Background: Culture of Mycobacterium tuberculosis currently represents the closest "gold standard" for diagnosis of tuberculosis (TB), but operational data are scant on the impact and cost-effectiveness of TB culture for human immunodeficiency (HIV-) infected individuals in resource-limited settings.
Methodology/principal findings: We recorded costs, laboratory results, and dates of initiating TB therapy in a centralized TB culture program for HIV-infected patients in Rio de Janeiro, Brazil, constructing a decision-analysis model to estimate the incremental cost-effectiveness of TB culture from the perspective of a public-sector TB control program. Of 217 TB suspects presenting between January 2006 and March 2008, 33 (15%) had culture-confirmed active tuberculosis; 23 (70%) were smear-negative. Among smear-negative, culture-positive patients, 6 (26%) began TB therapy before culture results were available, 11 (48%) began TB therapy after culture result availability, and 6 (26%) did not begin TB therapy within 180 days of presentation. The cost per negative culture was US$17.52 (solid media)-$23.50 (liquid media). Per 1,000 TB suspects and compared with smear alone, TB culture with solid media would avert an estimated eight TB deaths (95% simulation interval [SI]: 4, 15) and 37 disability-adjusted life years (DALYs) (95% SI: 13, 76), at a cost of $36 (95% SI: $25, $50) per TB suspect or $962 (95% SI: $469, $2642) per DALY averted. Replacing solid media with automated liquid culture would avert one further death (95% SI: -1, 4) and eight DALYs (95% SI: -4, 23) at $2751 per DALY (95% SI: $680, dominated). The cost-effectiveness of TB culture was more sensitive to characteristics of the existing TB diagnostic system than to the accuracy or cost of TB culture.
Conclusions/significance: TB culture is potentially effective and cost-effective for HIV-positive patients in resource-constrained settings. Reliable transmission of culture results to patients and integration with existing systems are essential.