To assess the effectiveness and cost-effectiveness of treating HIV-infected injection drug users (IDUs) and non-IDUs in Russia with highly active antiretroviral therapy HAART.
A dynamic HIV epidemic model was developed for a population of IDUs and non-IDUs. The location for the study was St. Petersburg, Russia. The adult population aged 15 to 49 years was subdivided on the basis of injection drug use and HIV status. HIV treatment targeted to IDUs and non-IDUs, and untargeted treatment interventions were considered. Health care costs and quality-adjusted life years (QALYs) experienced in the population were measured, and HIV prevalence, HIV infections averted, and incremental cost-effectiveness ratios of different HAART strategies were calculated.
With no incremental HAART programs, HIV prevalence reached 64% among IDUs and 1.7% among non-IDUs after 20 years. If treatment were targeted to IDUs, over 40 000 infections would be prevented (75% among non-IDUs), adding 650 000 QALYs at a cost of USD 1501 per QALY gained. If treatment were targeted to non-IDUs, fewer than 10 000 infections would be prevented, adding 400 000 QALYs at a cost of USD 2572 per QALY gained. Untargeted strategies prevented the most infections, adding 950 000 QALYs at a cost of USD 1827 per QALY gained. Our results were sensitive to HIV transmission parameters.
Expanded use of antiretroviral therapy in St. Petersburg, Russia would generate enormous population-wide health benefits and be economically efficient. Exclusively treating non-IDUs provided the least health benefit, and was the least economically efficient. Our findings highlight the urgency of initiating HAART for both IDUs and non-IDUs in Russia.
To allocate HIV prevention resources effectively, it is important to have information about the effectiveness of alternative prevention programs as a function of expenditure. We refer to this relationship as the "production function" for a prevention program. Few studies of HIV prevention programs have reported this relationship. This paper demonstrates the value of such information. We present a simple model for allocating HIV prevention resources, and apply the model to an illustrative HIV prevention resource allocation problem. We show that, without sufficient information about prevention program production functions, suboptimal decisions may be made. We show that epidemiologic data, such as estimates of HIV prevalence or incidence, may not provide enough information to support optimal allocation of HIV prevention resources. Our results suggest that good allocations can be obtained based on fairly basic information about prevention program production functions: an estimate of fixed cost plus a single estimate of cost and resulting risk reduction. We find that knowledge of production functions is most important when fixed cost is high and/or when the budget is a significantly constraining factor. We suggest that, at the minimum, future data collection on prevention program effectiveness should include fixed and variable cost estimates for the intervention when implemented at a "typical" level, along with a detailed description of the intervention and detailed description of costs by category.
