PURPOSE: South Africa's TB control program, developed according to the World Health Organization's “DOTS” framework, has achieved broad population coverage and high rates of case detection, but the program's failure to contain the HIV-driven TB epidemic, a low rate of treatment success, and the emergence of drug resistant TB cases motivate efforts to improve the program. We investigated the cost-effectiveness of alternative strategies for improving tuberculosis control in South Africa.

METHODS: A mathematical model of interacting epidemics of TB and HIV was developed and calibrated to match epidemiological data from South Africa. TB control program alternatives consisted of unique combinations of program components that correspond to specific approaches to case detection, diagnostic testing, and treatment, including several that target HIV-infected persons. We used deterministic simulation to estimate program impact on TB incidence, prevalence of drug-resistant TB among TB cases, TB deaths, and costs. The efficient allocation of resources to program components was calculated for a wide range of program budgets, and cost-effective strategies were identified.

RESULTS: Our analysis suggests that if the current TB program in South Africa were maintained, TB incidence would steadily decline, largely due to the underlying trend in the HIV epidemic; but, the fraction of TB that is drug-resistant would grow steadily. The status quo strategy averts 4560 deaths per 100,000 population over 20 years compared to doing nothing, with an incremental cost-effectiveness ratio of $460 per death averted. We found several efficient TB control strategies that are incrementally more intensive than current policy in South Africa. However, as the TB control budget is expanded, we found that the order in which enhancements to program components should be implemented was sensitive to assumptions regarding the distribution of TB, drug resistant TB and HIV within the population, adherence to preventive therapy, misapplication of preventive therapy to active TB cases, and the feasibility of reducing treatment default rates.

CONCLUSIONS: Strategies to improve TB control beyond the status quo exist at a cost per death averted that compares favorably to South Africa's per capita GDP. These strategies, if implemented, can be expected to contain drug-resistance while accelerating the decline in TB incidence and mortality that would occur under the status quo strategy.