Costs of Scaling HIV Programme Activities to a National Level in Sub-Saharan Africa: Methods and Estimates

HOME       WHAT'S NEW     SITE MAP     DISCUSSION LIST     CONTACTS      LINKS  

African Development Forum 2000           AIDS: The Greatest Leadership Challenge

  About ADF 2000

  Partners

  ADF '99

Home >  Documents > Costs of Scaling HIV Programme Activities to a National Level in Sub-Saharan Africa: Methods and Estimates

PREVIOUS PAGE < CONTENTS > NEXT PAGE

V. COSTS OF SCALING-UP HIV-PREVENTION AND CARE ACTIVITIES

In the previous chapter, estimates were made of the current level of coverage of HIV-prevention activities among the PTGs by type of intervention. Potential levels of coverage that were thought to be feasible to achieved by the year 2005 were then set for each of the HIV-prevention strategies. These target levels of coverage for the year 2005 reflected the currently low levels of coverage for many of the activities. For care and treatment activities, estimates of the potential increase in coverage in five years were also made. In each case, the identified levels were based on what was felt to be realistically possible to achieve in a five-year time-period, given the capacity constraints.

A. Unit costs of HIV/AIDS interventions

This section explores how the nature of costs might change as the level of coverage or the scale of implementation varies. Economic theory suggests that the level of costs and efficiency of different activities will change as the size of activities increases. However, given the dearth of actual cost data, there is very little practical guidance as to whether and how these costs will change for each of the different HIV/AIDS programme activities that we consider. This raises fundamental questions about how to model the costs of scaling-up activities. This is analysed in the first section. In the ideal world, data on marginal and average costs of interventions could be used to discuss the costs of scaling-up to reach additional people. However, given substantial data limitations, a unit-cost approach to modeling is followed in that report.

Possible bias due to the assumption of constant unit cost

Ideally, estimates of the costs or resources needed for scaling-up HIV/AIDS interventions would be based on existing cost information and extrapolated for likely changes in the cost-structure, delivery mechanisms and technologies. However, a key constraint is the lack of data. There are currently only about 30 published cost-studies from countries in Sub-Saharan Africa for all HIV/AIDS interventions. Furthermore, because of the low national coverage of many interventions, even when cost information is available, it is generally obtained at the individual facility or project level, operating on a small-scale (e.g. community or district).

Given these data constraints, the most common approach is to take the unit cost associated with an intervention and multiply it by a factor reflecting activity at a larger scale. However, this is clearly at odds with what economic theory suggests about the changes in costs as output increases. Starting from a low coverage level, scaling-up activities are likely to be accompanied by a decrease in unit costs (or average cost). Once a certain coverage level is reached, coverage cannot be increased further without additional investment in infrastructure. At that point, unit costs are likely to rise.

Figure 5.1: Possible Direction of Bias due to Constant Unit Costs

As shown by Figure 5.1, the assumption that unit costs would remain constant can lead to either over or underestimating total costs. At a low coverage level, the average cost curve is above the straight line representing unit cost. As a result, total costs are underestimated by the assumption of constant unit cost. Similarly, at high coverage levels total costs are underestimated when they are computed by using constant unit cost rather than the actual average cost. But in both cases, the bias is less when medium rather than low unit costs are used for estimating total costs. Thus, given the different cost scenarios (low and medium) that are used by the model, the assumption of medium unit cost may be more relevant at the lowest and highest levels of coverage.

Uncertainty

A fundamental issue is the uncertainty affecting much of the data used to estimate unit costs. Because many of the strategies could be implemented in a number of different ways, low and medium-cost estimates are presented for each form of intervention. This range reflects both uncertainty about the cost-structure as well as how the specific interventions may be delivered. However, while the model also uses behavioral, epidemiological and intervention-specific inputs, a unique value rather than a range was used for these parameters. This approach was adopted because uncertainty about the cost structure is much greater than for the other parameters.

Costs and income levels

Cost structures will reflect the different levels of prices in low-and high-income countries. With respect to the health sector, a key difference in prices between low and high-income countries relates to the salaries of health personnel. The model allows for differences between low-and high-income countries for strategies that are heavily reliant on health personnel such as clinical management and prevention of OI. The unit costs of prevention and care used in the modelling are summarized in table 5.1, 5.2 and 5.3. Costs are in US dollars atS 2000 prices.⁵

Table 5.1: Unit Costs for HIV-prevention Activities 1/ (Annual $US Costs in 2000 Prices)

Note: Drawn from review of HIV/AIDS costs from Kumaranayake and Watts (2000b).

Table 5.2: Unit Costs for Care and Treatment Activities (Annual $US Costs in 2000 Prices)

Note: See Annex 2 for the definition of low and medium-costs, and the source of data. High-income SSA countries are considered to be Botswana, Djibouti, Gabon, Mauritius, Namibia, South Africa, and Swaziland.

Table 5.3: Cost per Capita for Institutional Strengthening (Annual $US Costs in 2000 Prices)1/

1/ Categorization of countries is presented in Table 4.2.

B. Cost of HIV/AIDS interventions for sub-Sarahan Africa

In this section, we present estimates of the costs of scaling-up prevention, care and treatment strategies for SSA. Using data from SSA countries, the Resource Determination Model (RDM) was used to estimate the additional costs of reaching these higher levels of coverage, from the assumed levels of existing coverage. Some 37 Sub-Saharan Africa countries were selected for analysis⁶. They account for about 95 percent of the total SSA population. For each country, estimates of the current and future levels of coverage of each form of HIV/AIDS intervention among the specified PTG were made. They are based on the estimated number of people with HIV/AIDS in the year 2000 (regardless of when the diagnosis was made). From the individual country estimates, the costs of scaling-up for the whole of SSA were extrapolated.⁷

The annual costs of scaling-up activities are presented in table 5.4. These costs give an idea of the magnitude of resources that need to be spent annually, if one hopes to reach the scale of the target coverage levels in 2005. In total, between $1.5 billion and $2.2 billion would be required each year to scale-up prevention and basic care by the year 2005. The costs of scaling-up treatment by HAART would be between $1.5 billion and $2.4 billion.

Table 5.4: Annual Cost of Scaling-Up HIV/AIDS Activities For Sub-Saharan Africa
($US millions; 2000 prices)

Note: 1/ Total is less than the sum of the sub-totals. This is due to the fact that care and treatment strategies were costed taking into account the potential for double-counting of activities. If all care and treatment activities were scaled-up jointly, there would be some duplication of activities.

2/ Assumes that the prices of drugs will be reduced to $US 1,400 for the low cost estimate and $US2,635 for the medium-cost estimate - about 14% and 27% of current drug prices in the USA. Further discussion of this is found in annex 2.

3/ Figures are non-country specific and are based on estimates for regional surveillance, monitoring and evaluation costs by UNAIDs.

Interpretation of results

The cost results shown in table 5.4 should be interpreted as the annual additional cost of implementing activities to a specified level of coverage in the countries considered. What does this mean?

• Costs are annualized costs. They indicate the resources that must be spent annually to achieve and maintain a particular level of coverage. Just how many years are required to achieve this level of coverage depends very much on what is the baseline level of coverage and what is the 2005 target level of coverage. In practice, the annualized costs will provide a good approximation of the overall resource requirements for an intervention over the lifetime of the project.

• Costs are additional. Costs of scaling-up are estimated by using the difference between current and projected (2005) coverage of HIV/AIDS interventions. As a result, the estimated country costs represent the additional expenditures required to attain a particular coverage level by 2005.

• Costs are different from budgetary costs. The actual budgetary requirements may be different from the estimated annualized costs depending on the extent of the capital and start-up activities that are required at the beginning of the intervention.

• Costs exclude organizational and capacity constraints to scaling-up. While in theory the costs of projects should be scaled-up for any level of activity or target group for each form of intervention, we have only attempted to cost what may be potentially feasible levels of implementation. Thus, for example, MTCT prevention activities were costed only for those women having access to ante-natal and health services. Hence, the levels of coverage and interpretation of cost estimates are always in relation to the intervention-specific PTG.

• Costs are a function of a particular level of coverage. As seen in chapter 4, the PTGs for different interventions differ substantially. For some forms of strategy a 50% coverage may reach a substantial number of people (for example, for youth interventions). For others, especially those that are dependent upon access to health or other services, a 50% coverage may benefit far fewer people. In particular, the PTG for treatment was defined in a very restrictive way (as being HIV-infected who are symptomatic with access to health services). As a consequence, a 50% treatment coverage will reach far fewer people than 50% coverage of the HIV-prevention activities. It is also important to note that where infrastructure and existing capacity are low, even with high levels of coverage among the PTG, there may still be a large section of the population that are not reached by HIV activities.

• Costs may include some double counting. The current cost estimates are produced by separately considering how each intervention may reach a particular section of the population. However, where more than one form of intervention may provide the same service to the same sub-group, this method of estimating costs may inherently `double count' the provision of particular services. For example, a man having sex with a commercial sex worker could potentially obtain a condom from a sex worker intervention, a health clinic, his workplace, or a condom social marketing programme. If each of these is considered in isolation, there is the potential to substantially over-estimate the resources required to meet the potential demand for condoms. In practice, however, given the low levels of current coverage, the likelihood of double-counting is limited.

Relative costs of prevention and care

As shown by Table 5.4, care costs are lower than prevention costs. However, this should not be interpreted as meaning that countries should invest in care activities rather than in prevention. This result is due to the following four factors:

• For most SSA countries, access to health services is relatively low, resulting in a limited access to many care activities. As a consequence, the target groups for care activities are small, which result in relatively low cost of care.

• At high coverage levels, the HIV-prevention costs will include substantial double-counting of condom and STD treatment services, and so will over-estimate the true costs of, for example, meeting the population needs for condoms.

• The cost of care excludes antiretroviral treatment. If the cost of providing antiretroviral triple-therapy were added to the cost of care, the costs of care and treatment would greatly exceed prevention costs at all levels of coverage.

• The relative costs of intervention and care reflect the size of the potential target groups (PTG) as well as the unit costs. The size of the PTG is particularly important on the prevention side where youth, STD, VCT and workplace groups have relatively large PTG at 50% levels of coverage. In contrast, despite the unit costs of care being relatively high, the total costs of care are relatively smaller given the size of the PTG that can be covered within existing infrastructure.

Figure 5.2 shows the relative costs of different prevention and care strategies. The large proportion of costs attributable to STD treatment is due to the high incidence of STDs as well as the high coverage targets. The next most important category is youth interventions. Palliative care and clinical management of opportunistic infections account for about 17% of costs, followed by orphan care, which reflects the large number of orphans (related to the high HIV prevalence rates in SSA). The shares of the other components of care are smaller due to either their small unit-cost and coverage (e.g. psycho-social support) or the capacity constraints which mean that higher coverage for clinical management was not feasible.

Figure 5.2: Relative Costs of Prevention and Care Interventions (Medium Costs)

C. Costs of Scaling-up Programmes by Countries

The costs of scaling-up prevention and care vary substantially across countries. The variation is shown in figure 5.3 for the medium-cost scenario. It reflects differences in the size of population as well as in the initial coverage of the existing HIV/AIDS programmes. The costs of providing treatment (antiretroviral therapy) are shown in Figure 5.4.

Note: Estimates are for the medium-cost scenario. See Table 2, Annex 2 for the low cost estimates.

Figure 5.4: Annual Costs of Scaling-up HIV/AIDS Treatment (HAART) ($US Million; 2000 Prices)

Not surprisingly, the costs of scaling-up HIV/AIDS intervention vary greatly depending on the stage of the epidemic. This can be seen by classifying countries by the HIV adult prevalence rate.

Prevalence and HIV/AIDS costs

In the model, nearly all the costs of HIV-prevention activities are independent of HIV prevalence with only the costs of MTCT intervention being affected by HIV prevalence. In contrast, the costs of all care and treatment activities are dependent upon the country HIV prevalence rates and the estimated number of orphans. To illustrate this relation, the estimated per capita costs of HIV/AIDS interventions were plotted against the adult HIV prevalence rate for Sub-Saharan countries. This relationship is shown in figure 5.5 for prevention and care and figure 5.6 for all interventions (including HAART).

At the beginning of the epidemic, when the HIV/AIDS prevalence rate is below 5% in the adult population, the annual cost of scaling-up programmes is about $US4 per person on average. This is because most of the programme activities consist of prevention, with a relatively low cost per capita. As the HIV prevalence rate reaches 15%, the per capita cost of prevention and care increases sharply to about $US10-14 per capita. What explains the steep increase in costs is the rapid rise in the costs of treating opportunistic infections such as tuberculosis.

Figure 5.5: Per Capita Costs of Scaling-up Prevention Care and Care and HIV Prevalence Rates

The evolution of the cost per capita of prevention, care and treatment (antiretroviral therapy) is shown in figure 5.6. As previously, when the HIV/AIDS prevalence rate is below 5%., the per capita cost of HIV/AIDS interventions, including triple antiretroviral therapy, is relatively low, of the order of $US4-5. However, once the HIV/AIDS prevalence rate exceeds 10%, the per capita cost of HIV/AIDS interventions rise rapidly due to the increase in the number of AIDS patients. As shown by figure 5.6, the per capita cost can reach $US30.

Figure 5.6: Per Capita Costs of Scaling-up Prevention, Care and Treatment in SSA Countries

In addition to these financial costs, the HIV epidemic imposes broad costs in terms of the loss of young adults in their most productive years. In particular, existing levels of health infrastructure and human capital will also depreciate with the HIV/AIDS epidemic. For example, it has been estimated that a country with a stable 5% HIV prevalence can expect that each year between 0.5 and 1% of its health care providers will die from AIDS. In contrast, a country with a 30% prevalence would lose 3-7% of health workers to the epidemic each year (World Bank, 1997a). About 50-70% of medical beds in large hospitals are taken by patients with HIV-related illnesses. This suggests the need for complementary actions to sustain and expand current levels of infrastructure.

The education sector is also likely to be seriously affected.⁸ A recent survey in Malawi found that the rate of HIV infection among school-teachers was 30% (UNICEF, 1999). This will seriously compromise the ability and sustainability of prevention efforts within schools, where trained teachers have a critical role.

D. The Benefits of early prevention

There is a substantial body of evidence which shows the importance of acting early to prevent the spread of the epidemic (World Bank 1997a; UNAIDS 1999a). When prevention measures are put in place early, the spread of HIV is contained, which reduces the number of opportunistic infections and AIDS cases that would otherwise have occurred.

Figure 5.7 illustrates the cost of providing comprehensive package of interventions (including antiretroviral therapy) for a typical SSA countries with a per capita income of $US300. When the adult HIV prevalence rate is less than 5%, the annual cost of prevention, care and treatment is about 1.3 % of GDP. But once the HIV prevalence rate reaches 30%, the annual cost can reach 10% of GDP.⁹

While countries with a relatively high-income per capita can afford to provide AIDS care to a large group, this is unlikely to be the case for low-income countries. The only low-income countries that can financially afford antiretroviral therapy, even at the much lower prices recently announced, are those that implement early on a comprehensive package of HIV-prevention measures. The reason is that early intervention leads to a reduction in the future number of AIDS patients. As a result, the total cost of HIV/AIDS interventions remains affordable despite the relatively high cost of treatment and care.

E. Policy implications for priority setting

The costs presented above are based on what may be potentially feasible levels of implementation, given current levels of infrastructure and capacity development. These costs give an idea of the magnitude of resources that need to be spent next year, if the target coverage levels are to be reached in 2005.

Resource mobilization. In relation to current health expenditures or GDP, the additional resources for scaling-up activities represent a substantial investment. For some countries, it could require a doubling of the health budget. Despite the devastating impact of HIV in SSA, resources to address HIV/AIDS have been limited compared with those for other priority areas (Watts and Kumaranayake, 1999). In 1998, external spending on HIV/AIDS was about US $165 million, less than a third of the $650 million spend on childhood immunization programmes (UNICEF, 1999). Substantial additional resources would therefore need to be mobilized to support the implementation of scaled-up national programmes.

Priority setting. Clearly, the required level of coverage for different interventions will vary from setting to setting. The exact mix of interventions will vary by country depending upon factors such as the stage and characteristics of the epidemic, the cost-effectiveness of particular interventions in a setting, and the level of resources that are available and the level of support and implementation capacity within a country (Drayton, 1998). The model provides estimates of resources, but priority setting also needs consideration of the impact or effectiveness of the interventions for each setting. When implementation capacity is weak, a core set of programme activities to be scaled-up will likely be substantially more effective than attempting to expand a number of activities simultaneously (Ainsworth and Teokul, 2000).

Impact assessment. Just as information on costs was limited, information on impact and cost-effectiveness is even scarcer. While costs are important in examining resource requirements for different activities, a comparison different activities should also consider their likely and differentiated impact. It is only when both costs and effectiveness are considered together, that we can compare amongalternatives in terms of their success in averting HIV infections. While it is difficult to estimate the impact of prevention strategies, nevertheless, one can start to compare comparable outputs between interventions (e.g. the number of condoms distributed and used). Gathering information on this level of intervention outcome is essential for more refinement within the priority-setting and resource-allocation process.

Prevention activities. Regardless of the stage of the epidemic, there is still a need to sustain HIV-prevention efforts, even though the care needs are becoming more substantial. Ultimately, limiting the transmission of HIV is the long-term solution to mitigating the epidemic. There may be different prevention priorities at different stages of the epidemic. As the epidemic becomes more generalized, HIV incidence increasingly concentrates amongst youth. Thus-in-and-out-of-school interventions become increasingly more important. Resource considerations in such a situation may need to consider how best to reach different sections of the youth population (e.g. how can we reach the greatest number in the cheapest manner) or designing a package of key/essential services.

Infrastructure constraints. Focus on the feasibility of scaling-up also means that priority setting may need to consider which activities can be rapidly scaled-up, given the need to intervene quickly. The task ahead is complex. There is a patchwork of mostly small-scale, prevention and care activities being implemented across SSA by public-sector and private profit and non-profit organizations. There is little experience of replication and expansion, so strategic and creative thinking is needed.

In the short term, effective activities that can be scaled-up quickly need to be identified. For rapid scaling-up, the potential to use the existing infrastructure to achieve widespread coverage must be maximized. For example, with current enrolment rates, a quarter of youths aged 12-16 years could potentially be reached each year through interventions based on secondary schooling. However, an additional 10% could be reached if prevention activities were also undertaken in the last year of primary school (Watts and Kumaranayake, 1999). The substantial potential to use private-sector and informal networks must not be overlooked. Experience from rural development suggests that scaling-up is an incremental process, where one starts by covering an entire district or region, and then replicating the district or regional model across the country (Binswanger, 2000).

Specific interventions may be needed for countries that were classified as having very low programme strength. Many of these countries face disruptions due to conflict situations. Thus, a real question is how to implement activities when there are constraints which are more than limited infrastructure. This has been outside of the current analysis, but is crucial to consider in the implementation of policy.

Key gaps in information also need to be addressed. Knowledge is lacking about the relative quality, efficiency and cost-effectiveness of various interventions as their implementation is scaled-up. It is clear that we cannot wait for better information before implementation, given the need to act quickly. However during implementation, key gaps in information need to be addressed. Thus it is crucial to document the costs, cost-effectiveness and operational learning so that experiences can be shared and successes quickly replicated.

5 The original studies presented figures for different years. Prices were converted into constant US dollars for the year 2000 using the average annual inflation rate, as measured by the GDP deflator (World Bank, 1997b) for the period between 1985-1995. The GDP deflator for this period was 3.2% for the USA, over this period.

6 The selection of countries was based on the availability of data for these countries.

7 The missing countries were found in the very low and low programme strength countries.

8 "Exploring the Implications of the HIV/AIDS Epidemic for Educational Planning in Selected African Countries: the Demographic Question". ACTAfrica, World Bank, August 2000.

9 For simulation purpose it is assumed that: (a) 10% of the target group of the people living with AIDS would have access to antiretroviral drugs, and (b) the cost of antiretroviral drugs would be reduced to $US2,600 (medium-cost estimate).

 [Go to Top]

PREVIOUS PAGE < CONTENTS > NEXT PAGE

3-7 December 2000, Addis Ababa, Ethiopia

Copyright  © 2000 Economic Commission for Africa (ECA)
For questions regarding this web site contact the webmaster
Last updated: January 05, 2001 14:40.