Spatial distribution and characteristics of HIV clusters in Ethiopia

OBJECTIVES: Ethiopia’s HIV prevalence has decreased by 75% in the past 20 years with the implementation of antiretroviral therapy, but HIV transmission continues in high‐risk clusters. Identifying the spatial and temporal trends, and epidemiologic correlates, of these clusters can lead to targeted interventions. METHODS: We used biomarker and survey data from the 2005, 2011 and 2016 Ethiopia Demographic and Health Surveys (DHS). The spatial‐temporal distribution of HIV was estimated using the Kulldorff spatial scan statistic, a likelihood‐based method for determining clustering. Significant clusters (P < 0.05) were identified and compared based on HIV risk factors to non‐cluster areas. RESULTS: In 2005, 2011 and 2016, respectively, 219, 568 and 408 individuals tested positive for HIV. Four HIV clusters were identified, representing 17% of the total population and 43% of all HIV cases. The clusters were centred around Addis Ababa (1), Afar (2), Dire Dawa (3) and Gambella (4). Cluster 1 had higher rates of unsafe injections (4.9% vs. 2.2%, P < 0.001) and transactional sex (6.0% vs. 1.6%, P < 0.001) than non‐cluster regions, but more male circumcision (98.5% vs. 91.3%, P < 0.001). Cluster 2 had higher levels of transactional sex (4.9% vs. 1.6%, P < 0.01), but lower levels of unsafe injections (0.8% vs. 2.2%, P < 0.01). Cluster 3 had fewer individuals with> 1 sexual partner (0% vs. 1.7%, P < 0.001) and more male circumcision (100% vs. 91.3%, P < 0.001). Cluster 4 had less male circumcision (59.1% vs. 91.3%, P < 0.01). CONCLUSIONS: In Ethiopia, geographic HIV clusters are driven by different risk factors. Decreasing the HIV burden requires targeted interventions.