The effects of changes in household wealth on HIV risk in Manicaland, Zimbabwe

We aim to analyse relationships and trends between socio-economic status and HIV infection rates while accounting for demographic and behavioural factors using data over several rounds of a large population-based cohort study in Manicaland, Zimbabwe, over the period 1998 to 2011.
Socio-economic status will be estimated from summed asset ownership scores for sellable, non-sellable and all assets. Multilevel random-effects models will be used to analyse associations between wealth and HIV. Sexual behaviour data will be incorporated in the models to assess for a potential mediating role. Results will be compared over several study rounds to assess for trends and dynamics.
Preliminary results showed that mean asset scores decreased between rounds, reflecting a shift from moderate to poorer absolute wealth. Whilst ownership of sellable assets declined, there was an increase in non-sellable assets over time. Asset scores varied by location and were highest (and rising) in towns and lowest (and falling) in subsistence farming areas. Concurrently, adult HIV prevalence dropped significantly from 22.4% to 16.2%. In men, we found no significant associations between household wealth and HIV prevalence across all rounds. In women, multivariate analyses showed associations between greater poverty and higher HIV prevalence up to 2003-2005.