Estimating HIV incidence in voluntary counseling and testing clients in Uganda (1992-2003)

Objectives: To estimate HIV incidence from first-time testers among voluntary counseling and testing (VCT) clients in Uganda. Methods: Data on 203,000 VCT clients tested from 1992 through 2003 were adjusted for temporal changes in the testing population. Differential mortality rates by HIV status were used to derive expected prevalence at future times from baseline prevalence within 5-year birth cohorts. Incidence was computed as the proportion of HIV-uninfected persons who seroconverted divided by the proportion negative at baseline. Results: Annual HIV incidence per 100 uninfected persons increased from 0.9 (95% confidence interval [Cl]: 0.8 to 1.1) in 1993 to 2.3 (95% CI: 2.2 to 2.5) in 2003 (chi(2) test for trend, P < 0.001). Prevalence decreased from 23% to 13% in 1999 to 2000 and increased to 15% in 2003. Women had a higher incidence. Peak incidence shifted to older age groups over time. Conclusions: Estimating incidence from routine data presents a practical way of tracking HIV incidence and is a useful tool in targeting and evaluating the impact of prevention programs. Our analysis reveals a new phase of the HIV epidemic in Uganda: decreasing prevalence and increasing incidence, especially among middle-aged persons. These findings support the need for intensified prevention interventions among middle-aged persons in Uganda.