The impact of HIV infection on adult mortality in some communities in Zambia: a cohort study.

OBJECTIVE To examine the contribution of human immunodeficiency virus (HIV) to adult death rates between 1995/1996 and 1998/1999 in some urban and rural populations in Zambia. METHOD A cohort of 2592 out of 3013 adults aged 15-49 years enrolled in a baseline survey in 1995/1996 in areas of Chelston Township (HIV prevalence 26.2%) and Kapiri Mposhi district (HIV prevalence 15.6%) were followed up in 1998/1999. Person years of observation (PYO) were calculated for those known to have been alive, dead and migrated between the surveys. Fixed covariates Cox proportional hazard models were fitted to assess survival prospects. RESULTS Death rates per 1000 PYO in the rural cohort were 38.1 among those with HIV, 4.8 among those without HIV and 9.8 for all; the population attributable fraction of deaths among those with HIV was 52.3%. Respective estimates in the urban cohort were 53.8, 4.6, 17.2 and 73.6%. The hazard rate (HR) ratio of death for persons infected with HIV was 8.7 times [HR 8.65; 95% confidence interval (CI) 5.28-14.15] higher than for the uninfected after adjusting for age, area of residence, marital status, self-rated health and years spent on education. HR among those who rated their health good/excellent was 40% less, i.e. HR 0.60 (95% CI: 0.40-0.91) than those who rated it to be poor/fair. CONCLUSIONS HIV infection was the most important factor accounting for adult deaths in the communities.