the specific pattem o fhealth-care use for children (ARIanddiarrhea), care-seeking levels appear extremelylow (table 2). The level o fuse o fthe richest quintile inEthiopia is lower than that o fthe poorest quintile inmany SSA countries. Twenty one percent o f children inthe poorest quintile and 43 percent among the richest quintile sought treatment for diarrhea, among the lowest rates inthe world. Table 2: ARI and diarrhea incidence and care-seeking behavior (2000) Uganda 61.4 48.2 Zambia 70.7 53.9 Urban households have a significant advantage in terms of geographic access to health facilities. Residence affects outcomes most while income affects use of services most. 37. Inabsolute terms, outcomes andutilizatian ofhealthservices remain low even inurbanareas andamongthe richest households. Urban-rural differences are greater than rich-poor differences interms o f illness prevalence suggesting an important effect o f lack o f sanitation and other environmental factors on ruralwomen and children's health status. Despitebetter access to services for urbandwellers, differences inutilization (for example, coverage rates for fully immunized children, DPT3, ANC, assisted deliveries, and family planning) are higherbetweenpoor andrichhouseholds than urban andrural dwellers (Figures 9 and 10). J c3 t t--I .3 1 -18- Public andprivate health services are used equally, but mainly by wealthier groups. 38. Nearly45 percent sought care inapublic facility, while the rest sought care from a private facility. Except for the richest quintile, health stations and clinics seem to be the mainproviders o f care, followed byhealth centers. Bothpublic andprivate hospitals are frequented more often by the richest quintileo f households. Households in the poorest quintile are more likely to use public clinics, pharmacies and other trained private providers instead o fpublic hospitals (Figure 11). The poor/rich ratio i s lowest in the case o f treatment o f a sick child and for immunizations, while it i s highest for obtaining information about sexually-transmittedillness (Table 3). Figure 11: Health facility-wise distribution of utilization by wealth quintiles 0Poorest 02nd Poorest IIMiddle 0 2 n d Richest Richest Others Pnarmacy Other lra ned MassonlMGO Pr valeHospiC'i p1.0 .C POSl P A c C nc P A c hC PUD IC Hospila IO I 13 Table 3: Reasonsmentioned for use of health facility for various services by wealth quintiles Poorest 2nd Middle 2nd Richest Average Rich-to-poor Poorest Richest ratio Treatment o f sick child 26.4 24.0 29.2 37.3 38.4 30.9 1.5 Immunization 19.8 18.5 21.6 30.1 32.1 24.3 1.6 Family planning 6.4 5.1 8.9 11.6 19.0 10.2 3.0 Prenatal, postnatal, and delivery 4.1 4.1 6.5 7.0 12.1 6.7 2.9 Information on STI prevention 4.5 4.3 7.8 9.8 19.2 9.1 4.3 Information on breastfeeding and 4.1 3.2 6.4 6.9 14.3 7.0 3.5 infant feeding Any service 37.4 35.0 41.9 52.6 56.3 44.4 1.5 -20- The two main reasonsfor choosing afacility are availability/access (38percent) and quality of care (23 percent). Figure 13. Distributionof reasonsfor choosing a facility Positive experiencefrom previous consultations ERecommendationfromotherperson OAvailable nearby ECheaper than other I free of charge EBetterqualitythanother OShort time of waiting Others Total Pharmacy Mission/ NGO Pnvate Hospital Public health post Publicclinic Publichealth center Public Hospital 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% The performance of health services Transportation costs represent a highproportion of health care costs. 42. The average cost o ftransportationto a health provider is Birr 15. The cost o f transportation is highest for government hospitals (Birr 22) and lowest for health posts (Birr 5). The cost o ftransportationvaries widely across regions: it is as highas Birr73 inGambellaandas low as 4 BirrinDireDawa. Public sewices are experiencing an extensive decentralization process. 43. Ethiopia has gone through two stages o f decentralization over the last few years. Duringthe first stage, functions were decentralized from the centralto regional level. In the new health sector organizational framework, the Ministryo fHealth(MOH) i s mainly responsible for the formulation o fpolicies and supervision o f implementation, determination o f standards, issuance o f licenses and qualification o fprofessionals, establishing standards for research andtraining, and coordination o f external loans and grants. Duringa second stage, decentralizationwas expanded to the woredas (districts with an average population o f about 100,000), which are now receiving block grants to ensure keypublic service functions. Woreda Health Bureaus have also been giventhe authority to hire, fire and manage health personnel. Woredas, however, still depend on regional and central levels for many health system functions, including the recruitment and allocation o f healthpersonnel, and the procurement and distribution o f supplies. Ingeneral, institutional capacity at the Woreda level for the planning and implementation o f health programs and other programs i s a concern. -21- The Government still runs most of theformal healthfacilities. 44. The Government runs most health facilities existinginEthiopia today, andthe public network has expanded dramatically over the most recent years. Seventy-one percent ofhospitals, 94 percent o fhealth centers, 82 percent o f health stations and all health posts are currently runbythe Government (table 4). Onthe other hand,the pharmaceutical sector is dominatedbythe private sector (including NGOs): 85 percent o fpharmacies, 81 percent o f drug shops and all rural drug vendors are private1y-owned. 45. The number o fprivate sector providers has been growing rapidly inrecent years. Between 1996 and2002, the number of private clinics has increasedfrom 541 to 1,235, but this formal private sector is present almost exclusively inurbanareas. Twenty sevenpercent of all private clinics inEthiopia are also located inAddis Ababa, where 50 percent o f the hospitals are privately-owned. Outsideurbanareas, only private drug vendors-not always formal-can be found inthe private sector. Between 1996 and 2000, the number o fpharmacies increased from 541 to 1,235, drug shops by 108percent from 148 to 309, andrural drugvendors by 27 percent from 1,460 to 1,856. There may also be some public healthworkers providing services "on the side," as reported by some anecdotal information, but there are no firm data to support this. Faith-based organizations andNGOs are also numerous. There are 225 NGO projects inthe health sector, health beingthe single largest sector for NGO involvement. NGOs are also involved inwater supply, sanitation, and environmental areas which contribute to health outcomes. Table 4: Distributionof HealthFacilities by Ownershipand by Region,2002/03*** ***Facilitiesowned by NGOs (Non-Govemment agencies) and OGA (Other govemment agencies) ***Includes 5 central hospitals (St. Paul, St. Peter, Amanuel, ALERT, and Black Lion) ****Includes private hospitals Regions are ranked based on the development index used by the GOE (annex 1.1 explains the criteria and also presents other types of indices such as poverty and revenueicapita). Addis rankshighest interms o f development and Somali the lowest -22- The number of facilities has increased yet geographical access to health-care services remains one of the lowest in the world,particularlyfor clinical care. 46. Between 1995/1996 and 2002/03, the number o fhealth facilities has grown rapidly (Figure 14). The number ofhospitals has increased from 87 to 119, health centers from 257 to 451,andhealth posts from 0 to 14312. The number o f health stations has stayed fairly stable, only decreasing slightly from 2,45 1to 2,396. Figure 14: Number of Health Facilities by Type from 1996/97 to 2002/03* I 3000 2500 +Hospitals health centers 2000 health Stations 1500 Health Posts +I+PrivateClinics 1000 +Pharmacies 500 - -I- Drug Shops Rural Drug Vendors 0 Source: PPD MOH. HealthrelatedIndictors, 2002/03. *Note: 2002/03 data are not available for private clinics., drug shops, andrural drugvendors 47. The average distance to the nearest health facility was 7.7 kilometers in2000. Seventy percent o fhouseholds, however, reside less than 10 km away from a health facility (Table 5) while only about 40 percent ofhouseholds have access to formal clinical care at less than 5 km or one hour's walk, the usual standard to measure access (Table 6). More than 90 percent o f households travel on foot, evenwhen the facility is hrtherthan 10kms. Ruralhrbandifferentials are very large, withthe nearest health facility in2000 being 1.4 k m s away inurban areas and 8.8 kms inrural areas. Regional differentials are also significant, with distances as low as 1.3 kms inAddis and as far as 9.8 k m s inAfar (table 6). Income differentials, however, are less striking although, on average, the poorest households live further away from a health facility than richer households. -23- Table 5: Average distances to hospitals/health centerdhealth clinics (kms) 1995 2000 M e a n Std. Dev. M e a n Std. Dev. Total 8.8 9.3 7.7 8.1 Urban rural Rural 10.2 9.3 8.8 8.2 Urban 0.9 2.3 1.4 3.4 Income Quintiles Poorest 10 10 8.5 9.5 2'* poorest 10.1 10.2 8.1 8 Middle 9.2 9.4 7.6 7.5 2nd richest 8.7 8.8 7.5 7.6 Richest 7.0 8.0 6.1 7.4 I RegionTable 6: Access to nearest hospitaYhealth centedhealth clinic by region I 4kms I 1-4kms I 5-9kms I 10-14km I15-19km I20+kms I Tigray 7 31 31 19 9 4 Urban 39.8 54.4 5.2 .I .2 Rural 4.2 26.8 35.3 17.6 10 6.1 Total 9 31 31 15 9 5 -24- The human resource base is very limited and disproportionately distributed across regions. 48. The humanresource base supporting health services is very limited. Ethiopia has one o f the lowest ratio o f doctors to population inthe world. The ratio o fnurses to population is more favorable but this number includes large numbers of "junior" or "assistant" nurses who have only one year's training. 49. As manyhealth staff operate inurbanareas, rural areas face a continuous shortage o fhumanresources (Table 7). Inthe three largest regions (Oromia, Amhara and SNNPR), less than one doctor is available per 44,000 people and one nurse per 8,000 people. Afar and Gambellahave no specialist physicians. 50. The shortage also includes frontline service providers. Midwifery skills are particularly lacking inlarge regions such as Oromia and SNNPR-these regions have less than one midwifeper 100,000 people. Oromia has the lowest frontline health worker to population ratio. 51. The number o f administrative staff tends to be greater thanthe number o fhealth workers with a ratio o f 2: 1. Inaddition to the relatively large number o f administrative andsupport staff, considerable numbersoftrained healthworkers occupy non-clinical positions, which may not be very efficient inthe context o f a shortage o f clinical skills. For example, inBenshagul-Gumuz, only 5 out o f 18 health officers (27.7 percent) are in service delivery positions while the rest occupy either teaching or administrativeposts. InAmhara, 38 percent ofhealthworkers are employed inWoreda andzonal offices. Table 7: Population to Physicians, Nurses, Health Assistants Ratios by Region, 2002/03 NGO 51 433 103 360 OGA 380 4,015 301 4927 Private** 390 37 0 Total 69,129,021 2,663 25958 14,160 10,083 6856 10,083 14507 4,765 "Non-public health workers were not categorizedby region. **Data for private sector only available for Addis -25- There are concrete efforts to train more people and the number of health personnel are increasing. 52. Between 1996/97 and 2002/03, there was a significant increase inthe number o f health officers and nurses, from 30 to 631 for health officers, and from 4,774 to 14,160 for nurses. Para-medicalstaff increased by 159percent, from 1,788 to 4,641. The number o fphysicians increasedby 37 percent, from 1,483 to 2,032 (Figure 15). Yet, the populatiodpersonnel ratio has changed only marginally over time. Between 1996 and 2002, for example, the midwife/personnel ratio increased inTigray, Afar and Addis, but remained almost unchanged inAmhara and Oromia, nor did the populatioddoctor ratios change inany o f the large rural regions. Inthe context of rapid population growth and increasing attrition inthe private sector andmigration, changing this situationwill require drastic changes inthe number o ftrained workers. Figure 15: Trends in Growth of Health Professionals,1996/97-2002/03 16,000 14,000 12,000 +Physicians 10,000 Health officers 8,000 6,000 Health assistants 4,000 2,000 0 I I Source: MOHBPD. Health and Health-related indicators The health care workforce is male-dominated, particularly in rural areas. 53. The health care force i s male-dominated. Only about 12 percent o fphysicians, 13 percent o fhealth officers, and46 percent o fnurses are female. Even among frontline workers, only 58 percent are female (Figure 16). Inaddition, the large majority o f female health-care workers are located inthe urban areas. Addis and Tigray have the highest femalelmale ratio (more than60 percent) among health workers. Addis has the highest female/male ratios for doctors andnurses while Tigray has the highest female/male ratios for frontline health workers. Except for Tigray, most rural areas-where the need for matemal and child health services i s the most acute--are mostly served by men. Attracting more females into the health profession faces several hurdles: the rate o f secondary education is still low among girls inEthiopia; married women tend to follow their husbands andnot make independent professional moves; and unmarried women face security problems when living andworking inrural areas. -26- Figure 16. Percentage of Females by health worker category inEthiopia, 2002/03 c 12 5 Bphysicians W health officers 0nurses 46 health assistants 36 Source: data fromPPD, MOH. Health and Health Related Indicators. Addis Ababa. 2002103. A large number of health worker categories may provide some staffing flexibility to the regions but makes evaluation ofperformance and quality of curricula dijjjcult to manage. 54. The public health system has recently undergone a transition from a six-tier to a four-tier system (Figure 17), and the current staffing status i s a mix betweenthe old and the new systems. Some aspects remain ill-defined. For example, staffing norms and the actual status of health stations are unclear because they were supposed to bephased out and replacedwith health posts. Yet regions have continued to construct health posts that offer bothpreventive and clinical services. Hence some regions have adapted the standard to their own requirements. While this approach allows some flexibility, evaluating the efficiency o fhumanresources has been complicatedby the difficulty in tracking the relationship between the various types o f training andthe levels o f performance. -27- Figure 17: Structure of the Public Health Delivery System I 55. At present, the Qualification Requirementsfor HealthProfessionals(1999) contains salary scales and career structures for formally trained healthworkers, andjob descriptions havebeen developed for at least 90 health cadres andpost descriptions for senior positions. There i s some overlap among the differentjob categories (e.g., there are at least 10nurse categories) andthese canbe somewhat confusing given the relatively minor differences between some o f them. For each type o f cadre, there are inmost casesfive definedcareerpaths. A newhealthextension worker category has been recently added as a result o f establishment o f the Health ExtensionProgram. Training capacity remains inadequate relative to training targets. 56. There are about 30 training institutions located inseveral regions, which i s quite a limitednumber for a country o f close to 70 million people. Overall training capacity remains inadequate relative to training needs and objectives. For doctors and health officers, there are only five universities or higher-education colleges. Twelve nursing schools provide an annual training output o f about 2,226 nurses. The actual numbers o f trained doctors, health officers, midwife nurses and radiographers are even significantly lower-less than 50 percent for midwives-than targets plannedby the Ministry o f Health. Trainingcurricula are not well aligned with intended objectives. 57. Current training i s also not well aligned with some objectives, for example, the reduction o f matemal mortality. None o f the various levels o fmidwives meet the intemationally accepted definition o f a midwife. Significant changes need to be made in midwifery training for it to be more in line with the FDRE's public health policies. Existing evaluations indicate that training for health officers i s generally well conceived interms ofcommunity orientation, and focuses onhealthpromotion, illness prevention -28- and essential medical services. It prepares graduates to be managers inhealth centers and as "extensions" o fphysicians. However, there i s a clear need to modify the curricula for junior, mono-disciplinary nursing and midwifery, and frontline cadres inorder to better equip them with practicalclinical skills, especially ifthey are expected to function in rural peripheral health units. The newly developed health extension worker's curriculum aims at rapidly scaling up a program deliveringkey outreach services. 58. One important policy measure recently adopted by the MOH in2002/03 was the development o fthe Health Services Extension Package Initiative. This program seeks to provide healthpromotion and extension services to communities, and is being piloted in five regions. Ithas been recently revisedbased on discussions betweenthe MOH, regions, and donors inorder to become more responsive to regional/community needs. Different packages of services have been discussed, but health extension services are likely to include immunization, micronutrient supplementation, and family planning, and will linkwith community promoters' programs as well as clinical referral care. Salaries of health workers are high relative to GDP,although low in absolute value and in comparison with the international market. 59. Ethiopianmedical specialists and general practitioners are paid significantly less indollar terms thanphysicians from other countries. The average salary for amedical specialist is equivalent to about US$236 a month. This makes migration very attractive for doctors with prized skills on the international market, particularly surgeons and obstetricians. Yet public salaries for Ethiopian healthworkers appear to be relatively higher than the average inSSA (Table 8). While doctors typically receive about 8-12 times GDP as an annual remuneration inSSA, Ethiopian doctors typically receive about 18-22 times. Country Generalpractioner Diploma Nurse Chad 10.3-18.8 5-10.6 Burkina Faso 7.3-23.5 4.2-13.5 Mauritania 5.67-9.45 3.2-5.7 Cameroon 4.7 1.7 Niger 10.6-20.8 5.3-12.0 Ethiopia 18.3-30 11.7-27 60. Salaries have also increasedovertime, by at least 21 percent innominal terms and about 36 percent inreal terms (using 1995 prices) from 1999 to 2003. The HSDPI evaluation (2003) indicates that there were few recent complaints about salaries and remuneration. Health sector salaries seem to be more or less inline with the minimum cost-of-living increases, and favorable relative to other sectors. -29- While there have been improvements in salaries and allowances, there are other issues to be addressed in order to improvejob satisfaction among health workers. 61. Provision o f staffhousing has been repeatedly mentioned as a possible motivating factor for higher qualified personnel to stay inremote areas (HSDP reviews and focus group discussions). Training and career development are also important considerations. Onthe other hand, there is a general concern o f contracting HIV/AIDS andthis seems to beincreasingly a factor indeterminingwhether to work inclinical services. A survey on quality o f health services management (PHRD 1998) also mentioned the following as key problems: work overload, staff shortages, unclear or misunderstoodjob descriptions, budget shortages, unfair promotions, not gettingannual leave at the right time, lack o f transport facilities, lack of a safety policy andprotective materials, and inadequate care for sick healthworkers. Low absenteeism, high motivation and relatively strong work ethics still characterize the Ethiopian health worvorce. 62. Despite some observations o f absenteeism and highstaff turnover among doctors moving from hardship areas, as well as among staff inRegional HealthBureaus, the rate o f absenteeism inEthiopia is relatively low compared to other countries (Figure 18). This may be explainedby recent salary increases andthe implementation o f incentive schemes including hardship allowances. Insights from recent focus groups indicate that there i s still a relatively highwork ethic among health workers inEthiopia. Non-salary motivating factors include the more visible opportunities for upgrading and post basic training, as well as an impressionthat "things were gradually improving". Figure 18: Health Care Provider Absence Rate across 7 countries Ethiopia Uganda Bangladesh PNG India Indonesia Per" Country Ranked from Pwmsl to Richest IQHealth Care ProviderAbsence Rate I Source: Ethiopiafacility based survey, draft findings from Chaudhuryet al. 2004 -30- Health workers, however, are often not willing to be posted in rural areas and complain about their living conditions when out-posted. 63. According to the preliminary results o f a contingent valuation exercise undertaken by Semeels et al. (2004), a mark-up o f about 50 percent inmonthly wages would adequatelymotivate about 80 percent o f the current nursingstudents to serve in rural areas. Approximately 60 to 70 percent o f the current 4th year medical doctor students would be willing to serve inrural areas iftheir current starting salary increases by 50 percent. 64. Humanresource policies andpracticeswould also benefit from better enforcement andtransparency. Healthworker focus group discussions have expressed concems regarding favoritism with respect to postings, training, promotionand obtaining release from the public sector. Itwould be important to address these concems because they influence how healthworkers make career andlabor market choices Focus group discussions among health workers also underscore the need to improve accountability in bothprivate andpublic health sectors. 65. The public sector has recently introduced a new system o fperformance management which links salary increases, promotions, and training opportunities to healthworker performance. It would be important to assess its impact on healthworker behavior andperformance, andhow it can be improved further to make healthworkers accountable to users o f health services.' The deepening decentralization to the woredas may encourage more community or bottom-up approaches to improve accountability (Lindelow et a12004) Ethiopia still has very few hospital beds, and these are largely concentrated in Addis Ababa. 66. Ethiopia has a very low number o f hospital beds inrelation to its population. There are approximately 0.17 beds per 1,000 people, less than a fifth o f the average for SSA o f 1.1beds per 1,000 population (WDI2003). Addis, the most developed region, has the largest number of beds, followed by the larger regions (Tigray, Oromia, SNNP, and Amhara). However interms o fbedperpopulation ratios, 5 regions (Oromia, SNNP, Amhara, Somali, and Afar) have b e d 1000population ratios equal to or lower than 0.10 per 1000. Half the existing healthfacilities need serious repair or upgrading. 67. Ina detailed assessmentofa sampling ofhealth facilities in 1995, over 50 percent reported leakingroofs, electrical problems, plumbingand sanitary problems. An overall assessment of buildingconditions showed that 28.8 percent and 15.1 percent needed major repair or total replacements respectively. The availability ofwater was *Some users in focus group discussions have expressed their dissatisfaction with lengthy waiting times and impolite treatment from health workers. -31- inadequate and scarce inabout 30 percent o fhealth centers. Minimalpower supply and attempts for alternativepower supply for the facilities didnot yield good results. Rehabilitation o fhealth facilities i s often limitedto preventive maintenance, such as replacing the damaged part rather than rehabilitating functional flaws, correcting the shortcomings for service delivery and adapting it to new health care technology or anticipated future expansions. Availability of common medicines is generally good but remains aproblem in rural areas. 68. Ingeneral, it appears that the availability o fessential drugs hasimproved, but shortages o f drugs are still common. There are significant differences between various missions and evaluations as to the status o f drug and equipment supply inthe public sector. A 2002 study indicates that govemment hospitals have the highest number o f selected indicator drugs (82 percent), followed byprivate drugretail outlets andhealth centers (72 percent for both). Results for private facilities were unexpectedly low. Private hospitals also have a lower availability o f essential drugs than public hospitals (63 percent compared to 72 percent). Shortage of qualified pharmacists and druggists in public facilities also remains a serious problem. 69. There i s a shortage o fpharmacists and druggists ingovemment facilities; they tend to seek employment inthe private sector where salaries are more attractive. In2000, approximately 500 pharmacists were working inthe pharmaceutical sector, o f which only 121were inthe public sector. The number o fpharmacy technicians increased significantly inthe public sector to compensate for the lack o f trained pharmacists. About 26 percent o f the surveyed govemment hospitalpharmacies are beingmanagedby druggists or pharmacy technicians, while another 22 percent are beingmanagedby either nurses or health assistants. Forty- eight percent o f surveyed health center drugs stores are managed by either nurses or health assistants, and another 7 percent are being managed bynon-healthprofessionals. However, 98 percent o fprivate pharmacies are managedby pharmacists, and about 78 percent o f private drug stores are managed by the required druggists or pharmacy technicians. The number of health-carefacilities has increased, but utilization of curative services has not kept pace. 70. The trend o futilization o fservices has not matchedthe steady increase in facilities. Outpatient visits slightly increased by 3 percent from 27 visits per 100persons in2001 to about 29 consultations per 100persons in2002/03. Utilizationis less than a thirdo fthe HSDPgoal for 2004/05 o f 1.O visit perpersonper year. The top 10 leading causes o f outpatient visits in2002/03 account for 50.5 percent o f total visits. They include consultations related to major causes o f mortality such as malaria (15.5 percent o foutpatient visits), pneumonia (5.6 percent o foutpatient visits), respiratory system infections (5.7. percent), and dysentery (4.8 percent of outpatient visits). Few consultations occur for diarrhea which remains the main cause o f under-five mortality. -32- 71. The national average bedoccupancy rate (BOR) data i s 41.4 percent, based only on data from 5 regions (Tigray, Somali, BenshangulGumuz, SNNPR, andHarari) and the Center in2002/03. No datawas available for Addis in2002/03 but it was the region with the highest BOR at 175.2 percent in2001/02. Somali has the lowest BOR (7.7 percent) (MOHPPD 2002/03 and2000/01). Thequality of care is highly variable. 72. Representativedata on technical quality of care is quite limited. A qualityindex drawn from the 2000 DHS2000 basedon 7 types of services (measurement o fweight, heightmeasured, bloodpressuretaken, blood sample taken, urine sample taken, told about pregnancy complications, andtold where to go for pregnancy related complications) a pregnant women receivedduringthe ante-natal care, shows that quality of antenatal care differs significantlybetweenurban andrural areas. Table 9 shows that out of the highest possible score o f 7.0, the quality o f care index inurbanareas was 4.5 while only 2.4 inrural areas. This index also variedby region: Addis hadthe highest (5.1), while Afar hadthe lowest (2.0). Poor women also received a lower quality o f care, although this difference was less than the geographic variation. Table 9: Variation in quality of care of maternal health services* by wealth, region and urban residence WEALTH QUINTILES -33- In general, clients are satisfied with the care they receivedfrom NGO,private andpublic health facilities. However, a higher percentage considers the care they receive from publicfacilities to be below average. 73. In2001,about 52percent ofrespondentsperceivedthe quality o fcare they received as good. However, about 30 percent o f households who visited a government facility consider the quality o f care they received to be below average. A lower percentage o fhouseholds who obtained care from NGOs (14%) andprivate facilities (12%) consider the quality of care they received to bebelow average (Table 10). Table 10: Respondentsperceptionof quality for different providers Clients mainly complain about the lack of facilities and drugs, and the lengthy wait. 74. The main difficulties cited were lack o f facilities (42.7 percent o f the individuals), financing (33.8 percent), and essential drugs (23 percent). A consistent and sufficient availability o f drugs was considered an important indicator o f quality o f service. About 37 percent o fhouseholds who visited public health facilities were dissatisfiedbecause drugswere not consistently available. Other main reasons cited for dissatisfactionwith the quality o f care inpublic health care facilities included: inadequate skillshowledge, lack o f courtesy on the part o fpersonnel, inconvenience o f lengthyprocedures, inadequate availability of diagnostic facilities, andthe lengthy waiting time. The waiting time between arrival andbeingseen i s very high, averaging 7 hours at government hospital outpatient departments, 6.2 hours at NGO facilities, and 2.7 hours inother private facilities. However, private providers were also considered inferior to large government hospitals because they provide limited laboratory and x-ray and surgical services. They were also cited for excessive charges by requiringmore tests and expensive drugs. -34- Publicandprivatespendingon health Based on National Health Accounts data, bothprivate andpublic spending have been on the increase between 1995/96and 1999/2000. 75. Ethiopia's total health expenditure remains dramatically low; the per capita health expenditure has been estimated at aroundUS$5.6 in2000. Expenditures on health have substantially increasedbetween 1995 and 2000, from U S $ 4to US$ 5.60, andthis increase occurredinbothpublic andprivate spending. However, public spending grew faster largely due to extemal sources, mainlyfrom donors. Table 11: National Health Accounts--Evolution of total, publicand private spending 1996/1997-1999/2000 NHA 1: NHA 2: Variation 1995196- 1995-1996 1999-2000 1999/2000 (%) Total spendingas a share of GDP 4.1 percent 5.3 percent +34.1 Total spending US$ 4 5.6 +40 Total spending PPP 25 2.7 +32 Public spendingas a share of GDP 1.7 percent 2.74 +61.1 Public spendingUS$ per capita 1.65 2.77 67.8 Public spendingPPP 10.3 1.2 +57.2 Private spendingas a share of GDP 2.4 percent 2.8 +16.6 Private spendingUS$ per capita 2.3 2.82 +22.6 Private spendingPPP 14.4 16.5 +14.5 Source: FMOH, NHA, 1995-1996;FMOH, NHA, 1999/2000 However,per capita health spending remains among the lowest in the world. 76. The overall per capita health spending inEthiopia is amongthe lowest inthe world (Figure 19) and i s significantly lower than the SSA average ofUS$42. The recent increase has only slightly narrowed the gap. This low level o f spending mainly reflects a very low resource base or GDP. Ethiopia's total health spendingas apercentage of GDP (4.1 percent) i s comparable to the low-income countries' (LIC) average. Ethiopia's private health spending share o f GDP o f 2.8 percent remains close to the average SSA experience (2.6 percent). However, the private health expenditure share o f GDP in Ethiopia i s rather on the highside when compared to the L I C average of 1.1percent. -35- Figure 19: Per capita expenditure on health invarious countries vis-&-vis GDP loooo1 Unitedstates O B 100 1000 10000 100000 GDP percapita (constant US t 1995) Funding of the health sector in Ethiopia is shared equally between thepublic and the private sectors. 77. According to the most recent "HA (1999/2000), public spending inthe health sector, both domestic and from external sources, represents the largest share of total spending (49 percent) and amounts to US$2.77 per capita. Private consumption through out-of-pocket spendingalso represents a large share o f this spending (36 percent) or US$ 1.96. Out-of-pocket spendingby individuals includes direct payments to private practitioners, traditional healers, private pharmacies, and government facilities inthe form o f user charges. NGOs contribute a much lower although not trivial amount, their contribution reachingnearly 10percent o f all health spending. However, the contribution of private enterprises accounts for only 5 percent o fhealth spending (Table 12). -36- Table 12: Ethiopia: NationalHealth Accounts Data for 1999/2000 Sources: Authors' estimates based on various sources including NHA 1999/2000,Min.of Health, Min. ofFinance, IMFStatistics, PER2003. Majority of total health spending is allocated to curative care. 78. According to the recent NHA, about 64 percent o f total healthresources was spent on curative care and 25 percent on promotive andpreventive health care (PPHC) (Table 13). Administration costs, which mainly include those of federal and regional health administrations, comprise 8 percent o f total spendingwhile very few resources were spent on training (2 percent) and research and development (1percent). Pharmaceuticals andmedical supplies constitute about 38 percent o ftotal expenditures 74 percent o f which were financedbyhouseholds. Table 13: Expenditure by M a j o r Functions (amount and as % of total expenditures) Birr (millions) YOShare Service Functional Classifications Delivery Expansion Total Total Administrative Expenditure 221.2 7.0 228.2 8% Curative Expenditures 1673.4 211.2 1884.6 64% Inpatient 254.3 153.9 408.2 14% Outpatient 295.4 57.3 352.7 12% Pharmaceuticals and medical supplies 1,123.7 1,123.7 38% Promotive and Primary Health Care 516.9 214.0 730.9 25% Research and Development 21.5 21.5 1% Training 57.5 8.5 66.0 2% 2,490.4 440.8 Total (85%) (15%) 2,931.2 100% Source: FMOH.NHA, 199912000. -37- Public expenditures in Ethiopia have increased significantly in recent years. 79. Based on MOFEDdata, overall public spendinghas beenincreasing steadily over the last few years. Total government expenditures increasedby 90 percent in nominal terms and 70 percent inreal terms from 1995 to 2002. The share o fpublic expenditures to revenue has beenat about 30 percent o f GDP on average from 1997to 2002, placing Ethiopia among the SSA countries with the largest level o frelative public spending. Public spending on health has also increased albeit at a slower pace as compared to totalpublic spending. 80. The public expenditures monitoring system also captures a marked increase in public spending on health. Public spending on health increased by 80 percent innominal terms and by 58 percent inreal terms (using 1995 as the base year) between 1995 and 2002. Per capitahealth expenditures inreal terms increased by 10percent over the same period. However, public spending on health remains low at approximately 5% o ftotal public spending. Both capital and recurrent expenditures on health have increased although the rate of increase of capital expenditures is greater than recurrent expenditures from 1990-2002. Questions remain as to whether recurrent expenditures are appropriately allocated among investments made. 81. According to MOF data, capital expenditures increased by 128percent and 102percent inreal terms from 1995-2002.Average capital spending per capita remains low at about US$ 0.45 from 1995-2002,peaking at US$ 0.47 per capita in2001/02 in nominal terms and US$ 0.42 inreal terms. Duringthe same period, recurrent spending increased more modestly by 59 percent innominal terms and 42 percent inreal terms. Similar to capital expenditures per capita, average annual recurrent spendingper capita remains low at about US$0.88 from 1995-2002, reaching US$ 0.91 in2001/02 in nominal terms and US$0.81inreal terms. These figures are substantially lower than the average per capita recurrent expenditures o f low-income countries (US$2.50). The combined increase o fboth capital andrecurrent expenditures is a positive development. However, the latest NHA results (1999/2000) indicate that capital expenditures (expansion o fhealth facilities and equipment) comprise the largest share o f total public expenditures (27 percent). HSDP reviews indicate that facility expansion tends to be unaccompanied by sufficient medical supplies and health workers to provide good quality health services. -38- Expenditures on inputs have been relatively stable, but there are some indications that allocations to somepriority programs, such as immunization, have decreased. 82. The Public Expenditures Review (2003) shows a relatively stable distributiono f public spending over time. Wages and salaries represent the largest item o fpublic spending. Between 1995196and 2000/01,there was a progressive shift inspending composition, with an increase inthe percentage allotted to salaries. As a result, in 2000/01, approximately 61 percent o f the recurrentbudget paid for salaries, while 26 percent went to medical materials and supplies, including medical equipment. 83. Fundingofbasic immunization services appearsto have declined over the last five years.' A large share o fpublic money (about 40 percent) i s allocated to hospital care each year. This pattern o f higher spending on hospitals is often found worldwide and the Ethiopian case is not the most extreme, with many countries spending60 percent and more on hospital care. Giventhat most diseases that impose a heavy burdenon Ethiopians are those that can be preventedor treated on an outpatient basis inprimary care settings, it is likely that additional activities at the primarylevel will have a larger impact on the health status than at the hospital level. As the Government's budget now projects a shift inemphasis toward primary level care as a result o f the recently established Health Extension Program, it will be useful to monitor whether future actual expenditures reflect the shift toward the newly established health extension program (which include the immunization program). Spending rates are still low in all regions,justibing the reluctance of the Government to increasepublicfunding for health. 84. However, even when budgets allocated to health increase, low budget execution often undermines service delivery. Only about 41 percent o f regional budgetsranging from 26 percent inSomali to slightlyover 54 percent inAfar and Tigray were spent in 2000/01. Similarly, based on the available information from 9 out o f 11regions for 2001/02, execution rates ranged from 30 percent inSomali to over 92 percent inAddis andAmhara. Almost all regions spent a larger share of their recurrent budgets compared with their capital budgets while the Federal MOH spent a signlficantly larger share of its capital budget. 85. With the exception o fAfar, all the other regions performed significantly better inspending their recurrentbudgets comparedto their capitalbudgets in2002/03 (Table 14). Funds for vaccines was reported to be Birr 20 million or less than 7 percent o f govemment recurrent health expenditures) in EFY89 (1997198) and except for EY92 has been lower in every year since. In EFY94 (2001/02), expenditures on vaccines was only Birr 13 million, representing less than 3 percent of the govemment's recurrent health expenditure. It is possible that vaccines financed by external agencies in the capital budget have to some extent substituted for domestic funding, but the available data make it difficult to draw conclusions (PER 2003). -39- Somali 113 24 52 Tigray 99 25 78 Federal MOH 92.9 702.3 270 I *Expenditures are based on pre-actual (preliminary information) 86. This could meanthat capital needs relative to implementation capacity inthe regions tend to be overestimated, andor there could be factors such as donor processes which impede faster execution o f capital budgets at the regional level. The other possible causes o f over-all underspending include inadequate capacity for program planninghudgeting and management at the regional, zonal, and Woreda levels. The problem o f under-reporting could also be a contributing factor. The PER (2003) mentions that donor inflows tend to be overestimatedinthe budget at the beginning o f the year but actual expenditures are under-reported. Regional spending varies considerably 87. In2000/01, per capitaexpenditures across regionsrangedfrom Birr4.7 (US$ 0.55) inSomali to Birr 45.3 (US$ 5.3) inGambella. Spendinghasbeen highinBenishangul-Gumuz, Gambella andHarari. These are smallregions that are also less populated than the others andtheir highper capitahealth expenditures could reflect the costs o f administrative overhead. Expenditures per capita inthe three urbanregions (AddisAbaba, Dire Dawa, and Harari) are also relatively highbecause these regions have a relatively larger number o fhospitals and serve as referral points for service seekers from other regions. The most populatedregions such as Amhara, SNNPR, and Oromia also have relatively lower per capita health expenditures. Among the populated yet rural regions, Tigray spends the most on health on a per capita basis. Tigray seems successful in translatingpublic health expenditures into results. 88. No systematic relationship canbe found between the levelo f expenditures and the level o futilization o f services. For example, immunization coverage inBenshangul and Harari are low, while attended births only marginally increased over the last years for both regions as well as for Gambella. The three largest regions (Oromia, SNNPR, and -40- Amhara) display a similar low level o fperformance with only Birr 6-7 per capita o f public funds to finance their health services. Tigray appears more successful in translating health expenditures into improvements inhighimpact health interventions such as immunization and attended deliveries. Its performance can be attributed to its emphasis on community-oriented approaches-it has the largestnumber o f frontline workers compared to all the regions. The region also revised its health center staffing standards based on location, having thrice as many surgery-related staff assigned to remote HCs while physicians and nurses could be added inthe urbanHCs to deal with outpatient needs. However, stunting and underweight rates among under-five-year old children inTigray are among the highest across the regions. Donor funding has been flowing through extra-budgetary channels and is difJicult to capture. 89. From 1997-2001, the health sector received a yearly average o fUS$ 57 million or 9.5 percent o fthe total aid available to all sectors. Meanwhile, agriculture hadthe highest share o f annual aid, at 16.2 percent (US$ 85 million). External assistance reaches the government systemalong three channels. Loans are included inthe budget and accounts, andmost budgets support non-earmarked grants and some other grant funds. This is usually done on the basis o f commitments presented by donors duringbudget preparation andare often not reflected inthe govemment account. An unknown amount o f donor funds are provided inkindsuch as medicines; these resources are usually not captured inthe budgetprocess. This affects the GOE's ability to accurately determine whether it is allocating an appropriate amount o fits ownbudget for a specific item or budget category. Cost-recovery represents a small share of expenditures in thepublic health system. 90. Cost recovery has beenpart o f Ethiopia's health system since the early 1950s. Nominal amounts charged range from small fees (Birr 1to 5) for outpatient registrations, consultations, laboratory tests, andother routine diagnostic procedures, and inpatient beds, to higher fees (Birr 10 and above) for prescription drugs and inpatient surgical procedures. No changes were made to this policy until 1998, with the initiation of the new Health Care and Financing Strategy. When originally introduced, fees recovered a substantial portion ofthe total costs o fproviding the services. However, the level o f fees remainedunchanged for almost 50 years and today it has become almost symbolic. Moreover, close to 60 percent o f users receive exemptions. As a proportion o f GOEhealth expenditures, fee remittances to the MOFhave declined from 16percent in 1986 to less than 6 percent in 1995/97. With the exception o f special pharmacies and some hospitals inSNNPR, all fees collected are remitted to Regional Finance Bureaus who forward them to the MOF where they are accounted for as general government revenue. -41- Thepoorest groups benefit littlefrom public spending. 91. Table 15 shows that the richest to poorest ratios interms o f utilization are lowest for healthposts andhealth stations/clinics (0.2 and 0.5), slightly higher for health centers (1.6), andhighest for hospitals (about 6). However, the combined share o fhealth centers and clinics interms o fpublic recurrent expenditures is about 30 percent in 2000/0l--lower than public recurrent expenditures allocated to hospitals (38 percent). There are marked differentials by income quintile across households inthe use o fbasic health services. The poorest households consistently have the lowest utilization rates for immunization, assisted deliveries, and antenatal care by a trained professional. The richest to poorest ratio (27) is highest for the use o f assisted deliveries. About 24 percent o fwomen inthe richest households have had an assisted delivery comparedwith less than 1percent o fwomen inthe poorest households. Nonetheless it is also important to note that while income differentials with regard to access are high, absolute levels of use are still low even among the richest households. While some incremental increase canbe gained byreallocating spending from hospitals to PHC,over-all public health expenditures would need to be increased to improve the quality and accessibility o f health services. Table 15: Use of health facilities and services: national data, per quintile 4 1 42 43 4 4 45 Richest to Poorest Ratio %Use o f health Dosts ( W M S P 4.8 5.1 10.0 5.2 0.9 .2 % Use o f healthstations (WMS) 1I49.5 53.3 44.5 45.9 26.2 .5 %Use o fhealthcenters 26.3 20.3 23.4 28.5 41.9 1.6 (WMS) Hospitals 3.2 5.2 6.7 4.9 18.4 5.7 (WMS) Use o f immunization, all 6.7 5.6 15.4 15.1 33.3 4.9 vaccinations received (DHSOO) I Use o f assisted deliveries II 0.9 II 1.5 I 1.4 I 4.8 II 24.3 II 27 I (DHSOO) Use o f antenatal care by trained 15.3 16.4 20.6 28.7 58.2 3.8 professional (DHSOO) -42- Towards the future: progressingtowards the MDGs Identifiing pathwaysfor health services to contribute to the MDGs. 92. The Government o f Ethiopia has been assessing the potential contribution o fthe planned expansion o fhealth services inorder to reach the MDGs. The cost andbenefits o f the various policy options envisioned by the Government for delivering high-impact health interventions have been analyzed by examining: 0 the cost and potential impact o f strengthening the delivery o f and demand for high impact interventions under different health service delivery options, taking into consideration the current physical access and human resource constraints faced by Ethiopia; 0 various health financing scenarios based on reasonable assumptions o f economic growth, increase in public revenue, allocation o f public funds to health services, andpotential contributions fi-om households and the NGO sector; and 0 implementation issues inthe context o f decentralization, particularly the role to be played by each level o f the public sector, as well as the contribution o f the private sector. 93. This analysis was conducted using a specific tool developedby the World Bank, UNICEF and WHO: "Marginal Budgeting for Bottleneck (MBB)," created specifically to assess the marginal cost and impact o f service packages delivered through three major service delivery modes: Population Oriented "Outreach" (Health services extension package), Community/ Family Oriented services (Community Health Promoters), and clinical care (hospitals andhealthcenters and stations). 94. Table 16 shows that the three delivery mode packages respond differentlyto health coverage increases and therefore show varied returns for reductions inboth child andmaternal mortality. The cost requiredbyeach deliverymode also differs substantially. The SDRP-proposed health services extension package offered through health posts is affordable, which may lead to significant decreases in under-five mortality and number of maternal deaths. It may help serve as the backbone for more accessible health services. 95. Expanding access to key professional preventive services delivered by two female health extensionworkers, trained for one year through the outreach component o f the health services extension package, will serve as the "backbone" o f the Ethiopian health system. The health extension (HSEP) "Population Oriented" delivery approach" could potentially be cost-effective, reducing under-five mortalityby 9 percent, at a cost of US$0.34. However, its impact would likely be limited on the maternal mortality ratio (only 2 percent) because antenatal care contributes only marginally to maternal mortality as defined inthe MDG. The use o f this indicator obscures the highcontribution o f HSEP lo HSDPtargets,reducingthebottlenecksby90%inoutreach. For -43- interventions on the lifetime risk o fmothers dying (i.e., family planning) with outreach services potentially reduces this lifetime risk by 27 percent. Only a substantial effort in supporting outreach- and community-based services will address the most important causes of child mortality and mothers' llfetime risk of dying. 96. IfanincrementalUS$1.OO percapitaper year canbemobilizedfor Ethiopia's health services, investments would bebetter channeled towards the outreach and community services (HSEP and Community Promoters package) that were tested with success inTigray and SNNPR. Together those strategies may potentially reduce both the under-five mortality ratio andmothers lifetime risk o f dyingby approximately 30 percent, based on realistic assumptions regardingincreasing access along HSDPII targets. The outreach (Health Extension) package inEthiopia will also builda foundation for community-based activities. However, the interventions that are delivered through these delivery modes will have very little effect on the maternal mortality ratio. Including clinical care-with assisted deliveries and emergency obstetrical care would substantially contribute to the reduction o f the maternal mortality ratio andmothers' lifetime risk o f dyingby 30 percent each, andreduce under-five mortality by another 10percent. As indicatedpreviously, this would, however, imply substantial investments inclinical services on top o f the other approaches. When employed within theframework of current Ethiopian health policies, and using the health services extension objectives of HSDPII, the contribution of health services to reaching the MDGs would be considerable. 97. For a cost equivalent o fbetween 50 percent to 200 percent o f the current public spending on health, health services would contribute to reducing child mortality and the life time risk o f mothers dyingby approximately 40-45%. Yet this would not be sufficient to reach the MDGs. More ambitious objectives interms o f coverage with health service delivery that would allow Ethiopia to reach all the MDGs, would require significantly more resources-mainly human and financial as shown inTables 16 and 17. Reaching the health MDGsimplies not only a dramatic expansion o f the production o f keyhighimpact health services, but also the implementationo fmechanisms to ensure adequate demand for and use o fthose services. Five steps of further service expansion have been considered, assuming there are no financial constraints. Each step allows for the progressiveupgrading o f services, strengtheningboth supply and demand for high impact,services. 98. Limitingthe spread ofHIV, along withreducing childmortality bytwo-thirds, may be achieved at a cost o fUSS5 per capita, doubling current total health expenditures. Reaching the malaria MDGs for adults would add about US$1.72 per capita. Reaching the maternal healthMDGs would probably be the most challenging interms of mobilizing humanresources-requiring an additional US$ 3.5 per capita. Finally, expanding referral and hospital care, and adding the treatment o f chronically illpatients- including Highly Effective Anti-Retrovirals-for HIV patients, would be the most expensive component amounting to an additional US$9.8. -44- Table 16: Estimatedimpact and cost of different service delivery arrangementsfor all three modes: Health services extension package, Community Promoters,Clinical Care Scenario 1: Scenario 2: Scenario 3: Reduce demand, quality and Reduce demand, quality and Increase demand, quality and continuity continuity bottlenecks by 90%, base continuity bottlenecks by 90% , bottlenecks by 90% ,Access maximifeed Delivery Intervention access mode package Cost USMR (per Reduct. I I I capita) (1) 0% 0.2% 0% 1% ."Health `romoters reventive Infant trategy" 16% I o% I 0% 16% `amily/ :hild Care : o m u n i t y )riented 8% #emices llness 'otal Family Zommunity Care 20% 0.2% 0% $0.47 23% `reventive care for . "Health idolescent Girls & ervices Yomen of 5% 0% 19% 6% xtension :hildbearing age iackage" `reventive Care for 'opulation `regnant and 0% 1% 1% 1% hiented Jewbom lutreach or cheduled `reventive Infant & 1% 0% 0% 3% linic :hild Care essions .otal Outreach 6% 2% 21% $0.20 9% Preventive /** rlatemal & 3% 4% 4% 5% Jeonatal Care Yinical Primary .eve1 Illness 5% 0% 0% 10% 1. Clinical rlanagement + ndividual dented Car Xnical first needs to be eferral 2% 7% 7% 3% :ontinuousl) PHC/CHC) Illness wailable) vlanagement Yinical second I I I I I I eferral(FRU/DH) 0% 11% 11% 16% 16% 0% 18% 18% llness management rota1Clinical 9% 21% 21% $0.96 13% 30% 30% J 1.96 17% 34% 34% $2.65 rota1Three Modes I 33% I 22% I 37% $1.63 41% 32% 49% $3.08 49% 37% 58% $4.87 ;ource: authors' calculations 1) deaths of children less than 5 over one thousand live births (2) pregnancyrelated deaths o f women over hundred thousand live births (3) regnancy related deaths of women over hundred thousand women 15-49 (4) marginal cost per capita in US$ -45- Thehealth sector's contribution to achieving the health MDGs is large but would need inputsfrom other sectors. 99. According to the MBB simulations, the predictedprogress inhealth services would speed up the pace o funder-five mortality ratio reduction. Yet the current government targets are insufficient inachieving the twokhirds reduction goal o f under- five mortality ratio inMDGs through enhancing supply o f and demand for health services. Past experience has shown that despite plannedimprovements inefficiency o f the sector, most likelyimprovements inthe health sector alone under SDPRP objectives would not achieve boththe child andmaternalmortality MDGs. -46- Table 17: Scaling up coveragewith health services inEthiopia: resour :implications,c ts and potentialbenefit Ethiopia's strategy Expansion of Human Resources Average EstimatedImpact vIDG achieved facilities Implications incremental cost (private facilities (for both public per capita per year included) and private sector) over 2005-2015 Step 1: Information Increase in number Communication USS1.5 HIV incidence ieverse trend in and social o fradios at kebele specialists and peer decreasefrom -IIV incidence mobilization for levelby 308,239 educatorsincrease 3.66 to 0.55 per Stabilize trend in behavior change by 5 fold 100,000 3IV prevalence Reduction of U5MR O f5-1O% Step 2: Health # of healthposts Health Promoters US$3.54 Reduction of child Decrease Under 5 Services Extension increasesfrom (2 weeks training mortalityby 60- md Infant program 1,386 to 13,635 per year) increase 70% nortality by two (9.8 fold increase) from 14,527 to Reduction of hird 260,000 ( 17 fold) matemal mortality by 10% HealthExtension Reductionof Worker from 2800 matemaldeaths in2005 to 23,225 (LifetimeRisk of in2015 (8.3 fold) dying) by 40% #nurse midwives Decrease child and increasesfrom maternal mortality 1,559 to 10,590 due to malaria (5.8 fold) Step 3: First level # of Health Centers Nursemidwives US%1.72 Reduction of Further decrease Clinical Upgrade: increases from 423 increasesfrom mortality due to morbidity and to 2590 (6 fold) 1,559 to 15 088 malaria by more mortality due to (8.7 fold) than 50% malaria Reducemorbidity due to STI Step 4: # of Health Centers Number o fnurse US$ 3.50 Reduction of Reduce maternal Comprehensive offering CEOC midwives increase matemal mortality mortality by 75% emergency increases from less from to 1559to by 75% obstetric care than 100 to 3121. 19,443 (by 11.5 expansion and fold) Reduction of child upgrade Number o f health mortality by 70%- officers increase 80% from 632 to 4,154 (by 5.6 fold) Step 5: Referral # of secondreferral Number ofnurse US$9.79 Reducemortality expansion and hospitals increases midwives increase of HIV+ patients upgrade from 36 to 419. from 1559 to (11.6 fold) 22,964 Reducechild ( b y 14 fold) mortality by 75- Number o fmedical 85% doctors increase from 2,032 to 9,626 by 4.7 fold Total: US$20.05 NB:Health Extension workers: 1year oftr; ling invocational schools; Nursemidwives ree years training B. :vel; Health -47- I n moving forward, Ethiopia would need to build on its existing strengths. The Government is committed to poverty reduction and recognizes that health is a vital component of its Sustainable Development and Poverty Reduction Program (SDPRP). 100. Despitebeingone o f the poorest countries inthe world, Ethiopia has been able to reduce infant and child mortality rates, as well as malnutrition rates. The level o f stuntingo funder-five children is still, however, amongthe highest inthe world. Ithas been able to increase coverage rates for certain interventions such as polio immunization and vitamin A supplementation, as well as creatingawareness for familyplanningand HIVIAIDS. Whileit has made achievements in these areas, it has not been able to make the same inroads in other low-cost interventions such as the use of ORT during diarrhea episodes and use of bed nets. Improvements are needed ininformation and service delivery channels. 101. Chapter 8 outlines the following policy issues for discussion with the GOE: (a) Although utilization rates of health services vary based on income levels, absolute utilization rates are still low across all income quintiles. Thus there is the needto provide both supply and demand side interventions to address low utilization rates o f health services. The budgeting-impact simulations inchapter 7 indicate that investing in demand side interventions may buy more for the money than expanding access and quality only. Creating awareness can help ingenerating demandfor services. What other demand side interventions could also bepotentially promising in Ethiopia? Can cash transfers or vouchers be undertaken on a trial basis? Can these transfers be linkedto the use o f other activities e.g. a free bednet for women who attend antenatal care, or packets o f ORT given to children who come for immunization? (b) Healthpersonnel: (i) the skill-mix of health personnel given that Ethiopia would have to rely on apredominantly lower-skill basedpopulation in the short to medium term and the medium to long-term strategyfor upgrading their skills; (ii) the needfor a strategy to attract and retain staffin rural areas. This would need to include career development and training opportunities, and also address staff deployment and rotation issues. Incentives are needed to attract and retainhighly skilled healthpersonnel. Take for example, SNNPR, which has provided additional allowances to staff working in remote areas while also crediting a year o f service inthese areas as two years o f service. (4 Roles o f theprivate sector and NGOs vis-&vis thepublic sector and a strategy for providing a more enabling environment for NGOs and the private sector (d) There is general agreement that Ethiopia's per capita allocations and spending on health are very low and there is a needfor increased investments in health. The Marginal Budgeting for Bottlenecks Model makes a very good case for increasing fundingto health services. For example, just an increase inUSD 1.6per capita invested incommunity-based healthpromotion activities couldreduceunder-five mortality by 20 percent while increasing financing by USD4.87 to finance health extensiodoutreach, family-basedcommunity based, and clinical services could reduce under-five mortality by33 percent andmaternalmortality by22 percent. -48- (e> There is a need to address the reasonsfor low budget execution rates in order to ensure that additionalfunds earmarkedfor the sector are used to meet sectoral objectives. First, it has beenpointed out that one possible reason for underspending i s donor proceduresthat cause delays inspending. Second, underreporting is another factor, along with the weak capacity for programplanninghudgeting and management at the regional, zonal, and Woreda levels. Third, i s the need to change or restructure a costing system that i s based on historical expenditure trends rather than strategic directions outlined inHSDP. Expenditure tracking analysis can help determine the factors that contribute to underspending. (f) Within the context of deepeningdecentralization, what can be done to motivate woredas tofocus their attention on healthpriorities? Will performance-based agreements andmatching grants bepractical? SNNPR, for example, has started piloting performance based contracts with its Woreda officers and it would be important to learn from these experiences. (g> Assessment of resettlement conditions in2004 indicates that conditions were positive in areas where adequate planning, preparation, and resources hadbeen available. However, other areas, especially inOromiya where most resettlement has taken place, were facing inadequate food andwater supplies, as well as health care that required urgent action. (h) Coordinating with other sectors actions that will be critical to the success of health sector interventions,particularly those taken to improve the status and role of women in Ethiopian society. This report finds a strong correlationbetween a mother's level o f education (as well as improved knowledge gathered through other sources such as media) to her children's: infant mortality rate andnutritional status. Additionally, a mother's educational level i s indicative o f the likelihoodo f her availing o fhealth services such as immunization, antenatal care, and family planning services.