Federal Democratic Republic of Ethiopia. Funding for the NHA activities, and printing of this NHA report was provided by the United States Agency for International Development (USAID) through Health Sector Financing Reform/Health Finance &. Executive Summary The health sector in Ethiopia is financed from multiple sources such as federal.
The World malaria report 2017 presents a comprehensive state of play in global progress in the fight against malaria up to the end of 2016. It tracks progress in investments in malaria programmes and research, malaria prevention, diagnosis and treatment, surveillance, trends in malaria disease burden, malaria elimination, and threats in tackling malaria and safeguarding the investments made.

Overview

Investments in malaria programmes and research

Malaria control and elimination investments

In 2016, an estimated US$ 2.7 billion was invested in malaria control and elimination efforts globally by governments of malaria endemic countries and international partners.
The majority (74%) of investments in 2016 were spent in the WHO African Region, followed by the WHO regions of South-East Asia (7%), the Eastern Mediterranean and the Americas (each 6%), and the Western Pacific (4%).
Governments of endemic countries contributed 31% of total funding (US$ 800 million) in 2016.
The United States of America (USA) was the largest international source of malaria financing in 2016, providing US$ 1 billion (38%), followed by the United Kingdom of Great Britain and Northern Ireland (United Kingdom) and other international donors, including France, Germany and Japan.
More than half (57%) of resources in 2016 were channelled through the Global Fund to Fight AIDS, Tuberculosis and Malaria (Global Fund).

Investment outlook

Although funding for malaria has remained relatively stable since 2010, the level of investment in 2016 is far from what is required to reach the first milestone of the GTS, which is a reduction of at least 40% in malaria case incidence and mortality rates globally when compared to 2015 levels.
To reach this milestone, the GTS estimated that annual funding would need to increase to US$ 6.5 billion per year by 2020. The US$ 2.7 billion invested in malaria in 2016 represents less than half (41%) of that amount.
Stepping up investments in malaria research and development is key to achieving the GTS targets. In 2015, US$ 572 million was spent in this area, representing 83% of the estimated annual need for research and development.

Deliveries of malaria commodities

Insecticide-treated mosquito nets
Between 2014 and 2016, a total of 582 million insecticide-treated mosquito nets (ITNs) were reported by manufacturers as having been delivered globally.
Of this amount, 505 million ITNs were delivered in sub-Saharan Africa, compared with 301 million bednets in the preceding 3-year period (2011–2013).
Data from national malaria control programmes (NMCPs) in Africa indicate that, between 2014 and 2016, 75% of ITNs were distributed through mass distribution campaigns.
Rapid diagnostic tests
An estimated 312 million rapid diagnostic tests (RDTs) were delivered globally in 2016. Of these, 269 million were delivered in the WHO African Region.
The number of RDTs distributed by NMCPs increased between 2010 and 2015, but fell from 247 million in 2015 to 221 million in 2016. The decrease was entirely in sub-Saharan Africa, where distributions dropped from 219 million to 177 million RDTs over the 2015–2016 period.
Artemisinin-based combination therapy
An estimated 409 million treatment courses of artemisinin-based combination therapy (ACT) were procured by countries in 2016, an increase from 311 million in 2015. Over 69% of these procurements were reported to have been made for the public sector.
The number of ACT treatments distributed by NMCPs to the public sector increased from 192 million in 2013 to 198 million in 2016. Most of the NMCP distributions of ACTs (99%) in 2016 occurred in the WHO African Region.

Preventing malaria

Vector control

Across sub-Saharan Africa, household ownership of at least one ITN increased from 50% in 2010 to 80% in 2016. However, the proportion of households with sufficient nets (i.e. one net for every two people) remains inadequate, at 43% in 2016.
More people at risk of malaria in Africa are sleeping under an ITN. In 2016, 54% of the population was protected by this intervention, an increase from 30% in 2010.
Fewer people at risk of malaria are being protected by indoor residual spraying (IRS), a prevention method that involves spraying the inside walls of dwellings with insecticides. Globally, IRS protection declined from a peak of 5.8% in 2010 to 2.9% in 2016, with decreases seen across all WHO regions. In the WHO African Region, coverage dropped from 80 million people at risk in 2010 to 45 million in 2016.
The declines in IRS coverage are occurring as countries change or rotate insecticides to more expensive chemicals.

Preventive therapies

To protect women in areas of moderate and high malaria transmission in Africa, WHO recommends “intermittent preventive treatment in pregnancy” (IPTp) with the antimalarial drug sulfadoxinepyrimethamine. Among 23 African countries that reported on IPTp coverage levels in 2016, an estimated 19% of eligible pregnant women received the recommended 3 or more doses of IPTp, compared with 18% in 2015 and 13% in 2014.
In 2016, 15 million children in 12 countries in Africa’s Sahel subregion were protected through seasonal malaria chemoprevention (SMC) programmes. However, about 13 million children who could have benefited from this intervention were not covered, mainly due to a lack of funding. Since 2012, SMC has been recommended by WHO for children aged 3-59 months living in areas of highly seasonal malaria transmission in this subregion.

Diagnostic testing and treatment

Accessing care

Prompt diagnosis and treatment is the most effective means of preventing a mild case of malaria from developing into severe disease and death. Among national-level surveys completed in 18 countries in sub-Saharan Africa between 2014 and 2016 (representing 61% of the population at risk), a median of 47% (interquartile range [IQR]: 38–56%) of children with a fever (febrile) were taken to a trained medical provider for care. This includes public sector hospitals and clinics, formal private sector facilities and community health workers.
More febrile children sought care in the public sector (median: 34%, IQR: 28–44%) than in the private sector (median: 22%, IQR: 14–34%). However, the surveys from Africa also indicate that a high proportion of febrile children did not receive medical attention (median: 39%, IQR: 29–44%). Possible reasons include poor access to health-care providers or lack of awareness among caregivers.

Diagnosing malaria

Among 17 national-level surveys completed in sub-Saharan Africa between 2014 and 2016, the proportion of children with a fever who received a finger or a heel stick – suggesting that a malaria diagnostic test may have been performed – was greater in the public sector (median: 52%, IQR: 34–59%) than in both the formal and informal private sector.
Testing of suspected cases in the public health system increased in most WHO regions since 2010. The WHO African Region recorded the biggest rise, with diagnostic testing in the public health sector increasing from 36% of suspected cases in 2010 to 87% in 2016.

Treating malaria

Among 18 household surveys conducted in sub-Saharan Africa between 2014 and 2016, the proportion of children aged under 5 years with a fever who received any antimalarial drug was 41% (IQR: 21–49%).
A majority of patients (70%) who sought treatment for malaria in the public health sector received ACTs, the most effective antimalarial drugs. Children are more likely to be given ACTs if medical care is sought at public health facilities than in the private sector.
To bridge the treatment gap among children, WHO recommends the uptake of integrated community case management (iCCM). This approach promotes integrated management of common life-threatening conditions in children – malaria, pneumonia and diarrhoea – at health facility and community levels. In 2016, 26 malaria-affected countries had iCCM policies in place, of which 24 had started implementing those policies. An evaluation from Uganda found that districts with iCCM experienced a 21% increase in care-seeking for fever compared with districts without an iCCM policy in place.
Outside the WHO African Region, only a handful of countries in each of the other regions reported having such policies in place, though data on the level of implementation are unavailable for most countries.

Malaria surveillance systems

Effective surveillance of malaria cases and deaths is essential for identifying the areas or population groups that are most affected by malaria, and for targeting resources for maximum impact. A strong surveillance system requires high levels of access to care and case detection, and complete reporting by all health sectors, whether public or private.
In 2016, 37 out of 46 countries in the WHO African Region indicated that at least 80% of public health facilities had reported data on malaria through their national health information system. Rates vary within other WHO regions. For example, in the WHO Eastern Mediterranean Region, only 3 out of 8 countries had 80% or more public health facilities reporting in 2016.
Among 55 countries where the burden of malaria was estimated, 31 countries have a malaria case reporting rate by surveillance systems of less than 50%. This includes the high-burden countries of India and Nigeria.

Global and regional malaria trends in numbers

Malaria cases

In 2016, an estimated 216 million cases of malaria occurred worldwide (95% confidence interval [CI]: 196–263 million), compared with 237 million cases in 2010 (95% CI: 218–278 million) and 211 million cases in 2015 (95% CI: 192–257 million).
Most malaria cases in 2016 were in the WHO African Region (90%), followed by the WHO South-East Asia Region (7%) and the WHO Eastern Mediterranean Region (2%).
Of the 91 countries reporting indigenous malaria cases in 2016, 15 countries – all in sub-Saharan Africa, except India – carried 80% of the global malaria burden.
The incidence rate of malaria is estimated to have decreased by 18% globally, from 76 to 63 cases per 1000 population at risk, between 2010 and 2016. The WHO South-East Asia Region recorded the largest decline (48%) followed by the WHO Region of the Americas (22%) and the WHO African Region (20%).
Despite these reductions, between 2014 and 2016, substantial increases in case incidence occurred in the WHO Region of the Americas, and marginally in the WHO South-East Asia, Western Pacific and African regions.
Plasmodium falciparum is the most prevalent malaria parasite in sub-Saharan Africa, accounting for 99% of estimated malaria cases in 2016. Outside of Africa, P. vivax is the predominant parasite in the WHO Region of the Americas, representing 64% of malaria cases, and is above 30% in the WHO South- East Asia and 40% in the Eastern Mediterranean regions.
New data from improved surveillance systems in several countries in the WHO African Region indicate that the number of malaria cases presented in this year’s report are conservative estimates. WHO will review its malaria burden estimation methods for sub-Saharan Africa in 2018.

Malaria deaths

In 2016, there were an estimated 445 000 deaths from malaria globally, compared to 446 000 estimated deaths in 2015.
The WHO African Region accounted for 91% of all malaria deaths in 2016, followed by the WHO South- East Asia Region (6%).
Fifteen countries accounted for 80% of global malaria deaths in 2016; all of these countries are in sub-Saharan Africa, except for India.
All regions recorded reductions in mortality in 2016 when compared with 2010, with the exception of the WHO Eastern Mediterranean Region, where mortality rates remained virtually unchanged in the period. The largest decline occurred in the WHO regions of South-East Asia (44%), Africa (37%) and the Americas (27%).
However, between 2015 and 2016, mortality rates stalled in the WHO regions of South-East Asia, the Western Pacific and Africa, and increased in the Eastern Mediterranean and the Americas.

Malaria elimination

Globally, more countries are moving towards elimination: in 2016, 44 countries reported fewer than 10 000 malaria cases, up from 37 countries in 2010.
Kyrgyzstan and Sri Lanka were certified by WHO as malaria free in 2016.
In 2016, WHO identified 21 countries with the potential to eliminate malaria by the year 2020. WHO is working with the governments in these countries – known as “E-2020 countries” – to support their elimination acceleration goals.
Although some of E-2020 countries remain on track to achieve their elimination goals, 11 have reported increases in indigenous malaria cases since 2015, and 5 countries reported an increase of more than 100 cases in 2016 compared with 2015.

Challenges to achieving a malaria-free world

Some of the challenges impeding countries’ abilities to stay on track and advance towards elimination include lack of sustainable and predictable international and domestic funding, risks posed by conflict in malaria endemic zones, anomalous climate patterns, the emergence of parasite resistance to antimalarial medicines and mosquito resistance to insecticides.
WHO is supporting malaria emergency responses in Nigeria, South Sudan, Venezuela (Bolivarian Republic of) and Yemen, where ongoing humanitarian crises pose serious health risks. In Nigeria’s Borno State, WHO supported the launch of a mass antimalarial drug administration campaign that reached an estimated 1.2 million children aged under 5 years in targeted areas. Early results point to a reduction in malaria cases and deaths in this state.

Funding

In 34 out of 41 high-burden countries, which rely mainly on external funding for malaria programmes, the average level of funding available per person at risk in the past 3 years (2014–2016) reduced when compared with 2011–2013. Exceptions were Democratic Republic of the Congo, Guinea, Mauritania, Mozambique, Niger, Pakistan and Senegal, which recorded increases.
Among the 41 high-burden countries, overall, funding per person at risk of malaria remains below US$ 2.

Histidine-rich protein 2 deletions

In some settings, increasing levels of histidine-rich protein 2 gene (HRP2) deletions threaten the ability to diagnose and appropriately treat people infected with falciparum malaria. An absence of the HRP2 gene enables parasites to evade detection by HRP2-based RDTs, resulting in a false-negative test result. Although the prevalence of HRP2 gene deletions in most high-transmission countries remains low, further monitoring is required.

Drug resistance

ACTs have been integral to the recent success of global malaria control, and protecting their efficacy for the treatment of malaria is a global health priority.
Although multidrug resistance, including artemisinin (partial) resistance and partner drug resistance, has been reported in five countries of the Greater Mekong subregion (GMS), there has been a massive reduction in malaria cases and deaths in this subregion. Monitoring the efficacy of antimalarial drugs has led to timely treatment policy updates across the GMS.
In Africa, artemisinin (partial) resistance has not been reported to date and first-line ACTs remain efficacious in all malaria endemic settings.

Insecticide resistance

Of the 76 malaria endemic countries that provided data for 2010 to 2016, resistance to at least one insecticide in one malaria vector from one collection site was detected in 61 countries. In 50 countries, resistance to 2 or more insecticide classes was reported.
In 2016, resistance to one or more insecticides was present in all WHO regions, although the extent of monitoring varied.
Resistance to pyrethroids – the only insecticide class currently used in ITNs – is widespread. The proportion of malaria endemic countries that monitored and subsequently reported pyrethroid resistance increased from 71% in 2010 to 81% in 2016. The prevalence of confirmed resistance to pyrethroids differed between regions, and was highest in the WHO African and Eastern Mediterranean regions, where it was detected in malaria vectors in over two thirds of all sites monitored.
ITNs continue to be an effective tool for malaria prevention, even in areas where mosquitoes have developed resistance to pyrethroids. This was evidenced in a large multicountry evaluation coordinated by WHO between 2011 and 2016, which did not find an association between malaria disease burden and pyrethroid resistance across study locations in 5 countries.

Related links

Abstract
?Mathematical formulae have been encoded as MathML and are displayed in this HTML version using MathJax in order to improve their display. Uncheck the box to turn MathJax off. This feature requires Javascript. Click on a formula to zoom.

Abstract

Ethiopia's recent improvements in health outcomes benefited from the large increase in development assistance for health received in recent years, most of which supported its primary care system. Increased domestic resource mobilization for health will be needed to sustain progress given recent and likely future declines in external support. We estimate a projection model of Ethiopian government domestic resource mobilization potential compared with future health care delivery costs in order to assess Ethiopia's ability to finance its planned primary health care system. For the period of Ethiopia's current five-year health sector plan (2016–2020), the projection model indicates that if real external resources remain at the levels of 2011, domestic resource mobilization may only provide half of the estimated funds needed. Including out-of-pocket spending currently captured as retained user fees, Ethiopia is more likely to successfully finance continued delivery of primary care. Over 20 years, 2016–2035, the future sustainability of primary care services without increasing contributions from households will depend largely on significant economic growth and more government funding for primary care. Our modeling suggests that Ethiopia can substantially support further development of primary care services, even in the face of declining external support. However, it is unlikely to achieve its goals solely through “business as usual.” Ethiopia is already moving forward with timely adoption of sound strategies to maintain progress. External partners should support these trends to enable transition plans to greater domestic funding with minimal disruption to positive progress that has been and is being made.

Keywords: Ethiopia, health care costs, health finance, projection model, resource mobilization

INTRODUCTION

Global health progress has benefited from a rapid and large increase in development assistance for health beginning in the 1990s and extending to the mid 2010s. Development assistance for health was around seven billion USD in 1990 and exceeded 30 billion USD by 2015 globally (2015 USD).1 Institute for Health Metrics and Evaluation (IHME). Financing global health 2016: development assistance, public and private health spending for the pursuit of universal health coverage. Seattle (WA): Institute for Health Metrics and Evaluation; 2017 [accessed 2017Nov18]. http://www.healthdata.org/sites/default/files/files/policy_report/FGH/2017/IHME_FGH2016_Technical-Report.pdf.[Google Scholar] Recent evidence suggests, however, that this trend has leveled off or slightly declined in aggregate.1 Institute for Health Metrics and Evaluation (IHME). Financing global health 2016: development assistance, public and private health spending for the pursuit of universal health coverage. Seattle (WA): Institute for Health Metrics and Evaluation; 2017 [accessed 2017Nov18]. http://www.healthdata.org/sites/default/files/files/policy_report/FGH/2017/IHME_FGH2016_Technical-Report.pdf.[Google Scholar]

Individual countries benefited in different ways from these increased financial flows. In some sub-Saharan African countries, external funds make up a large share of total health spending or of total government health spending. In Ethiopia, the 2010 National Health Accounts (NHA) estimated that external funds accounted for 50% of total health expenditure, of which 63% went toward total government health expenditure.2 Ethiopia Federal Ministry of Health. Ethiopia's fifth national health accounts, 2010/2011. Addis Ababa (Ethiopia): The Federal Ministry of Health; 2014 [accessed 2015Jan15]. http://apps.who.int/nha/database/DocumentationCentre/Index/en.[Google Scholar] In addition, within countries, external funds could be an even larger share of government spending on specific services. For example, in Ethiopia, external funding is estimated to make up 73% of government primary health care spending (authors' estimates using data collated from Ethiopia Federal Ministry of Health2 Ethiopia Federal Ministry of Health. Ethiopia's fifth national health accounts, 2010/2011. Addis Ababa (Ethiopia): The Federal Ministry of Health; 2014 [accessed 2015Jan15]. http://apps.who.int/nha/database/DocumentationCentre/Index/en.[Google Scholar]), reflecting the priority given by external funders to supporting Ethiopia's successful efforts to achieve the health-related Millennium Development Goals by 2015.3 Ruducha J, Mann C, Singh NS, Gemebo TD, Tessema NS, Bashieri A, Friberg I, Zerfu TA, Yassin M, Franca GA, et al. How Ethiopia achieved Millennium Development Goal 4 through multisectoral interventions: a countdown to 2015 case study. Lancet Glob Health. 2017;5(11):e1142–e1151. doi:10.1016/S2214-109X(17)30331-5.[Crossref], [PubMed], [Web of Science ®], [Google Scholar]

Ethiopia has received international praise for its focus on developing primary health care services with a bottom-up strategy that emphasized community engagement through volunteer community health workers, lower-level facilities with locally recruited health extension workers, and development of health centers and primary hospitals. The country has ambitious plans to advance toward universal health coverage through a primary health care approach, continuing to strengthen this system by expanding coverage, quality, and scope of services.

External partners have been a major source of funding for Ethiopia's government primary care system. External resources for health increased since 1995 from six million USD to 883 million USD in 2014 (2012 USD).4 Mann C, Ng C, Akseer N, Bhutta ZA, Borghi J, Colbourn T, Hernández-Peña P, Huicho L, Malik MA, Martinez-Alvarez M, et al. Countdown to 2015 country case studies: what can analysis of national health financing contribute to understanding MDG 4 and 5 progress? BMC Public Health. 2016;16(Suppl. 2):66–81. doi:10.1186/s12889-016-3403-4.[Crossref], [Google Scholar],5 Federal Democratic Republic of Ethiopia Ministry of Health. Ethiopia health accounts, 2013/14. Addis Ababa (Ethiopia): The Federal Ministry of Health; 2017 [accessed 2018Feb12]. http://apps.who.int/nha/database/DocumentationCentre/Index/en.[Google Scholar] However, Ethiopia is now experiencing a decline in some external support for health. For example, cuts in Global Fund (GFATM) and the President's Emergency Plan for AIDS Relief (PEPFAR) funding for the country resulted in decreases of their support by 50% by 2012 from a 2010 peak.6Glassman A. Ethiopia's AIDS spending cliff. Washington (DC): Center for Global Development; 2012 [accessed 2018Feb13]. https://www.cgdev.org/blog/ethiopia's-aids-spending-cliff.[Google Scholar],7 Health Works Collective. PEPFAR funding allocations: FY2011–FY2013. Washington (DC): Health Works Collective; 2012 [accessed 2018Feb13]. https://www.healthworkscollective.com/wp-content/uploads/2012/09/PEPFAR-Country-Allocations-FY11-131.pdf.[Google Scholar] Despite this, overall external support was still increasing through 2014.

Domestically, Ethiopia has benefited from rapid economic growth, with an average of 9% real growth in gross domestic product (GDP) from 1999 to 2012. It is deemed one of the fastest growing economies in Africa.8Alebachew A, Yusuf Y, Mann C, Berman P. Ethiopia's progress in health financing and the contribution of the 1998 health care and financing strategy in Ethiopia. Addis Ababa (Ethiopia) and Boston (MA): Harvard T.H. Chan School of Public Health and Breakthrough International Consultancy, PLC; 2015 [accessed 2016Dec5]. https://cdn1.sph.harvard.edu/wp-content/uploads/sites/1578/2013/09/HCFS-Review-Report.HSPH-BIC.-FINAL-June-1-2015.pdf.[Google Scholar] The government of Ethiopia (GoE) has increased its funding for health from own resources over this period as well, almost doubling public health expenditure from 1995 to 2011 (2012 USD),4 Mann C, Ng C, Akseer N, Bhutta ZA, Borghi J, Colbourn T, Hernández-Peña P, Huicho L, Malik MA, Martinez-Alvarez M, et al. Countdown to 2015 country case studies: what can analysis of national health financing contribute to understanding MDG 4 and 5 progress? BMC Public Health. 2016;16(Suppl. 2):66–81. doi:10.1186/s12889-016-3403-4.[Crossref], [Google Scholar] although government spending from own sources still is relatively low at 4.4% of GDP in 2011.9Berman P, Mann C, Ricculli M-L. Financing Ethiopia's primary care to 2035: a model projecting resource mobilization and costs. Boston (MA): The Harvard T.H. Chan School of Public Health; 2015 [accessed 2018Feb16]. https://cdn1.sph.harvard.edu/wp-content/uploads/sites/1325/2013/01/Finance-Projection-Model-Report-Feb-19-2016.pdf.[Google Scholar]

The GoE recognizes the need to increase domestic resource mobilization for health and has stated this as one of the key goals in its recent health financing strategy and health sector plan.10 Federal Ministry of Health. Health care financing strategy 2017–2025. Addis Ababa (Ethiopia): The Federal Ministry of Health; 2017.[Google Scholar],11 Federal Ministry of Health. Health sector transformation plan (2015/16–2019/20). Addis Ababa (Ethiopia): Federal Ministry of Health; 2015 [accessed 2017Nov29]. https://www.globalfinancingfacility.org/sites/gff_new/files/Ethiopia-health-system-transformation-plan.pdf.[Google Scholar] Working with Ethiopia's Federal Ministry of Health, a research team from Harvard T.H. Chan School of Public Health developed a health financing projection model to explore the implications of future trends in external funding on Ethiopia's primary health care spending and the potential for domestic resources to sustain the growth of Ethiopia's primary health care system. We empirically estimate scenarios exploring Ethiopia's ability to finance its primary health care system through the development of a projection model for resources mobilized and costs. The work presented in this article provides empirical data on an important question: Can Ethiopia finance the needed health investments to sustain its health progress in primary health care? This question has relevance beyond Ethiopia because other developing economies are also facing changes in the availability of external resources while they seek to achieve future health gains.

SETTING AND METHODS

A more detailed report on this work is available in Berman et al.,9Berman P, Mann C, Ricculli M-L. Financing Ethiopia's primary care to 2035: a model projecting resource mobilization and costs. Boston (MA): The Harvard T.H. Chan School of Public Health; 2015 [accessed 2018Feb16]. https://cdn1.sph.harvard.edu/wp-content/uploads/sites/1325/2013/01/Finance-Projection-Model-Report-Feb-19-2016.pdf.[Google Scholar] which provides specific content on the current health financing trends (according to the National Health Accounts) and comparison with other lower-income countries with a focus on sub-Saharan Africa, trends in economic development and general government expenditures, more detail on the resource mobilization projections and their assumptions, and a step-by-step process for the cost function approach. This section provides details of the different elements of the model starting with the conceptual framework of the projection model, estimating spending on primary care (versus nonprimary care), highlighting the methods and assumptions projecting future spending and estimating future costs of government primary care, and comparing the future cost estimates with the Ethiopian Federal Ministry of Health's estimates of primary care costs according to their five-year health sector plan.

Study Setting

Ethiopia is the second most populous sub-Saharan African nation, with an estimated population of 99.4 million people in 2016.12 The World Bank Group. Ethiopia, economic overview. Washington (DC): The World Bank Group; 2016 [accessed 2016Nov5]. http://www.worldbank.org/en/country/ethiopia/overview.[Google Scholar] Assuming an annual average population growth of 2%,13 Central Statistical Agency. Population projections for Ethiopia 2007–2037. Addis Ababa (Ethiopia): Central Statistical Agency; 2013.[Google Scholar] Ethiopia's health system will have to serve 62% more people by 2035 compared to 2012 levels (authors' estimates based on the estimated population growth from the Central Statistical Agency13 Central Statistical Agency. Population projections for Ethiopia 2007–2037. Addis Ababa (Ethiopia): Central Statistical Agency; 2013.[Google Scholar]; Table 1). Over the last 20 years, the Ethiopian government has focused on rapidly expanding its public health system with a focus on primary care service delivery through the Health Extension Program.3 Ruducha J, Mann C, Singh NS, Gemebo TD, Tessema NS, Bashieri A, Friberg I, Zerfu TA, Yassin M, Franca GA, et al. How Ethiopia achieved Millennium Development Goal 4 through multisectoral interventions: a countdown to 2015 case study. Lancet Glob Health. 2017;5(11):e1142–e1151. doi:10.1016/S2214-109X(17)30331-5.[Crossref], [PubMed], [Web of Science ®], [Google Scholar],4 Mann C, Ng C, Akseer N, Bhutta ZA, Borghi J, Colbourn T, Hernández-Peña P, Huicho L, Malik MA, Martinez-Alvarez M, et al. Countdown to 2015 country case studies: what can analysis of national health financing contribute to understanding MDG 4 and 5 progress? BMC Public Health. 2016;16(Suppl. 2):66–81. doi:10.1186/s12889-016-3403-4.[Crossref], [Google Scholar] The frontline primary health care workers (health extension workers) increased from less than 3000 in 2001 to 35,000 by 2012, and health posts density increased from less than 0.2 per 10,000 people in 2001 to 2 per 10,000 people by 2012.3 Ruducha J, Mann C, Singh NS, Gemebo TD, Tessema NS, Bashieri A, Friberg I, Zerfu TA, Yassin M, Franca GA, et al. How Ethiopia achieved Millennium Development Goal 4 through multisectoral interventions: a countdown to 2015 case study. Lancet Glob Health. 2017;5(11):e1142–e1151. doi:10.1016/S2214-109X(17)30331-5.[Crossref], [PubMed], [Web of Science ®], [Google Scholar] Continued investments are planned to expand the scope and improve the quality of the health care system over the next 20 years.11 Federal Ministry of Health. Health sector transformation plan (2015/16–2019/20). Addis Ababa (Ethiopia): Federal Ministry of Health; 2015 [accessed 2017Nov29]. https://www.globalfinancingfacility.org/sites/gff_new/files/Ethiopia-health-system-transformation-plan.pdf.[Google Scholar]

Published online:
21 June 2018

TABLE 1. Ethiopian Demographic and Macroeconomic Indicators

Ethiopia has been able to maintain a high level of real GDP growth. Reporting more than an average of 10% of annual real GDP growth rate since 2003, it has become one of the fastest growing economies in sub-Saharan Africa.4 Mann C, Ng C, Akseer N, Bhutta ZA, Borghi J, Colbourn T, Hernández-Peña P, Huicho L, Malik MA, Martinez-Alvarez M, et al. Countdown to 2015 country case studies: what can analysis of national health financing contribute to understanding MDG 4 and 5 progress? BMC Public Health. 2016;16(Suppl. 2):66–81. doi:10.1186/s12889-016-3403-4.[Crossref], [Google Scholar] Despite this recent rapid economic growth, Ethiopia continues to be one of the poorest countries, with a gross national income per person of 590 USD in 2016.12 The World Bank Group. Ethiopia, economic overview. Washington (DC): The World Bank Group; 2016 [accessed 2016Nov5]. http://www.worldbank.org/en/country/ethiopia/overview.[Google Scholar]

Projection Model Methods

The conceptual framework developed for our modeling is shown in Figure 1. Our focus is on government-provided primary health care. For the most recent year with actual expenditures, we assume that the government's reported spending on primary care is equal to the costs of services delivered. However, when making future projections, we estimate costs as the resources needed to deliver the services that the government intends to provide, and we compare that with estimates of the resources the government is likely to mobilize. Both of these estimates are based on specific assumptions and prior empirical data. For future projections, we are interested in estimating the differences between likely resources available and the costs of services intended to be provided.

Published online:
21 June 2018
FIGURE 1. Conceptual Framework: Projections for the Future.
FIGURE 1. Conceptual Framework: Projections for the Future.

“Primary care” for this model is based on the Ethiopian Federal Ministry of Health's (FMOH) working definition. Primary care is defined as all health services delivered by health care providers located at the woreda (district) level or below. This is defined in Ethiopia as the primary health care unit (PHCU) that provides promotive, preventative, outpatient curative care services, and some inpatient services (in some health centers and recently established “primary hospitals”). PHCU providers include health posts, health centers, and primary hospitals, as well as health promotion and education activities of volunteer community health workers known as the “health development army.” In addition, we include outpatient primary care services offered at secondary and tertiary facilities (hospitals) and an estimate of the share of health system administration costs at federal, regional, and woreda levels associated with those services.

“Nonprimary care” consists of providers of specialized care (e.g., elderly and mental health), research and education training institutions, pharmacies, and suppliers of medical supplies. It also includes FMOH's definition of secondary and tertiary health care services (excluding the outpatient primary care services offered at these facilities). An estimate of the share of health system administration costs at federal and regional levels associated with these services was also included.

The model explores two projection periods. One time period is the short term of five years, aligned with the Health Sector Transformation Plan (HSTP; 2016–2020). The second time period is aligned with the government's strategic planning period of 20 years (2016–2035).

Estimating Actual Spending on Government-Provided Primary Care

Current actual spending on government provided primary care is based on Ethiopia's 2010–2011 NHA.2 Ethiopia Federal Ministry of Health. Ethiopia's fifth national health accounts, 2010/2011. Addis Ababa (Ethiopia): The Federal Ministry of Health; 2014 [accessed 2015Jan15]. http://apps.who.int/nha/database/DocumentationCentre/Index/en.[Google Scholar],[a][a] Ethiopia's most recent NHA was not included in this work because it was released in the fall of 2017, after this study was completed.View all notesFigure 2 shows the shares of spending from different financing sources going to primary care providers (as defined above). The main funders of government primary care are government itself, households, and “rest of the world.” Sixty-four percent of total health expenditures went to primary care and the remaining 36% was for nonprimary care.9Berman P, Mann C, Ricculli M-L. Financing Ethiopia's primary care to 2035: a model projecting resource mobilization and costs. Boston (MA): The Harvard T.H. Chan School of Public Health; 2015 [accessed 2018Feb16]. https://cdn1.sph.harvard.edu/wp-content/uploads/sites/1325/2013/01/Finance-Projection-Model-Report-Feb-19-2016.pdf.[Google Scholar] We combined the NHA breakdowns on sources, providers, and health care functions to estimate the shares of government, household, and rest of the world primary care spending that went to government primary care providers.

Published online:
21 June 2018
FIGURE 2. Estimated allocations for primary care and non-primary care expenditures using NHA 5 (2010–2011).
FIGURE 2. Estimated allocations for primary care and non-primary care expenditures using NHA 5 (2010–2011).

Monetary values of projections are presented in nominal Ethiopian birr (ETB) to avoid complications of estimating changes in exchange rates between ETB and the USD. This allows for inflation to be easily removed and obtain real values in ETB as well as estimate the purchasing power parity for ease of comparing values across countries. In some cases, monetary values are also presented in current USD using the 2015 exchange rate of 20.7 ETB to 1 USD14 International Monetary Fund. Staff report for the 2015 Article IV consultation, Federal Democratic Republic of Ethiopia. Washington (DC): International Monetary Fund; 2015 [accessed 2016Nov2]. https://www.imf.org/external/pubs/ft/scr/2015/cr15300.pdf.[Google Scholar] in parentheses after reported ETB.

Projecting Future Resource Mobilization for Government Primary Care

The health care finance projection model makes estimates using two approaches and for each approach, two scenarios for future resources mobilized from different financing sources for government primary care. The two approaches are referred to as historically based and the GDP-based projections. The historically based projections extrapolate previous trends in expenditure growth for financing sources—government, external, household out-of-pocket (OOP), and other private. We modify these trends based on certain assumptions. For the GDP-based approach, we assume that the financing sources will grow based on projected growth rates for GDP according to the International Monetary Fund (IMF) Article IV Consultations and estimates of expenditure elasticities in relation to GDP growth. (The GDP growth rates used in Berman et al.9Berman P, Mann C, Ricculli M-L. Financing Ethiopia's primary care to 2035: a model projecting resource mobilization and costs. Boston (MA): The Harvard T.H. Chan School of Public Health; 2015 [accessed 2018Feb16]. https://cdn1.sph.harvard.edu/wp-content/uploads/sites/1325/2013/01/Finance-Projection-Model-Report-Feb-19-2016.pdf.[Google Scholar] are slightly different than the more up-to-date estimates used for this article.) Because we are particularly interested in the impact of changes in external financing, we incorporate explicit assumptions about that source of funding.

Table 2 presents the data sources used under each projection approach, main assumptions, and more detail about the two scenarios. For each scenario, we are primarily interested in the impact of changes in external contributions for health and therefore the two scenarios will be referred to as external current and external decrease for each projection approach. For external current funding, we assume growth only at the current U.S. level of inflation from 2011 (i.e., no growth in real international value). For the external decrease scenario, we assume the external current scenario until 2020, at which time a gradual reduction to 50% of 2020 levels would occur by 2035.

Published online:
21 June 2018

TABLE 2. Summary of Key Inputs for Historically Based and GDP-Based Projection Approaches under the Resource Mobilization Model

Reflecting increased attention to domestic resource mobilization, we assume that government health expenditure would increase to the highest historical level of 6.7% of general government expenditure (GGE) by 2025 and remain at that proportion through 2035 for both resource mobilization scenarios. Household OOP spending is based on past average per capita health spending on primary health care services and population projections,13 Central Statistical Agency. Population projections for Ethiopia 2007–2037. Addis Ababa (Ethiopia): Central Statistical Agency; 2013.[Google Scholar],15Purvis G, Alebachew A, Feleke W. Ethiopia health sector financing reform midterm project evaluation. Washington (DC): United States Agency for International Development; 2011 [accessed 2015Aug9]. http://pdf.usaid.gov/pdf_docs/Pdact293.pdf.[Google Scholar] whereas other private spending assumptions are based on the previous trends for the historically based approach. For the GDP-based approach, household OOP is based on the estimated elasticity of health spending for low-income countries (LICs) in sub-Saharan Africa16Avila C, Connor C, Amico P. Universal coverage of essential health services in sub-Saharan Africa. Bethesda (MD): Health Finance & Governance Project, Abt Associates, Inc.; 2013 [accessed 2015Aug1]. https://www.hfgproject.org/wp-content/uploads/2013/09/Universal-Coverage-of-Essential-Health-Services-in-Sub-Saharan-Africa-Pr.pdf.[Google Scholar] and influenced by assumed GDP growth. In Ethiopia, private health sector development is mainly an urban phenomenon in a predominantly rural setting. The elasticity of health spending for “other private” (for-profit and not-for-profit health spending not including households) in Ethiopia used the real GDP growth estimates from the World Bank World Development Indicators DataBank17 World Bank. World Development Indicators. Washington (DC): The World Bank Group; 2016 [accessed 2016Mar5]. http://databank.worldbank.org/data/reports.aspx?source=world-development-indicators.[Google Scholar] and other private expenditures from the NHAs was 0.42. According to Avila et al.,16Avila C, Connor C, Amico P. Universal coverage of essential health services in sub-Saharan Africa. Bethesda (MD): Health Finance & Governance Project, Abt Associates, Inc.; 2013 [accessed 2015Aug1]. https://www.hfgproject.org/wp-content/uploads/2013/09/Universal-Coverage-of-Essential-Health-Services-in-Sub-Saharan-Africa-Pr.pdf.[Google Scholar] the elasticity of health spending for other private among LICs in sub-Saharan African countries was 1.26. We used the estimated elasticity of other private health spending using Ethiopian data given that the Ethiopian private health sector is not expanding as rapidly compared to other LICs within sub-Saharan Africa and with continued emphasis on the government health system in strategies and policies for the future health sector development.10 Federal Ministry of Health. Health care financing strategy 2017–2025. Addis Ababa (Ethiopia): The Federal Ministry of Health; 2017.[Google Scholar],11 Federal Ministry of Health. Health sector transformation plan (2015/16–2019/20). Addis Ababa (Ethiopia): Federal Ministry of Health; 2015 [accessed 2017Nov29]. https://www.globalfinancingfacility.org/sites/gff_new/files/Ethiopia-health-system-transformation-plan.pdf.[Google Scholar]Table 3 summarizes the annual nominal ETB growth rate in the different quantities estimated for total health expenditure. A fixed percentage allocation for each financing source is used to obtain resources mobilized for primary care.9Berman P, Mann C, Ricculli M-L. Financing Ethiopia's primary care to 2035: a model projecting resource mobilization and costs. Boston (MA): The Harvard T.H. Chan School of Public Health; 2015 [accessed 2018Feb16]. https://cdn1.sph.harvard.edu/wp-content/uploads/sites/1325/2013/01/Finance-Projection-Model-Report-Feb-19-2016.pdf.[Google Scholar]

Published online:
21 June 2018

TABLE 3. Nominal Annual Percentage Change for Each Approach and Scenario for Estimated Total Health Expenditure

We find that our resource mobilization model is not that different from the health expenditures documented in the most recent NHA (2013–2014) released in 2017. Total health expenditure in 2013–2014 was 49.6 billion birr, whereas our estimates were lower at 35.4 billion birr and 40.3 billion birr under the historical approach and GDP-based approach, respectively. However, the methods used to calculate the more recent figures may include quantities not included in our figures.

Estimating Future Resource Needs (Costs) of Government Primary Care

Our main approach to estimating the resource need forward-looking projection was to develop a cost function. Data from the FMOH's annual woreda-based planning exercise provided estimates of annual spending in PHCUs for 808 woredas for 2011, the most recent year of data availability at the time of this study. We combined this information with other woreda-level data on population and outputs and coverage of key primary care services to identify a regression model that predicted costs based on key essential primary care services at the woreda level. The model specification consists of primary care per capita costs, key coverage indicators (that provided the best predictive model according to the adjusted R2), socioeconomic status (in the form of the poverty rate or percentage of people living below the poverty line), and control for regional variations.(1) y2011i=α+β1lnSBA2011i+β2PMTCT2011i+β3OPD2011i+β4LLIN2011i+β5Poverty2011i+β6RegDumri+⋯+β15RegDumri+εi(1)

All data in the model were from 2011 and at the woreda level (i). This was the model with the best goodness of fit (highest adjusted R2). On the left-hand side of the equation, the dependent variable (y) is the natural log of primary care costs per capita. Per capita costs were used in order to account for population. The natural log was taken because that variable was not normally distributed, and this violated a model assumption that all variables are normally distributed. The same was done for skilled birth attendants (lnSBA), which is the percentage of eligible mothers whose delivery was attended by an SBA. All other variables used in this model were normally distributed. Preventative and promotive care offered within the primary care system include the percentage of eligible people counseled and tested on the prevention of mother-to-child transmission (PMTCT) of the human immunodeficiency virus and the percentage of long-lasting insecticide nets (LLIN) distributed based on the number of eligible population. The outpatient department rate per capita (OPD) accounts for basic curative services offered within the primary care system. This is controlled for socioeconomic status, or the percentage of people living below the poverty line (Poverty) of a woreda (i) and regional fixed effects with a regional dummy (RegDum) for each region, r, at the woreda level i. Summary statistics of the variables used in this model are reported in the Appendix.

This model was used to project future costs based on changes in key parameters—health coverage, socioeconomic status, population growth, scope of services, and inflation—forecasted over 20 years. Adjustments were also made to include some future capital expenditures for health facilities not captured in 2011 expenditure data,[b][b] In some instances, regions used their own treasury funds to provide additional financial resources directly to health centers. Unique cases such as this are not captured in the primary care cost model, and this is a caveat within this projection, leading the model to underestimate such costs.View all notes although other key factors, such as future quality improvements, are not modeled.

The cost function estimates for primary care are also compared to a recent government costing exercise that was produced for Ethiopia's 2015–2020 HSTP11 Federal Ministry of Health. Health sector transformation plan (2015/16–2019/20). Addis Ababa (Ethiopia): Federal Ministry of Health; 2015 [accessed 2017Nov29]. https://www.globalfinancingfacility.org/sites/gff_new/files/Ethiopia-health-system-transformation-plan.pdf.[Google Scholar] in Table 4. The HSTP costing was developed using the World Health Organization's OneHealth tool and is based on normative service cost estimates for a standard set of services to be provided in contrast to our cost function approach, which is based on reported actual expenditures. It provides two costing scenarios: base case and best case. The base case scenario is based on achieving the government's priority primary care targets from 2016 to 2020, and the best case scenario is based on achieving all targets in the HSTP as well as having high coverage of noncommunicable disease interventions by 2020.11 Federal Ministry of Health. Health sector transformation plan (2015/16–2019/20). Addis Ababa (Ethiopia): Federal Ministry of Health; 2015 [accessed 2017Nov29]. https://www.globalfinancingfacility.org/sites/gff_new/files/Ethiopia-health-system-transformation-plan.pdf.[Google Scholar]

Published online:
21 June 2018

TABLE 4. Comparison of HSTP Cost Estimates with Cost Function Approach

These two approaches are likely to give different results. The cost function approach is based on actual spending on services adjusted for increases in coverage, scope, and time-varying factors. This estimate incorporates any underspending from existing gaps in utilization, quality differences, and scale and scope efficiencies in delivery. In addition, the woreda-based costs probably suffer from data quality problems. Data quality concerns consist of possible reporting errors in the HMIS (e.g., due to multistep aggregation from paper-based records before data are entered electronically) and incomplete woreda data on spending because of off-budget spending or in-kind provisions either from community contributions or external support. The HSTP estimates are based on normative unit costs of services multiplied by planned utilization. The base case scenario for HSTP is about 58% higher on average than our cost estimates for primary care between 2015–2016 and 2019–2020.

Though we found no ideal way to project future costs of government primary care in Ethiopia, we feel that the two cost approaches (HSTP or cost function approach) likely represent high or low estimates of future resources needed to deliver primary care services. The HSTP cost estimates are potentially overestimates of the resource need because they are based on normative costs and standards to provide primary care. The cost function approach is an underestimate of the resources needed due to limited inclusion of capital investments, future changes in services offered among primary care facilities to meet changes in health needs, and future improvements that may be made in quality of services provided. We believe that the best estimate value probably lies between these two projections.

RESULTS

Applying our methods, approaches, and scenarios, we produced two sets of cost and resource mobilization projections. The first projection focuses on the five-year planning period for HSTP, whereas the second presents the longer planning period of 20 years.

Primary Care Financing and Cost Projections (2016–2020)

The first set of projections uses the HSTP costings and the cost model estimates and compares them with the two resource mobilization approaches. Figure 3 shows the resource mobilization results for both the historical and GDP-based approaches, where within the first five years the external decrease scenario is the same as the external current scenario as described in the Methods section. This suggests that, even if external resources remained at the previous high levels in 2011 (adjusted for annual inflation), Ethiopia may only raise about half the estimated funds needed to meet HSTP objectives from increases in domestic resource mobilization excluding household contributions.

Published online:
21 June 2018
FIGURE 3. Five-year Projection Period Estimates (2016–2020).
FIGURE 3. Five-year Projection Period Estimates (2016–2020).

The graph on the right-hand side in Figure 3 adds estimates of household OOP spending that finance user fees in government primary health care facilities into the resource mobilization projections. Ethiopia has a robust system of collecting fees for “nonexempt” services (services not provided free of cost to all). These fees are retained by facilities and woredas and are used mainly to purchase consumables for primary care. Incorporating these estimates in the projections shows that Ethiopia is less likely to be successful in financing continued delivery of primary care without household contributions even if external funding remains at 2011 levels (adjusted for inflation). Using the historical approach, even including household OOP, primary care financing does not reach the resources needed for the HSTP costing scenarios. If domestic resource mobilization and household OOP align more with GDP growth and both government and household spending contribute, domestic funds may be sufficient to finance the “best” case HSTP cost projections toward the end of the projection period.

Ethiopia is currently rapidly expanding a program of community-based health insurance (CBHI), which would create a prepaid fund at the district level with some external government subsidies. This major new initiative is to pool and channel household contributions largely in support of primary care by covering user fees charged at government health facilities. It would entitle members to use government services without user fees, essentially substituting prepayment for out-of-pocket spending at the time of service use. Early evidence indicates a reduction in user fees being paid by CBHI members, and this has reduced the effect of impoverishment among CBHI households because of its reduction in payment at the time of service use.18 Ethiopian Health Insurance Agency. Evaluation of community-based health insurance pilot schemes in Ethiopia: final report. Addis Ababa (Ethiopia): Ethiopian Health Insurance Agency; 2015 [accessed 2018Feb12]. https://www.hfgproject.org/wp-content/uploads/2015/05/CBHI-Evaluation-5-2015.pdf.[Google Scholar] This risk pooling approach does not substantially alter our modeling incorporating out-of-pocket spending in domestic resource mobilization. It would only do so if it has a large effect on demand for services, raising overall program costs. This may happen, but we did not incorporate expectations to that effect in our model.

One can compare these projections to estimates of spending on government primary health care at the beginning and end of this period. Our models estimate that at the beginning of the HSTP period, Ethiopia was spending ETB 15.30 billion (740 million USD), or ETB 161 per capita (8 USD per capita), on government provided primary care, of which 24% was from government's own funds, 66% was from external funds, and 9% was from household out-of-pocket payments under the historical approach and spending ETB 29.07 billion (1.41 billion USD), or ETB 306 per capita (14.8 USD per capita), under the GDP-based approach, of which 18% was from government's own funds, 35% from external sources, and 47% from household out-of-pocket payments.[c][c] The historical approach estimates are most likely underestimating the amount of household health spending for public primary care services, whereas the GDP approach is likely overestimating it with optimistic high economic growth rates. We think that household health spending for public primary care services is in between these two projected estimates.View all notes At the end of the period in 2020, including household out-of-pocket contributions, we project that this amount would be ETB 19.780 billion (957 million USD) using the historical approach and ETB 44.47 billion (2.15 billion USD) using the GDP-based approach, an increase in nominal terms of 29% and 53%, respectively. In 2020, the projected shares for different financing sources would be 34% government, 55% external, and 11% household using the historical approach and 26% government, 25% external, and 50% household using the GDP-based approach. If government wants to achieve our cost function level of spending without household spending, domestic resource mobilization for health would have to increase an annual average of 135% for both historical-based scenarios and 53% for both GDP-based scenarios in addition to the already assumed increased in government health spending over a five-year period (2016–2020). The additional amount of government funding needed to cover household contributions is even greater for both projections under the two HSTP cost scenarios.

Primary Care Financing and Cost Projection over a 20-Year Period (2016–2035)

This next set of projections follows an approach similar to that of the previous section except that it covers a longer period of time, we now only use our cost model projections because there are no normative cost estimates going this far ahead in time, and we are able to show the effects of sustaining or decreasing external resources. Of course, there are many uncertainties in this longer-term projection.

Figure 4 shows the 20-year projection without incorporating household out-of-pocket spending and with incorporating it. Both graphs suggest that the effects of maintaining external funding at current real levels or decreasing it does not matter that much given robust projections of economic growth in Ethiopia. Without government incorporating contributions from households (either out-of-pocket user fees or CBHI contributions), the future sustainability of primary care services will depend largely on significant economic growth and more ambitious channeling of that growth into government funding for primary care as shown in the GDP-based approach scenarios. Under the historical approach, Ethiopia would have an average annual deficit of 8.08 billion ETB (391 million USD) and 11.47 billion ETB (555 million USD) under the external current and external decrease scenarios, respectively, even with the inclusion of household contributions. Only under the GDP-based approach is Ethiopia likely to be able to finance primary health care expansion with continued efforts to supplement government funding with household contributions. Although our estimates of GDP growth accord with official government projections, it should be noted that this assumes a long-sustained period of rapid growth, which, because of the compounding of growth year on year, produces a high level of increased government revenue. This is a very optimistic projection. Under the historical approach, the Ethiopian government would need to increase health spending by an additional 91% and 120% for the external current and external decrease scenarios, respectively, over the 20-year period (2016–2035). This means that government health spending would have to be 8.4% of total government spending for the external current scenario and 9.4% for the external decrease scenario by 2035 under the historical approach. These quantities are a large increase for Ethiopia but have been attained by other sub-Saharan African countries.

Published online:
21 June 2018
FIGURE 4. 20-year Projection Period Estimates (2016–2035).
FIGURE 4. 20-year Projection Period Estimates (2016–2035).

DISCUSSION AND CONCLUSION

After several decades of robust economic growth, Ethiopia is still a relatively poor country. Despite these constraints, it has advanced substantially in developing a widely accessible government primary health care system that has produced substantial improvements in key health outcomes. Ethiopia took advantage of significant increases in external funding to support these efforts beginning in the 2000s and continuing to the mid 2010s. These external funds enabled the GoE to keep its own domestic spending on health care at modest levels and to focus its own-generated resources on other priorities while continuing to develop primary care services.

In recent years, the situation is changing as total external funding to health in Ethiopia is leveling off or declining. The GoE has recognized this situation and called for increased domestic resource mobilization to support the government health care system with a focus on primary care.10 Federal Ministry of Health. Health care financing strategy 2017–2025. Addis Ababa (Ethiopia): The Federal Ministry of Health; 2017.[Google Scholar]

Literature suggests that a government can increase fiscal space for health in five ways without jeopardizing its fiscal position or crowding out spending on other priority sectors to accomplish development goals.19–21Tandon A, Cashin C. Assessing public expenditure on health from a fiscal space perspective. Washington (DC): The World Bank; 2010 [accessed 2018Feb14]. http://documents.worldbank.org/curated/en/333671468330890417/pdf/560530WP0Box341penditureFiscalSpace.pdf.
Heller PS. IMF policy discussion paper: understanding fiscal space. Washington (DC); International Monetary Fund; 2005 [accessed 2018Feb14]. https://www.imf.org/external/pubs/ft/pdp/2005/pdp04.pdf.
Powell-Jackson T, Hanson K, McIntyre D. Fiscal space for health a review of the literature. London (UK) and Cape Town (South Africa): RESYST; 2012 [accessed 201814Feb]. https://assets.publishing.service.gov.uk/media/57a08a7740f0b649740005e8/Resyst-WP1.pdf. This can be done through (1) sufficient and sustained economic growth allowing greater revenue generation and low levels of debt; (2) increasing the share of health spending within the overall government budget; (3) raising new revenue earmarked for the health sector; (4) improving efficient use of health expenditures; and (5) increasing development assistance for health.19Tandon A, Cashin C. Assessing public expenditure on health from a fiscal space perspective. Washington (DC): The World Bank; 2010 [accessed 2018Feb14]. http://documents.worldbank.org/curated/en/333671468330890417/pdf/560530WP0Box341penditureFiscalSpace.pdf.[Google Scholar] Ethiopia can realistically increase funding through economic growth and efficiency gains, along with some reallocation of the general budget going to health, amid the decline in some external funding.6Glassman A. Ethiopia's AIDS spending cliff. Washington (DC): Center for Global Development; 2012 [accessed 2018Feb13]. https://www.cgdev.org/blog/ethiopia's-aids-spending-cliff.[Google Scholar],7 Health Works Collective. PEPFAR funding allocations: FY2011–FY2013. Washington (DC): Health Works Collective; 2012 [accessed 2018Feb13]. https://www.healthworkscollective.com/wp-content/uploads/2012/09/PEPFAR-Country-Allocations-FY11-131.pdf.[Google Scholar] Given Ethiopia's low income and limited tax-paying population, new dedicated taxes are unlikely to offer much advantage and, in addition, the government has not shown interest to increase the share of taxes in the economy overall.

Our modeling of future costs and resource mobilization for primary health care suggests that Ethiopia can substantially support the further development of primary care services but that this is unlikely through “business as usual”; that is, only by extending historical trends of government financing. Rather, it is likely that three complementary strategies within the fiscal space framework will be needed to assure sustained progress in primary care. First, government must sustain robust economic growth and increase its own contributions to primary care through increased allocations of own-generated resources, which have been underused up until now given the large flows of external funds supporting primary care. Raising Ethiopia's government contributions even to levels similar to comparator country averages will provide important additional support. Some of this is already occurring, as made evident in the latest NHA (2013–2014), which shows an increase in government spending on health by more than 200% and a slight increase in the share of general government spending going to health compared to the previous NHA (2010–2011).5 Federal Democratic Republic of Ethiopia Ministry of Health. Ethiopia health accounts, 2013/14. Addis Ababa (Ethiopia): The Federal Ministry of Health; 2017 [accessed 2018Feb12]. http://apps.who.int/nha/database/DocumentationCentre/Index/en.[Google Scholar] Second, in keeping with government policy to not overly burden the economy with new taxes, government could expand mechanisms for citizen contributions to primary health care through formal risk-pooling mechanisms like CBHI. Third, government could improve efficient use of health expenditures at all levels of the health care system, potentially freeing up resources to provide the same level of care with less resources or increase service outputs or scope without increasing spending. The combination of increased government funding for primary care and increased household funding, with continued economic growth and efficiency gains, offers a high probability of being able to sustain primary care progress, including expanding coverage and scope of primary care services even in the face of even large declines in external support.

It is promising that these strategies, increased government funding for health from own-generated resources, increased citizen contributions through formal risk-pooling, and seeking efficiency gains, are already part of the GoE's health financing policies going forward. However, some government strategies, such as development of social health insurance that would provide significant increased funding for inpatient services, might also result in pressures on government to shift funding away from primary health care unless managed well.

Our modeling suggests that there are plausible paths going forward to sustain health for a still poor country like Ethiopia anticipating declining external support. External partners should be encouraged by Ethiopia's timely adoption of sound strategies to maintain progress. Ethiopia and its partners should accelerate their efforts to collaborate on short- and medium-term strategies to achieve a transition in financing health with minimal disruption to the positive progress that has been and is being made.

DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST

Nha Of Ethiopia 1 To 7 Comparison Summary Report

The authors declare that they have no competing interests. The authors alone are responsible for the views expressed in this article and they do not necessarily represent the views of the organizations listed.

APPENDIX

Published online:
21 June 2018

APPENDIX. Summary Statistics for Variables in Cost ModelaaN = number of observations. y = natural log of primary care costs per capita, dependent variable. lnSBA = natural log of skilled birth attendants, which is the percentage of eligible mothers whose delivery was attended by an SBA. PMTCT = percentage of eligible people counseled and tested on the prevention of mother-to-child transmission of the human immunodeficiency virus. OPD = outpatient department rate per capita. LLIN = percentage of long-lasting insecticide nets distributed based on the number of eligible population. Poverty = percentagea of people living below the poverty line.View all notes

TABLE 1. Ethiopian Demographic and Macroeconomic Indicators
Characteristics20122035
Total population (million)aa Data from From Ethiopian Central Statistical Agency population projections based on 2007 census.13 Central Statistical Agency. Population projections for Ethiopia 2007–2037. Addis Ababa (Ethiopia): Central Statistical Agency; 2013.[Google Scholar]View all notesbb Authors' estimates based on Central Statistical Agency population projections and population growth rates.13 Central Statistical Agency. Population projections for Ethiopia 2007–2037. Addis Ababa (Ethiopia): Central Statistical Agency; 2013.[Google Scholar]View all notes83.74135.67
Nominal GDP in billion ETBcc Data from Ministry of Finance and Economic Cooperation audited National Accounts data for 201222 Ministry of Finance and Economic Cooperation. National accounts data, GDP by economic activity 1999/00–2012/13. Addis Ababa (Ethiopia): Ministry of Finance and Economic Cooperation; 2015.[Google Scholar] and authors' forecasted estimate for 2035 using national accounts data and real GDP growth rate and inflation rates from IMF Article IV Consultations.14 International Monetary Fund. Staff report for the 2015 Article IV consultation, Federal Democratic Republic of Ethiopia. Washington (DC): International Monetary Fund; 2015 [accessed 2016Nov2]. https://www.imf.org/external/pubs/ft/scr/2015/cr15300.pdf.[Google Scholar],23 International Monetary Fund. Federal Democratic Republic of Ethiopia, staff report for the 2014 Article IV Consultation. Washington (DC): International Monetary Fund; 2014 [accessed 2016Nov2]. https://www.imf.org/external/pubs/ft/scr/2014/cr14303.pdf.[Google Scholar],24 International Monetary Fund. Staff report for the 2013 Article IV Consultation, Federal Democratic Republic of Ethiopia. Washington (DC): International Monetary Fund; 2013 [accessed 2016Nov2]. https://www.imf.org/external/pubs/ft/scr/2013/cr13308.pdf.[Google Scholar]View all notes738.627.12
Inflation rate (GDP deflator, in percent)dd From IMF Article IV Consultations.14 International Monetary Fund. Staff report for the 2015 Article IV consultation, Federal Democratic Republic of Ethiopia. Washington (DC): International Monetary Fund; 2015 [accessed 2016Nov2]. https://www.imf.org/external/pubs/ft/scr/2015/cr15300.pdf.[Google Scholar],23 International Monetary Fund. Federal Democratic Republic of Ethiopia, staff report for the 2014 Article IV Consultation. Washington (DC): International Monetary Fund; 2014 [accessed 2016Nov2]. https://www.imf.org/external/pubs/ft/scr/2014/cr14303.pdf.[Google Scholar],24 International Monetary Fund. Staff report for the 2013 Article IV Consultation, Federal Democratic Republic of Ethiopia. Washington (DC): International Monetary Fund; 2013 [accessed 2016Nov2]. https://www.imf.org/external/pubs/ft/scr/2013/cr13308.pdf.[Google Scholar]View all notes33.58.6
Nominal government expenditure in billion ETBee From IMF Article IV Consultations24 for 2012 data and forecasted estimate from Berman et al.9Berman P, Mann C, Ricculli M-L. Financing Ethiopia's primary care to 2035: a model projecting resource mobilization and costs. Boston (MA): The Harvard T.H. Chan School of Public Health; 2015 [accessed 2018Feb16]. https://cdn1.sph.harvard.edu/wp-content/uploads/sites/1325/2013/01/Finance-Projection-Model-Report-Feb-19-2016.pdf.[Google Scholar] for 2035 data.View all notes124.423945.29
TABLE 2. Summary of Key Inputs for Historically Based and GDP-Based Projection Approaches under the Resource Mobilization Model
Historically Based ApproachGDP-Based Approach
Data sourcesBased on five rounds of NHA data trends by financing source that occurred over the 15-year time period (1996–2011).NHA 5 (2010–2011) is used as the baseline expenditure and then builds in assumptions on projected GDP growth rates obtained from Ethiopia's Ministry of Finance and Economic Cooperation and the IMF via 'Article IV Consultation'.
Main approach assumptionsHistorical resources mobilized are equal to expenditures. Not tied to potential influencing factors for health financing such as GDP growth, with the exception of OOP, which is tied to population growth. Estimated allocations for primary care and nonprimary care for each financing source are assumed to remain constant over time.Variable annual GDP growth projections are used. The estimated average GDP growth rate over 20 years is 15% (9% inflation and 7% real growth). Government health expenditure assumed to gradually increase to previous highest share of general government expenditure of 6.7% by 2035 from 4.4% in 2011. External resource contribution not tied to GDP; built-in assumptions of stagnation or reduction over time. Household OOP based on elasticity of health spending from Avila et al.16Avila C, Connor C, Amico P. Universal coverage of essential health services in sub-Saharan Africa. Bethesda (MD): Health Finance & Governance Project, Abt Associates, Inc.; 2013 [accessed 2015Aug1]. https://www.hfgproject.org/wp-content/uploads/2013/09/Universal-Coverage-of-Essential-Health-Services-in-Sub-Saharan-Africa-Pr.pdf.[Google Scholar] Other private expenditures based on authors' estimated Ethiopian elasticity of health spending (although not used in the final government primary care model). Estimated allocations for primary care and nonprimary care for each financing source are assumed to remain constant over time.
External currentExternal contributions remain at the current 2011 levels and increase only by 1.83%, or average US inflation. Government health contributions gradually increase to 6.7% of GGE by 2025 and maintain until 2035, with an average annual increase of 12%. Current trajectory of spending for OOP based on average per capita household contributions for services from Purvis et al.15Purvis G, Alebachew A, Feleke W. Ethiopia health sector financing reform midterm project evaluation. Washington (DC): United States Agency for International Development; 2011 [accessed 2015Aug9]. http://pdf.usaid.gov/pdf_docs/Pdact293.pdf.[Google Scholar] and population growth from the Central Statistical Agency.13 Central Statistical Agency. Population projections for Ethiopia 2007–2037. Addis Ababa (Ethiopia): Central Statistical Agency; 2013.[Google Scholar] Current trajectory of spending for other private. OOP experiences an average annual increase of 7%. Other private experiences an average annual increase of 6%.External contributions are not tied to GDP growth rates and are the same as the historically based approach. Government contributions gradually increase to 6.7% of GGE by 2025 and maintain until 2035, with an average annual increase of 19%. OOP elasticity of spending for health is 1.098, with an average annual increase of 12%. Other private elasticity spending for health is 0.42, with an average annual increase of 7%.
External decreaseExternal funding is same as external current scenario until 2020, which gradually decreases to 50% less by 2035. Government, OOP, and other private health contributions are the same as the external current scenario.External contributions are not tied to GDP growth rates and are the same as the historically based approach. Government, OOP, and other private contributions are the same as the external current scenario under the GDP-based approach.
TABLE 3. Nominal Annual Percentage Change for Each Approach and Scenario for Estimated Total Health Expenditure
Historical TrendGDP Based
Financing SourceExternal Current (%)External Decrease (%)External Current (%)External Decrease (%)
Government1219
External2−22−2
Household712
Other private67
TABLE 4. Comparison of HSTP Cost Estimates with Cost Function Approach
Estimated Costs (billion birr, current)
Approach20162020Percentage Change
Cost Function19.2524.9129
HSTP
Base case30.8337.8318
Best case35.2743.2823
APPENDIX. Summary Statistics for Variables in Cost ModelaaN = number of observations. y = natural log of primary care costs per capita, dependent variable. lnSBA = natural log of skilled birth attendants, which is the percentage of eligible mothers whose delivery was attended by an SBA. PMTCT = percentage of eligible people counseled and tested on the prevention of mother-to-child transmission of the human immunodeficiency virus. OPD = outpatient department rate per capita. LLIN = percentage of long-lasting insecticide nets distributed based on the number of eligible population. Poverty = percentagea of people living below the poverty line.View all notes
VariableNMeanMedianMinimumMaximumStandard Deviation
y7125.0465.0002.7637.1710.446
lnSBA21,398−0.937−0.830−4.6050.0000.827
PMTCT21,7620.8671.0000.0001.0000.296
OPD20,7481.1290.7600.0006.8481.123
LLIN21,7590.5540.7300.0001.0000.447
Poverty18,7200.1670.1400.0020.8620.102

Notes

[a] Ethiopia's most recent NHA was not included in this work because it was released in the fall of 2017, after this study was completed.
[b] In some instances, regions used their own treasury funds to provide additional financial resources directly to health centers. Unique cases such as this are not captured in the primary care cost model, and this is a caveat within this projection, leading the model to underestimate such costs.
[c] The historical approach estimates are most likely underestimating the amount of household health spending for public primary care services, whereas the GDP approach is likely overestimating it with optimistic high economic growth rates. We think that household health spending for public primary care services is in between these two projected estimates.