Mobile HIV testing in South Africa: maximizing yield through data-guided site selection

BACKGROUND: Mobile community HIV testing can effectively reach undiagnosed people living with HIV in southern Africa. Variable yield in HIV test positivity has been a challenge with high- and low-yield sites often being closely situated. We sought to test whether easy-to-identify, site-level characteristics were associated with HIV positivity yield in a routine mobile HIV testing program. METHODS: We used routine testing program test data augmented with site-level characterization, either of the community or shopping site at which HIV testing was offered. Specifically, we described the local environment and interviewed key informants to gain additional information regarding the availability of HIV and other services in the locale. RESULTS: We included 122 residential and 26 shopping sites with median HIV-positive test yields of 7.6% and 6.9%, respectively. The range for community sites was from 2 to 55% with high and low yields at geographically proximal sites. Factors related to lower income and marginalization, including informal housing and the absence of name-brand stores in shopping venues, were associated with higher HIV-positive testing yield. CONCLUSIONS: Characterization of sites, particularly identifying factors related to marginalization, lack of services, and poverty, can aid in identifying sites with higher HIV-positive yield.