RESUMO
BACKGROUND: HIV rapid diagnostic test (RDT) algorithms have been successfully employed worldwide to accelerate critically important HIV testing. Deviations from the algorithm and processing errors have been associated with inaccurate algorithm results. Positive RDT algorithm results should be confirmed prior to HIV clinic enrollment, but compliance varies. We sought to retest HIV status of patients in three West African military HIV clinics. SETTING: Military HIV clinics in Lome, Togo; Freetown, Sierra Leone; and Monrovia, Liberia METHODS: Patients coming for routine HIV clinic visits were approached for enrollment. Consenting participants completed a 15-minute questionnaire and provided blood samples for both national and WHO-recommended HIV RDT algorithms, and HIV ELISA (plus HIV PCR if HIV ELISA negative). RESULTS: In total, 817 participants provided data: 374 in Togo, 360 in Sierra Leone, and 83 in Liberia. One participant from Liberia was HIV-negative (although follow-up testing was positive). Two of 807 participants on antiretroviral treatment (ART) had inconclusive algorithms, while 2 of 10 participants not on ART had algorithms, for 4 total based on the WHO-approved algorithm. Using the national algorithms, only 3 were inconclusive. A substantial proportion of the cohort had taken ART for over 6 years (25-46%, depending on the site). CONCLUSION: HIV RDT retesting in three military HIV clinics did not uncover significant numbers of misclassified HIV patients. There was no significant difference between national and WHO-recommended RDT algorithms, although the study was underpowered to detect a difference. Antiretroviral treatment was not associated with increased rates of inconclusive RDT algorithm results.