SIMULATION OF HIV INCIDENCE DYNAMICS IN THE RAKAI POPULATION-BASED COHORT,UGANDA

In countries with high but declining HIV prevalence an HIV incidence of 2-3% and more HIV-related deaths than new infections it is often assumed that prevalence is not an adequate surrogate measure of incidence. Simulation models offer the potential to clarify the complex relationship between HIV incidence prevalence and mortality over time. Current patterns in HIV prevalence and mortality reflect cumulative past incidence rates linked by long incubation periods from infection to death. These dynamics were explored through simulation of an HIV epidemic in a population-based cohort of 2600 from Rakai Uganda. Model parameters included epidemic onset in 1980 with a peak in HIV incidence in 8 years and a peak in prevalence of 22% in 1989 an 8-year median incubation from infection to death and a 65% incidence reduction from baseline among youth aging into sexual activity over 2 years from 1989. HIV prevalence among pregnant women in an actual 1990-92 cohort of pregnant Rakai women was similar to that generated by the model suggesting that the simulation was valid. The simulation results indicate that although HIV prevalence is declining and mortality is exceeding new infections incidence declined substantially from an earlier peak of 8.3% among those 15-24 years of age. Moreover the modelling approach provided evidence that the stabilization of prevalence and its subsequent decline can be largely attributed to earlier declines in incidence not merely increases in mortality.