Abstract
HIV rapid testing algorithms (RTAs) using any two orthogonal rapid tests (RTs) allow for on-site confirmation of infection. RTs vary in performance characteristics therefore the selection of RTs in an algorithm may affect identification of infection, particularly if acute. National HIV Behavioral Surveillance (NHBS) assessed RTAs among men who have sex with men recruited using anonymous venue-based sampling. Different algorithms were evaluated among participants who self-reported never having received a positive HIV test result prior to the interview. NHBS project areas performed sequential or parallel RTs using whole blood. Participants with at least one reactive RT were offered anonymous linkage to care and provided a dried blood spot (DBS) for testing at CDC. Discordant results (RT-1 reactive/RT-2 non-reactive) were tested at CDC with lab protocols modified for DBS. DBS were also tested for HIV-1 RNA (VL) and antiretroviral (ARV) drug levels. Of 6500 RTAs, 238 were RT-1 reactive; of those, 97.1% (231/238) had concordant results (RT-1/RT-2 reactive) and 2.9% (7/238) had discordant results. Five DBS associated with discordant results were available for confirmation at CDC. Four had non-reactive confirmatory test results that implied RT-1 false reactivity; one had ambiguous confirmatory test results which was non-reactive in further testing. Regardless of order and type of RT used, RTAs demonstrated high concordant results in the population surveyed. Additional laboratory testing on DBS following discordant results confirmed no infection. Implementing RTAs in the context of anonymous venue-based HIV testing could be an option when laboratory follow-up is not practicable.
