237Spatial distribution and correlates of HIV prevalence in South Africa.

Background South Africa has a high HIV prevalence and generalized HIV epidemic. It is now well established that the HIV epidemic and its drivers are highly heterogeneous, even in generalized HIV epidemic settings. Methods This study uses data from South African HIV/AIDS, Behavioural Risks, Sero-status, and Mass Media Impact (SABSSM), 2005, 2008 and 2012 surveys. To identify spatial clusters, we used the spatial scan statistic method in SaTScan, assuming discrete Poisson distributions. Poisson regression models were used to explore the municipality-level correlates of HIV prevalence and a logistic regression model was used to determine individual-level correlates of HIV infection. Results Between 2005 and 2012, There was significant geographical variation in estimated HIV prevalence (range = <1.0%–27.5%). Eight, five and six significant overlapping high-risk spatial clusters of high HIV prevalence were detected in 2005, 2008 and 2012, respectively. HIV prevalence is clustered in the central and north-eastern regions of South Africa. Living in municipalities with high percentage of black South Africans, higher poverty index, higher population aged 25-49, and higher early sexual debut were associated with HIV prevalence, while living in municipalities with higher percentage male circumcision and a high percentage married were associated with low risk of HIV. Logistic regression revealed race, sex and mobility as correlates of HIV infection. Conclusions HIV prevalence is highly spatially heterogenous and affected by various municipal-level factors. Key messages Identification of the spatial clusters of HIV prevalence and contextual factors should inform targeted interventions that are necessary to bringing HIV infections under control. [ABSTRACT FROM AUTHOR]