Malaria Expenditure Analysis

Tanzania Case Study

Prepared for DFID-EA (Tanzania) and the
Roll Back Malaria Initiative
March 2000

Matthew Jowett
Nigel Miller
with Nainkwa Mnzava

International Programme
Centre for Health Economics
University of York
York YO10 5DD

Tel: +44 1904 433718
Fax: +44 1904 432701

This report was commissioned by DFID-EA (Tanzania) and prepared by:

Matthew Jowett, Research Fellow and Programme Manager, International Programme, Centre for Health Economics, University of York, U.K. Email: mj14@york.ac.uk

Nigel Miller, Research Fellow, International Programme, Centre for Health Economics, University of York, U.K. Email: njm11@york.ac.uk

With the assistance of:
Nainkwa Mnzava, Senior Economist, Budget Division, Department of Health Policy and Planning, Ministry of Health, Dar-es-Salaam, Tanzania.

Many thanks to staff in the DFID-EA (Tanzania) office for their support, and to all those who provided the information used to compile this report. Details of meetings held can be found in Appendix 7.

NOTES:

The data and commentary presented in this report refers to mainland Tanzania only, Information regarding malaria expenditures in Zanzibar is collected and reported independently, and is not within the scope of this report.

Throughout this report an exchange rate of US$ 1 = TSh. 800 was used for currency conversions. This was the rate at the time the study was conducted, in February 2000.

REPORT STRUCTURE

Executive Summary
1.   Introduction and study rationale
2.   Background to Tanzania
2.1   Economic situation
2.2   Health status and the burden of disease
2.3   Health facilities and health expenditures
3.   Results
3.1.    Total malaria expenditures
3.2.   Government expenditures on malaria
3.3   Donor expenditures on malaria
3.4   Private expenditures on malaria

Appendix 1: The distribution of voluntary health facilities in Tanzania
Appendix 2: The distribution of private health facilities in Tanzania
Appendix 3: Regional income differentials 1998
Appendix 4: Relationship between regional income and private sector provision
Appendix 5: Extrapolations of aggregate health expenditures from facility level ('000s US$)
Appendix 6: Estimates of total incidence of malaria/fever by region
Appendix 7: Interviews conducted as part of the study
Appendix 8: Mapping Malaria Risk in Africa (MARA)
Appendix 9: Literature
Appendix 10: Terms of reference

TABLES, FIGURES AND CHARTS

Table 1: Key economic and demographic indicators in Tanzania
Table 2: Key health Indicators in Tanzania
Table 3: The burden of disease from malaria in Tanzania (GoT facilities)
Figure 1: Percentage of life years lost by disease (morogoro 1992-95)
Figure 2: Incidence of malaria / fever by region (1996)
Table 4: Health facilities in Tanzania by ownership and level
Table 5: National health expenditures 1990 and 1998
Table 6: Total malaria expenditures in Tanzania 1998/9 (US$)
Figure 3: Total malaria expenditures by source 1998/99
Figure 4: Total malaria expenditures by input 1998/99
Figure 5: Total malaria expenditures per capita, by region 1998/99 (US$)
Table 7: Total malaria expenditures by inputs, facility type and ownership
Figure 6: GoT malaria expenditures by input
Figure 7: GoT malaria expenditures by system level
Table 8: Unit Costs of Malaria Outpatient and Inpatient by Facility
Table 9: Medical stores expenditure on malaria drugs
Table 10: Principle Donor Projects in Malaria 1998
Table 11: Private health expenditures
Table 12: Estimated fever attack rate per annum, by age group
Table 13: Estimated fever attack rate per annum, by age group
Table 14: malaria drug costs per treatment
Table 15: Private Expenditures on bednets
Table 16: Expenditures on coils and sprays

ABBREVIATIONS

AMMP      -   Adult Morbidity and Mortality Project
DDH      -   District Designated Hospital
DfID      -   UK Department for International Development
GDP      -   Gross Domestic Product
GoT      -   Government of Tanzania
HMIS      -   Health Management Information System
JICA      -   Japanese Development Agency
MARA      -   Mapping Malaria Risk in Africa
MoF      -   Ministry of Finance
MoH      -   Ministry of Health
MSD      -   Medical Stores Department
NIMRI       -   National Institute for Medical Research
NMCP      -   National Malaria Control Programme
PER      -   Public Expenditure Review
RBM      -   Roll Back Malaria
SWAp      -   Sector Wide Approach
TEHIP       -   Tanzania Essential Health Investment Programme
UNICEF   -   United Nations Children Fund
WHO      -   World Health Organisation

Executive Summary

This report presents the findings of 17-day mission to Tanzania, conducted in February 2000. The request for the mission originated from the UK Department for International Development as part of the multi-agency Roll Back Malaria Initiative. The mission purpose is primarily to generate information on malaria expenditures in light of the re-emergence of malaria as a major burden of disease, and the difficulties faced by policy-makers in accounting for the increasing role of the private sector in the provision of health care. The Terms of Reference specify three key objectives. These are;

• To provide a bird's eye-view of the level, allocation, application, and use of resources in support of malaria control activities;
• To provide a better understanding of malaria financing and constraints faced by program managers and decision makers;
• To make actionable recommendations for the improvement of financing of malaria activities in the context of the changing environment.

This study is one of four case studies on malaria financing in Sub-Saharan Africa. The other proposed countries are Senegal, Kenya and Ethiopia. Accurate data are scarce in Tanzania, and readers are advised that many of the estimates reported in this study are perforce derived from modelling exercises. However, in light of this, we have conducted, for most estimates, extensive cross-referencing.

We estimate that Tanzania spends approximately $US 2.14 per person per annum on malaria services. This is approximately 15% of spending on health in the average developing country. In Tanzania it represents approximately 39% of all health expenditures. Malaria accounts for 30% of the total burden of disease. Approximately three-quarters of malaria expenditures are household expenditures in the formal and informal private sector. Government contributes 20%; donors 9%. Of total malaria expenditure one-third is spent on anti-malarial drugs, and almost half on bednets, insecticides and coils. Anecdotal evidence suggests that the use of traditional healers for malaria treatment is small.

Whilst the primary objective of the mission was to generate quality data we believe that our evidence at least points towards certain issues and implications for policy-makers. These are;

• The huge proportion of resources devoted to malaria demonstrates that there is demand for malaria services way ahead of any other health service, and greater than is indicated by the evidence of Burden of Disease.
• The overwhelming majority of resources are generated in the private sector. This confirms the importance attached in health sector reform to understanding and evaluating any policy in light of its potential impact on the private sector.
• We estimate that 80% of fever incidents are treated outside of the public sector. A large proportion of these is not accompanied by any or adequate diagnosis or prescription, and this has serious implications for health, and is likely to accelerate the rate of emergence of drug resistance and increase future costs. Strategies to improve the quality of self-medication are needed. Regulation and training of the private sector are a priority.
• Our estimates assume that chloroquine is the first-line drug for malaria. Given the high proportion of malaria expenditures on drugs, the change in drug policy and introduction of more expensive drugs is likely to drive up the cost of treatment.
• The high level of expenditure on preventative services and the exponential growth in sales of bednets in urban areas (despite relatively high retail prices) suggests that there is potential for expansion of the market for bednets, coils and insecticides. Sales are largely concentrated in Dar-es-Salaam and other urban areas. Strategies to extend the bednet market to rural areas should be considered.
• In estimating private expenditures on anti-malarial drugs we assume that all episodes of fever are treated as malaria. We justify this assumption on the basis of the substantial household survey evidence from Africa that points to a very close association between incidence of fever and self-diagnosis of malaria. In fact, studies suggest that only approximately 40% of fevers are due to malaria. This implies a wastage of approximately $US 8 million, or one-eighth of total malaria expenditures. Tanzania has reasonable coverage in terms of access to health facilities, although the key issue here is access to quality diagnostic.
• The greatest burden of disease from malaria falls on infants. The primary reason for this is an under-developed level of resistance to malaria, and the relationship between malaria and still-birth, low birth weight and anaemia. The implication is that mothers and pregnant women should be at the forefront of any efforts against malaria, and that any policy recommendations should consider the impact they have on the behaviour and attitudes of mothers. There is evidence that the majority of pregnant women who suffer malaria are self-treating, and that significant numbers believe that anti-malarials may cause abortion.

Future allocations of malaria expenditures – the role of economic evaluation

It is appropriate to bring to the attention of policy-makers the evidence on cost-effectiveness of alternative malaria interventions and consider to what extent current expenditures are in line with what is known about the cost-effectiveness of malaria interventions.

There is a lack of quality data and consequently a huge range in estimates. By some margin, the most cost-effective intervention is drug treatment. Depending on the degree of resistance the cost per child death avoided ranges between US$1.47 to $2.56 for chloroquine, $2.35 to $2.89 for amodiaquine, and $1.70 for SP. On the other hand, chemoprophylaxis for children and for pregnant women costs $167 per child death, and from $81-$950. Insecticide treatment of bednets costs between $219 to $800. Provision and treatment of bednets costs between US$ 2,000 and US$ 3,000.

The evidence is in line with the great proportion of expenditures that are spent on drugs in Tanzania (approximately 33%). The evidence suggests that the change in drug regime to sulphadoxine-pyrethamine will be cost-effective. A number of the studies in the literature were conducted in regions in which chloroquine resistance is not a problem to the extent it is in Tanzania, which is especially interesting. In light of the evidence the large expenditures on bednets is surprising, until we recognise that these are overwhelmingly concentrated in urban areas. In rural areas only approximately 5% of households have bednets.

Information gaps

It is useful to highlight where there is greatest shortage and need for quality data in the study of malaria expenditures. No reliable estimates of the total incidence of malaria are available, with reported statistics an underestimate. An extensive study in this area is ongoing and readers are referred to Appendix 8 for details. Few quality data exist on household health expenditures. This is a major obstacle in the effective evaluation of health policy in a sector in which household expenditures and the private sector are of great, and increasing, importance. Quality household surveys are thus a priority. Finally, there is a need for donors to pool information on their contributions to the health sector. It is anticipated that Sector-Wide Reform will help in this regard.

BACKGROUND TO THE STUDY AND
CONTEXTUAL ANALYSIS

1.   Introduction and study rationale

In 1998 the Global Initiative to Roll Back Malaria (RBM) was launched by a range of partner organisations, including the World Bank (together with WHO, UNICEF and UNDP), and the governments of malaria-affected countries. Also included in the partnership are bilateral donors, OECD countries, the research community, industry, the private sector, and non-governmental organisations (NGOs). The key motivation for establishing the RBM was to galvanise international support and attention into concerted action against the resurgence of malaria as a cause of the burden of disease, and increasing resistance to anti-malarial drugs.

This report is one of four in-depth country case studies regarding expenditures on malaria commissioned by RBM. In order to maximise the impact of those services that aim to prevent and cure malaria, given budget constraints, programme managers require certain information. An analysis of the flow of expenditures to malaria services informs managers in several ways. The source of current funding for malaria services gives an indication of the stability of future revenue streams (i.e. a more diverse funding base is less susceptible to rapid), as well as identifying on whom the burden of financing malaria services currently falls. In addition as with any health service, each financing mechanism employed creates a particular set of incentives and penalties, both for providers and consumers, influencing both the quantity and quality of care consumed. A detailed picture of the allocation of malaria resources, when juxtaposed with evidence of cost-effectiveness facilitates improvements in future budget allocations, and the improved targeting of assistance.

The current shift away from vertical programming, and towards sector-wide approaches, as in Tanzania, has implications for the budgeting process, often reducing the transparency of allocations to individual interventions. As highlighted in the Terms of Reference (see Appendix 10), the move towards SWAps will change the way that malaria activities are planned and budgeted, as well as the way in which finances are sought. It is common for a broader Ministry of Health planning, procurement, budgeting and disbursement process, to replace the direct relationship with external financiers that existed under vertical programming. Budget cycles must be anticipated and followed, with access to funds for unplanned activities more difficult to seek out. At this point in time in Tanzania, a review of malaria expenditures also provides useful benchmark information, allowing the evaluation of the future allocations to malaria services.

2.   Background to Tanzania

2.1   Economic situation

Tanzania is one of the poorest countries in the world with a per capita income of around US$ 210 per capita. The World Development Report 2000 ranks Tanzania 199^th out of 210 countries for which income data is available. Once adjustments for purchasing power parity are made, per capita income increases to $490, but results in a fall in the rankings to 209^th. Purchasing power data provide a more accurate picture of what can actually be afforded, for a given sum of money, in any given country.

Despite the low level of incomes in Tanzania the economy has been growing substantially over the past decade. Table 1 shows data for selected years, with GDP growth remaining at between 3.5% and 4% over the past 10 years, and indeed since the mid 1980s. The majority of this growth has come in the tourism and industry sectors. Currently agriculture constitutes around 46% of GDP, with industry and manufacturing 21%, and services 33%. A Public Expenditure Review published in November 1999 projects GDP to increase to 5.7% over the next three years. Tanzania is a Heavily Indebted Poor country and is currently undergoing a process of structural adjustment.

Table 1: Key economic and demographic indicators in Tanzania

Source: World Development Indicators 1997
Public Expenditure Review Vol. I and Vol. II, November 1999; GoT
and World Bank.

Whilst the economy has been growing steadily in absolute terms, the picture in per capita terms is more varied. In 1990 income per capita was estimated at US$ 150, although other sources suggest this to have been as low as US$ 100 (World Development Report 1993). Estimates for 1998, as mentioned, are significantly higher at US$ 210. Across the 1990s population growth rates have fallen from 3.2% to 2.8%, which in turn reduces the drag on rises in per capita income. In 2000 Tanzania has a population of around 33 million. If population growth continues to fall and GDP continues to rise, then substantial rises in income per capita can be expected over the next decade. Even if GoT allocations to health services remain constant, it can be expected that per capita spending in the public sector will rise over the coming decade.

2.2   Health status and the burden of disease

Tanzania performs fairly well on a range of health status indicators relative to its level of income. Life expectancy at birth is 51 years, just below the average for sub-Saharan African countries (see Table 2). The infant mortality rate of 85 per 1,000 live births is below the average for Sub-Saharan Africa. The total fertility rate is currently 5.8, a fall from 6.3 in 1992. The prevalence of HIV infection ranges from 10-14% amongst adults.

Table 2: Key health Indicators in Tanzania

2.2.1   Burden of disease resulting from malaria

Data on the relative importance of malaria as a cause of ill health and death in Tanzania were collected from the Health Management Information System (HMIS) of the Ministry of Health. The unit managing this system produces an annual Statistical Abstract, which incorporates information on inpatient and outpatient attendances resulting from malaria. Several points should be noted with regard to this source of information. First the level of reporting varies from year to year. For example in the Table 3, which uses Statistical Abstracts for 1998, eight of Tanzania’s twenty regions did not report outpatient figures, with seven failing to report inpatient information. The authors impute data for these regions (see details in footnote 8). Secondly the usual concerns regarding the quality of the data apply. All fevers presented to health facilities that are not attributed to other causes are generally recorded as malaria, when in fact they may not be. Local experts suggest that over-reporting may be as much as 20%. Despite the problems with the data, they nevertheless provide the best estimate of the national burden of disease resulting from the malaria. Malaria is currently the largest single cause of morbidity and mortality in Tanzania, accounting for 34% of deaths of under 5s and 23% of deaths of over 5s in 1996. Table 3 presents official data for both outpatients and inpatient for 1996, and by region.

Table 3: The burden of disease from malaria in Tanzania (GoT facilities)

Source: Health Statistics Abstract 1998, MoH, Dar-es-Salaam, July 1998

In terms of malaria outpatient cases in GoT health facilities, Iringa Region has by far the greatest rate with 291 cases per 1,000 population. Kigoma, Mbeya and the Coast regions have the next highest incidence of malaria, ranging between 240 and 250 cases per 1,000 population per year. Figure 1 illustrates the burden of disease from malaria relative to other diseases, using data for 1992-1995 in Morogoro region. The results confirm malaria as the greatest single cause of death and illness, explaining over 30% of the total number of life years lost.

Figure 1: Percentage of life years lost by disease (Morogoro 1992-95)

   Source: MoH, Tanzania & AMMP Team, 1997

Given that around one-third of health facilities in Tanzania are non-governmental (see section 2.3), and that it is widely recognised that a significant proportion of malaria is self-treated, reported cases in government facilities are clearly a substantial underestimate of the total. In order to obtain a more accurate picture of total incidence (thus enabling a more accurate picture of overall demand for malaria treatment), we use information from the Tanzanian Essential Health Investment Programme (TEHIP), which is a collaborative partner in the MATA project. A more detailed explanation of the methodology used is reported in Section 3.4.1. We estimate that the total number of supposed malaria cases in Tanzania is almost 33.5 million, with a relatively small proportion (1 in 5), presenting at GoT facilities. Figure 2 reports annual incidence estimates by region.

Figure 2: Incidence of malaria / fever by region (1996)

Source: HIMS, Statistical Abstract 1998; NIMRI

Local health professionals indicate that almost all people suffering from a fever would take some action. In our calculations we thus assume that everyone with a fever seeks health care, whether this is self-treatment, seeking care at a voluntary or private sector facility, or visiting a traditional healer. This information is used in the estimation of private expenditures on malaria, outside the public sector (see Section 3.4).

2.3   Health facilities and health expenditures

2.3.1   Supply of health services

An analysis of the supply of health services has several purposes. First it provides contextual information with regard to access to health care, and the choice of services available, two factors which influence both levels and types of general health, and malaria-specific, expenditures. Information regarding the supply of private health services is used in this report to estimate private expenditures (see Section 3.4).

Increasingly Tanzanians have a choice of facilities at which they can seek health care. An estimated 95% of the national population live within 5kms of a health facility, a remarkably high level of coverage. Table 4 summarises the number of health facilities by ownership, and level. The Church sector has played a significant role in health service delivery for many years, and comprises the majority of facilities referred to as ‘voluntary’. Twenty-two hospitals owned in the voluntary sector are officially incorporated into the GoT health system as District Designated Hospitals (DDHs). Historically the Church sector has provided services in under-served rural areas, and typically DDHs exist where the GoT has little presence. This has proved an effective way of extending and maintaining access to health services for the Tanzanian population. The GoT provides the majority of recurrent funding for DDHs. Further analysis of the distribution and density of voluntary health facilities is provided in Appendix 1.

Table 4: Health facilities in Tanzania by ownership and level

Perhaps the most significant development in the supply of health services over the past decade is the growth in private sector provision. The number of private health facilities registered with the GoT has proliferated since their legalisation in 1992. According to official GoT statistics (MoH Statistical Abstracts 1998) there are currently 780 private dispensaries, 11 private health centres and 45 private hospitals registered in Tanzania. Further data regarding the distribution of these facilities by region are presented in Appendix 2. Dar-es-Salaam has, unsurprisingly, the greatest density of private facilities out of the twenty regions.

Income levels alone does not entirely explain demand for private services. Professor Gaspar K. Munishi who conducted one of the few studies into private health care in Tanzania, found that the regularity of household income across a year, and levels of education levels provide a better explanation. He found for example that whilst income levels in Kagera (US$138 per capita) were lower than in Dodoma (US$145 per capita), there were a higher number of private health facilities. In Dodoma, which has a drier climate, income fluctuates significantly across the seasons. In addition levels of formal education are lower in Dodoma, as is demand for modern health services. Traditional health services remain popular in Dodoma. The study found that official numbers of private health facilities were under-reported by 20-25%. Many private facilities operate without licences, under temporary licences, or with licences obtained from local, rather than national government. In some cases several facilities were operating under one licence. For further information on regional income variations see Appendix 3, and for discussion of the relationship between levels of income and the density of private health facilities by region see Appendix 4.

2.3.2   Health expenditures

Universal and free access to public health facilities was maintained until the early 1990s in Tanzania, when financial pressures, expanded demand for health services, and declining service quality forced policy change. The GoT has been a major provider and financier of health services, while voluntary organisations or NGOs (primarily Church-based) have been an important partner, especially in rural areas. The role of for-profit private providers has been growing rapidly, mostly in urban areas, since the legalisation of private practice in 1991. Table 5 provides summary data on key indicators of health financing for 1990 and 1998.

Table 5: National health expenditures 1990 and 1998

The reporting of national health expenditure data for Tanzania is sparse in publications providing such statistics (e.g. World Bank, World Development Reports). As part of this report an estimation of total health expenditure was made, at approximately US$ 5.5 per capita. This is an increase on the US$ 4 per capita reported in both 1990 and 1995, making Tanzania one of the countries with the lowest level of health spending in the world. Relatively little is spent on health care relative to other developing countries. We estimate the contribution of the private sector is 61% of total health expenditure. This figure is slightly higher than the estimated average for all developing countries in 1995.

METHODOLOGY AND RESULTS

3.   Results

3.1.    Total malaria expenditures

Total expenditure on the prevention and treatment of malaria is the sum of both GoT, donor, and private expenditures related to malaria. For both sources, certain items are easily identifiable, such as expenditures made by the National Malaria Control Programme, and household expenditures on bednets. The majority of expenditures however is not so easily identifiable, and must be estimated by allocating a proportion of total expenditures to malaria. For example the PER provides information on total personal emoluments at primary level health facilities. The methodology used is presented in section 3.2.1. A proportion of this figure must be allocated to malaria-related services as part of estimating malaria-related expenditures in GoT facilities. Table 6 presents a summary of the author’s estimates of total malaria-related expenditures in Tanzania.

Table 6: Total malaria expenditures in Tanzania 1998/9 (US$)


Source: Author calculations

It is estimated that approximately US$ 65 million is spent on both preventing and treating malaria in Tanzania, the equivalent of US$ 2.14 per capita, 39% of total health expenditures, and just under 1.1% of GDP. That malaria expenditures constitute such a high proportion of total health spending is extraordinary, but reflects the fact that malaria is the primary cause of death and illness, as highlighted in section 2.2.1.

Seventy-one percent of total malaria spending is from private sources (i.e. household expenditures), with the GoT representing a further 20%, and the remainder attributable to donor funding (see Figure 3). The dominance of private expenditures is primarily a function of the high estimates of household expenditures on pharmaceuticals (see Section 3.4). The increased availability of medicines, largely as a result of greater private sector supply, has been the major driving force behind this trend.

Figure 4 presents total malaria expenditures disaggregated by key inputs. The two major expenditure items are anti-malarial drugs (33%), and on bednets and insecticides (46%). Wages comprise a further 16%. The high level of drug expenditures, together with estimates of private spending on bednets and coils and sprays, reduces wages to a comparatively low proportion of total inputs.

Figure 4: Total malaria expenditures by input 1998/99

Figure 5: Total malaria expenditures per capita, by region 1998/99 (US$)

Figure 5 presents total malaria expenditures by region. Annual per capita malaria spending is unsurprisingly, highest in Dar-es-Salaam at around US$ 7.5, which reflects the relative wealth of the region. On average other regions typically spend between US$1.5 and US$ 2 per capita annually.

Table 7 provides a summary of total malaria expenditures broken down by level of health facility, source of expenditure, and by input. GoT expenditures include wages in GoT health facilities, including DDHs, payments for medicines and other key supplies typically at the central level, and an allocation to the National Malaria Control Programme. The data in Table 7 are analysed in more detail in the following sections.

Table 7: Total malaria expenditures by inputs, facility type and ownership

3.2.   Government expenditures on malaria

To estimate GoT malaria expenditures, we first take total expenditure data for key inputs to health services, such as labour and medicines, and apportion part of each to malaria related activities. The main source of GoT expenditure data is the Public Expenditure Review 1999. The method used to apportion from total expenditures to malaria-related expenditures is described in detail in section 3.2.1. Our estimates are cross-referenced with actual figures where available.

It is estimated that the GoT spends approximately US$ 12.89 million per anum on malaria-related activities. This is equivalent to US$ 0.4 per capita, or 20% of total malaria expenditures. Figure 6 further analyses GoT allocations to malaria by type of input. Data on total wages and other recurrent costs were obtained from the PER 1999, and apportioned to malaria services as per the formula in section 3.2.1.

Figure 6: GoT malaria expenditures by input

We estimate that over three-quarters of all GoT spending on malaria is on salaries, whilst medicines comprise a further 19%. Wages probably comprise a slightly higher proportion of the total than one might expect. This could reflect either the large number of malaria cases relative to other diseases, or relatively low allocations of GoT medicines to malaria. Figure 7 examines GoT malaria expenditures by system level, which is estimated from data on wage expenditures at urban and rural health facilities, and the distribution of malaria medicines by system level. It is estimates that 59% of GoT expenditure takes place at primary health facilities.

Figure 7: GoT malaria expenditures by system level

3.2.1 Apportioning aggregate expenditures to malaria

As mentioned in Section 3.1 it is necessary to apportion certain expenditures, which are only available in aggregate form, to malaria. This is done in two ways. The first approach is to establish the proportion of all outpatients and inpatients diagnosed with malaria in health facilities, and to use this figure to apportion expenditure data. For example in 1997 and 1998, of all outpatients who attend government facilities 31.4% were recorded as suffering from malaria. Using this approach 31.4% of the total wage bill would be apportioned to malaria. This approach suffers from two obvious weaknesses. The first is the availability and quality of this data. Only information from GoT facilities is reported in the Statistical Abstracts, leaving a gap for the almost 1,700 voluntary and private facilities. Many regions do not report at all. The second problem with this method is that it assumes malaria services are provided at the average unit cost.

The second approach builds on a detailed costing study conducted in four hospitals, four health centres and four dispensaries in both the public and voluntary sectors. Cost and activity data are used to derive estimates of unit costs for all the major health services, including malaria. Using this information it is possible to impute the proportion of total costs related to malaria for the facilities examined, and applied to known national and regional expenditures at each level of facility. The proportions are 37% for dispensaries, 30% for health centres, and 27% for hospitals. Table 8 presents the unit costs taken from the study. These are then applied to national and regional expenditure data.

All estimates derived from apportionment and presented in the report are based upon the unit cost approach. This is our preferred approach because the quality of data is clearly superior to the quality of aggregated data. Whilst the HERA study reports inconsistencies in data recording, wherever these appear they are reported and cross-referenced with elementary primary data collection. A weakness of this approach is that, in extrapolating upwards from facility level, it is assumed that we have unit cost and activity data of the typical facility. Clearly the sample size is relatively small.

Table 8: Unit Costs of Malaria Outpatient and Inpatient by Facility

To investigate this and to cross-reference aggregate data on government spending, and on cost-sharing, we took average facility expenditures on personal emoluments, drugs, other recurrent, and capital items by GoT, donor and household, and extrapolated upwards by numbers of facilities. The implied aggregate national expenditures are remarkably close to actual expenditures, and suggest that our average facility unit is as close to average as it needs to be. These results are presented in Appendix 5. We are able to cross-reference our estimates of malaria drug expenditure and donor expenditure based on apportioning with direct estimates.

3.2.2 GoT expenditure on medicines

Each month the Medical Stores Department supplies GoT dispensaries and health centres with yellow essential drugs kits, and GoT hospitals with blue essential drugs kits. Yellow kits at the primary level contain chloroquine and paracetomol which are used initially to treat patients diagnosed as having malaria, or a fever. At the hospital level, second and third line anti-malaria drugs, sulfadoxine-pyrimethamine and quinine, are provided. Table 9 reports expenditures for 1998. As data regarding donor drug supplies was unavailable from official sources, we estimate this by apportioning donor expenditures on all drugs.

Table 9: Medical stores expenditure on malaria drugs

3.3   Donor expenditures on malaria

Information on bilateral and multilateral donor support to malaria services was collected from a variety of sources. Several donor agencies fund projects specifically aimed at preventing and treating malaria, including DfID, JICA, WHO and UNICEF. Data on these expenditures were collected directly from original project documents, and annualised where they lasted for a number of years. A summary is provided in Table 10, with expenditures amounting to an equivalent of US$ 5.41 million in 1998.

Table 10: Principle Donor Projects in Malaria 1998

This data can be cross-referenced with the figure of US$ 5.589 million in Table 7, confirming the accuracy of the estimates. In addition to these specific malaria projects, further support is given to malaria, indirectly, through general support to the health sector. Estimates of total donor support to the health sector were obtained from various sources, but varied widely however, from a total of $US 21 million to $US 87 million. It is felt however that given the level of problems with accessing quality data, it was not possible to make estimates on this component. We thus expect our total figure to be on the low side.

In addition to problems in estimating broader support to the health sector from bilateral and multilateral donors, it was not possible to obtain information on financial expenditures in voluntary sector health facilities. This could include money from a range of sources including donations from abroad, and money collected from congregations. It was suggested, based on evidence from neighbouring countries that donations from abroad tend not to be substantial. A proportion of total expenditure in voluntary facilities should in principle be allocated to malaria in the same way as is done for GoT facilities. However due to the lack of any data no estimations were made, and hence again we expect our figure to be an underestimate.

3.4   Private expenditures on malaria

Estimating private expenditures on malaria is perhaps the most challenging task of all, given that limited information is available from official GoT data sources. Household expenditure surveys typically provide the richest source of information on private health expenditures, but somewhat surprisingly, almost no such data was available in Tanzania. Because of this we crosscheck by estimating both the outflow of funds from private facilities, and the inflow of household expenditures. The estimates are remarkably close and suggest we can have confidence in them.

To estimate outflow we take data on total GoT recurrent expenditure (excepting drug expenditure) and the number of GoT facilities to determine the average expenditure per GoT facility in each region. To this we make three adjustments to account for differences between private and public. We first inflate by 20% to account for differences in private and public sector wages, by a further 20% to account for payments to profit-makers, and then weight using an index of regional per capita income (regional per capita income over mean regional per capita income - see Appendix 3). The reason for inflating by regional income is to take account of the significant degree of under-reporting of private sector activity, which we assume to be correlated to regional income. To determine total private expenditures less drug expenditures we multiply up by reported numbers of facilities.

Table 11: Private health expenditures

To determine expenditures on malaria we apportion in the way described in section 3.2.1, and cross-reference with those malaria expenditures that can be directly estimated. The final figure for malaria expenditures come to $US 40,512,540.

Household expenditures are also estimated from survey data. The most recent household survey in 1992 found that total per capita expenditure on health was $US 3.5, of which 63% was spent on drugs. We then take population data to aggregate up to the national level, and deduct household contribution to public facilities. From this we estimate private sector household expenditures at US$ 103,735,967, with drug expenditures, $US 63,000,000. We accept that these figures may be an under-estimate, due to changes in the relative importance of the private sector since 1992. However, what is remarkable is that the estimate of private expenditures less drugs from the household survey (the difference between 103million and 63 million) is very close to our estimate from facility cost data. We cross-reference with direct estimates of malaria drug expenditure and the key non-drug malaria expenditures, on bednets, coils and sprays.

3.4.1 Private expenditures on anti-malaria medicines

It is widely believed that a significant majority of malaria cases are self-treated by drugs purchased in the formal or informal private sector. However, despite the potentially massive expenditures there is as yet no accurate estimate of the extent of these expenditures. In this section we describe a simple model to estimate private sector malaria drug expenditures, based on the incidence estimates presented in Figure 2 and Appendix 6, and the cost of medicines to treat one case of malaria (Table 14).

According to information obtained in Tanzania, the risk of malaria transmission is relatively similar across all 20 regions. A malaria risk-mapping exercise is currently on going in Tanzania by TEHIP. When data becomes available, a more accurate estimate of total fevers can be used to restimate private expenditures. However it is considered that the risk of fever across Tanzania is broadly similar to Kisumu in Kenya, where a more detailed risk-mapping exercise has been conducted. Table 12 presents the estimate of attack rate in Kisumu by age-group, reflecting the fact that children under 5 years tend to be at the highest risk. By combining these data with demographic data on age and population it is possible to estimate the total number of fevers in Tanzania by region in one year. There is a very close association between fever and malaria diagnosis so this translates to an approximation of malaria incidence in Tanzania (see Figure 2).

Table 12: Estimated fever attack rate per annum, by age group

From total incidence is subtracted the reported number of malaria cases in GoT health facilities (see Table 3), to give an estimate of cases falling in the private sector (either self-treatment or with private providers), assuming all fevers are treated. According to our estimates 82% of all fevers, which we use as a proxy for malaria, fall outside the public health sector (see Table 13).

Table 13: Estimated fever attack rate per annum, by age group

We develop a model of private expenditures on malaria drugs using these estimates. We assume that all fevers not reported at government facilities are treated by the 1^st line and by far the cheapest readily available drug, chloroquine. There is significant demand for 2^nd line malaria drugs due to high levels of chloroquine resistance, which is estimated to be as high as 50% in Tanzania, and for fevers not caused by malaria that do not respond to chloroquine. We assume 30% of all fevers proceed to the 2^nd line drug, sulfadoxine-pyrimethamine, and 5% of all fevers are severe cases which are treated by quinine. The cost of drugs used in the calculations is the retail price for one adult treatment discounted by 35% to account for smaller child dosage and non-compliance (Table 14).

Table 14: malaria drug costs per treatment

The model results in a final estimate of household malaria drug expenditure in the private sector of US$ 15,500,000 (see Table 7). This is 25% of total drug expenditure estimated from household survey data. This seems reasonable in light of the Burden of Disease, and is possibly an under-estimate.

3.4.2 Private expenditures on bednets, coils and sprays

Due to the lack of household expenditure information, we estimate private expenditures on bednets through an analysis of supply. We quantify the volume of bednets coming onto the market in Tanzania, and together with retail price data, estimate expenditure. In order to collect this information interviews were held with the three main local producers of bednets, namely A to Z Ltd, Sunflag Ltd. and TMTL Ltd. Based on this information we estimate that approximately 2.3 million bednets are produced locally, of which around 750,000 are exported (see Table 15).

Table 15: Private Expenditures on bednets

Official data on the volume of imports was unavailable in the time available. We use rough estimates provided by local producers market analysis, namely that a similar volume of bednets comes to the market from outside the country, as from domestic sources. We approached the principal importers but were unable to confirm this estimate. It should hence be treated with caution.

There are two sources of information regarding the sale of coils. There is anecdotal evidence that annual sales in Dar-es-Salaam are approximately $US 12 million which is approximately $US 6 per person. DfiD’s SMITN project (Social Marketing of Insecticide Treated Nets) estimates that average annual household expenditure in urban areas is $24 which is approximately $US 3-4 per person.

Table 16: Expenditures on coils and sprays

This may reflect the relative income of Dar-es-Salaam. To estimate national expenditures we assume that there are negligible expenditures in rural areas, and extrapolate upwards by population in urban areas, approximately 15% of total population. This delivers an estimate of just over $US 21 million (Table 16).

CONCLUSIONS AND RECOMMENDATIONS

4)   Conclusions

Readers are also referred to the Executive Summary for a thorough breakdown of the principal findings and recommendations of this report. This section aims to conclude with some comments to stimulate further discussion and help to identify further research.

We estimate that in total over 33 million incidents of fever occur in Tanzania each year, approximately one incident for each citizen. Around 6 million cases of malaria are reported in GoT facilities annually. Whilst malaria is the principal cause of fevers, many are actually unrelated to malaria. On these calculations a further 27 million fevers fall outside the GoT sector.

Since 1991, when the prohibition of private medical practice was lifted, Tanzania has seen an exponential growth in private facilities, with an estimate 1,500 currently in operation. Given the significant gap between demand for health service, and GoT supply, this development has positive aspects. However, there are two worrying aspects to the growth in private health services. One relates to the quality of care delivered. In contrast to the traditional image of private practitioners in OECD countries, of highly paid staff providing high quality service, in Tanzania and many other low-income countries, the private sector tends to provide relatively low cost, low quality care. In 1996, due to concerns over service quality, the GoT suspended the licences of almost half of all private facilities.

Secondly they only represent 20-25% of all facilities, and these are concentrated in urban areas. Let us assume that they care for 7 million of the 27 million fever cases falling outside the GoT sector. On this basis the remainder, and vast majority, of fever cases self-diagnose and self-treat. Many households keep a stock of medicines, treating fever, correctly or incorrectly, as malaria. This implies that a large proportion of incidents of a disease that kills over 1 million per year, is treated without diagnosis and with drugs that may be out of date, and taken in an inappropriate dosage. Anecdotal evidence indicates that, however income constrained households may be, they find the resources to buy medicines. The issue is not then entirely about resources, or the availability of medicines, but rather health-seeking behaviour, and the cost of accessing quality of care. Patients require access to quality diagnostic services in order to tackle the main cause of illness and death in Tanzania.

Future allocations of malaria expenditures – the role of economic evaluation

It is important appropriate to bring to the attention of policy-makers the evidence on cost-effectiveness of alternative malaria interventions and consider to what extent current expenditures are in line with what is known about the cost-effectiveness of malaria interventions.

There is a lack of quality data and consequently a huge range in estimates. By some margin, the most cost-effective intervention is drug treatment. Depending on the degree of resistance the cost per child death avoided ranges between US$1.47 to $2.56 for chloroquine, $2.35 to $2.89 for amodiaquine, and $1.70 for SP. On the other hand, chemoprophylaxis for children and for pregnant women costs $167 per child death, and from $81-$950. Insecticide treatment of bednets costs between $219 to $800. Provision and treatment of bednets costs between US$ 2,000 and US$ 3,000.

The evidence is in line with the great proportion of expenditures that are spent on drugs in Tanzania (approximately 33%). The evidence suggests that the change in drug regime to sulphadoxine-pyrethamine will be cost-effective. A number of the studies in the literature were conducted in regions in which chloroquine resistance is not a problem to the extent it is in Tanzania, which is especially interesting. In light of the evidence the large expenditures on bednets is surprising, until we recognise that these are overwhelmingly concentrated in urban areas. In rural areas only approximately 5% of households have bednets.

Information gaps

It is useful to highlight where there is greatest shortage and need for quality data in the study of malaria expenditures. No reliable estimates of the total incidence of malaria are available, with reported statistics an underestimate. An extensive study in this area is ongoing and readers are referred to Appendix 8 for details. Few quality data exist on household health expenditures. This is a major obstacle in the effective evaluation of health policy in a sector in which household expenditures and the private sector are of great, and increasing, importance. Quality household surveys are thus a priority. Finally, there is a need for donors to pool information on their contributions to the health sector. It is anticipated that Sector-Wide Reform will help in this regard.

Appendix 1: The distribution of voluntary health facilities in Tanzania

Appendix 2: The distribution of private health facilities in Tanzania

Appendix 3: Regional income differentials 1998

Levels of income are analysed in order to provide a weighting index used in the estimation of private expenditures. Data is presented for 1998.

• As expected Dar-es-Salaam has the highest income per capita, at US$ 550, or TSh 440,000. The other large urban areas including Arusha and Iringa are also amongst the more wealthy regions. The poorest region with US$ 138 per capita, or TSh 110,500, is Kagera. Dodoma, Kagera and Kilimanjaro are also amongst the poorest regions.
• Health expenditures are generally considered to be elastic i.e. that as income increases, so do levels of health expenditures. Given that Tanzania is currently undergoing a structural adjustment programme, it is likely that the main growth in health expenditures is from private, out-of-pocket sources.
• Private expenditures manifest themselves in a variety of ways. These include payments for health services at GoT facilities, in the form of user charges, and unofficial charges, which are more difficult to estimate. Expenditures in the increasing number of private health facilities in Tanzania, as well as self-treatment, primarily through the purchase of medicines are the other main types of private payment.

Appendix 4: Relationship between regional income and private sector provision

In recent years tighter regulations implemented by GoT have made it more difficult to register a private health facility. Higher standards must now be met, for example in terms of tax reporting. In addition the GoT has strengthened the regulation of service quality in private facilities. The Table below analyses the relationship between demand (proxied by income) and the supply of health services in Tanzania. For each region the density of health facility, by ownership type, is calculated using population data. Regions are then ranked. This information is then juxtaposed with the income-rank of a region.

It is expected that regional income has a positive effect on the number of private-for-profit facilities. The data presented above confirms that Dar-es-Salaam, with the highest income per capita, also has the greatest density of private-for-profit health facilities. However the relationship does not hold in many regions. For example Rukwa Region, which has the 3^rd highest level of income, ranks 19^th in the density of private facilities. Kilimanjaro Region, one of the poorest ranked 17^th, is second only to Dar-es-Salaam in terms of private facility-density (see Appedix 1 for more details). There is substantial anecdotal evidence that private facilities are generally considered to be of lower cost and quality than both GoT and voluntary health facilities, and that it is predominantly the poor that use these facilities.

The region with the lowest density of private facilities is Singida. In terms of income levels, which one would usually assume to be positively correlated with demand for private health services, Singida actually ranked 8^th out of 20 regions, with a per capita income of US$ 193, or TSh 154,184 in 1998.

In terms of voluntary facilities it expected that Church and charity based facilities locate in areas of higher poverty, or isolation, often with a specific mandate to serve the poor. The table shows that the greatest density of voluntary facilities is in Kilimanjaro Region, with one facility per 17,000 population. The region is also one of the poorest in the country, ranked 17th. The overall pattern is mixed, but generally opposes the expected pattern. The 2nd, 4th and 5th richest regions have some of the greatest number of voluntary facilities given their population size, whilst the five poorest regions have relatively low densities of voluntary facilities with the exception of Kilimanjaro (see Appendix 2).

Appendix 5: Extrapolations of aggregate health expenditures from facility level (‘000s US$)

Appendix 6: Estimates of total incidence of malaria/fever by region

Source: Based on expected fever rate in Kisumu. See Appendix 8.

Appendix 7: Interviews conducted as part of the study

Dr. A. N. Mwita, Coordinator, National Malaria Control Programme, Ministry of Health. 31^st January 2000.

Paul Smithson, Health and Population Advisor, DFID Eastern Africa, Tanzania. 31^st January 2000.

Ritha John A. Njau. National Professional Officer, Malaria Control. WHO, Dar-es-Salaam. 2^nd February 2000.

Kara Hanson, Researcher, London School of Hygiene and Tropical Medicine, London.

Sandra Baldwin, Health and Population Advisor, DFID Eastern Africa, Tanzania. 3^rd February 2000.

Mwele Ntuli Malecela, Director of Research and Training, National Institute for Medical Research, Dar-es-Salaam, Tanzania. 3^rd February 2000.

Don de Savigny, Tanzania Essential Health Interventions, National Institute for Medical Research, Dar-es-Salaam, Tanzania, 3^rd February 2000.

Graham Reid, Tanzania Essential Health Interventions, National Institute for Medical Research, Dar-es-Salaam, Tanzania, 3^rd February 2000.

Dr. Jane E. Miller, Project Manager, Social Marketing of Insecticide Treated Nets, Population Services International. 3^rd February 2000.

Salim Abdulla, Epidemiologist, IFAKARA - Health Research and Development Centre, PO Box 53, Ifakara, Tanzania, 4^th February 2000.

Vincent O’Neill, Health Advisor, Irish Aid. 4^th February 2000.

Dr. A. N. Mwita, Coordinator, National Malaria Control Programme, Ministry of Health. 5^th February 2000.

Kazuko Hashimoto, Health Co-operation Planning Advisor, Ministry of Health, Dar-es-Salaam. (Seconded from JICA). 7^th February 2000.

Dr. Isiye Ndombe, Senior Programme Co-ordinator, UNICEF and Riitta Poutiainen, Project Officer Health, UNICEF. Dar-es-Salaam, Tanzania. 10^th February 2000.

Mr. P.A. Ilomo. Head, Cost-Sharing Implementation Unit, MoH, Dar-es-Salaam, Tanzania. 11^th February 2000.

Christopher Msemo, Director of Procurement, Medical Stores Department, Dar-es-Salaam, Tanzania. 11 February 2000

TELEPHONE CONVERSATIONS HELD

Dr. Ahmed Hingora, Head, Primary Health Care Secretariat, Ministry of Health, Dar-es-Salaam. 7^th February 2000.

Mr. Binesh Haria, Director, A to Z Ltd. Arusha. (Manufacturers of bednets). 8^th February 2000.

Mr. Mohan, Finance Director, Sunflag Ltd. Arusha. (Manufacturers of bednets). 8^th February 2000.

Vanessa Herd. DFID Advisor, Ministry of Finance, Dar-es-Salaam, Tanzania. 9^th February 2000.

Mr. Kibaji, Ministry of Finance. Dar-es-Salaam, Tanzania. 9^th February 2000.

Professor Gaspar K. Munishi, Faculty of Arts and Social Science, University of Dar-es-Salaam, PO Box 35051, Dar-es-Salaam, Tanzania. 10^th February 2000.

Philip Setel, Adult Morbidity and Mortality Programme. NIMRI, Dar-es-Salaam, Tanzania. 10^th February 2000.

CONTACTED BY EMAIL

Torben Vestergaard Frandsen, Disease Control Textiles, Vestergaard Frandsen A/S, Akseltorv 4b, 6000 Kolding, Denmark. Tel. ++45 75 50 30 50. Fax. ++45 75 50 30 44

Siam Dutch. Bednet Manufacturers, Thailand. WWW.SIAMDUTCH.COM

Appendix 8: Mapping Malaria Risk in Africa (MARA)

PLEASE NOTE:   THE FOLLOWING INFORMATION HAS BEEN TAKEN DIRECTLY FROM THE MARA WEBSITE ON: http://www.mara.org.za/trview_e.htm

What is MARA/ARMA?

The MARA/ARMA collaboration was initiated to provide an atlas of malaria for Africa, through the use of a Geographic Information System (GIS), by integrating spatial malaria and environmental data sets, and producing maps of the type and severity of malaria transmission.

The initiative is non-institutional and runs in the spirit of an open collaboration in which parallel, international and regional efforts contribute towards achieving the overall objectives.

Background

Sub-Saharan Africa carries the highest per capita burden of disease in the world of which malaria is the single most important cause. Of global deaths attributed to malaria 90% now occur in sub-Saharan Africa. Recent advances in public health are offering new opportunities to make significant reductions in the burden of disease. However, many factors, especially endemicity, affect the choice of control methods. This requires us to rethink how we define endemicity, and how we may map malaria risk in order to better support planning and programming of malaria control.

Detailed mapping of malaria risk and endemicity has never been done in Africa. Accurate estimates of the burden of malaria at regional or district level remain largely unknown. In the absence of such data it is impossible to rationalize allocation of limited resources for malaria control.

Objectives

The objectives ofMARA/ARMA's are:

1. to collect comprehensively available malaria data and also to characterise risk categories in terms of non-malaria data (eg. climatic, environmental data);
2. to develop a mask layer of factors which exclude malaria (eg. absence of population, high altitude, deserts etc.);
3. to highlight areas of no data and provide further work on geographical modelling to extrapolate to such areas of no data, using among others the environmental stratification outlined in 1;
4. to develop a base-map of malaria risk in Africa, down to second administrative unit (district), from available geographic, demographic and malariometric data using a GIS and make this available to national, regional and international organisations.

Data Collection

Five regional centres, at existing institutions throughout the continent, and supervised by the co-investigators at those institutions are responsible for gathering malaria data (parasite ratios and incidence data) and incorporating them into the MARA/ARMA data base.

The main MARA/ARMA malaria data collection is yielding numerous data points across the continent. Many of these originate from unpublished reports and were obtained through country visits. A separate highland malaria data collection, by the Highlands Malaria Project which relates to MARA/ARMA, has acquired many reports of malaria epidemics outside the stable malaria areas. A collection of mosquito surveys has also been incorporated and is providing information on vector distribution.

Appendix 9: Literature

AFTH1 (1997). Tanzania Social Sector Expenditures Review

DfID (1998) East Africa Network for Monitoring Anti-Malarial Treatment Efficacy Project. Project Memorandum

DfID (1998) Local Initiatives for Integrated Malaria Control in Tanzania. Project Memorandum

DfID (1999) Social Marketing of Mosquito Nets and Insecticides for Net Treatment Project Memorandum

Goodman C, Mills A (1999) ‘The Evidence Base on the Cost-Effectiveness of Malaria Control Measures in Africa’, Health Policy and Planning, v14(4) pp.301-312

Government of Tanzania (1999) The United Republic of Tanzania Public Expenditure Review

Health Research for Action -HERA (1999) Health Care Financing in Tanzania Costing Study of Health Services Final Report

Hill, J (1996)A Situation Analysis and Oppurtunities for Malaria Control Support in Tanzania, Malaria Consortium, Liverpool School of Tropical Medicine

Jowett, M. (1999). Bucking the trend? Health care expenditures in low-income countries 1990-1995. International Journal of Health Planning and Management, 14, pp. 269-285.

Ministry of Health (1997) Health Statistics Abstract 1997

Ministry of Health (1998) Health Statistics Abstract 1998

Ministry of Health (1996) Implementation of Health Services User Fees in Tanzania: An Evaluation of Progress and Potential Impact

Ministry of Health (2000) National Guidelines on Malaria Diagnosis and Treatment Draft

Ministry of Health (1997) Plan of Action 1997-2000 of the National Malaria Control Programme

Ministry of Health (1999) The Health Sector Plan of Action July1999-June2000

Ministry of Health (1999) The Health Sector Reform Programme of Work July1999-June 2002

Oxford Policy Management (1997) Tanzania: Public Expenditure Review Budgeting Priorities in the Social Sectors

Pavignani E, (1998) Recurrent Costs in the Tanzanian Health Sector

PSI Tanzania (1999) SMITN Marketing Plan Jan1999-Dec1999

World Bank PER Team (1997) Report on the Tanzanian Public Expenditure Review in the Health Sector

Appendix 10: Terms of reference

Country Case Study on Malaria Financing and Expenditures (Tanzania)

Background
Context of the Study

Over the past few years, malaria has re-emerged as a major burden of disease in developing countries especially in sub-Saharan Africa. Estimates of the global malaria problem in 1998 place deaths from malaria to be 1.1 million (of which 87.4% was from Africa). The incidence of malaria during the same year was 272.9 million cases (of which 87.1% was from Africa), while disability-adjusted life years lost due to the disease was estimated to be 39.3 million DALYs (of which 87.9% was from Africa). Clearly, Africa accounts for a disproportionate burden of this disease.

Yet, during the same period that malaria has re-emerged worldwide and more so in Africa, dramatic shifts in health sector programming are occurring in sub-Saharan Africa with the increasing acceptance of sector-wide approaches (SWAps) in the health sector, and a wave of institutional and policy reforms ("Health Reforms") which are characterized by changes in the role of the public sector, shifts in the way in which health services are financed, integration and decentralization.

In a coordinated effort to revive malaria control and prevention worldwide, in 1998 the World Bank, together with its partner agencies (WHO, UNICEF and UNDP) launched the Global Initiative to Roll Back Malaria (RBM), with the target of reducing the number of deaths by the year 2010. RBM partnership comprises malaria-affected countries, UN agencies, development banks, bilateral donors, OECD countries, the research and control communities, industry, the private sector, and NGOs. RBM is not a financing mechanism, but rather aims to support countries through global partnerships to ensure that they have effective access to the information, technology and financial resources required to reduce their burden of malaria disease. The expenditure study would help inform the role and functions of RBM in supporting countries to control malaria.

Country Case Studies

Sector-wide Approaches promote a comprehensive program for the health sector, as opposed to individual projects, and donor support channeled in response to a government-defined expenditure program, rather than donor-designated program areas, such as malaria. The changes affect the ways in which Malaria Program Managers and/or staff implementing malaria activities plan, budget and seek funding for activities. In many settings, they must shift away from a direct relationship with external financiers to participating in a Ministry of Health planning, procurement, budgeting and disbursement process. The impact on planning is perhaps the greatest change, as budget cycles must be anticipated and adhered to, and it becomes more difficult to seek out sources for unplanned activities.

Health Reforms commonly encompass many of the following areas:

• Central Ministry of Health responsibility aims to assume a more selective role in setting standards and sectoral policy, over the provision of services. The private sector is increasingly recognized as a legitimate actor in service delivery, often receiving public financing for specific populations and/or services. This affects the role and functions of traditional Malaria Control Programs. In addition, the increasing privatization of health services and the widening role of the households and private providers in health care in many sub-Saharan African countries have made it extremely challenging to take account of non-governmental resources devoted to malaria prevention, control and treatment.
• A range of instruments to improve the financial sustainability, efficiency and incentives within the sector result in changes in the way in which activities are financed. Insurance, user fees, pre-payment, contracting out and public sector grant making affect the way in which malaria activities are financed, and influence the incentives to undertake activities.
• Integration, which aims to increase efficiency through sharing human, physical and financial resources, tends to blur the distinction between traditional vertically managed programs such as malaria control, so that it becomes increasingly difficult to understand the level, allocation, application, and/or uses of resources flowing into programs addressing malaria.
• Decentralisation affects the ways in which planning and budgeting occurs, and often affect change in the loci of responsibility for planning, budgeting, oversight, supervision, technical guidance and decision making.

Although decentralisation, integration, the diversification of actors involved within service delivery, increased government ownership, and a sector-wide perspective are, in themselves, steps and approaches that would presumably lead to better health care provision, it is incumbent for malaria program managers and decision makers to have a better grasp and understanding of the total flow of resources going to malaria, and the implications of these changes on how they work to ensure malaria activities are financed. Such information would assist in (a) the identification actors/players and sources of malaria financing; (b) the determination of resource gaps, if any, in the malaria subsector; (c) the analysis of resource use in the subsector; (d) the analysis of underspending on malaria and how to guide malaria program managers and technical agencies who advise them on how to improve the flow of resources; and (e) helping MOHs and the RBM partners in better understanding the role of the public sector in financing and supporting malaria control activities.

It is in this context that the RBM Partners intend to conduct four country case studies on the financing and expenditure on malaria in four sub-Saharan African countries. The four (tentative?) countries are: Senegal in West Africa, and Tanzania, Kenya, and Ethiopia in East Africa. The U.K. Department for International Development (DfID) is likely to fund the case studies for Tanzania and Kenya while the World Bank is likely to fund those for Senegal and Ethiopia. It is expected that an individual consultant will lead the case-study effort in each country. However, a generic set of data formats, analytical questions, and report outlines will be used so that these four country case studies will be standardized, and suitable comparisons can be made.

This TOR is addressed to one country case study, which in this case is __________, but this same TOR will be used for the three other countries.

Scope of Work
Objectives of the Case Study

The objectives of each of the case studies are:

• To provide a bird’s eye-view of the level, allocation, application, and use of resources in support of malaria control activities, the "Malaria subsector";
• To provide a better understanding of malaria financing and constraints faced by program managers and decision makers; and
• To make actionable recommendations for the improvement of financing of malaria activities in the context of the changing environment (SWAps and Reforms), without necessarily reverting to the traditional approach of vertical financing.

Coverage of the "Malaria Subsector" or "Malaria Interventions"

For purposes of this TOR, "malaria interventions" and "malaria subsector" is defined to cover the following:

Inputs: These will include the following (not necessarily mutually exclusive) cost or expenditure items:

• Cost of nets and cost of insecticides for nets;
• Cost of insecticides for residual spraying;
• Cost incurred by medical personnel (public and private) in the prevention, control and treatment of malaria;
• Cost incurred by laboratory technicians (public and private) in malaria case-finding and related activities;
• Cost of microscopes (+ slides, stains, other related supplies) used in malaria case finding;
• Cost incurred by traditional healers in the prevention and treatment of malaria;
• Cost incurred by pharmacists/pharmacy operators in dispensing malaria drugs (for self-curing patients, or patients referred by clinics/hospitals to these pharmacies for drugs);
• Cost of drugs for malaria treatment (first, second, and third line);
• Cost of inpatient care aside from malaria drugs;
• Cost of training on malaria, both technical and program management (development and implementation);
• Cost of IEC (information, education, and communication), inclusive of design of strategies and IEC delivery/campaigns; and
• Cost of surveillance system.

Activities: The above costs can be cross-referenced against the following (not necessarily mutually exclusive) malaria activities:

• Use of coils;
• Use of household sprays; residual spraying (conducted by the community, municipal/local government, or other government agency);
• Use of nets; treatment and/or re-treatment of nets;
• Prophylaxis, especially for pregnant women;
• Diagnosis of malaria cases;
• Treatment of uncomplicated disease episodes; treatment of severe disease episodes; treatment of children under 5 suspected of having malaria through the Integrated Management of Child Illnesses (IMCI); and
• Dedicated malaria surveillance.

Providers and funders: The above costs and activities should be cross-referenced and sorted according to the following sources of malaria funding:

• Donor agencies, inclusive of their contractors, grantees, or cooperating agencies;
• Central Government (Ministry of Health and other departments);
• Autonomous government units (with own budget line items) involved in malaria activities;
• Local governments using locally-generated revenues for malaria;
• Reimbursements to hospitals for malaria patients from insurance funds, mutual aid societies ("mutuelles" in West Africa), and the like;
• Non-government agencies (for-profit facilities and nonprofit facilities), inclusive of direct transfers/donations from abroad;
• Out-of-pocket household expenses;
• Direct social transfers or subsidies, if any (e.g., government provision of cheap nets or insecticides should also be reckoned in terms of their actual costs, and the sources of financing this subsidy should be identified as explicitly as possible).

Tasks of the Country Case Study

Conduct desk research and review of the literature related to malaria program costing.

• Collect, compile, and process available secondary data on the importation and local distribution of malaria inputs (foreign trade statistics or Ministry of Trade and Industry data). Social marketing activities, often funded by donors, should be included. Domestic production of these inputs, if any, should be accounted for.
• Collect, compile, and process available secondary data on:
• Donor expenditures on malaria – There may be a centralized record of these in-country, in which case suitable assumptions on malaria funding need to be made.
• Central government expenditures on malaria – Annual budget statistics (allocations vs. expenditures) should be analyzed for this purpose. Since malaria has ceased to be a vertical program in many African countries, a dedicated line item for malaria may have disappeared, although in certain cases, inputs that can be traced to malaria programs (e.g., malaria drugs, insecticides) can still be identified in the budget books. (Or, if this is not possible, tracing these back to the supporting budget preparation documents in the MOH and the Treasury may be helpful.) The consultant is expected to be familiar with government budgeting, and to make whatever reasonable adjustments, based on educated assumptions (see below), to tease away malaria financing from larger (gross) line items in the budget.
• Local government expenditures on malaria – The record of annual intergovernmental transfers (central to local governments) should be examined for this purpose. Many of the records of these intergovernmental transfers are not regularly updated; local research assistance may be helpful in this regard. The consultant is also expected to inquire into Local Governments’ own-generated revenues from local taxation, etc., and how much of these are being devoted to malaria activities.
• NGO expenditures on malaria – There may be an umbrella organization of mission hospitals that keep records on these. If not, selected interviews with key providers in the private sector may prove helpful.
• Household expenditures on malaria – Household expenditures on health are usually included in Poverty and Welfare Monitoring Surveys or in Household Income and Expenditure Surveys. If these or similar surveys are available, the consultant is expected to make reasonable assumptions about how much of the household expenditures are devoted to malaria.

Conduct field visits to confirm findings from Step 3, or to gather primary data if no secondary data are available. The following need to be highlighted under this step:

Donor expenditures on malaria – If data are not readily available, the consultant is expected to utilize the services of a local research assistant to conduct an inventory of major malaria donors, their respective malaria activities (classified according to the general headings specified in Section II.B), and funding (commitments vs. expenditures). Careful attention should be paid at differentiating the sources of funding of alternative expenditures. For instance, most interventions implemented by government and non-government organization are in part financed by donors. A "flow of funds" picture would be suitable in determining exactly how resources move from originator to final recipient of funds.

Community expenditures on malaria – If feasible, and if the amounts involved are significant, the consultant should estimate the resources provided by the community for malaria control. For non-financial contributions, reasonable assumptions should be made to translate them into monetary terms. Revenues from user fees (cost sharing programs) that are devoted to malaria activities – Selected interviews with Hospital Management Teams, District Health Management Teams, or District Health Boards managing these funds will prove useful.

NGO expenditures on malaria – The consultant is expected to undertake reasonable effort to generate as much valid information as feasible on malaria expenditures in this often-underestimated sector. Focus group discussions on the key players in this sector would be useful for generating educated estimates on malaria expenditures, and the key activities that are supported by NGOs.

Based on the review of literature and selected field visits, construct relevant assumptions and adjustment factors, including but not limited to the following. The consultant is expected to use these assumptions and adjustment factors to refine gross expenditures into finer and more suitable estimates of malaria funding. The proportion of malaria patients to total outpatients in government and NGO facilities; The proportion of malaria inpatient-days to total inpatient days in government and NGO facilities;

• The proportion of malaria patients to total patients treated by traditional healers;
• The allocation of joint costs (supplies, personnel time, etc.) on malaria-related activities.
• The depreciation of key (large capital) inputs in malaria programs; and
• Other relevant data adjustment factors.
• Organize the data generated into standard formats (tables or matrices) showing the sources and uses (or application) of malaria funds. The table formats should emulate as closely as possible the standard formats in National Health Accounts that are being conducted in a number of African countries.
• Use a suitable spreadsheet (Excel, Quattro Pro, or Lotus 1-2-3) or a dedicated software (such as Harvard’s National Health Accounts program), input the data collected, and generate summary tables. An illustrative Malaria Expenditure summary table for this consultancy would be like the one below:

Conduct analyses of the results and provide recommendations.

Prepare a report that summarizes the methods, data, and findings. In terms of methods, the consultant will carefully describe the sources of data, institutions surveyed and interviews held. The analysis should describe the structure of expenditures related to malaria control activities, by sources and uses of funds.

Level of Effort

Xx person-days of technical consultancy. This level of effort is expected to be allocated as follows:

Xx person-days home-based work (i.e., work in the consultant’s home country).

Xx person-days in the case-study country, of which xx days is anticipated to be in the capital, xx days spent in selected field visits, and xx days for report dissemination workshop.

Xx person-days of research assistance support, to be managed by the individual consultant.

Xx person-days for translation of French report into English.

Deliverables

A report, written in English, the main text (including tables) of which should not be longer than 20 pages.

A statistical annex containing data definitions, critical assumptions, data matrices.

A diskette containing the electronic version of the report and the statistical annex.

Timeline

The study is expected to be completed in 3 calendar months (mid-November 1999 to mid-February 2000. The consultant will present an analysis plan within the first two weeks of the contract

The draft report should be submitted to the World Bank not later than ________. World Bank comments will be provided within __ days upon receipt of the draft report. Based on these comments, the consultant is expected to submit the final report not later than _________.

Contact Persons

Julie Mclaughlin, task manager for the World Bank’s "Roll Back Malaria" Program; Oscar Picazo and David Robalino of AFTH1 will provide technical oversight for the economic and financing aspects.

Throughout the consultant’s country visits, s/he will be accompanied by a malaria specialist from WHO/AFRO.

Consultant Qualifications

At least Masters degree in economics, social-sector economics, health economics, finance, accounting, public health, or international health. Knowledge of malaria programs desirable but not required.

10 years demonstrated experience in the budgeting and costing of health programs and projects; public finance and/or government sector financing; donor financing of health programs; and household expenditures. Good understanding of African institutions and their data constraints, especially those in health care.

Proficiency in spreadsheet software (Excel, Quattro Pro, or Lotus 1-2-3), or NHA program.

Excellent skills in data retrieval, organization, adjustment, interpretation, and presentation. Familiarity with official statistics such as national income accounts, trade balances, and household expenditure surveys. Familiarity with national income accounting or more specifically, national health accounting, desirable but not required.