Diagnostic usefulness of five screening assays for HIV in an east African city where prevalence of infection is low.

Five commercial screening assays for HIV-1, evaluated for their usefulness in detecting infection in high-risk groups in the East African country of Djibouti, produced varying degrees of performance when compared to Western blot and immunofluorescence confirmatory assays. In this population with a low prevalence of HIV infection (16/599), two enzyme-linked immunosorbent assays (ELISA; Abbott and Elavia) and two rapid assays (cambridge latex agglutination and Du Pont's HIV-CHEK) exhibited less than optimal sensitivities. However, with the exception of Elavia, these assays displayed excellent specificities. The fifth test (Serodia gelatin agglutination) produced the highest sensitivity (0.938) and negative predictive value but the lowest specificity and positive predictive value of all five tests. False positive reactions, which occurred only with the Elavia and Serodia tests, could not be explained on the basis of early infection in patients or cross-reactions with the related retroviruses HIV-2 and HTLV-I. We conclude that none of the five tests alone is sufficient in this testing situation, but that a combination of tests could satisfy most requirements for identifying HIV-1 reactive and non-reactive sera.