TRENDS IN HIV-1 PREVALENCE MAY NOT REFLECT TRENDS IN INCIDENCE IN MATURE EPIDEMICS - DATA FROM THE RAKAI POPULATION-BASED COHORT, UGANDA

Objectives: To assess whether trends in serial HIV-1 prevalence reflec t trends in HIV incidence, and to decompose the effects of HIV-1 incid ence, mortality, mobility and compliance on HIV-1 prevalence in a popu lation-based cohort. Design: Two-year follow up (1990-1992) of an open cohort of all adults aged 15-59 years, resident in a sample oi 31 rep resentative community clusters in rural Rakai District, Uganda. Method s: A detailed household enumeration was conducted at baseline and in e ach subsequent year. All household residents were listed, and all deat hs and in- and out-migrations that occurred in the intersurvey year we re recorded. In each year, all consenting adults were interviewed and provided a serological sample; 2591 adults aged 15-59 years were enrol led at baseline. Results: HIV prevalence among adults declined signifi cantly between 1990 and 1992 123.4% at baseline, 21.8% in 1991, 20.9% in 1992; P < 0.05). Declining prevalence was also observed in subgroup s, including young adults aged 15-24 years (from 20.6 to 16.2% over 3 years; P < 0.02), women of reproductive age (from 27.1 to 23.5%; P < 0 .05), and pregnant women (from 25.4 to 20.0%; not significant). Howeve r, HIV incidence did not change significantly among all adults aged 15 -59 years [2.1 +/- 0.4 per 100 person-years of observation (PYO) in 19 90-1991 and 2.0 +/- 0.3 pei 100 PYO in 1991-1992], nor in population s ubgroups. HIV-related mortality was high (13.5 per 100 PYO among the H IV-positive), removing more infected persons than were added by seroco nversion. Net out-migration also removed substantial numbers of HIV-po sitive individuals. Conclusions: In this mature HIV epidemic, HIV prev alence declined in the presence of stable and high incidence. HIV-rela ted mortality contributed most to the prevalence decline. Prevalence w as not an adequate surrogate measure of incidence, limiting the utilit y of serial prevalence measures in assessing the dynamics of the HIV e pidemic and in evaluating the impact of current preventive strategies.