Accuracy and User-Acceptability of HIV Self-Testing Using an Oral Fluid-Based HIV Rapid Test

Background: The United States FDA approved an over-the-counter HIV self-test, to facilitate increased HIV testing and earlier linkage to care. We assessed the accuracy of self-testing by untrained participants compared to healthcare worker (HCW) testing, participants' ability to interpret sample results and user-acceptability of self-tests in Singapore. Methodology/Principal Findings: A cross-sectional study, involving 200 known HIV-positive patients and 794 unknown HIV status at-risk participants was conducted. Participants (all without prior self-test experience) performed self-testing guided solely by visual instructions, followed by HCW testing, both using the OraQuick ADVANCE Rapid HIV 1/2 Antibody Test, with both results interpreted by the HCW. To assess ability to interpret results, participants were provided 3 sample results (positive, negative, and invalid) to interpret. Of 192 participants who tested positive on HCW testing, self-testing was positive in 186 (96.9%), negative in 5 (2.6%), and invalid in 1 (0.5%). Of 794 participants who tested negative on HCW testing, self-testing was negative in 791 (99.6%), positive in 1 (0.1%), and invalid in 2 (0.3%). Excluding invalid tests, self-testing had sensitivity of 97.4% (95% CI 95.1% to 99.7%) and specificity of 99.9% (95% CI: 99.6% to 100%). When interpreting results, 96%, 93.1% and 95.2% correctly read the positive, negative and invalid respectively. There were no significant demographic predictors for false negative self-testing or wrongly interpreting positive or invalid sample results as negative. Eighty-seven percent would purchase the kit over-the-counter; 89% preferred to take HIV tests in private. 72.5% and 74.9% felt the need for pre- and post-test counseling respectively. Only 28% would pay at least USD15 for the test. Conclusions/Significance: Self-testing was associated with high specificity, and a small but significant number of false negatives. Incorrectly identifying model results as invalid was a major reason for incorrect result interpretation. Survey responses were supportive of making self-testing available.