A population-based cohort was recruited from a household census in 12 communities. A baseline survey was conducted between 1998 and 2000, with the first and second follow-up surveys occurring after 3 and 5 years, respectively. Using checklists of the resident population at the previous round, adult deaths were reported to enumerators by surviving household members or community informants.
HIV prevalence declined in Manicaland, eastern Zimbabwe, between 1998 and 2003. During this period, adult mortality in men was stable, whereas female mortality increased to levels similar to those of men. We examine the trends in mortality from 2003 to 2005.
Age-standardized adult mortality rates for men increased slightly but not significantly over time (1998-2000: 24/1000 person-years; 2001-2002: 26/1000 person-years; 2003-2005: 31/1000 person-years), reflecting a sharp rise in mortality among HIV-positive individuals (62, 79 and 105 per 1000 person-years). Female mortality rose sharply initially but levelled off after 2001 (15, 26 and 26 per 1000 person-years) also caused by the pronounced increase in mortality among HIV-positive women (35, 75 and 88/1000 person-years; 7/1000 person-years for HIV-negative men and women in all periods); 69% of adult male deaths and 74% of adult female deaths were attributable to HIV/AIDS in 2003-2005. In men, mortality was similar and stable in towns, estates, roadside business centres and subsistence farming areas. In women, mortality rose in towns and subsistence farming areas between 1998 and 2002 and was greater in towns than in other locations.
Recent data indicate that adult mortality may be stabilizing in eastern Zimbabwe after the recent downturn in HIV prevalence.