When the HIV/AIDS epidemic was first recognised in the early 1980s, treatment options for HIV-infected people were limited. While prophylaxis was available for some AIDS-associated opportunistic infections such as tuberculosis and pneumocystis carinii pneumonia, therapies which had been demonstrated to affect the behaviour of the virus itself were not available. Following the approval of the drug AZT by the USA’s Food and Drug Administration in 1987, however, its prescription to AIDS patients became common medical practice, especially when controlled trials showed it was an effective anti-retroviral agent and that it lengthened life expectancy. More recently, it has also become common medical practice to prescribe AZT (now called zidovudine) to HIV-infected patients who have not yet developed AIDS, both to patients who have HIV-related symptoms and to those who are asymptomatic (Cosler and Lambrinos, 1992). The drug is now widely used in North Amercia, Europe and Australia, along with a number of other anti-retrovirals (ARVs) which have now entered the market.

In contrast, ARVs are not yet widely available to HIV-infected people or AIDS patients in developing countries. While there are notable exceptions such as Thailand, in most countries of Africa, Asia, Latin America and the Caribbean, access to AZT and the newer drugs is only possible through the private sector (Santos et al, 1994). As the HIV epidemic worsens in these regions, policy development in the area of anti-retroviral therapy is increasingly warranted. This is especially so in light of recent suggestions that anti-retroviral combination therapies may not only prolong life but may even, if taken for life, be capable of preventing HIV-infected people from progressing to AIDS at all (Piot 1996); and evidence that provision of AZT to pregnant women reduces transmission to their children (Connor et al, 1994). In addition, with rising rates of infection among hospital patient populations, demand for AZT prophylaxis from health workers exposed to HIV is likely to increase.

The economic aspects of providing ARVs to HIV-infected people and exposed health workers are one of a number of important considerations which can be analysed to help inform policy development. This paper therefore addresses such aspects. It has seven main sections, which are:

• Cost of providing anti-retroviral therapy (1.), in which data on the cost of the drugs is provided and evidence concerning other costs associated with provision of therapy is summarized;
• Total cost implications of providing anti-retroviral therapy (2.), in which the cost data presented in (1.) are used in combination with region-specific data concerning the number of people with AIDS or HIV infection to estimate the total costs for different geographic regions of providing therapy;
• Total cost implications of providing anti-retroviral therapy in comparison with available resources (3.), in which region-specific GDP, health sector expenditures, National AIDS Programme budgets, and official development assistance are compared with the estimates for total costs of anti-retroviral therapy given in 2.;
• Cost-saving potential and cost-effectiveness of anti-retroviral therapy (4.), in which evidence concerning the impact of AZT on health system resource use and life expectancy are used in combination with estimates of the costs of medical care for HIV/AIDS to assess whether, despite the costs associated with their provision, ARVs might be either cost-saving or cost-effective;
• Cost, cost-saving potential, and cost-effectiveness of providing AZT therapy to pregnant women (5.), in which available data concerning the costs associated with providing AZT to pregnant women, the potential for cost-saving due to averted paediatric AIDS cases, and likely cost-effectiveness where therapy may not be cost-saving are discussed;
• Cost and Cost-effectiveness of Prophylaxis for Health Care Workers (6.); and
• Financing anti-retroviral therapy (7.), in which alternative financing options are assessed.

Conclusions (8.) and a bibliography are also included (9.).